case study statistical process control

Log in using your username and password

• Search More Search for this keyword Advanced search
• Latest content
• Current issue
• BMJ Journals More You are viewing from: Google Indexer

You are here

• Volume 16, Issue 5
• Application of statistical process control in healthcare improvement: systematic review
• Article Text
• Article info
• Citation Tools
• Rapid Responses
• Article metrics

• Johan Thor ,
• Jonas Lundberg ,
• Jakob Ask ,
• Jesper Olsson ,
• Cheryl Carli ,
• Karin Pukk Härenstam ,
• Mats Brommels
• Medical Management Centre, Karolinska Institutet, Stockholm, Sweden
• Correspondence to:
 Dr Johan Thor
 Medical Management Centre, Berzelius väg 3, 5th floor, Karolinska Institutet, S-171 77 Stockholm, Sweden; johan.thor{at}ki.se

Objective: To systematically review the literature regarding how statistical process control—with control charts as a core tool—has been applied to healthcare quality improvement, and to examine the benefits, limitations, barriers and facilitating factors related to such application.

Data sources: Original articles found in relevant databases, including Web of Science and Medline, covering the period 1966 to June 2004.

Study selection: From 311 articles, 57 empirical studies, published between 1990 and 2004, met the inclusion criteria.

Methods: A standardised data abstraction form was used for extracting data relevant to the review questions, and the data were analysed thematically.

Results: Statistical process control was applied in a wide range of settings and specialties, at diverse levels of organisation and directly by patients, using 97 different variables. The review revealed 12 categories of benefits, 6 categories of limitations, 10 categories of barriers, and 23 factors that facilitate its application and all are fully referenced in this report. Statistical process control helped different actors manage change and improve healthcare processes. It also enabled patients with, for example asthma or diabetes mellitus, to manage their own health, and thus has therapeutic qualities. Its power hinges on correct and smart application, which is not necessarily a trivial task. This review catalogues 11 approaches to such smart application, including risk adjustment and data stratification.

Conclusion: Statistical process control is a versatile tool which can help diverse stakeholders to manage change in healthcare and improve patients’ health.

• MRSA, methicillin resistant Staphylococcus aureus
• PEFR, peak expiratory flow rate
• QI, quality improvement
• RCT, randomised controlled trial
• SPC, statistical process control

https://doi.org/10.1136/qshc.2006.022194

Statistics from Altmetric.com

Request permissions.

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Quality improvement (QI) practices represent a leading approach to the essential, and often challenging, task of managing organisational change. 1 Statistical process control (SPC) is, in turn, a key approach to QI. 2 SPC was developed in the 1920s by the physicist Walter Shewhart to improve industrial manufacturing. It migrated to healthcare, first in laboratory settings (eg, Fisher and Humphries 3 ) and then into direct patient care applications, along with other approaches to QI. Before we report on our systematic review of the literature on how SPC has been applied to QI in healthcare, there is a need to define SPC and its role in QI.


 “Statistical process control (SPC) is a philosophy, a strategy, and a set of methods for ongoing improvement of systems, processes, and outcomes. The SPC approach is based on learning through data and has its foundation in the theory of variation (understanding common and special causes). The SPC strategy incorporates the concepts of an analytic study, process thinking, prevention, stratification, stability, capability, and prediction. SPC incorporates measurement, data collection methods, and planned experimentation. Graphical methods, such as Shewhart charts (more commonly called ‘control charts’), run charts, frequency plots, histograms, Pareto analysis, scatter diagrams, and flow diagrams are the primary tools used in SPC.” (Carey 4 , p xviii)

The terms “statistical process control” and “statistical quality control” are often used interchangeably, 5 although sometimes the latter is used to describe a broader organisational approach to quality management that evolved into the concept of total quality management. 6

One of the tenets of QI is that to improve healthcare performance we must change our way of working. 7 But change does not always mean improvement. To discriminate between changes that yield improvement and those that do not, relevant aspects of performance need to be measured. In addition, measurement guides decisions about where improvement efforts should be focused in the first place. SPC may facilitate such decision making. Control charts, central to SPC, are used to visualise and analyse the performance of a process—including biological processes such as blood pressure homoeostasis or organisational processes such as patient care in a hospital—over time, sometimes in real time. Statistically derived decision rules help users to determine whether the performance of a process is stable and predictable or whether there is variation in the performance that makes the process unstable and unpredictable. One source of such variation can be a successful intervention aimed at improvement that changes performance for the better. If the improvement is maintained, the process will stabilise again at its new level of performance. All of this can be easily determined by using SPC. 4

Although there are theoretical propositions that SPC can facilitate decision making and QI in healthcare (eg, Berwick, 8 Benneyan et al , 9 Plsek 10 ) it is not clear what empirical support there is in the literature for such a position 11 :


 “The techniques of statistical process control, which have proved to be invaluable in other settings, appear not to have realised their potential in health care. ... Is this because they are, as yet, rarely used in this way in health care? Is it because they are unsuccessful when used in this way and thus not published (publication bias)? Or is it that they are being successfully used but not by people who have the inclination to share their experience in academic journals?” (p 200)

The present systematic review aimed to answer these questions. We examined the literature for how and where SPC has been applied in QI of clinical/patient care processes and the benefits, limitations, barriers and facilitating factors related to such application.

MATERIALS AND METHODS

Drawing on the principles and procedures for systematic review of QI interventions 12 we searched for articles on the application of SPC in healthcare QI published between 1966 and June 2004 (see appendix A) in the following databases: Web of Science, Ovid Medline(R), EMBASE, CINAHL (Cumulative Index to Nursing and Allied Health Literature), PsycInfo, and the Centre for Reviews and Dissemination databases. We also included articles found by searching reference lists or from elsewhere which we were aware of, if they met our inclusion criteria: original empirical studies of SPC application in improvement of clinical/patient care processes in healthcare organisations, published in English. We excluded articles dealing with application of SPC in laboratory or technical processes (eg, film processing) and in surveillance/monitoring (unless they also contained empirical data about improvement efforts), as well as tutorials (unless they contained empirical case studies), letters, book reviews and dissertations.

We reviewed abstracts, when available, or else other information about the publication provided in the database (eg, publication type such as letters, book reviews or original articles). Articles that did not meet the inclusion criterion were excluded. We retrieved and read the full text of the remaining articles, again excluding the articles that did not meet the inclusion criterion.

We developed, pilot tested and modified a data abstraction form which we then used to consistently capture information of relevance to our review questions on reading the full text articles. The information recorded was: whether and how the article met the inclusion criterion; study objective(s); study design; whether the study empirically compared application of SPC with any other method for process data display and analysis; reported benefits, limitations, barriers and facilitating factors related to SPC; organisational setting; country where study was conducted; clinical specialty; unit of analysis; variables for SPC analysis; and other observations. Some questions in the form required a yes/no or brief response (eg, country where study was conducted) and others required answers in the form of direct quotes from the article or the a summary of the article written by the reviewer. Each article was read and data abstracted by one member of the review team (the coauthors of this review). Following this, all the data abstraction forms were reviewed by the first author, who solicited clarification and checked for any missing or incomplete data to ensure consistency in reporting across all articles reviewed. He also conducted the initial data synthesis, which was then reviewed by the entire team.

We determined the study design for each article and whether the investigators intended to test the utility of SPC application, alone or in combination with other interventions. In several articles, the study design or study objectives were not explicitly stated. Our determination of such intention in such cases was based on our reading of the full text papers.

Simple descriptive statistics—for example, the number of publications per year of publication or per country—were used to characterise the included studies. The qualitative nature of our research questions and of the abstracted data shaped our analysis and synthesis of findings regarding benefits, limitations, SPC variables, etc. 13 The abstracted data was reviewed one question at a time and data from each article was classified into one or more thematic categories, each with a descriptive heading. Informed by our present understanding of QI and healthcare, we developed these categories as we reviewed the data, rather than using categories derived a priori from theory. For data that did not fit into an existing category, we developed a new one. Thus the categories emerged as we synthesised the data. We report the categorised data in tabular form, illustrated with examples, and give the references of all the source studies.

To strengthen our review through investigator triangulation, 14 we sought feedback on an earlier version of this manuscript from two SPC experts: one was the most frequent coauthor in the included studies and the other was an expert on SPC application also in settings other than healthcare. Their comments helped us refine our data synthesis and distil our findings.

The database searches yielded 311 references. The initial review (abstracts etc.) yielded 100 articles which we read in full text form. Of these, 57 articles met the inclusion criteria and have been included in the review. 15– 71 To characterise the body of liferature, figure 1 shows the year of publication and whether the studies were conducted in USA or elsewhere (further specified below); table 1 gives the study designs and objectives—whether or not to test SPC utility.

• View inline

 Study design and objectives of the studies included in the systematic review*

• Download figure
• Open in new tab
• Download powerpoint

 The number of included articles by year of publication. (A total of 55 articles were published in 1990–2003; the two articles from 2004 are not included in this graph since the database searches were conducted in June 2004.) Black bars: studies conducted in the USA; grey bars: studies conducted outside the USA.

Most of the articles (45/57) concerned application of SPC in healthcare improvement in the USA. 15– 35, 37– 40, 42, 43, 45, 47, 49– 56, 59, 60, 63, 67– 71 While the first US-based article was published in 1990, the non-US articles were published between 1998 and 2003: three articles were from the UK, 61, 62, 66 three were from Switzerland, 36, 41, 46 and one each were from Australia, 58 Finland, 65 France, 57 Indonesia, 44 Norway 64 and Spain. 48 The intention to test the utility of SPC is exemplified by a study aiming to reduce the rate of acquisition of methicillin-resistant Staphylococcus aureus (MRSA) on wards and units at Glasgow Royal Infirmary hospitals. 61 Annotated control charts displaying data on MRSA acquisition were fed back monthly to medical staff, managers and hotel services. Sustained reductions in the rate of acquisition from the baseline, which could not otherwise be accounted for, started 2 months later. In contrast, investigators at a paediatric emergency department used SPC to demonstrate a decline in the rate of contamination following the introduction of a new approach to drawing blood for culture specimens, 68 but the study had no intention to test the utility of SPC per se.

To characterise the content of the articles, we first present how and where SPC has been applied to healthcare QI. Tables 2–4 present the study settings (ie, hospital etc. where SPC was applied; table 2), the field of healthcare (ie, specialties or forms of care; table 3), and the units of analysis (table 4). Table 5 enlists the 97 distinct SPC variables that have been reported. Tables 6–9 convey our synthesis of the reported benefits, limitations, barriers and facilitating factors related to SPC application. For each category, we have given explanations or examples and references to the source articles.

 How and where SPC was applied: study settings*

 How and where SPC was applied: fields of healthcare*

 How and where SPC was applied: units of analysis*

 SPC variables*

 Benefits of using SPC to improve clinical processes*

SPC has been applied to healthcare improvement in a wide range of settings and specialties, at diverse levels of organisations and directly by patients, using many types of variables (fig 1, tables 2–5). We found reports of substantial benefits of SPC application, as well as important limitations of, barriers to and factors that facilitate SPC application (tables 6–9). These findings indicate that SPC can indeed be a powerful and versatile tool for managing changes in healthcare through QI. Besides helping diverse stakeholders manage and improve healthcare processes, SPC can also help clinicians and patients understand and improve patients’ health when applied directly to health indicators such as PEFR in asthma or blood sugar concentrations in diabetes. In healthcare, the “study subject” can thus also be an active agent in the process, as when patients apply SPC to their own health. Several studies indicated the empowering effects this may have on patients. 35, 38, 40, 50 SPC application thus has therapeutic potential as it can help patients manage their own health. We agree with Alemi and Neuhauser 70 that this potential merits further investigation.

Most of the included articles concerned application of SPC in healthcare improvement in the USA. Articles from other countries appeared only towards the end of the study period (fig 1). We have no explanation for this finding, but we speculate that it is related to differences between US and other healthcare systems with regard to QI awareness and implementation. 73

Only 22 studies included in the review were intended to test the utility of SPC (table 1). Of the four controlled studies, only one included a control chart in the intervention (as a minor component which did not fully exploit the features of SPC). In 35 articles we did not find an intention to test the utility of SPC application. In those cases, SPC was applied for other reasons (ie, to evaluate the impact of other interventions). Even though many articles thus did not address the utility of SPC, all studies offered information—to varying degrees—relevant to our review’s question of how SPC has been applied to healthcare. The utility of SPC is reflected in benefits reported regarding SPC application (table 6).

SPC has been applied in over 20 specialties or fields of healthcare, at a wide range of levels (tables 3 and 4), suggesting that SPC has broad applicability in healthcare. The dominance of anaesthesia and intensive care can be explained in large part by the fact that many studies included their services in conjunction with other specialties. This reflects the way in which anaesthesia has a vital supporting role in many clinical care processes. The 97 SPC variables reported (table 5) demonstrate a diversity of situations in which SPC has been applied, ranging from process indicators of patients’ health to health outcomes and many aspects of healthcare processes and organisational performance. This indicates that SPC is a versatile QI tool.

The benefits of SPC application (table 6) mirror those given in books and tutorials on SPC (exemplified by the quote in the Introduction to this review). As noted in a report from a top-ranked healthcare system which has applied SPC widely:


 “Among the most powerful quality management tools that IHC [Intermountain Health Care, USA] has applied is statistical process control, SPC. Most notable among those tools are control charts. Under optimal conditions, these graphical depictions of process performance allow participants to know what is happening within their processes as ‘real time’ data enable them to make appropriate decisions. The capability of truly understanding processes and variation in a timely manner has resulted in the most dramatic, immediate, and ongoing improvements of any management technique applied at IHC.” (Shaha, 26 p 22)

The limitations of SPC application (table 7) identified by this review are important, and yet perhaps less emphasised than the benefits in books and tutorials on SPC. SPC cannot solve all problems and must be applied wisely. There are many opportunities to “go wrong”, as illustrated by the case where incorrect application was highlighted by other authors (limitation number 5 in table 7). In several cases, our own understanding of SPC suggested that investigators had not used it correctly or fully (eg, standard decision rules to detect special causes were not applied to identify process changes). In the worst case scenario, incorrect application of SPC could lead to erroneous conclusions about process performance and waste time, effort and spirit and even contribute to patient harm. In the more authoritative studies we reviewed, co-investigators included experts in industrial engineering or statistics or authors who otherwise had developed considerable expertise in SPC methodology. On the basis of these observations, we conclude that although SPC charts may be easy to use even for patients, clinicians or managers without extensive SPC training, they may not be equally simple to construct correctly. To apply SPC is, paradoxically, both simple and difficult at the same time. Its power hinges on correct and smart application, which is not necessarily a trivial task. The key, then, is to develop or recruit the expertise necessary to use SPC correctly and fully and to make SPC easy for non-experts to use, before using it widely.

 Limitations of SPC application in improvement of clinical processes*

Autocorrelation is another limitation of SPC highlighted by this review. Our review, and published books, offer limited advice on how to manage it:


 “There is no single acceptable way of dealing with autocorrelation. Some would say simply to ignore it. [Others] would disagree and suggest various measures to deal with the phenomenon. One way is to avoid the autocorrelation by sampling less frequently. ... Others argue against plotting autocorrelated data on control charts and recommend that the data be plotted on a line chart (without any centerline or control limits).” (Carey, 4 p 68)

Just over a quarter of the articles reported barriers to SPC application (table 8). The three broad divisions of barriers—people, data and chart construction, and IT—indicate where extra care should be taken when introducing SPC in a healthcare organisation. Ideas on how to manage the limitations of and barriers to SPC application can be found among the factors reported to facilitate SPC application (table 9). They deal with, and go beyond, the areas of barriers we found. We noted the prominence of learning and also of focusing on topics of interest to clinicians and patients. The 11 categories under the heading “Smart application of SPC can be helpful” contain valuable approaches that can be used to improve SPC application. Examples include risk adjustment 51, 52, 71 and stratification 30, 37, 59 to enable correct SPC analysis of data from heterogeneous populations of patients (or organisational units). Basic understanding of SPC must be taught to stakeholders and substantial skill and experience is required to set up successful SPC application. Experts, or facilitators, in healthcare organisations can help, as indicated in table 9, and as we have described for other QI methods. 74

 Barriers to SPC application*

 Factors or conditions facilitating application of SPC*

We found more information on SPC benefits and facilitating factors than on limitations and barriers, and this may represent a form of publication bias, as indicated by the quote in the Introduction. 11 We did not find any study that reported failed SPC application. We can speculate that there have been situations when SPC application failed, just as there must be many cases of successful SPC application that have not been reported in the literature. Studies of failed SPC application efforts, as well as studies designed to identify successful ways to apply SPC to manage change, would help inform future SPC application efforts. On the basis of this review, we agree with the argument that “medical quality improvement will not reach its full potential unless accurate and transparent reports of improvement work are published frequently and widely (p 319),” 75 and also that the way forward is to strengthen QI research rather than to lower the bar for publication. 76

Methodological considerations regarding the included studies

None of the studies we found was designed to evaluate the effectiveness quantitatively—that is, the magnitude of benefits—of SPC application. This would have required other study designs such as cluster randomised trials or quasi-experimental studies. 12 Although the “methods of evaluating complex interventions such as quality improvement interventions are less well described [than those to evaluate less complex interventions such as drugs]”, Eccles et al argue that the “general principle underlying the choice of evaluative design is ... simple—those conducting such evaluations should use the most robust design possible to minimise bias and maximise generalisability. [The] design and conduct of quantitative evaluative studies should build upon the findings of other quality improvement research (p 47).” 77 This review can provide such a foundation for future evaluative studies.

An important distinction is warranted here: we believe that SPC rests on a solid theoretical, statistical foundation and is a highly robust method for analysing process performance. The designs of the studies included in this systematic review were, however, not particularly robust with regard to evaluating the effectiveness of SPC application, and that was not their objective. This does not mean that SPC is not a useful tool for QI in healthcare, only that the studies reviewed here were more vulnerable to bias than more robust study designs, even if they do indicate many clear benefits of SPC application (table 6). Despite the studies not being designed to evaluate the effectiveness of SPC, many used SPC to effectively show the impact of QI or other change initiatives. In this way, SPC analysis can be just as powerful and robust as study designs often deemed superior, such as randomised controlled trials (RCTs). 77 The key to this power is the statistical and practical ability to detect significant changes over time in process performance when applying SPC. 9 On the basis of a theoretical comparison between control charts and RCTs, Solodky et al 38 argue that control charts can complement RCTs, and sometimes even be preferable to RCTs, since they are so robust and enable replication—“the gold standard” for research quality—at much lower cost than do RCTs. These points have been further elaborated in subsequent work. 78, 79

A curious methodological wrinkle in our review is: can you evaluate the application of a method (eg, SPC) using that same method for the evaluation? Several of the included studies used SPC both as (part of) an intervention and as a method to evaluate the impact of that intervention. For example, Curran et al used annotated control charts to feed information on MRSA acquisition rates back to stakeholders and used these same control charts to show the effectiveness of the feedback programme. 61

Relationship between monitoring and improvement

When SPC is applied for monitoring, rather than for managing change, the aims are different—for example, to detect even small but clinically important deviations in performance—as are the methodological challenges. 80, 81 This review focused on the latter. Thus although studies on SPC application for monitoring healthcare performance were excluded from this review, we recognise the importance of such monitoring. The demarcation between monitoring and improvement is not absolute. Indeed, there are important connections between measurement, monitoring and improvement, even if improvement does not follow automatically from indications of dissatisfactory performance. “To improve performance, organizations and individuals need the capability to control, improve, and design processes, and then to monitor the effects of this improvement work on the results. Measurement alone will not suffice (pp 1–35).” 82

Monitoring performance by way of control charts has been suggested as a better approach to clinical governance in the British National Health Service. Through six case studies, Mohammed et al demonstrate how control chart monitoring of performance can distinguish normal performance from performance that is either substandard or better than usual care. “These case studies illustrate an important role for Shewhart’s approach to understanding and reducing variation. They demonstrate the simplicity and power of control charts at guiding their users towards appropriate action for improvement (p 466).” 83

Comments on the review methodology

No search strategy is perfect, and we may well have missed some studies where SPC was applied to healthcare QI. There are no SPC specific keywords (eg, Medical Subject Headings, MeSH) so we had to rely on text words. Studies not containing our search terms in the title or abstract could still be of potential interest although presumably we found most of the articles where SPC application was a central element. We believe the risk that we systematically missed relevant studies to be small. Therefore, our findings would probably not have changed much due to such studies that we might have missed.

The review draws on our reading, interpretation and selection of predominantly qualitative data—in the form of text and figures—in the included articles to answer the questions in our data abstraction form. The questions we addressed, the answers we derived from the studies, and the ways we synthesised the findings are not the only ways to approach this dataset. Furthermore, each member of the review team brought different knowledge and experiences of relevance to the review, potentially challenging the reliability of our analysis. An attempt was made to reduce that risk by having one investigator read all data abstraction forms, and obtain clarifications or additional data from the original articles when needed. That investigator also conducted the initial data synthesis, which was then reviewed by the entire team and the two outside experts. Although other interpretations and syntheses of these data are possible, we believe that ours are plausible and hope they are useful.

The methods for reviewing studies based primarily on qualitative data in healthcare are less well developed than the more established methods for quantitative systematic reviews, and they are in a phase of development and diversification. 13, 84, 85 Among the different methods for synthesising evidence, our approach is best characterised as an interpretive (rather than integrative) review applying thematic analysis—it “involves the identification of prominent or recurrent themes in the literature, and summarising the findings of different studies under thematic headings”. 86 There is no gold standard for how to conduct reviews of primarily qualitative studies. Our response to this uncertainty has been to use the best ideas we could find, and to be explicit about our approach to allow readers to assess the findings and their provenance.

The main limitation of this review is the uncertainty regarding the methodological quality of many of the primary studies. Assessment of quality of qualitative studies is still under debate, and there is no consensus on whether at all, or, if so, how to conduct such assessments. 84 We reviewed all the studies that satisfied our inclusion criteria and made no further quality assessment. Therefore our findings should be considered as tentative indications of benefits, limitations, etc to be corroborated, or rejected, by future research. The main strength of this review is our systematic and explicit approach to searching and including studies for review, and to data abstraction using a standardised form. It has helped generate an overview of how SPC has been applied to healthcare QI with both breadth and depth—similar to the benefits of thematic analysis reported by investigators reviewing young people’s views on health and health behaviour. 87

In conclusion, this review indicates how SPC has been applied to healthcare QI with substantial benefits to diverse stakeholders. Although there are important limitations and barriers regarding its application, when applied correctly SPC is a versatile tool which can enable stakeholders to manage change in healthcare and improve patients’ health.

Database search strategy

Web of Science (1986 – 11 June 2004)

TS [topic search]  =  ((statistical process control or statistical quality control or control chart* or (design of experiment and doe)) and (medical or nurs* or patient* or clinic* or healthcare or health care))

We limited the search to articles in English only which reduced the number of hits from 167 to 159. We saved these 159 titles with abstracts in an EndNote library. Using a similar strategy, we searched the following databases through Ovid:

Ovid MEDLINE(R) (1966 to week 1, June 2004)

EMBASE (1988 to week 24, 2004)

CINAHL (1982 to week 1, June 2004)

PsycINFO (1985 to week 5, May 2004)

This yielded 287 hits, including many duplicates, which we saved in the same EndNote library as above.

Centre for Reviews and Dissemination (CRD)

We searched all CRD databases and found two articles which we also added to our EndNote library.

Acknowledgments

We thank Ms Christine Wickman, Information Specialist at the Karolinska Institutet Library, for expert assistance in conducting the database searches. We also acknowledge the pilot work conducted by Ms Miia Maunuaho as a student project at Helsinki University, supervised by Professor Brommels, which provided a starting point for this study. We thank Professor Duncan Neuhauser, Case Western Reserve University, Cleveland, Ohio, USA, and Professor Bo Bergman, Chalmers University of Technology, Gothenburg, Sweden, for their helpful comments on an earlier version of this manuscript. We thank Dr Rebecca Popenoe for her editorial assistance.

• ↵ Burnes B . Managing change: a strategic approach to organisational dynamics , 3rd edn. London: Financial Times/Prentice Hall, 2000 .
• ↵ Wheeler D J , Chambers D S. Understanding statistical process control , 2nd edn. Knoxville, TN: SPC Press, 1992 .
• ↵ Fisher L M , Humphries B L. Statistical quality control of rabbit brain thromboplastin for clinical use in the prothrombin time determination. Am J Clin Pathol 1966 ; 45 : 148 –52. OpenUrl PubMed Web of Science
• ↵ Carey R G . Improving healthcare with control charts: basic and advanced SPC methods and case studies . Milwaukee: ASQ Quality Press, 2003 .
• ↵ Daniels S , Johnson K, Johnson C. Quality glossary. 2006 [cited 28 June 2006], http://www.asq.org/glossary/s.html .
• ↵ Blumenthal D . Applying industrial quality management science to physicians’ clinical decisions. In: Blumenthal D, Scheck A, eds. Improving clinical practice: total quality management and the physician . San Francisco: Jossey-Bass Publishers, 1995 : 25 –49.
• ↵ Berwick D M . A primer on leading the improvement of systems. BMJ 1996 ; 312 : 619 –22. OpenUrl FREE Full Text
• ↵ Berwick D M . Controlling variation in health care: a consultation from Walter Shewhart. Med Care 1991 ; 29 : 1212 –25. OpenUrl CrossRef PubMed Web of Science
• ↵ Benneyan J C , Lloyd R C, Plsek P E. Statistical process control as a tool for research and healthcare improvement. Qual Saf Health Care 2003 ; 12 : 458 –64. OpenUrl Abstract / FREE Full Text
• ↵ Plsek P E . Quality improvement methods in clinical medicine. Pediatrics 1999 ; 103 (1 Suppl E) : 203 –14. OpenUrl Abstract / FREE Full Text
• ↵ Wilcock P M , Thomson R G. Modern measurement for a modern health service. Qual Health Care 2000 ; 9 : 199 –200. OpenUrl FREE Full Text
• ↵ Grimshaw J , McAuley L M, Bero L A, et al. Systematic reviews of the effectiveness of quality improvement strategies and programmes. Qual Saf Health Care 2003 ; 12 : 298 –303. OpenUrl Abstract / FREE Full Text
• ↵ Mays N , Roberts E, Popay J. Synthesising research evidence. In: Fulop N, Allen P, Clarke A, et al , eds. Studying the organisation and delivery of health services: research methods . London: Routledge, 2001 : 188 –220.
• ↵ Fulop N , Allen P, Clarke A, et al. Issues in studying the organisation and delivery of health services. In: Fulop N, Allen P, Clarke A, et al , eds. Studying the organisation and delivery of health services: research methods . London: Routledge, 2001 : 1 –23.
• ↵ Re R N , Krousel-Wood M A. How to use continuous quality improvement theory and statistical quality control tools in a multispecialty clinic [see comment]. Qrb Qual Rev Bull 1990 ; 16 : 391 –7. OpenUrl PubMed
• ↵ Schnelle J F , Newman D R, Fogarty T E, et al. Assessment and quality-control of incontinence care in long-term nursing facilities. J Am Geriatr Soc 1991 ; 39 : 165 –71. OpenUrl PubMed Web of Science
• ↵ Bluth E I , Havrilla M, Blakeman C. Quality improvement techniques: value to improve the timeliness of preoperative chest radiographic reports. AJR Am J Roentgenol 1993 ; 160 : 995 –8. OpenUrl PubMed Web of Science
• ↵ McKenzie L . Process management: two control charts. Health Care Superv 1993 ; 12 : 70 –81. OpenUrl PubMed
• ↵ Rollo J L , Fauser B A. Computers in total quality management. Statistical process control to expedite stats. Arch Pathol Lab Med 1993 ; 117 : 900 –5. OpenUrl PubMed Web of Science
• ↵ Dey M L , Sluyter G V, Keating J E. Statistical process control and direct care staff performance. J Ment Health Adm 1994 ; 21 : 201 –9. OpenUrl PubMed Web of Science
• ↵ Guinane C S , Sikes J I, Wilson R K. Using the PDSA cycle to standardize a quality assurance program in a quality improvement-driven environment. Jt Comm J Qual Improv 1994 ; 20 : 696 –705. OpenUrl PubMed
• ↵ Laffel G , Luttman R, Zimmerman S. Using control charts to analyze serial patient-related data. Qual Manag Health Care 1994 ; 3 : 70 –7. OpenUrl PubMed
• ↵ Nelson F E , Hart M K, Hart R F. Application of control chart statistics to blood pressure measurement variability in the primary care setting. J Am Acad Nurse Pract 1994 ; 6 : 17 –28. OpenUrl CrossRef PubMed
• ↵ Carey R G , Teeters J L. CQI case-study—reducing medication errors. Jt Comm J Qual Improv 1995 ; 21 : 232 –7. OpenUrl PubMed
• ↵ Oniki T A , Clemmer T P, Arthur L K, et al. Using statistical quality control techniques to monitor blood glucose levels. Proc Annu Symp Comput Appl Med Care 1995 : 586 –90.
• ↵ Shaha S H . Acuity systems and control charting. Qual Manag Health Care 1995 ; 3 : 22 –30. OpenUrl CrossRef PubMed
• ↵ Ziegenfuss J TJr , McKenna C K. Ten tools of continuous quality improvement: a review and case example of hospital discharge. Am J Med Qual 1995 ; 10 : 213 –20. OpenUrl FREE Full Text
• ↵ Johnson C C , Martin M. Effectiveness of a physician education program in reducing consumption of hospital resources in elective total hip replacement. South Med J 1996 ; 89 : 282 –9. OpenUrl CrossRef PubMed Web of Science
• ↵ Piccirillo J F . The use of patient satisfaction data to assess the impact of continuous quality improvement efforts. Arch Otolaryngol Head Neck Surg 1996 ; 122 : 1045 –8. OpenUrl CrossRef PubMed Web of Science
• ↵ Shahian D M , Williamson W A, Svensson L G, et al. Applications of statistical quality control to cardiac surgery. Ann Thorac Surg 1996 ; 62 : 1351 –8 discussion 89. OpenUrl CrossRef PubMed Web of Science
• ↵ Tsacle E G , Aly N A. An expert system model for implementing statistical process control in the health care industry. Computers & Industrial Engineering 1996 ; 31 : 447 –50. OpenUrl CrossRef
• ↵ Cornelissen G , Halberg F, Hawkins D, et al. Individual assessment of antihypertensive response by self-starting cumulative sums. J Med Eng Technol 1997 ; 21 : 111 –20. OpenUrl PubMed Web of Science
• ↵ Linford L H , Clemmer T P, Oniki T A. Development of a blood glucose protocol using statistical quality control techniques. Nutr Clin Pract 1997 ; 12 : 38 –41. OpenUrl CrossRef PubMed
• ↵ Ornstein S M , Jenkins R G, Lee F W, et al. The computer-based patient record as a CQI tool in a family medicine center. Jt Comm J Qual Improv 1997 ; 23 : 347 –61. OpenUrl PubMed
• ↵ Boggs P B , Wheeler D, Washburne W F, et al. Peak expiratory flow rate control chart in asthma care: chart construction and use in asthma care. Ann Allergy Asthma Immunol 1998 ; 81 : 552 –62. OpenUrl PubMed Web of Science
• ↵ Konrad C , Gerber H R, Schuepfer G, et al. Transurethral resection syndrome: effect of the introduction into clinical practice of a new method for monitoring fluid absorption. J Clin Anesth 1998 ; 10 : 360 –5. OpenUrl CrossRef PubMed Web of Science
• ↵ Nelson E C , Splaine M E, Batalden P B, et al. Building measurement and data collection into medical practice. Ann Intern Med 1998 ; 128 : 460 –6. OpenUrl CrossRef PubMed Web of Science
• ↵ Solodky C , Chen H G, Jones P K, et al. Patients as partners in clinical research—a proposal for applying quality improvement methods to patient care. Med Care 1998 ; 36 : AS13 –20. OpenUrl CrossRef PubMed Web of Science
• ↵ Vitez T S , Macario A. Setting performance standards for an anesthesia department. J Clin Anesth 1998 ; 10 : 166 –75. OpenUrl CrossRef PubMed Web of Science
• ↵ Boggs P B , Hayati F, Washburne W F, et al. Using statistical process control charts for the continual improvement of asthma care. Jt Comm J Qual Improv 1999 ; 25 : 163 –81. OpenUrl PubMed
• ↵ Konrad C , Gerber H, Schupfer G, et al. Detection of fluid volume absorption by end-tidal alcohol monitoring in patients undergoing endoscopic renal pelvic surgery. J Clin Anesth 1999 ; 11 : 386 –90. OpenUrl CrossRef PubMed Web of Science
• ↵ Levett J M , Carey R G. Measuring for improvement: from Toyota to thoracic surgery. Ann Thorac Surg 1999 ; 68 : 353 –8. OpenUrl CrossRef PubMed Web of Science
• ↵ Pollard J B , Garnerin P. Outpatient preoperative evaluation clinic can lead to a rapid shift from inpatient to outpatient surgery: a retrospective review of perioperative setting and outcome. J Clin Anesth 1999 ; 11 : 39 –45. OpenUrl CrossRef PubMed Web of Science
• ↵ Purba M . Use of a control chart to monitor diarrhoea admissions: a quality improvement exercise in West Kalimantan Provincial Hospital, Pontianak, Indonesia. J Qual Clin Pract 1999 ; 19 : 145 –7. OpenUrl CrossRef PubMed
• ↵ Schwab R A , DelSorbo S M, Cunningham M R, et al. Using statistical process control to demonstrate the effect of operational interventions on quality indicators in the emergency department. J Healthc Qual 1999 ; 21 : 38 –41. OpenUrl PubMed
• ↵ Bonetti P O , Waeckerlin A, Schuepfer G, et al. Improving time-sensitive processes in the intensive care unit: the example of “door-to-needle time” in acute myocardial infarction. Int J Qual Health Care 2000 ; 12 : 311 –7. OpenUrl Abstract / FREE Full Text
• ↵ Ratcliffe M B , Khan J H, Magee K M, et al. Collection of process data after cardiac surgery: Initial implementation with a Java-based intranet applet. Ann Thorac Surg 2000 ; 69 : 1817 –21. OpenUrl CrossRef PubMed Web of Science
• ↵ Saturno P J , Felices F, Segura J, et al. Reducing time delay in the thrombolysis of myocardial infarction: an internal quality improvement project. Am J Med Qual 2000 ; 15 : 85 –93. OpenUrl Abstract / FREE Full Text
• ↵ Sinanan M , Wicks K, Peccoud M, et al. Formula for surgical practice resuscitation in an academic medical center. Am J Surg 2000 ; 179 : 417 –21. OpenUrl CrossRef PubMed Web of Science
• ↵ Staker L V . Changing clinical practice by improving systems: the pursuit of clinical excellence through practice-based measurement for learning and improvement. Qual Manag Health Care 2000 ; 9 : 1 –13. OpenUrl CrossRef PubMed
• ↵ Alemi F , Oliver D W. Tutorial on risk-adjusted P-charts. Qual Manag Health Care 2001 ; 10 : 1 –9. OpenUrl CrossRef PubMed
• ↵ Alemi F , Sullivan T. Tutorial on risk adjusted X-bar charts: applications to measurement of diabetes control. Qual Manag Health Care 2001 ; 9 : 57 –65. OpenUrl PubMed
• ↵ Aronsky D , Kendall D, Merkley K, et al. A comprehensive set of coded chief complaints for the emergency department. Acad Emerg Med 2001 ; 8 : 980 –9. OpenUrl CrossRef PubMed Web of Science
• ↵ Caron A , Neuhauser D V. Health care organization improvement reports using control charts for key quality characteristics: ORYX measures as examples. Qual Manag Health Care 2001 ; 9 : 28 –39. OpenUrl PubMed
• ↵ Quinn D C , Graber A L, Elasy T A, et al. Overcoming turf battles: developing a pragmatic, collaborative model to improve glycemic control in patients with diabetes. Jt Comm J Qual Improv 2001 ; 27 : 255 –64. OpenUrl PubMed
• ↵ Roman S H , Chassin M R. Windows of opportunity to improve diabetes care when patients with diabetes are hospitalized for other conditions. Diabetes Care 2001 ; 24 : 1371 –6. OpenUrl Abstract / FREE Full Text
• ↵ Boelle P Y , Bonnet F, Valleron A J. An integrated system for significant anaesthetic events monitoring. J Am Med Inform Assoc 2002 ; 9 : S23 –7. OpenUrl Abstract / FREE Full Text
• ↵ Burnett L , Chesher D, Burnett J R. Optimizing the availability of “stat” laboratory tests using Shewhart “C” control charts. Ann Clin Biochem 2002 ; 39 : 140 –4. OpenUrl CrossRef PubMed Web of Science
• ↵ Carey R G . Improving patient satisfaction: a control chart case study. J Ambul Care Manage 2002 ; 25 : 78 –83. OpenUrl PubMed
• ↵ Chu J , Neuhauser D V, Schwartz I, et al. The efficacy of automated/electrical twitch obtaining intramuscular stimulation (atoims/etoims) for chronic pain control: evaluation with statistical process control methods. Electromyogr Clin Neurophysiol 2002 ; 42 : 393 –401. OpenUrl PubMed
• ↵ Curran E T , Benneyan J C, Hood J. Controlling methicillin-resistant Staphylococcus aureus: A feedback approach using annotated statistical process control charts. Infect Control Hosp Epidemiol 2002 ; 23 : 13 –8. OpenUrl CrossRef PubMed Web of Science
• ↵ Marshall T , Mohammed M A, Lim H T. Understanding variation for clinical governance: an illustration using the diagnosis and treatment of sore throat. Br J Gen Pract 2002 ; 52 : 277 –83. OpenUrl Abstract / FREE Full Text
• ↵ Stewart L J , Greisler D. Measuring primary care practice performance within an integrated delivery system: a case study. J Healthc Manag 2002 ; 47 : 250 –61. OpenUrl PubMed Web of Science
• ↵ Fasting S , Gisvold S E. Statistical process control methods allow the analysis and improvement of anesthesia care. Can J Anaesth 2003 ; 50 : 767 –74. OpenUrl PubMed Web of Science
• ↵ Hyrkas K , Lehti K. Continuous quality improvement through team supervision supported by continuous self-monitoring of work and systematic patient feedback. J Nurs Manag 2003 ; 11 : 177 –88. OpenUrl PubMed
• ↵ Marshall T , Mohammed M A. Understanding variation in quality improvement: the treatment of sore throats in primary care. Fam Pract 2003 ; 20 : 69 –73. OpenUrl Abstract / FREE Full Text
• ↵ Mertens W C , Higby D J, Brown D, et al. Improving the care of patients with regard to chemotherapy-induced nausea and emesis: the effect of feedback to clinicians on adherence to antiemetic prescribing guidelines. J Clin Oncol 2003 ; 21 : 1373 –8. OpenUrl Abstract / FREE Full Text
• ↵ Norberg A , Christopher N C, Ramundo M L, et al. Contamination rates of blood cultures obtained by dedicated phlebotomy vs intravenous catheter. JAMA 2003 ; 289 : 726 –9. OpenUrl CrossRef PubMed Web of Science
• ↵ Rosow E , Adam J, Coulombe K, et al. Virtual instrumentation and real-time executive dashboards. Solutions for health care systems. Nurs Adm Q 2003 ; 27 : 58 –76. OpenUrl CrossRef PubMed
• ↵ Alemi F , Neuhauser D. Time-between control charts for monitoring asthma attacks. Jt Comm J Qual Saf 2004 ; 30 : 95 –102. OpenUrl PubMed
• ↵ Hart M K , Robertson J W, Hart R F, et al. Application of variables control charts to risk-adjusted time-ordered healthcare data. [comment]. Qual Manag Health Care 2004 ; 13 : 99 –119. OpenUrl PubMed
• ↵ Nelson E C , Batalden P B, Huber T P, et al. Microsystems in health care: Part 1. Learning from high-performing front-line clinical units. Jt Comm J Qual Improv 2002 ; 28 : 472 –93. OpenUrl PubMed
• ↵ Olsson J , Elg M, Molfenter T. Reflections on transnational transferability of improvement technologies—a comparison of factors for successful change in the United States and northern Europe. Qual Manag Health Care 2003 ; 12 : 259 –69. OpenUrl PubMed
• ↵ Thor J , Wittlov K, Herrlin B, et al. Learning helpers: how they facilitated improvement and improved facilitation—lessons from a hospital-wide quality improvement initiative. Qual Manag Health Care 2004 ; 13 : 60 –74. OpenUrl PubMed
• ↵ Davidoff F , Batalden P. Toward stronger evidence on quality improvement. Draft publication guidelines: the beginning of a consensus project. Qual Saf Health Care 2005 ; 14 : 319 –25. OpenUrl Abstract / FREE Full Text
• ↵ Pronovost P , Wachter R. Proposed standards for quality improvement research and publication: one step forward and two steps back. Qual Saf Health Care 2006 ; 15 : 152 –3. OpenUrl FREE Full Text
• ↵ Eccles M , Grimshaw J, Campbell M, et al. Research designs for studies evaluating the effectiveness of change and improvement strategies. Qual Saf Health Care 2003 ; 12 : 47 –52. OpenUrl Abstract / FREE Full Text
• ↵ Diaz M , Neuhauser D. Pasteur and parachutes: when statistical process control is better than a randomized controlled trial. Qual Saf Health Care 2005 ; 14 : 140 –3. OpenUrl FREE Full Text
• ↵ Neuhauser D , Diaz M. Quality improvement research: are randomised trials necessary? Qual Saf Health Care 2007 ; 16 : 77 –80. OpenUrl FREE Full Text
• ↵ Benneyan J C , Borgman A D. Risk-adjusted sequential probability ratio tests and longitudinal surveillance methods [see comment] [comment]. Int J Qual Health Care 2003 ; 15 : 5 –6. OpenUrl FREE Full Text
• ↵ Lim T O . Statistical process control tools for monitoring clinical performance. Int J Qual Health Care 2003 ; 15 : 3 –4. OpenUrl FREE Full Text
• ↵ Berwick D M , James B, Coye M J. Connections between quality measurement and improvement. Med Care 2003 ; 41 (1 Suppl) : I30 –8. OpenUrl CrossRef PubMed Web of Science
• ↵ Mohammed M A , Cheng K K, Rouse A, et al. Bristol, Shipman, and clinical governance: Shewhart’s forgotten lessons. Lancet 2001 ; 357 : 463 –7. OpenUrl CrossRef PubMed Web of Science
• ↵ Mays N , Pope C, Popay J. Systematically reviewing qualitative and quantitative evidence to inform management and policy-making in the health field. J Health Serv Res Policy 2005 ; 10 (Suppl 1) : 6 –20. OpenUrl Abstract / FREE Full Text
• ↵ Pawson R , Greenhalgh T, Harvey G, et al. Realist review—a new method of systematic review designed for complex policy interventions. J Health Serv Res Policy 2005 ; 10 (Suppl 1) : 21 –34. OpenUrl Abstract / FREE Full Text
• ↵ Dixon-Woods M , Agarwal S, Jones D, et al. Synthesising qualitative and quantitative evidence: a review of possible methods. J Health Serv Res Policy 2005 ; 10 : 45 –53. OpenUrl Abstract / FREE Full Text
• ↵ Harden A , Garcia J, Oliver S, et al. Applying systematic review methods to studies of people’s views: an example from public health research. J Epidemiol Community Health 2004 ; 58 : 794 –800. OpenUrl Abstract / FREE Full Text

Funding: No dedicated funding was received for this study. All coauthors were supported by their respective employers in conducting this research as part of their work.

Competing interests: None.

Linked Articles

• Quality Lines Quality Lines David P Stevens BMJ Quality & Safety 2007; 16 322-322 Published Online First: 03 Oct 2007.

Read the full text or download the PDF:

• Guide: Statistical Process Control (SPC)

Daniel Croft

Daniel Croft is an experienced continuous improvement manager with a Lean Six Sigma Black Belt and a Bachelor's degree in Business Management. With more than ten years of experience applying his skills across various industries, Daniel specializes in optimizing processes and improving efficiency. His approach combines practical experience with a deep understanding of business fundamentals to drive meaningful change.

• Last Updated: March 24, 2024
• Learn Lean Sigma

Statistical Process Control is a key method used in making sure products are made to specification and efficiently, especially in making expensive products like cars or electronics where margins can be low and the cost of defects can eliminate profitaility. It uses a detailed, numbers-based way to monitor, manage, and improve processes where product are made or services are provided.

SPC focuses on carefully examining data to help businesses understand problems or issues in how they make products or provide services. It’s all about making decisions based on solid evidence rather than guesses or feelings, with the goal of finding and fixing any changes in the process that could affect the quality of the product.

Table of Contents

What is statistical process control.

Statistical Proces Control is a key method used in quality control, and Lean Six Sigma used to maintain and improve product quality and efficiency. It is used in various industries but is primarily used in manufacturing as a systematic, data-driven approach to uses statistical methods to monitor, control, and improve processes and process performance. 

SPC is a data-driven methodology that relies heavily on data to analyze process performance. By collecting and analyzing data from various stages of a manufacturing or service process, SPC enables organizations to identify trends, patterns, and anomalies. This data-driven approach helps in making informed decisions rather than relying on assumptions or estimations.

The main aim of SPC is to detect and reduce process variability. Variability is a natural aspect of any process, but excessive variability can lead to defects, inefficiency, and reduced product quality. By understanding and controlling this variability, organizations can ensure that their processes consistently produce items within desired specifications. SPC involves continuous monitoring of the process to quickly identify deviations from the norm. This systematic approach ensures that problems are detected early and can be rectified before they result in significant quality issues or production waste.

History and Background of the Development of SPC

Walter shewhart and control charts.

The foundations of SPC were laid in the early 20th century by Walter Shewhart working at Bell Laboratories. Shewhart’s primary contribution was the development of the control chart , a tool that graphically displays process data over time and helps in distinguishing between normal process variation and variation that signifies a problem. The control chart remains a cornerstone of SPC and is widely used in various industries to monitor process performance.

W. Edwards Deming and Post-War Japan

After World War II, W. Edwards Deming brought the concepts of SPC to Japan, where they played a key role in the country’s post-war industrial rebirth. Deming’s teachings emphasized not only statistical methods but also a broader philosophical approach to quality. He advocated for continuous improvement ( Kaizen ) and total quality management , integrating SPC into a more comprehensive quality management system.

Impact on Manufacturing and Beyond

The implementation of SPC led to significant improvements in manufacturing quality and efficiency. It allowed companies to produce goods more consistently and with fewer defects. The principles of SPC have since been adopted in various sectors beyond manufacturing, including healthcare, finance, and service industries, demonstrating its versatility and effectiveness in process improvement.

Fundamental Concepts of SPC

To understand and read what control charts are telling you it is first important to understand how variation and how it might be displayed on the chart. In everything, there is always a level of variation relative to what is being measured. But we must identify what acceptable variation and what is a variation that needs to be explored and addressed.

Understanding Process Variation

In SPC, process variation is categorized into two types: common causes and special causes.

• Common Cause Variation: These are inherent variations that occur naturally within a process. They are predictable and consistent over time. In the image below common cause variation is the variation within the control limits
• Special Cause Variation: These variations are due to external factors and are not part of the normal process. They are unpredictable and can indicate that the process is out of control. In the image below, the special cause variation is the data point outside the upper control limit 

Control Charts

Control charts are essential tools in SPC, used to monitor whether a process is in control.

• Graphical Representation: They graphically represent data over time, providing a visual means to monitor process performance.
• Control Limits: Control charts use upper and lower control limits, which are statistically derived boundaries. They help in distinguishing between normal process variation (within limits) and variations that require attention (outside limits).

Types of Control Charts

After understanding the types of variation you might find on a control chat, it is important to understand the types of control charts in SPC. This is crucial for effectively monitoring and improving various processes. These charts are broadly categorized based on the type of data they handle: variable data and attribute data. Additionally, the implementation of SPC in a process, from data collection to continuous improvement, is a systematic approach that requires diligence and precision. Let’s explore these aspects in more detail.

Variable Data Control Charts

Variable data control charts are used for data that can be measured on a continuous scale. This includes characteristics like weight, length, or time.

X-bar and R Chart

• Purpose: Used to monitor the mean (average) and range of a process.
• Application: Ideal for situations where sample measurements are taken at regular intervals and the mean and variability of the process need to be controlled.
• Structure: The X-bar chart shows how the mean changes over time, while the R chart displays the range (difference between the highest and lowest values) within each sample.

Individual-Moving Range (I-MR) Chart

• Purpose: Monitors individual observations and the moving range between observations.
• Application: Useful when measurements are not made in subgroups but as individual data points.
• Structure: The I chart tracks each individual data point, and the MR chart shows the range between consecutive measurements.

Attribute Data Control Charts

Attribute data control charts are used for data that are counted, such as defects or defective items.

P-chart (Proportion Chart)

• Purpose: Monitors the proportion of defective items in a sample.
• Application: Ideal for quality characteristics that are categorical (e.g., defective vs. non-defective) and when the sample size varies.
• Structure: It plots the proportion of defectives in each sample over time.

C-chart (Count Chart)

• Purpose: Tracks the count of defects per unit or item.
• Application: Used when the number of opportunities for defects is constant, and defects are counted per item or unit.
• Structure: It plots the number of defects in samples of a constant size.

Implementing SPC in a Process

Implementing SPC in a process is a structured approach that involves several key steps: data collection, establishing control limits, monitoring and interpretation, and continuous improvement. Each of these steps is critical to the successful application of SPC. Let’s explore these steps in more detail.

Step 1: Data Collection

First, data must be collected systematically to ensure it accurately represents the process. This involves deciding what data to collect, how often to collect it, and the methods used for collection. The selection of data is important. It should be relevant to the quality characteristics you want to control. For example, in a manufacturing process, this might include measurements of product dimensions, the time taken for a process step, or the number of defects.

The data should be representative of the actual operating conditions of the process. It means collecting data under various operating conditions and over a sufficient period.

The sample size and frequency of data collection should be adequate to capture the variability of the process. It’s a balance between collecting enough data for reliability and the practicality of data collection.

Step 2: Establishing Control Limits

Control limits are calculated using historical process data. They are statistical representations of the process variability and are usually set at ±3 standard deviations from the process mean.

These limits reflect what the process can achieve under current operating conditions.

To help you calculate your data control limits, you can use our Control Limits Calculator.

Control limits are not fixed forever. As process improvements are made, these limits may be recalculated to reflect the new level of process performance.

When significant changes are made to a process (like new machinery, materials, or methods), it might be necessary to recalculate the control limits based on new performance data.

Step 3: Monitoring and Interpretation

Regularly reviewing control charts is essential for timely detection of out-of-control conditions. Apart from individual points, it’s crucial to look for patterns or trends in the data, which could indicate potential issues.

When data points fall outside the control limits or exhibit non-random patterns, it triggers a need for investigation. The goal is to identify the root cause of the variation, whether it’s a common cause that requires a process change, or a special cause that might be addressed more immediately.

Step 4: Continuous Improvement

SPC is not just about maintaining control; it’s about continuous improvement. The insights gained from SPC should drive ongoing efforts to enhance process performance.

Based on SPC data, processes can be adjusted, improved, and refined over time. This might involve changes to equipment, materials, methods, or training.

In conclusion, SPC is a key tool in the aim for quality control and process improvement. Its strength lies in its ability to make process variability visible and manageable. From the seminal contributions of Walter Shewhart and W. Edwards Deming, SPC has evolved into a comprehensive approach that integrates seamlessly with various quality management systems.

By continuously monitoring processes through control charts and adapting to the insights these charts provide, SPC empowers organizations to maintain control over their processes and pursue relentless improvement. Thus, SPC not only sustains but also elevates the standards of quality, efficiency, and customer satisfaction in diverse industrial landscapes.

• Madanhire, I. and Mbohwa, C., 2016. Application of statistical process control (SPC) in manufacturing industry in a developing country .  Procedia Cirp ,  40 , pp.580-583.
• Gérard, K., Grandhaye, J.P., Marchesi, V., Kafrouni, H., Husson, F. and Aletti, P., 2009. A comprehensive analysis of the IMRT dose delivery process using statistical process control (SPC).   Medical physics ,  36 (4), pp.1275-1285.

Q: What is Statistical Process Control?

A : SPC is a method used to monitor, control, and improve processes by analyzing performance data to identify and eliminate unwanted variations.

Q: Why is SPC important?

A : SPC helps ensure processes are consistent and predictable. It aids in early detection of issues, reducing defects, and improving overall product or service quality.

Q: What is a control chart?

A : A control chart is a graphical representation used in SPC to plot process data over time, with control limits that help distinguish between common and special cause variations.

Q: How are control limits determined?

A : Control limits are typically set at three standard deviations above and below the process mean, based on historical data. However, these limits can be adjusted depending on the specific chart type and industry standards.

Q: What's the difference between common cause and special cause variation?

A : Common cause variation is the inherent variability in a process, while special cause variation arises from specific, unusual events and is not part of the normal process.

Daniel Croft is a seasoned continuous improvement manager with a Black Belt in Lean Six Sigma. With over 10 years of real-world application experience across diverse sectors, Daniel has a passion for optimizing processes and fostering a culture of efficiency. He's not just a practitioner but also an avid learner, constantly seeking to expand his knowledge. Outside of his professional life, Daniel has a keen Investing, statistics and knowledge-sharing, which led him to create the website learnleansigma.com, a platform dedicated to Lean Six Sigma and process improvement insights.

Download Template

Free Lean Six Sigma Templates

Improve your Lean Six Sigma projects with our free templates. They're designed to make implementation and management easier, helping you achieve better results.

Other Guides

Was this helpful.

Case study: How Automated SPC Helps One Manufacturer Consistently Improve

Quality data can help manufacturers to advance quality and streamline their operations – if they collect it properly. Instead of gathering plant floor data manually, which many large organizations still do, manufacturers can automate statistical process control (SPC) data collection and get a major leg up on possible quality issues.

Manual data collection increases the risk of errors. Humans make mistakes, and that means that valuable data can be typed in incorrectly or completely overlooked. When out-of-specification data is missed, manufacturers cannot immediately spot quality issues and prevent problems. Manual data collection also does not provide the long-term operational perspective that automated, insight-driven data collection can offer – knowledge about operational trends, which can ultimately drive continuous improvement.

Automated data collection can help quality professionals glean operational insights that lead to boosts in productivity and higher levels of customer satisfaction. Manufacturing companies that use automated data collection see benefits that range from improved product quality and compliance, reduced costs and risk, and the ability to make more strategic, data-driven business decisions.

For instance, one of Middleborough, Mass.-based Ocean Spray's plants began

using an automated data collection system in 2006 to visualize its manufacturing data. This quality team wanted access instead of paper-based records.

Due to the facility's success implementing automated data collection, the company's remaining food processing plants have also jumped onboard, using digital SPC to gain insights and boost quality.

In December 2018, Ocean Spray’s Markham, Wash. Plant had a process capability of 0.96 for a dried cranberries quality process. The following month, it was 1.46.

Prior to implementing SPC, operators at Ocean Spray’s food processing plants ran their operations on spec vs on target. They made individual adjustments to finished products at a cost to the company. In other words, the company spent additional money and used extra ingredients to achieve a desired quality level. With multiple plants producing the same products, achieving this consistency in both quality and consumer experience was expensive.

Once they made decisions with insights gleaned from SPC, the company manufactured products to target – not above. This ultimately saved the company money and still pleased clients.

Automated SPC ultimately helped Ocean Spray to see more clearly, and consistently into its sweetened dried cranberry manufacturing process. This insight made it easier to compare processes across plants and make immediate, significant fixes.

Collecting precise, consistent data became simpler, and that made it easier for Ocean Spray quality teams to create more streamlined and accurate reporting. At the end of every month, Ocean Spray seamlessly pulls data from its food plants, and can compare information just as smoothly.

In addition, its quality team spends much less time handling data, as all five of its plants now store information in the same format, in the same database, and using the same naming schemes.

With customer satisfaction and continuous improvement driving Ocean Spray Cranberries Inc.’s success, using automated SPC across its food and beverage manufacturing facilities has helped it remain a leader in its field.

Share This Story

Darryl Seland is Editor-in-Chief of Quality magazine. He can be reached at [email protected] .

Restricted Content

You must have JavaScript enabled to enjoy a limited number of articles over the next 30 days.

Related Articles

How SPC Software Helps Medical and Pharmaceutical Manufacturers Stay On Track

Case study: automated testing maintains reputation, case study: optics manufacturer decreases part failure rate by 30%, get our new emagazine delivered to your inbox every month., stay in the know with quality’s comprehensive coverage of the manufacturing and metrology industries..

Copyright ©2024. All Rights Reserved BNP Media.

Design, CMS, Hosting & Web Development :: ePublishing

Statistical Process Control (SPC)

• First Online: 06 January 2024

Cite this chapter

• Stefan Bracke 2  

252 Accesses

The controllability and quality capability of production processes is a precondition for ensuring the reliable manufacture of technically complex products. The term “reliability” refers here – in deviation from the definition in Chap. 3 – to the probability of a defect-free proportion with respect to a monitored product characteristic within a defined time interval of product manufacture. If the monitoring refers to a production process characteristic, “reliability” refers to the probability that the production process parameters of the production process range within the specification limits within a time interval and thus the prerequisite for fault-free production is fulfilled. Statistical process control (SPC) is an established method for ensuring controlled and quality capable manufacturing processes in industrial practice. Statistical process control is a way of maintaining an optimized manufacturing process in an optimized state through continuous monitoring and, if necessary, corrections.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

• Available as PDF
• Read on any device
• Instant download
• Own it forever
• Available as EPUB and PDF
• Durable hardcover edition
• Dispatched in 3 to 5 business days
• Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Author information

Authors and affiliations.

Lehrstuhl für Zuverlässigkeitstechnik und Risikoanalytik, Bergische Universität Wuppertal, Wuppertal, Germany

Stefan Bracke

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Stefan Bracke .

Rights and permissions

Reprints and permissions

Copyright information

© 2024 Springer-Verlag GmbH Germany, part of Springer Nature

About this chapter

Bracke, S. (2024). Statistical Process Control (SPC). In: Reliability Engineering. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-662-67446-8_16

Download citation

DOI : https://doi.org/10.1007/978-3-662-67446-8_16

Published : 06 January 2024

Publisher Name : Springer, Berlin, Heidelberg

Print ISBN : 978-3-662-67445-1

Online ISBN : 978-3-662-67446-8

eBook Packages : Engineering Engineering (R0)

Share this chapter

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Publish with us

Policies and ethics

• Find a journal
• Track your research

Statistical Process Control for the Food Industry by Sarina A. Lim, Jiju Antony

Get full access to Statistical Process Control for the Food Industry and 60K+ other titles, with a free 10-day trial of O'Reilly.

There are also live events, courses curated by job role, and more.

10 Case Studies

This chapter disclosed the case studies gathered from research and literature. Each case study is described in brief, focusing on the application of the specific expediting process control approach it illustrates, including the approach to choose pilot projects, the control of key quality of the process attributes, and the application of various type of control charts served for different types of data. This chapter displays how SPC has been successfully applied in several different situations.

The details and situation of the cases are varied, however in general each will provide the background of the cases, the data and suitable type of charts. The charts are interpreted, and interesting points are discussed in the case study. It is not necessarily to understand the calculation and detail calculations to benefit from this book. Implementing SPC in the organisation, it only requires a few number staff to understand the theory and formulae on the calculations of control charts.

10.1 Application of the Control Charts in the Industries

Control charts are typically taught in the classes and mostly when teaching the theory of SPC it is understandable to find the data for the control chart are the easiest to be understood. Theoretically, normality is expected in the food processing data; however, this is rarely the case in the food industry. In the real‐world data, neither SPC nor any other technique is the solution for all problems, where other quality tools may ...

Get Statistical Process Control for the Food Industry now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.

Don’t leave empty-handed

Get Mark Richards’s Software Architecture Patterns ebook to better understand how to design components—and how they should interact.

It’s yours, free.

Check it out now on O’Reilly

Dive in for free with a 10-day trial of the O’Reilly learning platform—then explore all the other resources our members count on to build skills and solve problems every day.

A statistical process control case study

Affiliation.

• 1 Department of Health Services and Information Management, School of Allied Health Sciences, East Carolina University, Greenville, NC 27858, USA. [email protected]
• PMID: 17047496
• DOI: 10.1097/00019514-200610000-00004

Statistical process control (SPC) charts can be applied to a wide number of health care applications, yet widespread use has not occurred. The greatest obstacle preventing wider use is the lack of quality management training that health care workers receive. The technical nature of the SPC guarantees that without explicit instruction this technique will not come into widespread use. Reviews of health care quality management texts inform the reader that SPC charts should be used to improve delivery processes and outcomes often without discussing how they are created. Conversely, medical research frequently reports the improved outcomes achieved after analyzing SPC charts. This article is targeted between these 2 positions: it reviews the SPC technique and presents a tool and data so readers can construct SPC charts. After tackling the case, it is hoped that the readers will collect their own data and apply the same technique to improve processes in their own organization.

• Data Interpretation, Statistical*
• Diffusion of Innovation*
• Efficiency, Organizational* / statistics & numerical data
• Health Facilities
• Outcome and Process Assessment, Health Care / methods*
• Outcome and Process Assessment, Health Care / statistics & numerical data
• Total Quality Management / methods
• Total Quality Management / statistics & numerical data*
• United States

A practical application of statistical process control to evaluate the performance rate of academic programmes: implications and suggestions

Quality Assurance in Education

ISSN : 0968-4883

Article publication date: 22 August 2022

Issue publication date: 28 September 2022

This study aims to properly and objectively assess the students’ study progress in bachelor programmes by applying statistical process control (SPC). Specifically, the authors focused their analysis on the variation in performance rates in business studies courses taught at a Spanish University.

Design/methodology/approach

A qualitative methodology was used, using an action-based case study developed in a public university. Previous research and theoretical issues related to quality indicators of the training programmes were discussed, followed by the application of SPC to assess these outputs.

The evaluation of the performance rate of the courses that comprised the training programs through the SPC revealed significant differences with respect to the evaluations obtained through traditional evaluation procedures. Similarly, the results show differences in the control parameters (central line and control interval), depending on the adopted approach (by programmes, by academic year and by department).

Research limitations/implications

This study has inherent limitations linked to both the methodology and selection of data sources.

Practical implications

The SPC approach provides a framework to properly and objectively assess the quality indicators involved in quality assurance processes in higher education.

Originality/value

This paper contributes to the discourse on the importance of a robust and effective assessment of quality indicators of the academic curriculum in the higher education context through the application of quality control tools such as SPC.

• Quality assurance
• Higher education
• Statistical process control
• Performance rate
• Evaluation process
• Academic programme

Gessa, A. , Marin, E. and Sancha, P. (2022), "A practical application of statistical process control to evaluate the performance rate of academic programmes: implications and suggestions", Quality Assurance in Education , Vol. 30 No. 4, pp. 571-588. https://doi.org/10.1108/QAE-03-2022-0065

Emerald Publishing Limited

Copyright © 2022, Ana Gessa, Eyda Marin and Pilar Sancha.

Published by Emerald Publishing Limited. This article is published under the Creative Commons Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, translate and create derivative works of this article (for both commercial & non-commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence may be seen at http://creativecommons.org/licences/by/4.0/legalcode

Introduction

The need to advance the modernisation of universities is a key element in developing a knowledge-based society and a more competitive economy. In this context, it is not surprising that quality assurance, has become one of the cornerstones for policy and decision makers in higher education institutions in the European Higher Education Area (EHEA) ( Curaj et al. , 2015 ).

Quality assurance in higher education has two primary functions. First, it establishes the legitimacy of an institution and the academic programmes it offers. Second, it informs the institutions’ stakeholders about program objectives and outcomes and the fulfilment of the expected quality standards ( Kinser, 2014 ).

In the European Union, national accreditation and quality assurance agencies play a key role, as they are in charge of quality management through evaluation, certification and accreditation of programmes, professors and universities. They design different programmes to guarantee internal and external quality, following the Standards and Guidelines for Quality Assurance in the EHEA (ESG), adopted by the European Association for Quality Assurance in Higher Education (ENQA) (2015) .

Under ENQA’s umbrella, universities apply procedures that facilitate both the improvement of the quality of their degrees and the external evaluation processes conducted by the corresponding competent institutions. In this sense, the indicators used for the evaluation of the academic curriculum performance constitute a key element for accountability and transparency.

Despite the experience accumulated over more than a decade in university degree programme evaluation processes, criticisms of approaches to definition and to operationalisation is recurrent in the literature ( Hanna et al. , 2012 ; Strang et al. , 2016 ). The main concerns across agencies relate, among others, to the consolidation of good practices and appropriate statistical principles when evaluating and analysing performance indicators.

subjectivity about what should be considered excellent, acceptable, or insufficient prevails in the analysis;

the lack of approaches that consider the context of universities, faculties, departments and even courses;

the ignorance of uncontrollable factors that underline the inherent variability in the educational processes; and

more importantly – the extended practice of comparing performance with averages or the use of arbitrary cut-off numbers without taking into consideration that educational processes have an inherent variability ( Bi, 2018 ).

An alternative approach to overcome these shortcomings and remove the barriers is the application of statistical process control (SPC), something which has been widely and successfully used for decades in the manufacturing industry, and later more slowly introduced in the service sector, in general and in the educative sector in particular ( MacCarthy and Wasusri, 2002 ; Sulek, 2004 ; Suman and Prajapati, 2018 ; Utley and May, 2009 ; Yang et al. , 2012 ). Contrary to the classical statistical methods that are developed for fixed populations and not for processes, SPC can be very effective in detecting process shifts and the dynamics of the process itself. Moreover, it enables hidden problems to be revealed therefore indicating the actions necessary for continuous improvement.

In any process, a certain amount of inherent or natural variability will always exist due to the cumulative effect of unavoidable causes. These sources of variability are called “chance causes” ( Montgomery, 2009 ). SPC helps to assess the variability of the educational processes, distinguishing between assignable (inappropriate educational resources and methodology; ineffective curriculum, etc). and random (student profile, family context, etc). causes ( Daneshmandi et al. , 2020 ; Nikolaidis and Dimitriadis, 2014 ).

This work aims to contribute to understanding the usefulness of the application of SPC, in the processes of accreditation and monitoring of university degrees, through the analysis of the variability of the performance rate associated with the procedures that constitute the quality assurance system (QAS). Specifically, the authors focussed their analysis on the variation in performance rates in business studies taught at a Spanish University. Based on this goal, the following research question is suggested: Are there any significant differences between the results obtained with a standard statistical analysis and the results obtained through the SPC application?

For this purpose, the next section outlines the theoretical framework of this study. Subsequently, the paper presents the methodology and results. The last section discusses the conclusions.

Theoretical framework

Quality assurance in spanish higher education systems.

Quality assurance has become one of the key issues for higher education systems (HES) around the world. The changes that have arisen in the European education area have forced university institutions to adopt new management models, with a priority being to guarantee the quality of the studies, as a contributing factor to the development of the economy and society [Red Iberoamericana para la Acreditación de la Calidad de la Educación Superior (RIACES), 2007]. The comparability and recognition of degrees from the perspective of quality management, strengthened by evaluation and assurance mechanisms and quality of higher education qualifications and certifications, has undoubtedly made the EHEA possible. The key instrument to make it possible was the ESG [ European Association for Quality Assurance in Higher Education (ENQA), 2015 ]. Specifically:

[…] they set a common framework for quality assurance systems for learning and teaching at European, national and institutional level; they enable the assurance and improvement of the quality of higher education in the EHEA; they support mutual trust, thus facilitating recognition and mobility within and across national borders; and they provide information on quality assurance in the EHEA [ European Association for Quality Assurance in Higher Education (ENQA), 2015 , p. 7].

“Higher education institutions have primary responsibility for the quality of their provision and its assurance;

quality assurance responds to the diversity of HES, institutions, programmes and students;

quality assurance supports the development of a quality of culture; and

quality assurance takes into account the needs and expectations of students, and other stakeholders and society” [ European Association for Quality Assurance in Higher Education (ENQA), 2015 , p. 8].

European agencies vary in their approaches to the implementation and adaptation of the ESG ( Alzafari and Ursin, 2019 ; Kohoutek et al. , 2018 ; Nascimbeni, 2015 ; Manatos and Huisman, 2020 ). In Spain, the National Agency for Quality Assessment and Accreditation (ANECA) has developed a set of programmes to conduct its activities of evaluation, certification and accreditation of teaching staff, universities and degrees ( Table 1 ). These programmes constitute a framework that makes it easier for universities to implement their internal QAS as well as the external evaluation processes conducted by the responsible regional agencies and universities.

performance rate in the qualification;

dropout rate;

efficiency rate; and

graduation rate.

The reports also include global data regarding the qualification and an analysis of the adequacy of the evolution of the indicators and their consistency with the target established in the qualification verification report.

The usual practice of universities and evaluation agencies when analysing these indicators is limited to the application of conventional descriptive statistics methods that involves the analysis of variation through measures such as mean, standard deviation, median etc. As a result, the analysis seems subjective, decontextualized and unsound, as it focusses on the evaluation of the outcomes of the educational processes and their comparison with outdated fixed reference values ( Andreani et al. , 2020 ; Klasik and Hutt, 2019 ). Therefore, this practice does not enable us to identify whether the process is under control and if it is not, to identify if this status is due to causes attributable to the processes themselves or to random causes, depending, in turn, on factors such as the branch of study, the type of centre, the university and the region ( Bi, 2018 ; Hanna et al. , 2012 ; Kember and Leung, 2011 ). An alternative practice to overcome this limitation could be the application of SPC charts, which have become a useful tool with which to achieve stability and improvement in the quality of educational service delivery by monitoring certain variables or attributes over time, such as the previously mentioned performance indicators. The application of SPC in an educational context is already a reality, and it is increasing, as shown in the next section. This will allow the progressive implementation of typical practices of quality management in the field of education, within the general framework of the continuous improvement of processes.

Statistical control of processes in education services

SPC allows the identification of different sources of variation and also enables detection of an “out of control” status ( Besterfield, 1995 ; Shewhart, 1936 ). A control chart “shows the value of the quality characteristics of interest as a function of time or sample number” ( Montgomery, 2009 ). Thus, the variability of a quality characteristic should be based on output, which involves estimating its statistical distribution and parameters ( Juran and Gryna, 1988 ).

The standard control chart shows a central line (CL), which is the mean of the variable being monitored, the upper control limit (UCL) and the lower control limit (LCL), which are typically set at  ± 3 standard deviations of the CL and its graphic representation is shown in Figure 1 .

When a plot violates a control limit, which is when some of the plot points fall outside the three-sigma limits (Zone B), it should be treated as a special cause of variation, attributable to the system. If special causes are present, it is said that the process is out of control. The zone between the UCL and the LCL (Zone A) shows the expected normal (common cause) plot point variation. Unlike special causes of variability, common causes are also called “natural” or “random variability” and are due to a large number of small sources of variation that are not easily identifiable. If assignable or special causes of variation are removed, characteristic parameters such as the mean, standard deviation and probability distribution are constant and process behaviour is predictable; the system is said to be “in a state of statistical control” or simply “under control”. However, certain abnormal patterns (trends, sudden shifts, systematic variation, cycles and mixtures) may alert one to the existence of special causes.

There are a great variety of quality control charts, depending on the type of quality characteristic to control (variable or attributes) and the number of variables to control (univariate or multivariate) ( Montgomery, 2009 ).

controlling student achievement;

monitoring the effectiveness of the teaching–learning process;

evaluating student satisfaction; and

identification abnormal patterns in certain educational processes.

Statistical process control charts to control students’ achievements

Objective and quantitative measures such as examination scores or grade point averages (GPAs) to monitor learning performance have been used to conduct numerous studies, under different approaches and educational levels. Thus, for example, Schafer et al. (2011) , in their work, used traditional Shewhart X-R charts to follow the performance of primary and secondary students in large-scale assessment programmes. In a similar way, Bakir and McNeal (2010) and Bakir et al. (2015) designed a non-parametric control chart based on the sign statistic to detect statistically significant shifts in students’ GPAs from a desired level. Milosavljevic et al. (2018) used attribute charts from the perspective of the number of passing exams. Other studies following a similar framework include Peterson (2015) , Zolkepley et al. (2018) , Hrynkevych (2017) , Aldowaisan and Allahverdi (2016) , Mazumder (2014) , Djauhari et al. (2017) , Cervetti et al. (2012) and Hanna et al. (2012) .

SPC charts have also been applied to monitor the consistency of scale scores or ratings over time. Lee and von Davier (2013) used cumulative sum charts (CUSUM) in a study conducted with data from 72 countries to detect changes in a measurement process of rating performance items in operational assessments. Omar (2010) proposed X-S charts to monitor the consistency of the scores in a co-assessment process. SPC charts that have also been applied in similar frameworks are Beshah (2012) , Edwards et al. (2007) and Savic (2006) .

Monitoring the effectiveness of the teaching–learning process

Another line of research aims to measure both teachers’ contributions to increasing student knowledge and student’s learning outcomes. The technique usually applied for this purpose consists of administering to students, prior to a lecture, a test known as the Background Knowledge Probe (BKP), which contains questions about concepts covered during the lecture, and after the lecture students have to answer the same questions. Rather than grading the students’ outcomes, the BKP should be understood as a resource to measure student’s gain. This would enable an improvement in the teaching programmes by virtue of detecting the student’s mistakes and gaps in knowledge transfer. Thus, Green et al. (2012) used traditional mean graphics (Graph X), whereas Agrawal and Khan (2008) , Grygoryev and Karapetrovic (2005a , 2005b ), Karapetrovic and Rajamani (1998) and Pierre and Mathios (1995) used graphics of non-conformity attributes ( p graphics).

Evaluating students’ satisfaction

SPC chart techniques have also been used to monitor and evaluate the level of student satisfaction regarding the quality of faculty members’ teaching and university service operation. Thus, Jensen and Markland (1996) published one of the first studies in this field, reporting the use of a Hotelling’s T multivariate control chart to detect shifts in satisfactory and unsatisfactory perceptions of computer services at a large university institution. Debnath and Shankar (2014) proposed the use of attribute control charts ( c -charts and u -charts) to evaluate the level of student satisfaction with their academic process. In their study, students were asked to provide information about various parameters, such as the grievance-handling process with respect to students’ admissions and results, facilities, practical orientation process.

On the other hand, it is worth mentioning works that, under the controversial field of Student Evaluation of Instruction, are addressed from the perspective of SPC, bringing a new approach that focusses not on educational outputs and outcomes, but on the quality of the underlying educational processes. Bi (2018) and Nikolaidis and Dimitriadis (2014) used X-S charts; Cadden et al. (2008) and Marks and O'Connel (2003) used X charts; Manguad (2006) , proposed the use of X-R charts; Ding et al. (2006) and Carlucci et al. (2019) applied attributes control chart ( p and u charts, respectively). Finally, Sivena and Nikolaidis (2019) through simulation, evaluate several popular types of control charts, identifying the most suitable among them.

Identification of abnormal patterns in certain educational processes

Another segment of this body of research has looked at the use of CUSUM charts to detect known items and outliers in computer adaptive testing ( Meijer, 2002 ; Veerkamp and Glas, 2000 ). This instrument allows the level of acquisition of different skills (linguistic, professional accreditations, etc). to be detected in the context of item response theory.

Given the aforementioned considerations, three facts have motivated this research. First, the existence of inadequate statistical approaches that university institutions have traditionally used to assess certain outcomes of educational processes. Second, the potential of SPC to assess different quality characteristics for higher education institutions has been demonstrated in the literature review. Third, the authors of this paper are active members of the Board for the Monitoring and Accreditation of Qualifications at a business faculty at a Spanish University. Moreover, some of the authors teach courses related to quality management. They share the aforementioned concerns and for the sake of “practice as they preach”, they advocate that some successfully industrial quality management techniques such as SPC which is occasionally taught in their courses, should be incorporated to assess and analyse the variation related to educative performance indicators. This application would improve the assessment of the quality of educative processes so that root causes can be detected and corrective actions can be taken by those who supervise instruction Hanna et al. (2012) .

Materials and method

Research context.

The study was conducted at the Faculty of Business Sciences and Tourism (FBST) of the University of Huelva (Spain). The FBST offers three bachelor’s programmes: Bachelor in Business Management (BM), Bachelor in Finance and Accounting (F&A) and Bachelor in Tourism. Table 2 shows the main characteristics of the study context.

Design of the study

The method adopted for this study is the action-based case study approach conducted at FBST. Previous research and theoretical issues related to SPC implementation and quality indicators of the training programmes proposed in Protocol for the Monitoring and Renewal of the Accreditation for Official University Qualifications were discussed. Then the authors built a conceptual framework that provided a step-by-step approach to SPC implementation as illustrated in Figure 2 .

In this first approach applying the SPC to assess the variability of progress indicators, we only focus on the performance rate.

Applied at course level and by academic year, the performance is indicated by relationship in the total number of ordinary credits passed by the students in a particular academic year and the total number of ordinary credits in which they are enrolled [ University Commission for the Regulation of Monitoring and Accreditation (UCRMA), 2010 ].

This rate provides a snapshot of the proportion of students passing a course in an academic year. Analysis of this rate, with the help of SPC charts, helps to identify root causes such as unmotivated students to further take actions such as monitoring the enrolment process.

For this purpose, the performance rates of the courses taught at the FBST of the University of Huelva between the years 2014 and 2020 were collected from reports on the QAS of the FBST.

To avoid skewing the results, the last academic year of the degrees was excluded from the study due to the type of courses taught. It included elective courses and credits given to students for the completion of curricular internships and the final degree project.

Applying statistical process control

To apply SPC, we followed a standard set of guidelines for setting up an SPC charting scheme, illustrated in Figure 2 .

As our main goal was to monitor the effectiveness of undergraduate academic programmes at faculty level, we chose an X-R chart and the quality characteristic to control was the performance rate of the courses in a given academic year for each programme taught at the FBST As already mentioned, to draw an X-R chart, it was necessary to calculate the CL, UCL and LCL ( Table 3 ).

To achieve the proposed goal, we used Minitab 17 statistical software.

Results and discussion

Bearing in mind the inherent variability of the educational processes, we are able to identify: first, courses that fall below the LCL underperform (Zone B, Figure 1 ), i.e. the percentage of students who passed the course was worryingly low. Second, courses that fall above the UCL (Zone B, Figure 1 ) indicate that the percentage of students who passed the course was extraordinarily high. Finally, courses that fall within the three-sigma limits had more stable performances, meaning that only common causes explain the variability presented between the control limits (Zone A, Figure 1 ).

In this work, several sets of X-R graphs are presented in which the parameters were calculated by programme, by academic year in which the courses were taught and by department. They were completed with a comparative analysis.

Control by academic programme

In Figure 3 , we provide the results of X-bar charts that plot the variability of the performance rate of the courses for the years 2014–2020, for each of the three programmes analysed, considering the courses within the programme grouped by academic year.

The graphs show similar behaviours for the BM and F&A programmes, with a similar under control performance average (60.54% and 59.49%, respectively). In addition, by academic year, the courses with a low performance rate, e.g. below the lower control limits, were concentrated in the first year and those with a high performance rate, e.g. above the LCS, were concentrated in the third year. The performance rates of the second-year courses are considered to be normal, as they were located within the control interval zone.

The performance rate of the courses within the Tourism Degree is 79.93%, which is substantially higher than the BM and F&A ratio. It is worth noting that the distance to the LCL of the out-of-control courses is much greater for the Degree in Tourism than for the BM and the F&A.

Rather than a particular high or low value for the performance rate, the wide range in the proportions observed between degrees and as shown in the graphs, evidences that while F&A show more stability in the proportion of students passing the courses, a worrying lack of stability is a common factor for B&M and Tourism.

This suggests that higher ratings are not necessarily better, and for this reason it is very important to contextualize the analysis. Because the correct application of X chart formulas requires that all performance ratings contribute to determining the three-sigma limits, the previous analysis lacks subjectivity when it categorizes performance ratings into three groups: high, normal and low.

Surprisingly, for the BM and the F&A, the means of the performance ratios calculated using the traditional approach are higher than those estimated applying SPC: 64.67 > 60.54 and 65.14 > 59.49, respectively. By contrast, the traditional mean of the Degree in Tourism (71.39), is lower than the mean calculated under statistic control (75.93) ( Table 2 ).

The differences pointed out above highlight the need to review the procedures for evaluating the results of the degree study programs, promoting methods such as SPC that allow a contextualized, robust and unbiased analysis.

However, it is important to bear in mind that the analysis could change if control chart parameters were calculated using the programme mean, the department mean or the programme year mean. Thus, to complement the SPC study, in the following sections we present the results according to academic year and department.

Control by academic year

Figure 4 shows the results of SPC application conducted independently for each of the academic years that make up each degree, and not in an aggregated form approach as in the previous section.

The results show differences in the control parameters (CL and control interval), depending on the programme itself and the academic year in which the courses were taught. As an example, the central values for the first year and the second year of BM and F&A were very similar (50.55% and 51.84%; 58.81% and 57.98%, respectively). However, these values increased significantly for courses taught in the third year and differed considerably between programmes (BM 79.19%, F&A 67.72%) ( Figure 4 ). Notably, the three programmes present an overall upward trend in the under control performance rate associated with the sequence of the courses within a structured curricular of the program and the wide range in the academic year to year performance ratio observed ( Table 4 ). This finding is consistent with the results of Hanna et al. (2012) who found a similar trend in a study of the proportion of students passing the class analysed by course and using p -charts. Taken together, these findings would suggest the need to identify the root causes of this trend such as the necessity to revise admission standards. Regarding possible reasons behind the high performance ratios values of the courses located in the second and third academic year of the program we suggest three possibilities. First, it is possible that the faculty and the administration were aware of the presence of special reasons behind the variation and responded by improving teaching strategies. Second, it is also possible that those students lacking motivation and ability were filtered out and did not pass to the subsequent academic year, thereby increasing student performance for higher courses. Third, a low performance rate in the first academic year coupled with a high dropout rate would have had the effect of lowering the standards for the high level courses.

Under this independent analysis, the results also show that the courses taught in the second year of BM and F&A were under statistical control as they were in the aggregate analysis, but with a lower average performance rate (58.81 and 57.98% versus 60.54 and 59.49% of BM and F&A, respectively). Surprisingly, for the courses that were out of control in BM and F&A, those of the first year were above the UCL limit, whereas for the third year, they were located below the LCL in clear contrast to what occurred in the aggregate analysis of the programme described in the previous section. In the case of the tourism degree, the courses outside the control interval were mainly concentrated in the last year, unlike the results obtained in the joint analysis, which were concentrated in the other years. In general, a progressive increase in the value of the CL (under control) of the performance rate was observed, to the extent that the courses were taught in higher academic years. This year-by-year contextualized analysis applying SPC demonstrates the usefulness of this approach by revealing hidden problems that affect variability, such as the profile of students who enrol in a degree for the first time or how students adapt to their university studies over time.

Finally, the use of the three-sigma limits also shows important differences with the results of the traditional evaluation (mean of the academic program) ( Table 4 ). This reinforces the argument previously made in the analysis by academic programs with SPC.

Control by department

Bearing in mind that the courses taught in the different degrees were assigned to different departments, in this section, we present the CL for the courses, grouping them by departments with the highest teaching load for each degree. Table 5 shows the results.

These new parameters imply changes with respect to those obtained in the previous applications. For example, courses which in the control graph built by the programme were under control, were out of control with this new parameterisation and vice versa. Figure 5 shows differences of under control means (CL) between the two departments with the highest teaching load allocation for each degree.

As a summary of the results, and for comparative purposes, Figure 6 shows the performance under control of the courses, grouped by program, academic year and department, comparing them in turn with the mean performance rate, which was precisely the one traditionally used to monitor the performance rate.

Conclusions

The results of this study show significant differences in the analysis of the variability of the performance rate between the scenarios described (degree, year and department), depending on the reference parameters (CL, UCL, LCL), focussing the analyses on the process itself rather than the result.

Similarly, in the analysis of the variability of the processes, the SPC approach does not only allow hidden problems to be revealed and their causes to be determined so that corrective actions can be taken to reduce oscillations and therefore improve the stability of the educational processes. Once SPC has been incorporated as an assessment tool to measure the variation relating to academic performance ratios, better information would be available to establish more suitable target goals.

The main conclusion of this study is that the traditional analysis which compares performance ratings with means of the degrees or arbitrary cut-off numbers overlooks the inherent variability of the educative processes. In addition to being useless for comparative purposes, this traditional approach lacks the objectivity and robustness necessary for its application in decision-making, in the accreditation and monitoring processes in which this work is framed.

By contrast, SPC allows a contextualized, robust and unbiased analysis of the variability of quality indicators involved in the accreditation and monitoring processes, providing valuable information for decision making to administrators, teachers and other stakeholders in HE.

This study is a first approach to SPC application for monitoring the control indicators proposed in the ESG [ European Association for Quality Assurance in Higher Education (ENQA), 2015 ]. However, the results are focussed on the analysis of just three programmes in a Spanish University, and thus should be interpreted with caution. In future studies, we will explore the application of multivariable charts and capability analysis, the inclusion of other programmes and time horizons, thereby widening the scope of this study.

Conventional control chart for monitoring the variability of a process

Application of the SPC in the scope of the study

Control results by academic programmes (2014–2020)

Control results by academic year

Results by departments (CL)

Comparative analysis related to performance rate

Evaluation programmes of Spanish University

Control parameters by departments

Agrawal , D.K. and Khan , Q.M. ( 2008 ), “ A quantitative assessment of classroom teaching and learning in engineering education ”, European Journal of Engineering Education , Vol. 33 No. 1 , pp. 85 - 103 .

Aldowaisan , T. and Allahverdi , A. ( 2016 ), “ Continuous improvement in the industrial and management systems engineering programme at Kuwait university ”, European Journal of Engineering Education , Vol. 41 No. 4 , pp. 369 - 379 .

Alzafari , K. and Ursin , J. ( 2019 ), “ Implementation of quality assurance standards in European higher education: does context matter? ”, Quality in Higher Education , Vol. 25 No. 1 , pp. 58 - 75 .

Andreani , M. , Russo , D. , Salini , S. and Turri , M. ( 2020 ), “ Shadows over accreditation in higher education: some quantitative evidence ”, Higher Education , Vol. 79 No. 4 , pp. 691 - 709 .

Bakir , S.T. and McNeal , B. ( 2010 ), “ Monitoring the level of students’ GPAs over time ”, American Journal of Business Education , Vol. 3 No. 6 , pp. 43 - 50 .

Bakir , S.T. , Prater , T. and Kiser , S. ( 2015 ), “ A simple nonparametric quality control chart for monitoring student's GPAs. SOP ”, SOP Transactions on Statistics and Analysis , Vol. 2015 No. 1 , pp. 8 - 16 .

Beshah , B. ( 2012 ), “ Students’ performance evaluation using statistical quality control ”, International Journal of Science and Advanced Technology , Vol. 2 No. 12 , pp. 75 - 79 .

Besterfield , D.H. ( 1995 ), Control de Calidad , 4th ed. , Prentice Hall , Mexico .

Bi , H.H. ( 2018 ), “ A robust interpretation of teaching evaluation ratings ”, Assessment and Evaluation in Higher Education , Vol. 43 No. 1 , pp. 79 - 93 .

Cadden , D. , Driscoll , V. and Thompson , M. ( 2008 ), “ Improving teaching effectiveness through the application of SPC methodology ”, College Teaching Methods and Styles Journal (CTMS) , Vol. 4 No. 11 , pp. 33 - 46 .

Carlucci , D. , Renna , P. , Izzo , C. and Schiuma , G. ( 2019 ), “ Assessing teaching performance in higher education: a framework for continuous improvement ”, Management Decision , Vol. 57 No. 2 , pp. 461 - 479 .

Cervetti , M.J. , Royne , M.B. and Shaffer , J.M. ( 2012 ), “ The use of performance control charts in business schools: a tool for assessing learning outcomes ”, Journal of Education for Business , Vol. 87 No. 4 , pp. 247 - 252 .

Curaj , A. , Matei , L. , Pricopie , R. , Salmi , J. and Scott , P. (Eds). ( 2015 ), The European Higher Education Area: Between Critical Reflections and Future Policies , Springer , Cham, Heidelberg, New York, NY, Dordercht, London .

Daneshmandi , A.A. , Noorossana , R. and Farahbakhsh , K. ( 2020 ), “ Developing statistical process control to monitor the values education process ”, Journal of Quality Engineering and Production Optimization , Vol. 5 No. 1 , pp. 33 - 54 .

Debnath , R.M. and Shankar , R. ( 2014 ), “ Emerging trend of customer satisfaction in academic process ”, The TQM Journal , Vol. 26 No. 1 , pp. 14 - 29 .

Ding , X. , Wardell , D. and Verma , R. ( 2006 ), “ An assessment of statistical process control‐based approaches for charting student evaluation scores ”, Decision Sciences Journal of Innovative Education , Vol. 4 No. 2 , pp. 259 - 272 .

Djauhari , M.A. , Sagadavan , R. and Lee , S.L. ( 2017 ), “ Monitoring the disparity of teaching and learning process variability: a statistical approach ”, International Journal of Productivity and Quality Management , Vol. 21 No. 4 , pp. 532 - 547 .

Edwards , H.P. , Govindaraju , K. and Lai , C.D. ( 2007 ), “ A control chart procedure for monitoring university student grading ”, International Journal of Services Technology and Management , Vol. 8 Nos 4/5 , pp. 344 - 354 .

European Association for Quality Assurance in Higher Education (ENQA) ( 2015 ), “ Standards and guidelines for quality assurance in the European higher education area (ESG) ”, Brussels .

Green , K.W. , Jr , Toms , L. and Stinson , T. ( 2012 ), “ Statistical process control applied within an education services environment ”, Academy of Educational Leadership Journal , Vol. 16 No. 2 , pp. 33 - 46 .

Grygoryev , K. and Karapetrovic , S. ( 2005a ), “ An integrated system for educational performance measurement, modeling and management at the classroom level ”, The TQM Magazine , Vol. 17 No. 2 , pp. 121 - 136 .

Grygoryev , K. and Karapetrovic , S. ( 2005b ), “ Tracking classroom teaching and learning: an SPC application ”, Quality Engineering , Vol. 17 No. 3 , pp. 405 - 418 .

Hanna , M.D. , Raichura , N. and Bernardes , E. ( 2012 ), “ Using statistical process control to enhance student progression ”, Journal of Learning in Higher Education , Vol. 8 No. 2 , pp. 71 - 82 .

Hrynkevych , O.S. ( 2017 ), “ Statistical analysis of higher education quality with use of control charts ”, Advanced Science Letters , Vol. 23 No. 10 , pp. 10070 - 10072 .

Jensen , J.B. and Markland , R.E. ( 1996 ), “ Improving the application of quality conformance tools in service firm ”, Journal of Services Marketing , Vol. 10 No. 1 , pp. 35 - 55 .

Juran , J.M. and Gryna , F.M. ( 1988 ), Juran’s Quality Control Handbook (4th edn) , McGraw-Hill. New York, NY .

Karapetrovic , S. and Rajamani , D. ( 1998 ), “ An approach to the application of statistical quality control techniques in engineering courses ”, Journal of Engineering Education , Vol. 87 No. 3 , pp. 269 - 276 .

Kember , D. and Leung , D. ( 2011 ), “ Disciplinary differences in student ratings of teaching quality ”, Research in Higher Education , Vol. 52 No. 3 , pp. 278 - 299 .

Kinser , K. ( 2014 ), “ Questioning quality assurance ”, New Directions for Higher Education , Vol. 2014 No. 168 , pp. 55 - 67 .

Klasik , D. and Hutt , E.L. ( 2019 ), “ Bobbing for bad apples: accreditation, quantitative performance measures, and the identification of low-performing colleges ”, The Journal of Higher Education , Vol. 90 No. 3 , pp. 427 - 461 .

Kohoutek , J. , Veiga , A. , Rosa , M.J. and Sarrico , C.S. ( 2018 ), “ The European standards and guidelines for quality assurance in the European higher education area in Portugal and the Czech republic: between the worlds of neglect and dead letters? ”, Higher Education Policy , Vol. 31 No. 2 , pp. 201 - 224 .

Lee , Y. and Von Davier , A.A. ( 2013 ), “ Monitoring scale scores over time via quality control charts, model-based approaches, and time series techniques ”, Psychometrika , Vol. 78 No. 3 , pp. 557 - 575 .

MacCarthy , B.L. and Wasusri , T. ( 2002 ), “ A review of non-standard applications of statistical process control (SPC) charts ”, International Journal of Quality and Reliability Management , Vol. 19 No. 3 , pp. 295 - 320 .

Manatos , M.J. and Huisman , J. ( 2020 ), “ The use of the European standards and guidelines by national accreditation agencies and local review panels ”, Quality in Higher Education , Vol. 26 No. 1 , pp. 48 - 65 .

Manguad , B.A. ( 2006 ), “ Using SPC to assess performance in a graduate course of business ”, The American Society of Business and Behavioral Sciences, 13th Annual Meeting , Las Vegas .

Marks , N. and O'Connel , T. ( 2003 ), “ Using statistical control charts to analyse data from students evaluations of teaching ”, Decision Sciences Journal of Innovative Education , Vol. 1 No. 2 , pp. 259 - 272 .

Mazumder , Q.H. ( 2014 ), “ Applying six sigma in higher education quality improvement ”, 21st ASEE Annual Conference and Exposition, Indianapolis, IN , pp. 15 - 18 .

Meijer , R.R. ( 2002 ), “ Outlier detection in high‐stakes certification testing ”, Journal of Educational Measurement , Vol. 39 No. 3 , pp. 219 - 233 .

Milosavljevic , P. , Pavlovic , D. , Rajic , M. , Pavlovic , A. and Fragassa , C. ( 2018 ), “ Implementation of quality tools in higher education process ”, International Journal of Continuing Engineering Education and Life-Long Learning , Vol. 28 No. 1 , pp. 24 - 36 .

Montgomery , D.C. ( 2009 ), Introduction to Statistical Quality Control , 7th ed ., Wiley , New York, NY .

Nascimbeni , F. ( 2015 ), “ The increased complexity of higher education collaboration in times of open education ”, Campus Virtuales , Vol. 3 No. 1 , pp. 102 - 108 .

Nikolaidis , Y. and Dimitriadis , S.G. ( 2014 ), “ On the student evaluation of university courses and faculty members' teaching performance ”, European Journal of Operational Research , Vol. 238 No. 1 , pp. 199 - 207 .

Omar , M.H. ( 2010 ), “ Statistical process control charts for measuring and monitoring temporal consistency of ratings ”, Journal of Educational Measurement , Vol. 47 No. 1 , pp. 18 - 35 .

Peterson , S.J. ( 2015 ), ““ Benchmarking student learning outcomes using Shewhart control charts ”, 51st ASC Annual International Conference Proceedings .

Pierre , C.B. and Mathios , D. ( 1995 ), “ Statistical process control and cooperative learning structures: a data assessment ”, European Journal of Engineering Education , Vol. 20 No. 3 , pp. 377 - 384 .

Red Iberoamericana para la Acreditación de la Calidad de la Educación Superior (RIACES) ( 2007 ), “ Proyecto ‘sistemas de garantía de calidad de las agencias de evaluación’ ”, Propuesta de ANECA .

Savic , M. ( 2006 ), “ P-charts in the quality control of the grading process in the high education ”, Panoeconomicus , Vol. 53 No. 3 , pp. 335 - 347 .

Schafer , W.D. , Coverdale , B.J. , Luxenberg , H. and Jin , Y. ( 2011 ), “ Quality control charts in large-scale assessment programs ”, Practical Assessment, Research and Evaluation , Vol. 16 No. 15 , p. 2 .

Shewhart , W.A. ( 1936 ), Statistical Method from the Viewpoint of Quality Control , Dover Publications , New York, NY .

Sivena , S. and Nikolaidis , Y. ( 2019 ), “ Improving the quality of higher education teaching through the exploitation of student evaluations and the use of control charts ”, Communications in Statistics-Simulation and Computation , pp. 1 - 24 .

Strang , L. , Bélanger , J. , Manville , C. and Meads , C. ( 2016 ), Review of the Research Literature on Defining and Demonstrating Quality Teaching and Impact in Higher Education , Higher Education Academy , New York, NY .

Sulek , J.M. ( 2004 ), “ Statistical quality control in services ”, International Journal of Services Technology and Management , Vol. 5 Nos 5/6 , pp. 522 - 531 .

Suman , G. and Prajapati , D. ( 2018 ), “ Control chart applications in healthcare: a literature review ”, International Journal of Metrology and Quality Engineering , Vol. 9 , p. 5 .

Tomas , C. and Kelo , M. ( 2020 ), “ ESG 2015-2018 ENQA agency reports: thematic analysis ”, ENQA Occasional Paper, 28 .

University Commission for the Regulation of Monitoring and Accreditation (UCRMA) ( 2010 ), “ Protocol for the monitoring and renewal of the accreditation of official university degrees ”, available at: http://deva.aac.es/include/files/universidades/seguimiento/Protocolo_CURSA%20_020710.pdf?v=2018927184158 ( accessed 22 November 2021 ).

Utley , J.S. and May , G. ( 2009 ), “ Monitoring service quality with residuals control charts ”, Managing Service Quality: An International Journal , Vol. 19 No. 2 , pp. 162 - 178 .

Veerkamp , W.J. and Glas , C.A. ( 2000 ), “ Detection of known items in adaptive testing with a statistical quality control method ”, Journal of Educational and Behavioral Statistics , Vol. 25 No. 4 , pp. 373 - 389 , pr9 .

Yang , S. , Cheng , T. , Hung , Y. and Cheng , S. ( 2012 ), “ A new chart for monitoring service process mean ”, Quality and Reliability Engineering International , Vol. 28 No. 4 , pp. 377 - 386 .

Zolkepley , Z. , Djauhari , M.A. and Salleh , R.M. ( 2018 ), “ SPC in service industry: case in teaching and learning process variability monitoring ”, AIP Conference Proceedings 1974 (1) , AIP Publishing LLC , p. 40026 .

Further reading

Agencia Nacional de Evaluación de la Calidad y Acreditación (ANECA) ( 2016 ), “ Support guide for the process of follow-up of official titles of bachelor and master ”, available at: www.aneca.es/eng/content/view/full/13282 ( accessed 2 December 2021 ).

Agencia Nacional de Evaluación de la Calidad y Acreditación (ANECA) ( 2021 ), “ Evaluation programmes ”, available at: www.aneca.es/Programas-de-evaluacion ( accessed 2 December 2021 ).

Minitab Inc ( 2014 ), Minitab Statistical Software Version 17 , Minitab Inc. , State College, PA , available at: www.minitab.com

Acknowledgements

Funding : This study was supported by the University of Huelva (Spain) – Andalusia Consortium (CBUA).

Corresponding author

Related articles, we’re listening — tell us what you think, something didn’t work….

Report bugs here

All feedback is valuable

Please share your general feedback

Join us on our journey

Platform update page.

Visit emeraldpublishing.com/platformupdate to discover the latest news and updates

Questions & More Information

Answers to the most commonly asked questions here

Quality America Inc.

SPC defined

Control Chart Properties

Rational Subgroups

Distributions

Defining Control Limits

Short Run Techniques

Interpretation & Calculations

Attributes Charts

Pareto Charts

X-Bar / Range Charts

X-Bar / Sigma Charts

Individual-X Charts

Moving Average Charts

CuSum Charts

EWMA Charts

Multivariate Control Charts

Histograms, Process Capability

Autocorrelation Charts

Measurement Systems Analysis

Applications

Getting Started with SPC

Key Success Factors for the Implementation of SPC

How to Study Process Capability

Multiple Steam Processes

Analysis of Batch Processes

SPC to Improve Quality, Reduce Cost

Struggling with Independence

SPC Analysis of MLB Home Runs

Use Of SPC To Detect Process Manipulation

Using Data Mining and Knowledge Discovery With SPC

Case Study: SPC to Improve Quality and Reduce Cost

One company that has been able to improve quality and reduce costs through the use of a computer based Statistical Process Control (SPC) software pro duct is Green Bay Packaging Inc. The Green Bay Packaging Inc. plant in Tulsa, Oklahoma, a corrugated shipping container manufacturer, has been using SPC for years. Recently, customers are requesting statistical reports to be forwarded to them in an ongoin g and timely manner. Initially, they responded by using handwritten control charts. This took additional time to perform and was a drain on their resources. It also did not allow them to perform alternate data analysis in a cost effective or timely basis.

G reen Bay Packaging Inc. quickly saw the need for SPC software and chose SPC-PC IV from Quality America, Inc. They found the product to have all the charting tools they needed as well as the flexibility and ease of use their employees demanded. SPC-PC IV i s now used in many divisions of the company. Not only are manufacturing plants utilizing the product, but the corporate quality staff uses SPC-PC IV to analyze monthly reports sent electronically by several divisions.

Green Bay Packaging Inc. then went one step further with SPC-PC IV and utilized its Real Time Gage Module to accept readings directly from gages. Prior to using the Real Time Module, they were recording measurements manually. They realized that this process can lead to a higher possibility of e rrors in reading the measurement from the gage, transcribing the measurements to paper, and entering the data into a computer. Besides reducing the possibility of errors, Green Bay Packaging Inc. main goal in reading measurements in real time was to provi d e immediate analysis and reduce data entry time.

They found that by using SPC-PC IV in real time they are now able to quickly and efficiently analyze the data using different tools. Green Bay Packaging Inc. is able to visualize the data in the form of hist ograms, something that they were unable to do consistently through their manual process. In addition, ability of SPC-PC IV to display their specification limits directly onto a histogram allows them to easily visualize their process in relation to specific ations. Green Bay Packaging also stores assignable causes directly in the software to perform Pareto analysis to determine what areas of their process they need to focus on. This is something that was extremely difficult to do with paper records.

Through t he use of SPC-PC IV, Green Bay Packaging has been able to monitor their process more effectively. This has lead to reductions in process variability, scrap material, and costs.

Learn more about the SPC principles and tools for process improvement in Statistical Process Control Demystified (2011, McGraw-Hill) by Paul Keller , in his online SPC Concepts short course (only $39 ), or his online SPC certification course ( $350 ) or online Green Belt certification course ( $499 ).

Lean Six Sigma Training Certification

• Facebook Instagram Twitter LinkedIn YouTube
• (877) 497-4462

Mastering CUSUM Charts: The Key to Detecting Small Process Shifts

May 30th, 2024

For processes where even slight variations carry major implications, promptly noticing and fixing minor shifts assumes critical priority, and this is where CUSUM charts come in play.

While common charts like Shewhart adequately track many scenarios, they can occasionally miss subtle process transformations.

This highlights the value of CUSUM charts. They provide a strong yet sensitive monitoring method, ensuring stability through small changes’ timely identification.

In this discussion, we explore CUSUM’s advantageous design. We’ll also profile some examples where CUSUM detection saved both quality and expenses.

My aim in sharing is to equip you to access CUSUM’s early warning intelligence.

So whether manufacturing, services, or elsewhere, consider optimizing deviation visibility within your domain. Timely issue illumination paves pathways to increased savings, satisfaction, and success.

Key Highlights

• Understanding the principle: Cumulative sum of deviations from target values
• Advantages over traditional Shewhart charts: Superior sensitivity to small process shifts
• Designing effective CUSUM charts: Choosing optimal parameters like reference value (k) and decision interval (h)
• Interpreting CUSUM signals: Identifying out-of-control conditions and root causes
• Integrating it with other SPC tools: EWMA charts, Shewhart charts, combined approaches
• Best practices and considerations: Training, data quality, process stability
• Limitations and alternatives: Large shift detection, non-normal distributions, specialized charts
• The future of process monitoring: Continuous improvement and evolving methodologies

Introduction to CUSUM Charts

Control charts have long been the go-to tool for quality professionals and engineers alike.

These powerful visualizations enable us to track process performance over time, detect abnormalities, and take corrective actions to maintain stability and consistency.

However, traditional control charts like the Shewhart charts can sometimes fall short when it comes to detecting small shifts in the process mean – and this is where CUSUM (Cumulative Sum) charts step in.

CUSUM charts are a specialized type of control chart designed specifically for monitoring small process shifts.

By cumulating the deviations of individual measurements or subgroup averages from a target value, these charts amplify the signal of even the most subtle process changes, making them an invaluable tool for industries where precision and tight process control are paramount.

Understanding the CUSUM Principle

At the core of CUSUM charts lies a simple yet powerful concept: the cumulative sum of deviations from a target value.

Unlike traditional control charts that analyze each data point independently, CUSUM charts incorporate information from both current and previous observations, effectively amplifying the signal of any process shifts or trends.

Advantages Of Shewhart Charts

While Shewhart charts are excellent at detecting large process shifts, their performance can be limited when it comes to identifying smaller, more gradual changes in the process mean.

This is where CUSUM charts truly excel, offering several key advantages:

Better small shift detection: By accumulating deviations over time, these charts are exquisitely sensitive to small process shifts, often detecting changes as small as 0.5 to 2 sigma from the target mean.

Average run length (ARL): CUSUM charts can be designed to optimize the average run length (ARL), which represents the average number of samples required to detect a specific shift in the process mean.

This allows for tailoring the chart’s sensitivity to match the practical needs of the process being monitored.

Sigma limits: Unlike Shewhart charts, which typically use three-sigma control limits , CUSUM charts employ decision intervals based on the desired ARL performance, providing greater flexibility and customization.

Designing an Effective CUSUM Chart

While the underlying principle is relatively straightforward, designing an effective chart requires careful consideration of several key parameters.

These parameters not only influence the chart’s sensitivity but also determine its ability to strike a balance between detecting genuine process shifts and avoiding false alarms.

Choosing Parameters for Optimal Performance

Two of the most critical parameters in the CUSUM chart design are the reference value (k) and the decision interval (h).

These values must be selected judiciously to ensure the chart’s performance aligns with the desired objectives:

Alpha and beta risks: The reference value (k) and decision interval (h) are often chosen based on the acceptable levels of alpha (false alarm) and beta (missed detection) risks, respectively.

Striking the right balance between these two risks is crucial for effective process monitoring .

Shift detection level: Another key consideration is the magnitude of the process shift you wish to detect quickly.

This shift, typically expressed as a multiple of the standard deviation (sigma), directly influences the selection of the reference value (k).

Vertex angle: In the visual representation of CUSUM charts (the V-Mask approach ), the vertex angle of the V-shaped mask is determined by the chosen reference value (k) and decision interval (h).

This angle impacts the chart’s sensitivity and responsiveness to process shifts.

CUSUM Chart Construction

Once the design parameters have been established, the actual construction of a CUSUM chart can be approached through two primary methods: the tabular (algorithmic) method or the visual V-Mask method.

While both techniques serve the same purpose, each offers its own unique advantages and considerations.

Interpreting CUSUM Chart Signals

Unlike traditional control charts, where points outside the control limits indicate an out-of-control condition, CUSUM charts rely on monitoring the cumulative sums themselves. Specifically:

Out-of-control conditions: If the cumulative sum exceeds the decision interval (h), it signifies that the process has likely shifted and is now operating at a different mean level.

Identifying shift points: By backtracking the cumulative sum plot, it is possible to pinpoint the exact sample or observation where the process shift likely occurred, aiding in root cause analysis and corrective action planning .

Root cause analysis: The ability to identify shift points makes CUSUM charts invaluable for investigating and addressing the underlying causes of process instability, enabling targeted improvement efforts and preventing future occurrences.

CUSUM Charts in Practice

Their versatility and power have made them indispensable tools across a wide range of industries and applications, particularly in scenarios where small process shifts can have significant consequences.

From pharmaceutical manufacturing to chemical processes and quality control in various sectors, CUSUM charts have proven their mettle time and again.

Integrating CUSUM with Other SPC Tools

While CUSUM charts excel at detecting small process shifts, they may not be as effective at identifying larger deviations.

To address this limitation, many practitioners recommend integrating these charts with other statistical process control (SPC) tools, such as:

EWMA charts: Exponentially Weighted Moving Average (EWMA) charts, like CUSUM charts, are designed to detect small process shifts but employ a different weighting mechanism for past data points.

Shewhart charts: Combining CUSUM charts with traditional Shewhart charts can provide a comprehensive monitoring solution, leveraging the strengths of both approaches for detecting small and large process shifts.

Combined approaches: Some advanced monitoring strategies involve the simultaneous use of multiple control chart types, such as CUSUM, EWMA, and Shewhart charts, to maximize the likelihood of detecting any process abnormalities, regardless of their magnitude.

Implementing CUSUM Charts

As with any statistical tool, the successful implementation of CUSUM charts requires careful planning, training, and the right resources.

While the mathematical foundations may seem daunting, modern statistical software and spreadsheet applications have made their construction and interpretation more accessible than ever before.

Case Studies and Examples

To illustrate the practical application of CUSUM charts, let’s explore a few examples:

Pharmaceutical manufacturing: In the pharmaceutical industry, where even minor deviations from established parameters can have severe consequences, CUSUM charts are routinely employed to monitor critical quality attributes, such as active ingredient concentrations and tablet weights, ensuring compliance and patient safety.

Chemical processes: Process industries, like chemical manufacturing, often leverage these charts to monitor key process variables, such as temperatures, pressures, and flow rates, enabling timely adjustments and maintaining optimal operating conditions.

Quality control applications: From automotive parts inspection to semiconductor manufacturing, CUSUM charts have proven invaluable in various quality control applications, helping to identify and address potential issues before they escalate, minimizing waste, and ensuring consistent product quality.

Best Practices and Considerations

While CUSUM charts offer numerous advantages, their effective implementation requires careful consideration of several best practices and potential limitations.

By understanding and addressing these factors, organizations can maximize the benefits of CUSUM charts while mitigating potential risks and challenges.

Limitations and Alternatives

As powerful as these charts are, they do have some inherent limitations that should be acknowledged:

Large shift detection: While excelling at detecting small process shifts, CUSUM charts may not be as responsive to larger deviations in the process mean.

In such cases, traditional Shewhart charts or alternative control chart types may be more appropriate.

Non-normal distributions: CUSUM charts are designed with the assumption of normally distributed data .

For processes with non-normal distributions or skewed data, alternative control chart methods, such as those based on non-parametric techniques, may be more suitable.

Specialized control charts: In certain specialized applications or industries, more tailored control chart techniques may be required to address unique process characteristics or regulatory requirements.

CUSUM charts have proven instrumental in detecting and addressing subtle process performance shifts.

By tracking accumulated deviations from metric targets, these specialized charts provide unmatched sensitivity.

This allows organizations to maintain tight operations control and ensure consistent quality excellence.

As industries continually progress, CUSUM chart roles in monitoring will likely magnify.

The strong emphasis today on insights-driven decisions and constant upgrading highlights detecting even nuanced transformations’ importance for retaining competitiveness advantage.

The concept of these charts also synergize seamlessly with constant improvement philosophies, recognizing perfection as ongoing and room for progress always present.

By fully harnessing CUSUM analysis, enterprises foster proactive tracking, rapid responses, and relentless dedication to excellence in cultures.

Whether a quality veteran, engineer, or specialist, mastering CUSUM techniques empowers illuminating process comprehension to new heights. This activates significant improvements and contributions towards operational excellence pursuits within your organization.

SixSigma.us offers both Live Virtual classes as well as Online Self-Paced training. Most option includes access to the same great Master Black Belt instructors that teach our World Class in-person sessions. Sign-up today!

Virtual Classroom Training Programs Self-Paced Online Training Programs

SixSigma.us Accreditation & Affiliations

Monthly Management Tips

• Be the first one to receive the latest updates and information from 6Sigma
• Get curated resources from industry-experts
• Gain an edge with complete guides and other exclusive materials
• Become a part of one of the largest Six Sigma community
• Unlock your path to become a Six Sigma professional

" * " indicates required fields

Loading metrics

Open Access

Peer-reviewed

Research Article

Leveraging conformal prediction to annotate enzyme function space with limited false positives

Roles Data curation, Investigation, Methodology, Software, Validation, Visualization, Writing – original draft, Writing – review & editing

Affiliation School of Computational Science and Engineering, Georgia Institute of Technology, Atlanta, Georgia, United States of America

Roles Investigation, Methodology, Software, Writing – original draft, Writing – review & editing

Roles Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing

* E-mail: [email protected]

• Kerr Ding, 
• Jiaqi Luo, 

• Published: May 29, 2024
• https://doi.org/10.1371/journal.pcbi.1012135
• Reader Comments

This is an uncorrected proof.

Machine learning (ML) is increasingly being used to guide biological discovery in biomedicine such as prioritizing promising small molecules in drug discovery. In those applications, ML models are used to predict the properties of biological systems, and researchers use these predictions to prioritize candidates as new biological hypotheses for downstream experimental validations. However, when applied to unseen situations, these models can be overconfident and produce a large number of false positives. One solution to address this issue is to quantify the model’s prediction uncertainty and provide a set of hypotheses with a controlled false discovery rate (FDR) pre-specified by researchers. We propose CPEC, an ML framework for FDR-controlled biological discovery. We demonstrate its effectiveness using enzyme function annotation as a case study, simulating the discovery process of identifying the functions of less-characterized enzymes. CPEC integrates a deep learning model with a statistical tool known as conformal prediction, providing accurate and FDR-controlled function predictions for a given protein enzyme. Conformal prediction provides rigorous statistical guarantees to the predictive model and ensures that the expected FDR will not exceed a user-specified level with high probability. Evaluation experiments show that CPEC achieves reliable FDR control, better or comparable prediction performance at a lower FDR than existing methods, and accurate predictions for enzymes under-represented in the training data. We expect CPEC to be a useful tool for biological discovery applications where a high yield rate in validation experiments is desired but the experimental budget is limited.

Author summary

Machine learning (ML) models are increasingly being applied as predictors to generate biological hypotheses and guide biological discovery. However, when applied to unseen situations, ML models can be overconfident and make enormous false positive predictions, resulting in the challenges for researchers to trade-off between high yield rates and limited budgets. One solution is to quantify the model’s prediction uncertainty and generate predictions at a controlled false discovery rate (FDR) pre-specified by researchers. Here, we introduce CPEC, an ML framework designed for FDR-controlled biological discovery. Using enzyme function prediction as a case study, we simulate the process of function discovery for less-characterized enzymes. Leveraging a statistical framework, conformal prediction, CPEC provides rigorous statistical guarantees that the FDR of the model predictions will not surpass a user-specified level with high probability. Our results suggested that CPEC achieved reliable FDR control for enzymes under-represented in the training data. In the broader context of biological discovery applications, CPEC can be applied to generate high-confidence hypotheses and guide researchers to allocate experimental resources to the validation of hypotheses that are more likely to succeed.

Citation: Ding K, Luo J, Luo Y (2024) Leveraging conformal prediction to annotate enzyme function space with limited false positives. PLoS Comput Biol 20(5): e1012135. https://doi.org/10.1371/journal.pcbi.1012135

Editor: Cameron Mura, University of Virginia, UNITED STATES

Received: September 2, 2023; Accepted: May 3, 2024; Published: May 29, 2024

Copyright: © 2024 Ding et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The dataset underlying this article was derived from sources in the public domain. We used the data downloaded from https://github.com/flatironinstitute/DeepFRI . Our code is publicly available at https://github.com/luo-group/CPEC .

Funding: This work is supported in part by the National Institute Of General Medical Sciences of the National Institutes of Health ( https://www.nih.gov/ ) under the award R35GM150890, the 2022 Amazon Research Award ( https://www.amazon.science/research-awards ), and the Seed Grant Program from the NSF AI Institute: Molecule Maker Lab Institute (grant #2019897) at the University of Illinois Urbana-Champaign (UIUC; https://moleculemaker.org/ ). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Machine learning (ML) algorithms have proven to be transformative tools for generating biological hypotheses and uncovering knowledge from large datasets [ 1 , 2 ]. Applications include designing function-enhanced proteins [ 3 , 4 ], searching for novel drug molecules [ 5 ], and optimizing human antibodies against new viral variants [ 6 ]. These discoveries often involve a combination of computation and experimentation, where ML-based predictive models generate biological hypotheses and wet-lab experiments are then used to validate them. This approach is beneficial as it greatly reduces the search space and eliminates candidates that are unlikely to be successful, thus saving time and resources in the discovery process. For example, in drug discovery, ML has become a popular strategy for virtual screening of molecule libraries, where researchers use ML models to predict the properties of molecules, such as binding affinity to a target, and identify the most promising candidates for downstream experimental validation and lead optimization [ 7 ].

To gain new insights into biological systems or make novel discoveries (e.g., designing new drugs), ML algorithms are often used to make predictions for previously unseen data samples. For example, to support the design of new vaccines or therapeutics for COVID-19, ML algorithms need to predict the potential for immune escape of future variants that are composed of mutations that have not yet been seen. Similarly, in drug screening, ML algorithms should be able to predict molecules that are structurally different from those in the training data, which helps scientists avoid re-discovering existing drugs. However, making predictions for samples that are under-represented in the training data is a challenging task in ML. While human experts can assess the success likelihood of generated hypotheses based on their domain knowledge or intuition, this ability is not naturally developed by an ML model and, as a result, the model could be susceptible to pathological failure and only provide overconfident or unreliable predictions. This can have critical implications in ML-assisted biological discovery, as unreliable ML predictions can guide experimental efforts in the wrong direction, wasting resources on validating false positives.

In this work, we aim to develop ML models that can generate hypotheses with limited false positives, providing confident and accurate predictions that can potentially help improve the yield rate in downstream validation experiments. Specifically, we use the function annotation problem of protein enzymes as an example to demonstrate our method. The underlying computational problem of function annotation is a multi-class, multi-label classification problem as a protein can have multiple functions. In computational protein function annotation, a model typically predicts a set of functions that the query protein may potentially have. The set of predicted functions, if validated by experiments, can be incorporated into existing databases to augment our knowledge of the protein function space. There is often a trade-off regarding the size of the prediction set: researchers prefer a set with a small size, containing a handful of very confident predictions, as it is not desirable to spend resources on too many hypotheses that ultimately turn out to be false positives; on the other hand, researchers may be willing to increase the budget to validate a larger set of predictions in order to improve the chance of discovering novel functions for under-studied proteins.

The above tradeoff is often captured by different notions of prediction score cutoff, which decides whether to assign a particular function label to a protein, in existing computational methods for function annotation. For example, when annotating protein functions using sequence-similarity-based tools such as BLAST [ 8 ], a cutoff of the BLAST E-value can be used to determine the significance of sequence match. However, the choice of E-value cutoff is often based on the user’s intuition and good cutoff values on a dataset may not generalize to another dataset. Recent ML methods for enzyme function annotation typically first predict the probability that the input protein has a particular function and annotate the protein with this function if the predicted probability is greater than 0.5 [ 9 – 11 ]. However, using an arbitrary cutoff such as 0.5 is problematic as the predicted probabilities do not always translate to the confidence of the ML model, especially when the model is not well-calibrated (e.g., a predicted function with probability 0.95 may still be an unreliable prediction if the model is overconfident and produces very high probability scores most of the time). Recently, Hie et al. [ 12 ] developed a framework that used the Gaussian process to estimate the confidence or uncertainty in the ML model’s predictions. While the framework was shown to be effective to guide biological discovery, it is unclear how the estimated uncertainty is related to the final false discovery rate (FDR) in experimental validation and how to set a cutoff on the uncertainty scores to achieve a desired FDR. Consequently, it is challenging to provide FDR estimates before the experimental validation, and FDR typically can only be assessed post-validation.

Here, we propose an ML method, called CPEC (Conformal Prediction of EC number), to achieve FDR-controlled enzyme function prediction by leveraging a statistical framework known as conformal prediction (CP) [ 13 ]. CPEC receives the sequence or structure of an enzyme as input and predicts a set of functions (EC numbers) that the enzyme potentially has. The unique strength of CPEC is that the averaged per-protein FDR (i.e., the number of incorrect predictions divided by the prediction set size for a protein) can be controlled by a user-specified hyper-parameter α . The CP framework theoretically guarantees that the FDR of our per-protein predictions is no larger than α with a very high probability. This equips researchers with foresight, offering success rate estimates even before experimental validation. In an ML-guided workflow of protein function discovery, researchers can specify the desired FDR level α based on the experiment budget or expectations. For example, setting α to a smaller value when only the most confident predictions are needed or the test budget is limited, or setting to a larger value when the goal is to discover novel functions and a slightly higher FDR and budget are acceptable. The base ML model of CPEC is PenLight2, an improved version of the deep learning model PenLight [ 14 ] for the multi-class multi-label protein function annotation problem, which uses a graph neural network to integrate 3D protein structure data and protein language model embeddings to learn structure-aware representations for function prediction. Benchmarked on a carefully curated dataset, we first found that CPEC outperformed existing deep learning methods for enzyme function prediction. We also demonstrated that CPEC provides rigorous guarantees of FDR and allows users to trade-off between precision and recall in the predictions by tuning the desired maximum value α of FDR. Additionally, we showed that CPEC consistently provides FDR-controlled predictions for proteins with different sequence identities to the training set, suggesting its robustness even in regimes beyond its training data distribution. Moreover, based on CPEC, we proposed a cascade model that can better balance the resolution and coverage for EC number prediction.

Materials and methods

Problem formulation.

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

(A) CPEC is a machine learning (ML) framework that leverages conformal prediction to control the false discovery rate (FDR) while performing enzyme function predictions. Compared to conventional ML predictions, CPEC allows users to select the desired FDR tolerance α and generates corresponding FDR-controlled prediction sets. Enabled by conformal prediction, CPEC provides a rigorous statistical guarantee such that the FDR of its predictions will not exceed the FDR tolerance α set by the users. The FDR tolerance α offers flexibilities in ML-guided biological discovery: when α is small, CPEC only produces hypotheses for which it has the most confidence; a larger α value would allow CPEC to afford a higher FDR, and CPEC thus can predict a set with more function labels to improve the true positive rate. Abbreviation: Func: function. Incorrect predictions in prediction sets are colored gray. (B) We developed a deep learning model, PenLight2, as the base model of the CPEC framework. The model is a graph neural network that receives the three-dimensional structure and the sequence of a protein as input and generates a function-aware representation for the protein. It employs a contrastive learning scheme to learn a vector representation for proteins, such that the representations of functionally similar proteins in the latent space are pulled together while dissimilar proteins are pushed apart.

https://doi.org/10.1371/journal.pcbi.1012135.g001

Conformal risk control

Overview of conformal risk control..

Conformal risk control guarantee for FDR control.

Calibration algorithm for FDR control.

Given the FDR control guarantee, the natural follow-up question would be how to find a valid parameter λ that can control the risk through the calibration step on calibration data. The Learn then Test (LTT) algorithm [ 22 ], which formulated the selection of λ as a multiple hypotheses testing problem, has been proposed to solve this question. CPEC adopts the LTT algorithm established upon the data distribution assumption that all feature-response pairs ( X , Y ) from the calibration set and the test set are independent and identically distributed (i.i.d.).

Algorithm 1: CPEC for FDR control

/* Calculation of Hoeffding’s inequality p-values { p 1 , …, p N }    */

for i ← 1 N do

  for j ← 1 n c do

while p i ≤ δ and i ≥ 1 do

  i ← i − 1;

Protein function prediction

Ec number prediction dataset..

We applied CPEC on the task of Enzyme Commission (EC) numbers [ 17 ] prediction to demonstrate its effectiveness. EC number is a widely used four-level classification scheme, which organizes the protein according to their functions of catalyzing biochemical reactions. In specific, a protein would be labeled with an EC number if it catalyzes the type of biochemical reactions represented by that EC number. For each four-digit EC number a . b . c . d , the 1st-level a is the most general classification level while the 4th-level d is the most specific one. We used the dataset that contains EC number-labeled protein sequences and structures, provided by Gligorijević et al. [ 10 ]. The protein structures were retrieved from Protein Data Bank (PDB) [ 27 ]. Protein chains were then clustered at 95% sequence identity using the BLASTClust function in the BLAST tool [ 8 ] and then organized into a non-redundant set which only included one labeled high-resolution protein chain from each cluster. The EC number annotations were collected from SIFTS (structure integration with function, taxonomy, and sequence) [ 28 ]. As the 4th-level EC number is the most informative functional label, we only kept proteins that have ground-truth level-4 EC numbers in our experiments. Eventually, the dataset we used has 10, 245 proteins and a train/valid/test ratio of roughly 7: 1: 2. The proteins in the test set have a maximum sequence identity of 95% to the training set. Within the test set, test proteins were further divided into disjoint groups with [0, 30%), [30%, 40%), [40%, 50%), [50%, 70%), and [70%, 95%] sequence identity to the training set. The lower the sequence identity to the training set, the more difficult the test protein would be for ML models to predict its functions. In experiments, we have used the more challenging test data group ([0, 30%)) to evaluate the robustness of our framework.

Contrastive learning-based protein function prediction.

For protein function prediction tasks, supervised learning has long been a popular choice in the deep learning community. Supervised learning-based methods take protein sequences or structures as input and directly map them into class labels. While the idea is simple and efficient, supervised learning has been suffering from a major drawback: its performance could be severely affected by the class imbalances of the training data, an unfortunately common phenomenon in protein function prediction tasks. For example, in the EC number database, some EC classes contain very few proteins (less than ten), while some other EC classes contain more than a hundred proteins. Those classes with more proteins would dominate the training, thereby suppressing the minority classes and degrading the performance of supervised learning. To overcome this challenge, a new paradigm called contrastive learning has become popular in recent years [ 29 ]. Instead of directly outputting class labels, contrastive learning-based models map the training proteins into an embedding space where functionally similar proteins are close to each other and functionally dissimilar pairs are far away. Our previously developed ML methods PenLight and CLEAN [ 14 , 30 ] have demonstrated the effectiveness of contrastive learning in enzyme function predictions. In each iteration of the contrastive learning process, the PenLight or CLEAN model samples a triplet including an anchor protein p 0 , a positive protein p + , and a negative protein p − , such the positive protein pairs ( p 0 , p + ) have similar EC numbers (e.g., under the same subtree in the EC number ontology) while the negative pairs ( p 0 , p − ) have dissimilar EC numbers. The objective of contrastive learning is to learn low-dimensional embeddings x 0 , x + , x − for the protein triplet such that the embedding distance d ( x 0 , x + ) is minimized while d ( x 0 , x − ) is maximized ( Fig 1B and S1 Text ). In the prediction time, the EC number of the training protein with the closest embedding distance to the query protein will be used as the predicted function labels for the query protein.

In this work, we developed PenLight2, an extension of our previous PenLight model [ 14 ] for performing multi-label classification of EC numbers. Similar to PenLight, PenLight2 is a structure-based contrastive learning framework that integrates protein sequence and structure data for predicting protein function. It integrated protein 3D structures and protein language model (ESM-1b [ 31 ]) embeddings into a graph attention network [ 32 ] and optimized the model using the contrastive learning approach, which pulled the embeddings of the (anchor, positive) pair together and the embeddings of the (anchor, negative) pair away. By naturally representing the amino acids as nodes and spatial relations between residues as edges, the graph neural network can extract structural features in addition to sequence features and generate function-aware representations of the protein. In this work, we shifted from the multi-class single-label classification approach used in PenLight [ 14 ] to a multi-class multi-label classification framework, which better aligns with the function annotation data of enzymes in which an enzyme can be labeled with multiple EC numbers. PenLight2 achieved two key improvements compared to PenLight: model training (triplet sampling strategy) and model inference (function transfer scheme and prediction cutoff selection):

1) Triplet sampling strategy. For training efficiency, PenLight takes a multi-class single-label classification approach and randomly samples one EC number for promiscuous enzymes when constructing the triplet in contrastive learning, considering that only less than 10% enzymes in the database used are annotated with more than one EC number. To enhance the effectiveness of contrastive learning for promiscuous enzymes, in this work, we adopt a multi-class multi-label classification approach, in which retain the complete four-level EC number annotations for an enzyme in the triplet sampling of PenLight2 ( Fig 1B ). Specifically, we thus generalized PenLight’s hierarchical sampling scheme to accommodate proteins with multiple functions in PenLight2: in each training epoch, for each anchor protein (every protein in the training set), we randomly choose one of its ground truth EC numbers if it has more than one and then follow original sampling scheme in PenLight for the sampling of the positive and the negative proteins ( S1 Text ). A filter is applied to ensure that the anchor and the negative do not share EC numbers.

3) Prediction cutoff selection. In contrast to the original PenLight model that only predicted the top-1 EC number for a query protein, PenLight2 implemented an adaptive method to achieve multi-label EC prediction. Following the max-separation method proposed in our previous study [ 30 ], we sorted the distances between the query protein and all EC clusters and identified the max difference between adjacent distances. PenLight2 then uses the position with the max separation as the cutoff point and outputs all EC numbers before this point as final predictions. This cutoff selection method aligns with the multi-label nature of the task.

With these improvements, we extended the original PenLight from the single-label classification to the multi-label setting. We denote this improved version as PenLight2.

We performed multiple experiments to evaluate CPEC’s prediction accuracy and ability of FDR control. We further evaluated CPEC using test data that have low sequence identities to the training data to demonstrate its utility for generating hypotheses (function annotations) for novel protein sequences.

CPEC achieves accurate enzyme function predictions

We first evaluated the prediction performance of PenLight2, the base ML model in CPEC, for predicting function annotations (EC numbers) of protein enzymes. The purpose of this experiment was to assess the baseline prediction accuracy of CPEC when the FDR control is not applied. We compared CPEC with three state-of-the-art deep learning methods capable of reliably predicting enzyme function on the fourth level of EC hierarchy, including two CNN-based (convolutional neural networks) methods DeepEC [ 9 ] and ProteInfer [ 11 ] that take protein sequence data as input and one GNN-based (graph neural networks) method DeepFRI [ 10 ] that takes both protein sequence and structure data as input. All these three methods applied the multi-class classification paradigm for function prediction: first predicting a score between 0 and 1 as the predicted probability that the input enzyme has a particular EC number and then generating all EC numbers with predicted probability greater than 0.5 (except for DeepFRI which used 0.1 as cutoff) as the final predicted function annotations for the input enzyme. We evaluated all methods using metrics F1 score, which assesses prediction accuracy considering both precision and recall, and the normalized discounted cumulative gain (nDCG) [ 33 ], which rewards higher rankings of true positives over false negatives in the prediction set ( S1 Text ). On a more challenging test set (test proteins with [0, 30%) sequence identity to training proteins), we further evaluated all methods by drawing the micro-averaged precision-recall curves.

The evaluation results showed that our method outperformed all the three state-of-the-art methods in terms of both F1 score and nDCG ( Fig 2A ). For example, PenLight2 achieved a significant improvement of 34% and 26% for F1 and nDCG, respectively, over the second-best baseline DeepFRI. The pronounced performance gaps between PenLight2 and other baselines also suggested the effectiveness of the contrastive learning strategy used in PenLight2. The major reason is that contrastive learning utilized the structure of the function space (the hierarchical tree of the EC number classification system) to learn protein embeddings that reflect function similarity, while the multi-class classification strategy used in the three baselines just treated all EC numbers as a flat list of labels and may only capture sequence/structure similarity but not function similarity. In addition, we observed that methods that incorporated protein structure data (PenLight2 and DeepFRI) achieved better than methods that only use sequence data as input (DeepEC and ProteInfer), suggesting that protein structure may describe features related to functions more explicitly and is useful for predicting protein function. Those results demonstrated that the design choices of PenLight2, including the contrastive learning strategy and representation learning of protein structure, greatly improve the accuracy of protein function prediction. To further analyze PenLight2’s prediction performance, we delineated its F1 score into precision and recall and observed that PenLight2 has slightly lower precision than other methods but substantially higher recall and F1 score ( S1 Fig ). We noted that other baseline methods such as ProteInfer achieved the high precision score at a cost of low coverage ( Fig 2A ), meaning that they did not predict any functions for a large number of query proteins due to their high uncertainties in those proteins. Additionally, we evaluated PenLight despite that it only performs single-label prediction, and we found that PenLight and PenLight2 had similar performances. As the fraction of promiscuous enzymes is low in the test set, we expected PenLight2 to be a more accurate predictor than PenLight in future enzyme function prediction tasks when promiscuous enzymes prevail.

(A) We evaluated DeepEC [ 9 ], ProteInfer [ 11 ], DeepFRI [ 10 ], and PenLight2 for predicting the 4th-level EC number, using F1 score, the normalized discounted cumulative gain (nDCG), and coverage as the metrics. Specifically, coverage is defined as the proportion of test proteins for which a method has made at least one EC number prediction. (B) We further evaluated all methods for predicting the 4th-level EC number on more challenging test proteins with [0, 30%) sequence identities to the training proteins and drew the micro-averaged precision-recall curves. For each curve, we labeled the point with the maximum F1 score (Fmax).

https://doi.org/10.1371/journal.pcbi.1012135.g002

On a more challenging test set which only includes test proteins with [0, 30%) sequence identities to training proteins, we also observed that PenLight2 robustly predicted the EC numbers of the test proteins and outperformed all baseline methods ( Fig 2B and S3 Fig ). The improvement of the micro-averaged Fmax value from the best baseline method ProteInfer to PenLight2 was 32%. In the high recall region, PenLight2 achieved a higher precision value than any of the baseline methods. The results here were consistent with the results on the entire test set, which further proved the effectiveness of PenLight2 for EC number prediction.

CPEC provides FDR control for EC number prediction

After validating its prediction performance, we integrated PenLight2 as the base model into the conformal prediction framework. Conformal prediction provides a flexible, data-driven way to find an optimal cutoff for PenLigth2 to decide whether to predict a function label for the input protein, such that the FDR on the test data is lower than the user-specified FDR upper bound α . Here, we performed experiments to investigate whether CPEC achieves the desired FDRs and how its prediction performance would change when varying α . For comparison, we compared CPEC to several other thresholding strategies for generating the prediction set, including 1) max-separation ( Methods ); 2) top-1, where only the EC number with the closest embedding distance to the input protein is predicted as output; and 3) σ -threshold, where all EC numbers with an embedding distance smaller than μ + 2 σ to the input protein are predicted as output, where μ and σ are the mean and standard deviation of a positive control set that contains the distances between all true protein-EC number pairs. Platt scaling [ 34 ], a parametric calibration method, was further included as a thresholding strategy for comparison. We also included our baseline DeepFRI, which outputs EC numbers if it predicts that the probability of the input having this EC number is greater than a cutoff of 0.1. The purpose of the experiment here is not to show CPEC can outperform all other methods under all metrics but to show that CPEC can achieve a desired tradeoff by tuning the interpretable parameter α and simultaneously provide a rigorous statistical guarantee on its FDR. In an evaluation experiment, we have further compared CPEC with two point-uncertainty prediction methods (Monte Carlo dropout [ 35 ] and RED [ 36 ]), demonstrating that CPEC provides precise FDR control prior to validation, whereas MC dropout and RED can only evaluate FDR post-validation ( S1 Text ).

Reliable FDR controls.

In theory, the conformal prediction framework guarantees that the actual FDR of the base ML model on the test data is smaller than the pre-specified FDR level α with high probability. We first investigated how well this property holds on our function prediction task. We varied the value of α from 0 to 1, with increments of 0.1, and measured CPEC’s averaged per-protein FDR on the test data. As expected, we observed that the actual FDR of CPEC ( Fig 3A , blue line) was strictly below the specified FDR upper bound α ( Fig 3A , diagonal line) across different α values. This result suggested that CPEC successfully achieved reliable FDRs as guaranteed by the conformal prediction. We have further performed an evaluation experiment to investigate the impact of the calibration set sizes on CPEC’s FDR control, and the results suggested that the FDR control performances of CPEC were robust to various calibration set sizes ( S1 Text ).

For FDR tolerance α from 0.1 to 0.9 with increments 0.1, we evaluated how well CPEC controls the FDR for EC number prediction. Observed FDR risks, precision averaged over samples, recall averaged over samples, F1 score averaged over samples, and nDCG were reported for each FDR tolerance on test proteins in (A-E). The black dotted line in (A) represents the theoretical upper bound of FDR over test proteins. Three thresholding strategies were assessed over PenLight2 as a comparison to CPEC, which includes 1) max-separation [ 30 ], 2) top-1, and 3) σ -threshold. The results of CPEC were averaged over five different seeds. DeepFRI was also included for comparison.

https://doi.org/10.1371/journal.pcbi.1012135.g003

Tradeoff between precision and recall with controlled FDR.

Varying the FDR parameter α allowed us to trade-off between the prediction precision and recall of CPEC ( Fig 3B and 3C ). When α was small, CPEC predicted function labels for which it has the most confidence, in order to achieve a lower FDR, resulting in high precision scores (e.g., precision 0.9 when α = 0.1). When CPEC was allowed to tolerate a relatively larger FDR α , it predicted more potential function labels for the input protein at the FDR level it can afford, which resulted in an increasing recall score as α was increasing. Similarly, the nDCG score of CPEC was also increasing with α ( Fig 3E ), indicating that CPEC not only retrieved more true function labels but also ranked the true labels at the top of its prediction list.

Interpretable cutoff for guiding discovery.

CPEC is able to compute an adaptive cutoff internally based on the user-specified FDR parameter α for deciding whether or not to assign a function label to the input protein. This allows researchers to prioritize or balance precision, recall, and FDR, depending on test budget or experiment expectations, in an ML-guided biological discovery process. In contrast, many existing methods that use a constant cutoff often have optimized performance in one metric but suffer in another. For example, in our experiment, DeepFRI and Platt scaling threshold strategy had the highest precisions but their recalls were the lowest among all methods; the σ -threshold strategy had a recall of 0.94 yet its FDR (0.75) was substantially higher than others ( Fig 3A–3C and S2 Fig ). Although some methods such as DeepFRI may achieve a better tradeoff between precision and recall by varying its probability cutoff from 0.1 to other values, they lack a rigorous statistical guarantee on the effect of varying the cutoff values. For example, if the cutoff of DeepFRI was raised from 0.1 to 0.9, one can expect that, qualitatively, it would lead to a higher precision but also a higher FDR. However, it is hard to quantitatively interpret the consequence of raising the cutoff to 0.9 (e.g., how high would the FDR be) until the model is evaluated using ground-truth labels, which are often unavailable before experimental validation in the process of biological discovery. In contrast, with CPEC, researchers are also able to balance the interplay between the prediction precision and recall by tuning the interpretable parameter α and assured that the resulting FDR will not be greater than α .

Overall, through these experiments, we validated that CPEC can achieve the statistical guarantee of FDR. We further evaluated the effect of varying the FDR tolerance α on CPEC’s prediction performances. Compared to conventional strategies for multi-label protein function prediction, CPEC provides a flexible and statistically rigorous way to better tradeoff precision and recall, which can be used to better guide exploration and exploitation in biological discovery with a controlled FDR.

Adaptive prediction of EC numbers for proteins with different sequence identities to the training set

The risk in our conformal prediction framework is defined as the global average of per-protein FDRs, which may raise the concern that the overall FDR control achieved by CPEC on the test set was mainly contributed by FDR controls on those proteins that are easy to characterize and predict, and it is possible that the model suffered from pathological failures and did not give accurate FDR controls on proteins that are hard to predict. To this point, we defined the prediction difficulty based on the level of sequence identity between test proteins and training proteins, following the intuition that it is more challenging for an ML model to predict the functions of a protein if the protein does not have homologous sequences in the training data. We first performed a stratified evaluation to analyze CPEC’s FDR-control performance at different levels of prediction difficulty. After examining the consistency of the FDR control across different difficulties, we explored an adaptive strategy for predicting EC numbers, which allows the ML model not to predict a too specific EC number than what the evidence supported and only predict at the most confident level of EC hierarchy.

Consistency of FDR control.

We first confirmed CPEC’s FDR-control ability across different levels of prediction difficulty. Specifically, we partitioned the test set into disjoint groups based on the following ranges of sequence identity to the training set: [0, 30%), [30%, 40%), [40%, 50%), [50%, 70%) and [70%, 95%]. We varied the values of α from 0.05 to 0.5 with increments of 0.05. For each level of FDR tolerance α , we examined the FDR within each group of test proteins. As shown in Fig 4A , CPEC achieved consistent FDR controls across different levels of train-test sequence identity and different values of α , where the observed FDR were all below the pre-specified FDR tolerance α . Even for the most difficult group of test proteins that only have [0, 30%) sequence identity to the training proteins, CPEC still achieved an FDR of 0.03 when tolerance α = 0.1. This is because a well-trained ML model would have low confidence when encountering difficult inputs, and CPEC would abstain from making predictions if the model’s confidence does not exceed the decision threshold. The results of this experiment built upon the conclusion of the previous subsection and validated that CPEC can not only control the FDR of the entire test set but also the FDR for each group of test proteins with different levels of prediction difficulty. We have performed an evaluation experiment to further assess CPEC’s FDR control on test proteins that do not belong to the same CATH superfamilies [ 37 ] as any of the training proteins. We found that CPEC provided effective FDR control for these test proteins from unseen superfamilies ( S5 – S7 Figs and S1 Text ), suggesting that CPEC can offer effective FDR-controlled EC number predictions even for test proteins that are very dissimilar to its training proteins.

(A) We reported the observed FDR for test proteins with different sequence identities to the training set (i.e. different difficulty levels) for FDR tolerance α from 0.05 to 0.5 with increments of 0.05. Test proteins were divided into disjoint groups with [0, 30%), [30%, 40%), [40%, 50%), [50%, 70%), and [70%, 95%] sequence identity to the training set. The smaller the sequence identity, the harder the protein would be for machine learning models to predict function labels. (B) We designed the procedure to first predict the EC number at the 4th level. If the model was uncertain at this level and did not make any predictions, we would move to the 3rd level to make more confident conformal predictions instead of continuing with the 4th level with high risks. We used the same FDR tolerance of α = 0.2 for both levels of CPEC prediction. For proteins with different sequence identities to the training data, we reported the hit rate of our proposed procedure. The hit rate on the 4th level, the hit rate on the 3rd level, the percentage of proteins with incorrect predictions on both levels, and the percentage of not called proteins for both levels were reported. The results were calculated as an average over 5 different seeds of splitting the calibration set.

https://doi.org/10.1371/journal.pcbi.1012135.g004

An adaptive strategy for EC number prediction.

The EC number hierarchy assigns four-digit numbers to enzymes, where the 4th-level label describes the most specific functions of enzymes whereas the 1st-level label describes the most general functions. In EC number prediction, ideally, a predictive model should not predict a too specific EC number than what the evidence supported. In other words, if a model is only confident about its prediction up to the 3rd level of an EC number for a protein, it should not output an arbitrary prediction at the 4th level. We first trained two CPEC models, where the first model, denoted as CPEC4, predicts EC numbers at the 4th level as regular, and the other, denoted as CPEC3, predicts the 3rd-level EC numbers. We then combine the two models as a cascade model: given an input protein and a desired value of α , we first apply the CPEC4 to predict the 4th-level EC numbers for the input protein with an FDR at most α . If CPEC4 outputs any 4th-level EC numbers, they will be used as the fine-level annotations for the input; if CPEC4 predicts nothing due to the FDR tolerance α being too stringent, we apply CPEC3 on the same input to predict EC numbers at the 3rd-level. If CPEC3 outputs any 3rd-level EC numbers, they will be used as the coarse-level annotations for the input; otherwise, the cascade model just predicts nothing for this input. The motivation of this adaptive prediction strategy is that even though 3rd-level EC numbers are less informative than 4th-level ones, it might be more useful for researchers in certain circumstances to acquire confident 3rd-level EC numbers than only obtaining a prediction set with a large number of false positive EC numbers at the 4th level.

To validate the feasibility of the above adaptive model, we evaluated CPEC3 and CPEC4 using the same FDR tolerance α = 0.2 on our test set. We reported the hit rate, defined as the fraction proteins for which our model predicted at least one correct EC number, for both the 4th-level and the 3rd-level EC numbers. We found that this adaptive prediction model, compared to the model that only predicts at the 4th level, greatly reduced the number of proteins for which the model made incorrect predictions or did not make predictions ( Fig 4B ). For example, on the test group with sequence identity [0, 30%) to the training data, around 60% proteins were correctly annotated with at least one EC number, while only 40% proteins were correctly annotated if the adaptive strategy was not used. This experiment demonstrated the applicability of CPEC for balancing the prediction resolution and coverage in protein function annotation.

Application: EC number annotation for low-sequence-identity proteins

Conformal prediction quantifies the ML model’s uncertainty in its predictions, especially for the predictions for previously unseen data. This is extremely useful in ML-guided biological discovery as we often need to make predictions for unseen data to gain novel discoveries. For example, in protein function annotation, the most challenging proteins to annotate are those previously uncharacterized or do not have sequence homologous in current databases. Conventional ML models that do not quantify prediction uncertainties are often overconfident when making predictions for the aforementioned challenging samples, leading to a large number of false positives in their predictions, which can incur a high cost in experimental validation without yielding a high true positive rate. Considering the importance of predicting previously uncharacterized data, here we designed an evaluation experiment to assess CPEC’s prediction performance on these challenging proteins. We created a test set that contains only proteins that have less than 30% sequence identity to any proteins in the training set, which simulated a challenging application scenario.

We varied the FDR tolerance α from 0.05 to 0.5 and counted the number of correct predictions, where assigning one EC number to a protein was counted as one prediction. We observed that CPEC had an effective uncertainty quantification for its predictions on this low-sequence-identity test set ( Fig 5A ). For example, when α = 0.05 which forced the model to only output the most confident predictions, CPEC was highly accurate, with a precision of nearly 0.97. At the FDR tolerance level of 0.1, CPEC was able to retrieve 25% (180/777) true protein-EC number pairs at a precision higher than 0.9. Keeping increasing the value of α allowed CPEC to make more correct predictions, without significantly sacrificing precision. For instance, at the level α = 0.5, CPEC successfully predicted 70% true protein-EC number pairs while maintaining a reasonable precision of 0.6 and an nDCG of 0.7. As a comparison, the baseline method DeepFRI correctly predicted 309 protein-EC pairs, out of the total 777 true pairs, with a precision of 0.89 and an nDCG score of 0.50, which roughly corresponds to CPEC’s performance at α = 0.2.

(A) CPEC was evaluated on difficult test proteins ([0, 30%) sequence identity to the training data). For FDR tolerance from 0.05 to 0.5, the total number of correct predictions, precision averaged over samples, and the normalized discounted cumulative gain was reported under five different seeds for splitting calibration data. Note that the upper bound of correct predictions, i.e. the ground truth labels, is 777. As a comparison, DeepFRI successfully made 307 predictions, with a sample-averaged precision of 0.8911 and an nDCG score of 0.5023. (B) An example of the prediction sets generated by CPEC for Gag-Pol polyprotein (UniProt ID: P04584; PDB ID: 3F9K), along with the prediction set from DeepFRI. CPEC used the chain A of the PDB structure as input. The prediction sets were generated under FDR tolerance α = 0.25, 0.3, 0.35. The sequence of this protein has [0, 30%) sequence identity to the training set and, therefore, can be viewed as a challenging sample. Incorrect EC number predictions are colored gray. (C) Boxplots showing the FDR@1st-hit metric, defined as the smallest FDR tolerance α at which CPEC made the first correct prediction for each protein. The evaluation was performed on five groups of test proteins, stratified based on their sequence identities to the training set.

https://doi.org/10.1371/journal.pcbi.1012135.g005

We again note that CPEC is more flexible than methods such as DeepFRI in that it provides an interpretable and principled way to tradeoff between precision and recall, which allows researchers to not only prioritize high-confidence predictions but also increase prediction coverage for improving the yield rate of true positives. To illustrate this, we visualized the prediction results of CPEC and DeepFRI in Fig 5B . We selected a protein that has multiple EC number annotations (UniProt ID: P04584). Using its default setting, DeepFRI predicted four labels for this protein, among which three were correct. For CPEC, we gradually increased the value of α and see how the prediction set was changing. Interestingly, we observed that CPEC gradually predicted more true EC numbers as α was increasing while maintaining a low FDR. In particular, when α = 0.25, CPEC outputted two EC numbers, both of which were correct predictions; when α was relaxed to 0.3, CPEC predicted one more EC number, which turned out to be also correct; when we further relaxed the FDR tolerance α to 0.35, CPEC predicted six EC numbers for the protein, and four of them were correct. This example illustrated CPEC’s utility in practice: researchers have the flexibility when using CPEC to guide experiments, where a small value of α prioritizes accurate and confident hypotheses, and a large value of α promotes a high yield of true positives while ensuring the number of false positives to be limited.

Having observed that CPEC was able to recover more true function labels as we were relaxing the FDR tolerance α , we asked one important question—at which value of α can CPEC output the first correct function label (“hit”) for the input protein. We referred to this α value as FDR@1st-hit. This metric can be viewed as a proxy of the experiment cost researchers need to pay before they obtain the first validated hypothesis. We computed the FDR@1st-hit value for all test proteins in each of the five disjoint groups partitioned by their sequence identity to training sequences ( Fig 5C ). We found that for the majority of the test sequences (the four groups out of five with sequence identity at least 30% to training sequences), CPEC was able to reach the first hit at an FDR lower than 0.15. For the most difficult group where all proteins share [0, 30%) sequence identity to training data, the median FDR@1st-hit was 0.3. This observation was consistent with our intuition and expectation, as low-sequence-identity proteins are more difficult for the ML model to predict, thus requiring a larger hypotheses space to include at least one true positive. Overall, CPEC achieved a reasonable FDR@1st-hit for function annotation, meaning that it produced a limited number of false positives before recovering at least one true positive, which is a highly desired advantage in ML-guided biological discovery.

Machine learning models play a vital role in generating biological hypotheses for downstream experimental analyses and facilitating biological discoveries in various applications. A significant challenge in the process of ML-assisted biological discoveries is the development of ML models with interpretable uncertainty quantification of predictions. When applied to unseen situations, ML models without uncertainty quantification are susceptible to overconfident predictions, which misdirects experimental efforts and resources to the validation of false positive hypotheses. Addressing this challenge becomes essential to ensure the efficiency and reliability of ML-assisted biological discovery.

In this work, we have presented CPEC, an ML framework that enables FDR-controlled ML-assisted biological discoveries. Leveraging the conformal prediction framework, CPEC allows users to specify their desired FDR tolerance α , tailored to the experiment budget or goals and makes corresponding predictions with a controlled FDR. We demonstrate CPEC’s effectiveness using enzyme function annotation as a case study, simulating the discovery process of identifying the functions of less-characterized enzymes. PenLight2, an improved version of PenLight optimized for multi-label classification is utilized as CPEC’s base ML model. Specifically, CPEC takes the sequence and structure of an enzyme as input and outputs a set of functions (EC numbers) that the enzyme potentially has. The conformal prediction algorithm in CPEC theoretically guarantees that the FDR of the predicted set of functions will not exceed α with high probability. The evaluation of CPEC on the EC number prediction task showed that CPEC provides reliable FDR control and has comparable or better prediction accuracy than baseline methods at a much lower FDR. Interpretable cutoffs were provided by CPEC for guiding the EC number annotations of proteins. Furthermore, CPEC demonstrated its robustness in making FDR-controlled predictions even for proteins with low sequence identity to its training set.

Quantifying uncertainties of ML model predictions is a key desideratum in ML-guided biological discovery. Although a few prior studies have investigated the uncertainty quantification of ML models [ 12 , 38 ], their uncertainty estimates are only indicative of prediction errors but do not translate to error-controlled predictions. In contrast, CPEC enables researchers to specify a maximum level of error rate and produces a set of predictions whose error rate is guaranteed to be lower than the specified level. Additionally, CPEC stands out by providing risk estimates, which delivers insights into the potential outcomes even before experimental validation and aids in the strategic allocation of experimental resources. One limitation of the CPEC framework is that when under covariate shift (i.e., P calib ( X ) ≠ P test ( X )), the data assumption of CPEC that the data in the calibration and test sets are i.i.d. is violated, which might lead to suboptimal FDR control performances ( S6 and S7 Figs). Although weighted conformal prediction frameworks have been proposed to address this limitation [ 39 ], the quantification and control of non-monotonic risk functions (e.g., FDR) under covariate shift remained a challenging problem. In this work, we define the error rate as the false discovery rate (FDR) to reflect the practical consideration in experiments where the goal is to maximize the success rate of hypothesis validation given a limited test budget. Nevertheless, the CPEC framework can be extended to support other forms of error rates, such as false negative rate [ 13 ]. In addition to protein function annotation, we expect CPEC to be a valuable tool for researchers in other biological discovery applications particularly when a balance between the experimental budget and the high yield rate is desired, such as drug target identification [ 40 ], material discovery [ 41 ], and virtual molecule screening [ 38 ].

Supporting information

S1 text. supplementary information..

Additional methodology, detailed experiment descriptions, and further evaluation experiments are included in the file.

https://doi.org/10.1371/journal.pcbi.1012135.s001

S1 Fig. Performance evaluation of representative baseline methods for EC number prediction.

We evaluated DeepEC, ProteInfer, DeepFRI, and CPEC (PenLight2) for predicting the 4th level EC number, using sample-averaged precision and recall as the metrics. DeepEC and DeepFRI were evaluated using the only trained model provided in their repositories, whereas ProteInfer was assessed using 5 different trained models. DeepFRI was trained on the same dataset as PenLight2 while DeepEC and ProteInfer were trained by their respective datasets. PenLight2 was trained using 5 different seeds.

https://doi.org/10.1371/journal.pcbi.1012135.s002

S2 Fig. CPEC achieves FDR control for EC number prediction.

Platt scaling [ 34 ], RED [ 36 ], and Monte Carlo dropout [ 35 ] were further evaluated as thresholding strategies, in comparison to CPEC. Due to the requirements of the methods, RED and MC dropout were applied on top of an MLP model. The results of CPEC and all of the thresholding strategies were averaged over five different seeds.

https://doi.org/10.1371/journal.pcbi.1012135.s003

S3 Fig. Performance evaluation of representative baseline methods for EC number prediction on test proteins with [0, 30%) sequence identities to training proteins.

We evaluated DeepEC, ProteInfer, DeepFRI, and CPEC (PenLight2) for predicting the 4th level EC number, using sample-averaged precision, recall, F1 score, nDCG, and coverage as the metrics. DeepEC and DeepFRI were evaluated using the only trained model provided in their repositories, whereas ProteInfer was assessed using 5 different trained models. DeepFRI was trained on the same dataset as PenLight2 while DeepEC and ProteInfer were trained by their respective datasets. PenLight2 was trained using 5 different seeds.

https://doi.org/10.1371/journal.pcbi.1012135.s004

S4 Fig. The FDR control of CPEC with different calibration set sizes.

The performances of CPEC’s FDR control were evaluated using calibration sets with various sizes (abbrev: calib. set size): 20%, 10%, 5%, and 1% of the total number of the training data. The same training data was used across all calibration set sizes to ensure consistency in the comparison. The black dotted line in the first panel represents the theoretical upper bound of FDR over test proteins. The results were averaged over five different seeds.

https://doi.org/10.1371/journal.pcbi.1012135.s005

S5 Fig. Performance evaluation of representative baseline methods for EC number prediction on test proteins that do not belong to the same CATH [ 37 ] superfamilies as the training proteins.

CPEC and three baseline methods (DeepEC, ProteInfer, and DeepFRI) were evaluated for predicting the 4th level EC number, using sample-averaged precision, recall, F1 score, nDCG, and coverage as the metrics. DeepEC and DeepFRI were evaluated using the only trained model provided in their repositories, whereas ProteInfer was assessed using 5 different trained models. DeepFRI was trained on the same dataset as CPEC, while DeepEC and ProteInfer were trained using their respective datasets. Training proteins not labeled in the CATH database were only removed from the training dataset of CPEC but not from the baseline methods’ training sets, which gave potential advantages to baseline methods. CPEC was trained using 5 different seeds.

https://doi.org/10.1371/journal.pcbi.1012135.s006

S6 Fig. Application of the FDR control for the EC number prediction of out-of-distribution (OOD) proteins.

The FDR control of CPEC was evaluated on a more challenging data split: no training and test proteins belong to the same CATH [ 37 ] superfamily. Training proteins not labeled in the CATH database were only removed from the training dataset of CPEC but not from the baseline methods’ training sets, which gave potential advantages to baseline methods. The results were averaged over five different seeds.

https://doi.org/10.1371/journal.pcbi.1012135.s007

S7 Fig. Application of the FDR control for the EC number prediction of out-of-distribution (OOD) proteins.

The FDR control of CPEC was evaluated on a more challenging data split: no training and test proteins belong to the same CATH [ 37 ] superfamily. A total number of 200 test proteins were sampled from the test set, and proteins that belong to the same superfamilies as the sampled test proteins were removed from the training set of CPEC. Training proteins not labeled in the CATH database were only removed from the training dataset of CPEC but not from the baseline methods’ training sets, which gave potential advantages to baseline methods. The results were averaged over five different seeds.

https://doi.org/10.1371/journal.pcbi.1012135.s008

S8 Fig. PenLight2, the base ML model of CPEC, outperforms the state-of-the-art methods for EC number prediction.

CPEC (PenLight2) and four baseline methods (including a baseline MLP model that takes the ESM-1b protein embeddings as the input) were evaluated for predicting the 4th-level EC number on more challenging test proteins with [0, 30%) sequence identities to the training proteins and the micro-averaged precision-recall curves were drawn. For each curve, the point with the maximum F1 score (Fmax) was labeled.

https://doi.org/10.1371/journal.pcbi.1012135.s009

S9 Fig. Evaluation of two point-uncertainty prediction approaches.

Two point-uncertainty prediction methods (Monte Carlo dropout (MC dropout) [ 35 ] and RED [ 36 ]) were evaluated in terms of uncertainty quantification. To make a fair comparison, a multi-layer perception taking ESM-1b protein embedding as the input was selected as the base ML model. The percentiles of the prediction variance (10th, 20th, 30th,…, and 100th percentiles) on the test set were used as the cutoffs. Predictions with variances larger than the cutoff were dropped. Observed false discovery rate (FDR), precision, recall, and coverage were used as metrics. The results were averaged over five different seeds.

https://doi.org/10.1371/journal.pcbi.1012135.s010

Acknowledgments

This work used the Delta GPU Supercomputer at NCSA of UIUC through allocation CIS230097 from the Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS) program ( https://access-ci.org/ ), which is supported by NSF grants #2138259, #2138286, #2138307, #2137603, and #2138296. The authors acknowledge the computational resources provided by Microsoft Azure through the Cloud Hub program at GaTech IDEaS ( https://research.gatech.edu/energy/ideas ) and the Microsoft Accelerate Foundation Models Research (AFMR) program ( https://www.microsoft.com/en-us/research/collaboration/accelerating-foundation-models-research/ ).

• View Article
• PubMed/NCBI
• Google Scholar
• 13. Angelopoulos AN, Bates S. A gentle introduction to conformal prediction and distribution-free uncertainty quantification. arXiv preprint arXiv:210707511. 2021;.
• 14. Luo J, Luo Y. Contrastive learning of protein representations with graph neural networks for structural and functional annotations. In: PACIFIC SYMPOSIUM ON BIOCOMPUTING 2023: Kohala Coast, Hawaii, USA, 3–7 January 2023. World Scientific; 2022. p. 109–120.
• 18. Vovk V, Gammerman A, Shafer G. Algorithmic learning in a random world. Springer Science & Business Media; 2005.
• 19. Papadopoulos H, Proedrou K, Vovk V, Gammerman A. Inductive confidence machines for regression. In: European Conference on Machine Learning. Springer; 2002. p. 345–356.
• 21. Angelopoulos AN, Bates S, Fisch A, Lei L, Schuster T. Conformal risk control. arXiv preprint arXiv:220802814. 2022;.
• 22. Angelopoulos AN, Bates S, Candès EJ, Jordan MI, Lei L. Learn then test: Calibrating predictive algorithms to achieve risk control. arXiv preprint arXiv:211001052. 2021;.
• 23. Vovk V, Gammerman A, Saunders C. Machine-learning applications of algorithmic randomness. 1999;.
• 25. Hoeffding W. Probability inequalities for sums of bounded random variables. In: The collected works of Wassily Hoeffding. Springer; 1994. p. 409–426.
• 32. Brody S, Alon U, Yahav E. How attentive are graph attention networks? arXiv preprint arXiv:210514491. 2021;.
• 33. Wang Y, Wang L, Li Y, He D, Liu TY. A theoretical analysis of NDCG type ranking measures. In: Conference on learning theory. PMLR; 2013. p. 25–54.
• 35. Gal Y, Ghahramani Z. Dropout as a bayesian approximation: Representing model uncertainty in deep learning. In: international conference on machine learning. PMLR; 2016. p. 1050–1059.
• 36. Qiu X, Miikkulainen R. Detecting misclassification errors in neural networks with a gaussian process model. In: Proceedings of the AAAI Conference on Artificial Intelligence. vol. 36; 2022. p. 8017–8027.

• Open access
• Published: 29 May 2024

Using 2% PVPI topical solution for serial intravitreous injections and ocular surface findings: a case control study

• José Henrique Casemiro 1 ,
• Ana Paula Miyagusko Taba Oguido 2 &
• Antonio Marcelo Barbante Casella 3  

International Journal of Retina and Vitreous volume  10 , Article number:  41 ( 2024 ) Cite this article

27 Accesses

Metrics details

The use of povidone-iodine for ocular surface asepsis is widespread for intravitreal injections. They became frequent procedures, leading to serial exposure of patients’ eyes to iodinated solutions. In this study, we investigate the changes in the ocular surface in patients submitted to repeated use of povidine for intravitreal injection of anti-VEGF asepsis, analyzing Ocular Surface Disease Index, non-invasive break up time, blinking quality, lipid layer, meniscus height and osmolarity.

This case-control study included 34 individuals (68 eyes), 14 males, 20 females aged 48 to 94. Inclusion criteria were individuals who received application of 2% povidone-iodine eyedrops for intravitreal injections treatment with the non-treated contralateral eye used as control. Ocular surface examinations were performed at a single occasion. A pre-intravitreal injection asepsis protocol with povidone-iodine was applied. All statistical analysis was performed using the STATA® 18.0 Software and a p-value = 0.05 was considered as the statistical significance value in all tests.

The median number of IVIs in treated eyes was 12 (range 6–20). The results in treated eyes compared with untreated eyes were respectively : median OSDI 16 (IQR 6–39) and 12.5 (IQR 8–39) ( p  = 0.380); mean NIBUT 10.30 (SD ± 2.62) and 10.78 (SD ± 2.92) ( s, p  = 0.476); median blinking quality 100 (IQR 100) and 100 (IQR 100 ) (%, p  = 0.188); median lipid layer 87 (IQR 77–90) and 86 (IQR 74–100) (nm, p  = 0.451); median meniscus height 0.22 (IQR 0.19-0,31) and 0.24 (IQR 0.20–0.27) (mm, p  = 0.862), median Meibomian gland atrophy 33 (IQR 24–45) and 31.5 (IQR 25–39) (%, p  = 0.524); and mean osmolarity 306.6 (SD ± 21.13) and 313.8 (SD ± 29) (mOsm, p  = 0.297). There was no statistically significant relationship between the repetitive use of 2% iodinated solution and signs or symptoms compatible with dry eye syndrome in this group of patients.

Conclusions

The findings suggest that 2% povidone iodine is a safe and efficacious agent for ocular surface antisepsis during intravitreal injections, not leading to substantial ocular surface modifications. This conclusion supports the continued use of povidone iodine in routine ophthalmic procedures without increased risk of inducing dry eye syndrome.

The use of povidone iodine (PVPI) for ocular surface asepsis is widespread, both for surgical procedures and intravitreal injections [ 1 , 2 , 3 , 4 , 5 ]. Surgeries for cataracts, glaucoma, and intravitreal injections have become common and frequent procedures in ophthalmology, leading to the serial exposure of patients’ eyes to iodine solutions [ 5 , 6 , 7 , 8 , 9 , 10 ]. These changes are directly associated with dry eye syndrome [ 11 , 12 ].

In particular, studies have demonstrated that intravitreal injections used to treat diabetic macular edema or age-related macular degeneration result in significant changes in the ocular surface, leading to dry eye syndrome and damage to homeostasis of the ocular surface [ 5 , 11 , 13 , 14 ].

Dry eye syndrome is a multifactorial disease of the ocular surface characterized by the loss of tear film homeostasis, hyperosmolarity, inflammation, damage and neurosensory abnormalities [ 11 , 15 , 16 , 17 , 18 ]. Its etiology is variable, ranging from nonspecific inflammation of the ocular surface to direct chemical or physical aggression, infections, and autoimmune diseases [ 11 , 12 , 15 , 19 ].

In addition to the most common symptoms, burning sensation, itching, speck, eye redness, excess tearing reflex, brightness sensitivity, and quality of vision loss are also frequent findings that affect efficiency at work and the quality of life of patients [ 11 , 12 , 15 , 19 , 20 ].

This study aimed to observe changes in the ocular surface and tear film due the serial use of 2% PVPI, the gold standard drug for asepsis of the ocular surface. As it is well known that pre-injection antisepsis of the ocular surface with PVPI has a toxic effect on the corneal epithelium, the aim is to identify changes in the tear film and ocular surface and avoid serious problem like dry eye syndrome [ 11 , 12 , 21 , 22 ].

A case-control study was conducted at the Ophthalmology and Psicology Clinic (APMTO MD) in Londrina, Paraná. The patients were recruited from the Retina and Vitreous Institute (AMBC MD) in Londrina, Paraná. The study included 34 individuals (68 eyes). 14 males, 20 females, aged 48 to 94 years. All participants signed the informed consent form, which allowed their participation in the study. Inclusion criteria were individuals who received application of 2% PVPI eyedrops for anti-VEGF IVIs treatment with the contralateral eye used as control, that had not been treated with any topical medication during the same period of applications and good comprehension of the Ocular Questionnaire Surface Disease Index (OSDI). Exclusion criteria were patients who could not understand the OSDI questionnaire; patients using antidepressant medicine, diuretics, sympathomimetics, eye drops for glaucoma, or eye lubricants; people with allergies to iodine; unfavorable clinical conditions to undergo the examination procedures for the study; inappropriate test quantity and quality; unsatisfactory images or unsatisfactory and inadequate data.

The study was approved by the Ethics and Research Committee Involving Human Beings of the State University of Londrina by N. 5.300.176.

The individuals underwent directed clinical and ophthalmological analysis, received explanations about the study, used their data, and signed consent forms. All clinical measures were performed using the IDRA equipment (SBSSISTEMI, Orbasano, Torino, Italy), at which time the OSDI questionnaire was also applied and tear osmolarity was collected using the I-PEN ® (I-MED PHARMA INC. Dollard-des-Ormeaux, QC, Canada). All examinations and administration of the questionnaire were performed by the same professional. No drops or medications that could cause changes in any subsequent measurements were used.

The variables analyzed were age, sex, date of the last PVPI application, number of PVPI applications, OSDI questionnaire, tear osmolarity, NBUT, tear film interferometry, tear meniscus height, percentage of meibomian gland loss, and blink quality. The sequence of procedures obeyed the following order: Explanation to the subject regarding the exams and questionnaire to which he would be submitted, guidance to the patient not to identify in any way the eye being treated and the eye not treated during data collection, nor during the questionnaire OSDI; patient positioning in the IDRA® equipment; capture of blinking quality video images; capture of tear film interferometry; capture of images to measure the height of the tear meniscus and immediate measurement; capture of tear film (NBUT); image capture for the percentage of meibomian gland loss by everting the lower eyelid with a cotton swab; positioning the patient outside the IDRA equipment; application of the I-PEN® electrode to capture tear osmolarity in the lower conjunctiva, first in the right eye, and subsequently in the left eye; application of the OSDI questionnaire.

All statistical analyses were performed using STATA® 18.0 Software and p-values ≤ 0.05 indicated statistical significance.

The Shapiro-Wilk test was used to verify data normality. Data that did not follow a normal distribution were analyzed using the Wilcoxon rank-sum test and were described as means and as medians and interquartile ranges. Data that showed normality were analyzed using the Student’s T test and presented as means and standard deviations. Descriptive, quantitative, and multivariate analyses compared treated (case) and untreated (control) eyes.

The average number of IVIs in treated eyes was 12 (range 6–20). The results in treated eyes compared with untreated eyes were respectively: median OSDI 16 (IQR 6–39) and 12.5 (IQR 8–39) ( p  = 0.380); mean NIBUT 10.30 (SD ± 2,62) and 10.78 (SD ± 2.92) ( s, p  = 0.476); median blinking quality 100 (IQR 100) and 100 (IQR 100 ) (%, p  = 0.188); median lipid layer 87 (IQR 77–90) and 86 (IQR 74–100) (nm, p  = 0.451); median meniscus height 0,22 (IQR 0.19–0.31) and 0.24 (IQR 0.20–0.27) (mm, p  = 0.862), median Meibomian gland athrophy 33 (IQR 24–45) and 31.5 (IQR 25–39) (%, p  = 0.524); and mean osmolarity 306.6 (SD ± 21.13) and 313.8 (SD ± 29) (mOsm, p  = 0.297).). The results revealed that the use of 2% PVPI did not affect the analyzed variables in a statistically significant way. All data is summarized on Table  1 .

These results are disposable on fig 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 and 9 as annexed.

Histogram showing the days of last application of IVIS ( intravitreal injections ) in treated eyes and the density showing the proportion of eyes in each period of time

Histogram showing the number of application ov IVIS ( intravitreal injections ) in treated eyes and the density showing the proportion of eyes in each amount of number of applications

Blue box plot showing score OSDI ( Ocular Surface Disease Index ) in treated eyes comparing with pink box plot showing score OSDI in fellow eyes

Blue box plot showing NIBUT ( non invasive break up time ) in seconds in treated eyes comparing with pink box plot showing NIBUT in seconds in fellow eyes

Blue box plot showing blink quality in treated eyes comparing with pink box plot showing blink quality in fellow eyes

Blue box plot showing lipid layer in treated eyes comparing with pink box plot showing lipid layer in fellow eyes

Blue box plot showing meniscus height in milimeters in treated eyes comparing with pink box plot showing meniscus height in milimeters in fellow eyes

Blue box plot showing Meibomian gland loss in treated eyes comparing with pink box plot showing Meibomian gland loss in fellow eyes

Blue box plot showing tear osmolarity in miliosmoles in treated eyes comparing with pink box plot showing osmolarity in miliosmoles in fellow eyes

Through multivariate analysis, we obtained some interesting outcomes as follows:

When controlling for NIBUT, meibomian gland atrophy, number of applications, and days of the last application according to treatment, sex was an important variable in explaining the variability in the OSDI score (coef = 15.63 | p-value = 0.003). On average, controlling for the abovementioned variables, being female contributed to an increase in the OSDI to 15.63 points.

After controlling for meniscus height and age according to treatment, tear osmolarity contributed significantly to variability in the lipid layer (coef = -0.266, p  = 0.004). In this sense, the addition of one unit in tear osmolarity led to a -0.266 drop in the lipid layer.

After controlling for meniscus height, OSDI, days since the last application, age, and sex according to the treatment, these factors contributed significantly to the variability in the lipid layer [(coef = 0.562 | p-value = 0.004) (coef = − 5.622 | p-value = 0.048)]. In this sense, the addition of one year of age led to a decrease of -0.562 on average. For the same treatment group, female sex led to a decrease of -5,622.

Age, lipid layer, meniscus height, sex according to treatment, age according to treatment, and sex were important factors for explaining the variability in tear osmolarity.

We noticed that a greater age correlated with lower tear osmolarity. However, being in the treated group reduced the decrease in tear osmolarity with advancing age.

Being female implied higher tear osmolarity. However, the increase in tear osmolarity was smaller in the treated group.

A greater height of the lipid layer and meniscus correlated with lower tear osmolarity.

The present study showed that the use of topical PVPI at 2% did not cause significant damage to the ocular surface when the findings of the ocular surface and tear film analyses were used.

Our results contradict some existing data indicating the toxicity of long-term iodine use on the ocular surface; we found two statistically relevant results that the application of iodine may improve the stability of the tear film in the elderly and women, since the eyes in older individuals and female patients that received iodine showed a smaller increase in tear osmolarity [ 4 , 14 , 23 , 24 ].

Moreover, the results of this study corroborated some hypotheses that the use of PVPI could be positive in some dry eye disease diagnostic features, such as the improvement of the tear film meniscus height and the decrease of the tear film osmolarity [ 25 , 26 ].

A localized anti-inflammatory surface effect of the anti-VEGF agent used in intravitreous injections should be considered and assessed in further studies [ 22 , 25 ].

The literature review also shows that there was an improvement in the tear function of some patients who used iodine in ocular asepsis [ 25 , 26 , 27 ], perhaps due to an antimicrobial action preventing the proliferation of bacterial flora that could produce harmful enzymes or cause meibomitis and blepharitis [ 25 , 26 , 28 ].

The cell regeneration mechanism might have satisfactorily recomposed the ocular surface or the tear homeostasis might have compensates for the damage caused by iodine in the cells in question; furthermore, these are just hypotheses.

We also determined that the risk factors for dry eye disease, age and female sex [ 10 , 16 , 29 , 30 ], were associated with the observed clinical data: greater ages lower the height of the tear meniscus, the greater the tear osmolarity, and the smaller the lipid layer of the tear film. The female sex was also associated with higher OSDI scores and fewer tear film lipid layers.

Regardless of the cause or consequence, the osmolarity and lipid layer of the tear film were inversely proportional.

Through multivariate analysis, we determined that the risk factors for dry eye syndrome, age, and female sex correlated with worse results in the tear meniscus measurement tests, OSDI questionnaire, and tear film interferometry, corroborating the literature implicating them as risk factors for dry eye disease [ 20 , 30 , 31 ].

Moreover, due to the sample size, false negatives, or simply because in practice, iodine in the amount and frequency used does not lead to histological damage that may reflect functional changes. The results did not discourage the use of iodine for ocular asepsis but also did not indicate its use for protocols with higher concentrations or more applications than those used in current protocols.

The strengths of the study are as follows: the same patient was the control and treated group, avoiding any environmental or medical bias. The number of injections administered was higher than that reported in other studies. No drops were used during the examination to avoid artificial changes to the tear film.

The limitations of this study were as follows: the study had a small sample size of 34 patients, resulting in 68 eyes being analyzed, which may have caused an analysis bias when using these data in the general population. We must remember that the analyzed population was from southern Brazil and had mostly descended from Italian, German, Spanish, and Portuguese immigrants; therefore, these data may only reflect the specific epidemiology of this population. The meibomian glands analyzed were located in the inferior tarsus.

The use of iodine on the ocular surface was not significantly associated with any of the evaluated parameters. There were no statistically significant correlations between the tests applied to the case eyes. The current study indicates that the application of 2% topical povidone-iodine (PVPI) does not inflict significant damage to the ocular surface, as evidenced by the analyses of the ocular surface and tear film. Notable strengths of this study include the use of the same patient as both the control and the treated subject, which minimizes potential biases from environmental or medical factors. Additionally, the absence of any artificial agents during the examination ensures that the tear film remains unaltered.

Contrary to previous concerns regarding the long-term toxicity of iodine on the ocular surface, our findings suggest potential benefits of iodine application in stabilizing the tear film, particularly in older individuals and female patients. This is supported by a smaller increase in tear osmolarity in these groups following iodine application. Furthermore, the study corroborates hypotheses that PVPI may positively affect certain Dry Eye Disease diagnostic features, such as improved tear film meniscus height and reduced tear film osmolarity.

Data availability

No datasets were generated or analysed during the current study.

Abbreviations

Povidine or polyvinylpyrrolidone-iodine

• Intravitreal injections

Vascular endothelial growth factor

Ocular Surface Disease Index

Non invasive break up time

Blink quality

Lipid layer

Standard deviation

Interquartile range

Oakley C, Allen P, Hooshmand J, Vote BJT. PAIN AND ANTISEPSIS AFTER OCULAR ADMINISTRATION OF POVIDONE-IODINE VERSUS CHLORHEXIDINE. Retina. 2018;38:2064–6.

Article   CAS   PubMed   Google Scholar  

Fukuda M, Shibata S, Shibata N, Fujita N, Miyashita H, Tanimura N, et al. Polyvinyl alcohol-iodine induced corneal epithelial injury in vivo and its protection by topical rebamipide treatment. PLoS ONE. 2018;13:e0208198.

Article   PubMed   PubMed Central   Google Scholar  

Grzybowski A, Brona P. Povidone-iodine is still a premium antiseptic measure in ocular surgery. Acta Ophthalmol. 2017;95.

Craig JP, Nichols KK, Akpek EK, Caffery B, Dua HS, Joo C-K, et al. TFOS DEWS II definition and classification report. Ocul Surf. 2017;15:276–83.

Article   PubMed   Google Scholar  

Craig JP, Nelson JD, Azar DT, Belmonte C, Bron AJ, Chauhan SK, et al. TFOS DEWS II Report Executive Summary. Ocul Surf. 2017;15:802–12.

Moss SE. Prevalence of and risk factors for Dry Eye Syndrome. Arch Ophthalmol. 2000;118:1264.

Jiang X, Wang Y, Lv H, Liu Y, Zhang M, Li X. Efficacy of intra-meibomian gland injection of the anti-VEGF agent bevacizumab for the treatment of meibomian gland dysfunction with lid-margin vascularity. Drug Des Devel Ther. 2018;12:1269–79.

Article   CAS   PubMed   PubMed Central   Google Scholar  

Kim S, Ahn Y, Lee Y, Kim H. Toxicity of povidone-iodine to the ocular surface of rabbits. BMC Ophthalmol. 2020;20:359.

Malmin A, Thomseth VM, Førland PT, Khan AZ, Hetland HB, Chen X, et al. Associations between serial Intravitreal injections and Dry Eye. Ophthalmology. 2023;130:509–15.

Audelan T, Legrand M, M’Garrech M, Best A-L, Barreau E, Labetoulle M, et al. Vieillissement de la surface oculaire : physiopathologie et conséquences pratiques pour la prise en charge. J Fr Ophtalmol. 2018;41:262–70.

Kanclerz P, Myers WG. Potential substitutes for povidone-iodine in ocular surgery. Eye. 2021;35:2657–9.

Reibaldi M, Avitabile T, Bandello F, Longo A, Bonfiglio V, Russo A, et al. The effectiveness of 0.6% povidone Iodine Eye drops in reducing the Conjunctival bacterial load and needle contamination in patients undergoing Anti-VEGF intravitreal injection: a prospective, randomized study. J Clin Med. 2019;8:1031.

Han SB. Prevalence of Dry Eye Disease in an Elderly Korean Population. Arch Ophthalmol. 2011;129:633.

Silas MR, Schroeder RM, Thomson RB, Myers WG. Optimizing the antisepsis protocol: effectiveness of 3 povidone–iodine 1.0% applications versus a single application of povidone–iodine 5.0%. J Cataract Refract Surg. 2017;43:400–4.

Zhang X, Wang L, Zheng Y, Deng L, Huang X. Prevalence of dry eye disease in the elderly. Medicine. 2020;99:e22234.

Donnenfeld ED, Holland EJ, Solomon KD, Fiore J, Gobbo A, Prince J, et al. A Multicenter Randomized Controlled Fellow Eye Trial of pulse-dosed Difluprednate 0.05% Versus Prednisolone acetate 1% in cataract surgery. Am J Ophthalmol. 2011;152:609–e6171.

Grzybowski A, Kanclerz P, Myers WG. The use of povidone–iodine in ophthalmology. Curr Opin Ophthalmol. 2018;29:19–32.

Bramantyo T, Roeslani RD, Andriansjah A, Sitorus RS. The efficacy of 1% Chloramphenicol Eye Ointment Versus 2.5% povidone-iodine Ophthalmic Solution in reducing bacterial colony in Newborn Conjunctivae. Asia-Pacific J Ophthalmol. 2015;4:180–3.

Article   CAS   Google Scholar  

Bourcier T, Acosta MC, Borderie V, Borra´s F, Gallar J, Bury T, et al. Decreased corneal sensitivity in patients with Dry Eye. Invest Opthalmology Visual Sci. 2005;46:2341.

Article   Google Scholar  

Niederkorn JY, Kaplan HJ. Rationale for Immune Response and the Eye. Immune Response and the Eye. Basel: KARGER; 2007. pp. 1–3.

Google Scholar  

Saedon H, Nosek J, Phillips J, Narendran N, Yang YC. Ocular surface effects of repeated application of povidone iodine in patients receiving frequent intravitreal injections. Cutan Ocul Toxicol. 2017;36:343–6.

Beni´tez-del-Castillo JM, Acosta MC, Wassfi MA, Di´az-Valle D, Gegu´ndez JA, Fernandez C et al. Relation between Corneal Innervation with Confocal Microscopy and Corneal Sensitivity with Noncontact Esthesiometry in Patients with Dry Eye. Investigative Opthalmology & Visual Science. 2007;48:173.

Yu K, Asbell PA, Shtein RM, Ying G-S. Dry Eye subtypes in the Dry Eye Assessment and Management (DREAM) study: a latent Profile Analysis. Transl Vis Sci Technol. 2022;11:13.

Dohlman TH, Lertsuwanroj B, D’Amico DJ, Ciralsky JB, Kiss S. Evaluation of signs and symptoms of ocular surface disease after intravitreal injection. Acta Ophthalmol. 2019;97.

Zoukhri D. Effect of inflammation on lacrimal gland function. Exp Eye Res. 2006;82:885–98.

Shelanski HASM. PVP-iodine: history, toxicity, and therapeutic uses. J Int Coll Surg. 1956;25:727–34.

CAS   PubMed   Google Scholar  

Kemeny-Beke A, Szodoray P. Ocular manifestations of rheumatic diseases. Int Ophthalmol. 2020;40:503–10.

Zamora JL. Chemical and microbiologic characteristics and toxicity of povidone-iodine solutions. Am J Surg. 1986;151:400–6.

Mantelli F, Lambiase A, Sacchetti M, Orlandi V, Rosa A, Casella P, et al. Cocaine snorting may induce ocular surface damage through corneal sensitivity impairment. Graefe’s Archive Clin Experimental Ophthalmol. 2015;253:765–72.

Bever GJ, Brodie FL, Hwang DG. Corneal Injury from Presurgical Chlorhexidine Skin Preparation. World Neurosurg. 2016;96:610.e1-610.e4.

Wolffsohn JS, Arita R, Chalmers R, Djalilian A, Dogru M, Dumbleton K, et al. TFOS DEWS II Diagnostic Methodology report. Ocul Surf. 2017;15:539–74.

Download references

Acknowledgements

APMO provided IDRA analysis, and was a contributor to design the study, revised, written and approved the manuscript. AMBC Applied intravitreal injections and provided patients for the study and was a contributor to design the study, revised, written and approved the manuscript.

No grant Money was received for this resesearch.

Author information

Authors and affiliations.

Postgraduate Program in Health Sciences, State University of Londrina, UEL, Londrina, Brazil

José Henrique Casemiro

Department of Health Sciences, Surgical Clinic, State University of Londrina, UEL, Londrina, Brazil

Ana Paula Miyagusko Taba Oguido

Londrina State University, Avenida Robert Koch, 60, Londrina, Paraná, 86038-440, Brazil

Antonio Marcelo Barbante Casella

You can also search for this author in PubMed   Google Scholar

Contributions

JHC analyzed and interpreted patient data, reviewed the literature and was a major contributor to the acquisition of data, interviewed the patients, sponsored supplies, designed the study and written the manuscript. APMO provided IDRA analysis, and was a contributor to design the study, revised, written and approved the manuscript. AMBC Applied intravitreal injections and provided patients for the study and was a contributor to design the study, revised, written and approved the manuscript.

Corresponding author

Correspondence to Antonio Marcelo Barbante Casella .

Ethics declarations

Ethics approval and consent to participate and consent for publication.

Approval for the anonymous analisys of Imaging data and medical records was obtained from the Institutional Review Board.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Casemiro, J.H., Oguido, A.P.M.T. & Casella, A.M.B. Using 2% PVPI topical solution for serial intravitreous injections and ocular surface findings: a case control study. Int J Retin Vitr 10 , 41 (2024). https://doi.org/10.1186/s40942-024-00557-1

Download citation

Received : 11 March 2024

Accepted : 07 May 2024

Published : 29 May 2024

DOI : https://doi.org/10.1186/s40942-024-00557-1

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Dry eye disease
• Povidone iodine
• Ocular surface disease

International Journal of Retina and Vitreous

ISSN: 2056-9920

• Submission enquiries: Access here and click Contact Us
• General enquiries: [email protected]

• Open access
• Published: 30 May 2024

Genetic association of Interleukin-17A polymorphism in Bangladeshi patients with breast and cervical cancer: a case-control study with functional analysis

• Md. Abdul Aziz 1 , 2 , 3 ,
• Subrina Chowdhury 1 ,
• Sarah Jafrin 1 , 2 ,
• Md Abdul Barek 1 , 2 ,
• Mohammad Sarowar Uddin 1 , 2 ,
• Md. Shalahuddin Millat 1 , 2 &
• Mohammad Safiqul Islam 1 , 2 , 3  

BMC Cancer volume  24 , Article number:  660 ( 2024 ) Cite this article

17 Accesses

Metrics details

Breast and cervical cancer are the two leading cancers in terms of incidence and mortality. Previous studies reported different interleukins, including interleukin-17A ( IL-17A ) to be responsible for the development and progression of these malignancies. Therefore, we speculated that the variants in this gene might be associated with these cancer developments in Bangladeshi population. For evaluating the hypothesis, we investigated the association of IL-17A rs3748067 polymorphism with the susceptibility of both breast and cervical cancer.

This case-control study was performed on 156 breast cancer patients, 156 cervical cancer patients, and 156 controls using the tetra-primer amplification refractory mutation system-polymerase chain reaction. The statistical software package SPSS (version 25.0) was applied for analyses. The genetic association was measured by the odds ratio (OR) and 95% confidence intervals (CIs). A statistically significant association was considered when p -value ≤ 0.05. Functional analysis was performed using GEPIA and UALCAN databases.

From the calculation of the association of IL-17A rs3748067 with breast cancer, it is found that no genotype or allele showed a statistically significant association ( p >0.05). On the other hand, the analysis of IL-17A rs3748067 with cervical cancer demonstrated that CT genotype showed a significant association (CT vs. CC: OR=1.79, p =0.021). In the overdominant model, CT genotype also revealed a statistically significant association with cervical cancer, which is found to be statistically significant (OR=1.84, p =0.015).

Our study summarizes that rs3748067 polymorphism in the IL-17A gene may be associated with cervical cancer but not breast cancer in Bangladeshi patients. However, we suggest studies in the future with a larger sample size.

Peer Review reports

Breast cancer became the leading cancer worldwide in 2020, with a reported 2.3 million new cases representing 11.7% of total cancer incidence. In terms of mortality, breast cancer is the fifth leading causes of mortality globally, with approximately 685,000 deaths [ 1 , 2 , 3 ]. Cervical cancer, in contrast, is the fourth most diagnosed malignancy and also the fourth major cause of mortality in females. In 2020 alone, about 604,000 new cervical cancer cases and 342,000 deaths were reported. Moreover, cervical cancer was found to be one of the top three cancers that affect females under the age of 45 in 146 countries, which accounts for 79% in 185 countries assessed [ 2 , 4 ].

Patients’ age, reproductive and hormonal factors (first birth or menarche at early age, fewer children, less breastfeeding, menopause at later age, menopausal hormone therapy, and oral contraceptives), personal or family history, genetic predisposition, environmental factors, and lifestyle factors (alcohol consumption, excessive body weight, and physical inactivity) have been correlated with an elevated risk for the development and progression of breast cancer [ 5 , 6 , 7 ]. Again, risk factors of cervical malignancy include both behavioral (sexual activity and lifestyle factors) and certain infectious (human papillomavirus) contributors [ 8 ]. Other risk factors are age at the first full-term pregnancy, diet, family history, immunosuppression, immune deficiency, oral contraceptives, parity, and smoking [ 9 , 10 , 11 ].

Interleukin-17 A (IL-17A) is one of the most intensively investigated interleukins from the IL-17 family which play a critical function in cancer development, progression, and control [ 12 , 13 ]. It is found in the human chromosome 6.12.2 and encodes a 155 amino acid containing protein (consisting of signal peptide with 23 amino acids and a mature peptide with 132 amino acids) [ 14 ]. In carcinogenesis, IL-17A has been reported to engage myeloid-derived suppressor cells (MDSCs) that repress anti-tumor activity [ 15 , 16 ]. IL-17A could also stimulate unnecessary tumor growth by influencing IL-6, which in turn activates tumorigenic signal transducer and activator of transcription (STAT3) signaling pathway and over-express genes associated with pro-survival and pro-angiogenesis [ 17 ].

Numerous studies have reported a higher expression of IL-17A in tumor cells, including breast cancer, colorectal carcinoma, gastric carcinoma, hepatocellular carcinoma, ovarian cancer, medulloblastoma, pancreatic cancer, non-small-cell lung cancer, and thyroid cancer [ 18 , 19 ]. Polymorphisms in the IL-17A gene have been investigated over time to find the possible association with cancers. A major single nucleotide polymorphism (SNP) in the IL-17A gene is rs3748067 which is found on the 3’-untranslated regions (UTR) in chromosome 6 location 52,190,541. The association of r3748067 polymorphism with various cancers has been extensively evaluated in the last decade that includes breast cancer [ 20 ], cervical cancer [ 21 , 22 , 23 , 24 , 25 , 26 ], colorectal cancer [ 27 , 28 ], gastric cancer [ 29 , 30 ], lung cancer [ 31 ], and others.

Although previous studies have evaluated the correlation of IL-17A gene rs3748067 polymorphism with the susceptibility of breast and cervical cancers, the results were incosistent. Besides, no study has been performed in Bangladeshi breast and cervical cancer patients to evaluate the association of rs3748067 polymorphism. Therefore, we conducted the present case-control study to analyze the association of the common SNP in the IL-17 A (rs3748067) gene with the susceptibleness of breast and cervical cancer.

Study settings

The reporting of the present retrospective case-control analysis conforms to the latest STROBE guidelines designed for case-control studies [ 32 ]. In this study, we recruited two groups of patients: one group with breast cancer and another group with cervical cancer. Both groups consisted of 156 patients, each of whom was appointed randomly from the National Institute of Cancer Research and Hospital (NICRH) during the period from July 2019 to June 2020. Again, for the control arm, we recruited 156 healthy volunteers, who visited the NICRH during the time of patient recruitment by matching their age and sex with the breast and cervical cancer patients. A predesigned study protocol and a consent form were used for the clinical investigation of breast and cervical cancer patients. Ethical permissions were obtained from the NIRCH (for breast cancer: NICRH/Ethics/2019/446 and for cervical cancer: NICRH/Ethics/2019/447) ethics committee. We used a standard questionnaire for collecting the details of the patients, including their sociodemographic details, clinicopathological history, and present status. Sociodemographic details of the controls were also recorded. We have selected patients who were free from other comorbidities such as liver, lung, and kidney diseases. This study was conducted at the Laboratory of Pharmacogenomics and Molecular Biology located at the Department of Pharmacy, Noakhali Science and Technology University.

Blood sample collection and DNA extraction

Each participant included in this case-control study donated about 3 ml of blood. The blood samples were collected via a 3 ml intact syringe and transferred immediately into an ethylene diamine tetra acetic acid (EDTA) containing plastic tube. The tubes were then stored in a -80ºC refrigerator until processed. The extraction of genomic DNA from whole blood was completed following the DNA extraction method described by Islam and colleagues [ 33 ] using a DNA extraction kit provided by Favorgen (Taiwan). The purity of extracted DNA was assessed by keeping the absorbance ratio A260:A280 and samples with a ratio of > 1.5 were considered pure DNA.

Primer design and genotyping

There are different online-based software available for designing primers. We have used the Primer1 software to design four required primers. For genotyping process, the tetra-primer amplification refractory mutation system–polymerase chain reaction (T-ARMS–PCR) was utilized as described by Aziz and colleagues [ 34 ]. To validate the method, we first carried out a gradient PCR at temperatures ranging from 60°C to 65°C by a continuous alteration of primer concentration and MgCl 2 concentration. After completing multiple PCRs, the intended PCR products for IL-17A rs3748067 were found at the temperature of 65°C. The genotyping of all samples was completed using the same formula of the PCR working master mix at the temperature of 65°C and visualized using ethidium bromide-stained 1.5% gel electrophoresis. The details of primers and conditions are listed in Table  1 , and the agarose gel images are shown in Fig.  1 (breast cancer samples) and Fig.  2 (cervical cancer samples). For controlling the quality of genotyping and ensuring repeatability, 20% of the samples were randomly assessed.

PCR products for IL-17A rs3748067 in breast cancer after 1.5% agarose gel electrophoresis

PCR products for IL-17A rs3748067 in cervical cancer after 1.5% agarose gel electrophoresis

Statistical analysis

The sociodemographic and clinicopathological characteristics were reported as percentages. The genotypes and allele frequencies were measured for the deviation from the Hardy-Weinberg equilibrium (HWE) by applying the χ 2 -statistic. The link of IL-17A gene rs3748067 polymorphism with breast and cervical cancer was calculated by logistic regression according to five genotypic models (additive model 1, additive model 2, dominant model, recessive model, and overdominant model) and allele model using odds ratio (OR) with 95% confidence intervals (CI). Best-fit model was determined using Akaike’s information criterion (AIC) and Bayesian information criterion (BIC) values. A statistically significant risk was considered in terms of p -value ≤ 0.05. Statistical calculations were done by the use of latest SPSS software version 25 (IBM Corp., Armonk, NY, USA).

Functional analysis

The Gene Expression Profiling Interactive Analysis (GEPIA) (available at http://gepia.cancer-pku.cn ) is a recently developed interactive web resource. It is utilized for examining the RNA sequencing expression data of 9,736 tumors and 8,587 normal tissues retrieved from The Cancer Genome Atlas (TCGA) along with the Genotype-Tissue Expression (GTEx) applying a standard processing pipeline. In this study, GEPIA was applied to evaluate the transcriptional level of the IL-17A gene expression of breast and cervical cancer tissues versus normal tissues and visualized using box plots. Hub genes with |Log 2 FC| ≥ 1 as well as p  ≤ 0.05 were considered statistically significant. Again, the UALCAN is a comprehensive interactive web server for investigating cancer OMICS data [ 35 ]. We have used this webserver to show the IL-17A expression based on the sample types, patient’s age, individual cancer stages, patient’s race, weight, and tumor grade for both breast and cervical cancer.

Characteristics of participants

The distribution of characteristic variables of breast cancer patients and healthy subjects is listed in Table  2 . It is found that 40.38% of patients were under 45 years old and 46.15% were between 45 and 60 years. In comparison with cases, controls consisted of 50.64% under 45 years, and 43.58% were between 45 and 60 years. The average age of the breast cancer patients and controls was 45.37 years and 40.02 years, respectively. Besides, the average body mass index (BMI) was 28.61 kg/m 2 and 22.57 kg/m 2 in the patient group and control group, respectively. About 97.44% of cases were married, and 92.95% of controls were married. Of the patients, 61.21% had invasive duct cell carcinoma and most of the patients had grade II breast cancer consisting of 65.45%. Around 56.83%, 41.01%, 22.30%, 19.42%, and 17.27% of patients have been diagnosed by USG, biopsy, CT, FNAC, and X-ray, respectively. Almost half of the patients received surgery and around 71% of patients received 4–8 cycles of chemotherapy. Patients with negative hormonal status were prevalent such as 34.88% were ER (-), 34.11% were PR (-), and 39.53% were HER2 (-).

The detailed characteristics of cervical carcinoma patients and healthy subjects are summarized in Table  3 . As the data show, about 48.08% of patients were under 45 years old and 39.74% were between 45 and 60 years. The average age of cervical malignancy patients was 41.12 years, and the average BMI was 26.93 kg/m 2 . Approximately 94.87% of cases were married. The menstruation cycle starting age of 86.43% of patients was ≤ 13 years, whereas 90.28% of controls had their first menstruation cycle at ≤ 13 years. Again, the age of menstruation cycle stopping of 77.61% of patients was ≤ 45 years compared to 68.33% of the controls. Almost 80% of patients conceived their first child before or under 18 years, whereas 75.73% of controls gave birth to their first child at this age. The history of contraceptives shows that 75.64% of cervical cancer patients took pills and 63.56% of them took the pill for less than or equal to 5 years. Around 85% of cervical cancer patients had squamous cell carcinoma, and most of the patients were at IIB (56.55%) tumor stage, while 68.59% had grade 2 cancer and 55.13% were with T1 tumor size. 83.33% of patients were with negative (+) lymph nodes, and the status of distant metastasis showed that 68.42% of patients were in Mx state.

Distribution of genotypes of rs3748067

The frequency of genotypes in breast cancer patients obeyed HWE (χ 2  = 2.85, p -value = 0.091) with a minor allele frequency of 20.51%. In controls, the genotype distribution did not show any deviation from HWE (χ 2  = 3.46, p -value = 0.063) and minor allele frequency was 17.31%. The distribution of genotypes in cervical cancer patients also showed no departure from HWE (χ 2  = 2.05, p -value = 0.152) and the frequency of minor allele was 21.15%, as shown in Table  4 .

Association between  IL-17A rs3748067 variant with breast cancer

Table  5 presents the association analysis of IL-17 A gene rs3748067 polymorphism with breast cancer. From the analysis, it is found that additive model 1 and additive model 2 showed increased risk but the associations were not statistically significant (CT vs. CC: OR = 1.25, p  = 0.394; TT vs. CC: OR = 1.35, p  = 0.545, respectively). Other genotype models, such as dominant, recessive, and over-dominant models, also showed a similar nonsignificant association (CT + TT vs. CC: OR = 1.26, p  = 0.332; OR = 1.27, p  = 0.628; OR = 1.22, p  = 0.441, respectively). In the allele model, minor allele T showed an enhanced risk association, and the association is not statistically significant (T vs. C: OR = 1.23, p  = 0.307).

Association between  IL-17A rs3748067 variant with cervical cancer

The correlation of IL-17A gene rs3748067 polymorphism with cervical cancer susceptibility (Table 5 ) demonstrated that two genetic association models, i.e., additive model 1 and over dominant model, showed a statistically significant association with cervical cancer (CT vs. CC: OR = 1.79, 95% CI = 1.09 to 2.92, p  = 0.021; OR = 1.84, 95% CI = 1.13 to 3.00, p  = 0.015). Other models did not show any significant association with cervical cancer (Additive model 2- TT vs. CC: OR = 0.58, p  = 0.394; Dominant model - CT + TT vs. CC: OR = 1.58, p  = 0.052; Recessive model: TT vs. CC + CT: OR = 0.49, p  = 0.248; Allele model: T vs. C: OR = 1.28, p  = 0.223).

Comparison of genotypes and risk association between breast and cervical cancer

The frequency of genotypes of IL-17 A rs3748067 and their comparison between breast cancer and cervical cancer patients are given in Fig.  3 . It is observed that CC homozygote frequency (CC = 102) is higher in breast cancer patients than in cervical cancer patients (CC = 94). The distribution of CT heterozygote and TT mutant homozygotes shows that the frequencies of CT genotypes are higher, but TT genotypes are lower in cervical cancer patients (CT = 44 vs. 58 and TT = 10 vs. 4).

Comparison of genotypes of IL-17 A rs3748067 between breast and cervical cancer

Besides that, the comparison of ORs for analyzing the risk association of IL-17A rs3748067 between breast and cervical cancer patients (Fig.  4 ) showed that the ORs were higher for two genetic association models of cervical cancer- additive model 1 and overdominant model compared to breast cancer (1.79 vs. 1.25 and 1.84 vs. 1.22, respectively) and associations were also statistically significant. Although other genetic models in breast cancer showed higher ORs than in cervical cancer, except for the allele model, these models were not statistically significant. The model that produced the lowest values of AIC and BIC was deemed to be the optimal fit. It may be that the recessive model would be the most suitable choice for breast cancer, although no significant association was found, whereas, in the case of cervical cancer, the overdominant model is the best-fit model (Table 5 ).

Comparison of risk association models of IL-17A rs3748067 between breast and cervical cancer population

IL-17A transcription levels

The level of IL-17A transcription in breast and cervical cancer tissues versus normal tissues is visualized in Fig.  5 . The box plots indicated that there is a significantly greater expression of IL-17 A in cervical carcinoma (CESC) tissues than in normal tissues. The expression level in breast carcinoma (BRCA) tissues and normal tissues was not statistically significant.

IL-17A gene expression of breast and cervical cancer tissues versus normal tissues based on the GEPIA ( http://gepia.cancer-pku.cn )

The IL-17A expression based on sample types, patient’s age, individual cancer stages, patient’s race, weight, and tumor grade from the UALCAN web server for cervical cancer and breast cancer is depicted in Fig.  6 and Supplementary Fig.  1 , respectively. The expression of IL-17A was found to be higher in cervical tumor samples (Fig.  6 a), 81–100 years of age (Fig.  6 b), cancer stage 1 and stage 2 (Fig.  6 c), African American patients (Fig.  6 d), obese and extremely obese patients (Fig.  6 e), and tumor grade 2 and grade 3 patients (Fig.  6 f). Again, in terms of breast cancer, no significant expression change was observed for sample types, patient’s age, cancer stages, patient’s race, and gender (Supplementary Fig.  7 a-e) except for the medullary subtype (Supplementary Fig.  7 f).

IL-17A expression based on the sample types, patient’s age, individual cancer stages, patient’s race, weight, and tumor grade of cervical cancer

Breast and cervical cancers are the two most commonly diagnosed malignancies in females worldwide [ 1 , 2 ]. Cytokines have been playing an indispensable role in tumor growth and progression. IL-17A is considered one of the most common cytokines from the IL-17 family that has been extensively studied due to its prominent role in carcinogenesis, especially in cervical and breast carcinoma besides inflammation [ 36 ]. A plethora of studies have described that IL-17A protein is greatly expressed within tumor tissues: for instance, gastric carcinoma, breast cancer, ovarian cancer, colorectal carcinoma, lung cancer, thyroid cancer, and hepatocellular carcinoma [ 18 , 19 ]. Again, the increased IL-17A levels in the blood are linked with the aggressiveness of pancreatic adenocarcinoma, non-small cell lung cancer, thyroid tumors, laryngeal squamous cell carcinoma, and colorectal carcinoma [ 37 , 38 , 39 , 40 ]. In addition, previous studies with rs3748067 variant in the IL-17A gene described its notable association with a variety of cancers in multiple ethnicities, such as breast cancer [ 20 ], cervical cancer [ 21 , 22 , 23 , 24 , 25 , 26 ], colorectal cancer [ 27 , 28 , 41 ], gastric cancer [ 29 , 30 ], and others. Based on the previous research, we performed this case-control study that reported the correlation of IL-17A rs3748067 polymorphism with the risk of cervical malignancy.

From the analysis in this study, we did not find any significant association of IL-17 A rs3748067 polymorphism with breast carcinoma in the studied population. However, our study found a higher frequency of the major allele C (79.49%) compared to the minor T allele (20.51). Again, the frequency of CC homozygous genotype was also greater (65.38%) than the heterozygous or mutant homozygous genotypes. The only previous study with this polymorphism in breast cancer also described similar findings. The study by Wang et al. (2012) showed no notable link between IL-17A gene rs3748067 variant and breast carcinoma in a Chinese case-control study in females. They also reported a higher frequency of a major allele (G allele) in their studied population [ 20 ]. To further explore the role of IL-17A in breast carcinoma, we analyzed the expression level in breast carcinoma (BRCA) tissues and normal tissues that were not statistically significant. Moreover, we did not observe any significant expression of IL-17A in terms of sample types, cancer stages, patient age, patient race, and gender.

Our study revealed a statistically significant link between IL-17A gene rs3748067 variant and cervical cancer. Our analysis demonstrated that CT genotype (OR = 1.79, p  = 0.021) and over-dominant model (OR = 1.84, p  = 0.015) are significantly correlated with cervical carcinoma risk. Besides, the frequency of the major allele (C allele: 78.85%) is greater than the minor allele (T allele: 21.15%). Moreover, we found that there is a significantly greater transcription level of IL-17A in cervical carcinoma (CESC) tissues than in normal tissues. The level of IL-17A expression was found to be higher in cervical tumor samples, 81–100 years of age, African American patients, obese and extremely obese patients, cancer stage 1 and stage 2, and in patients with tumor grade 2 and grade 3.

The correlation between IL-17A gene rs3748067 variant and breast carcinoma was examined for the first time in 2012 by Wang and colleagues [ 20 ] in Chinese Han women. The study recruited 491 breast cancer patients and 502 healthy individuals, and for genotyping, applied the SNaPshot technique. The study revealed that rs3748067 GG genotypes percentage was lower in PR-positive cases and was significantly correlated with PR hormonal status. They concluded that rs3748067 GG genotypes might be linked to poor prognosis and ineffective treatment. The link of this variant was not studied later in any other population. IL-17A rs3748067 has been studied in cervical cancer several times. The correlation was also evaluated in Chinese women by Niu et al. [ 22 ]. They explicated that subjects with the TT genotype and T allele were more prone to cervical carcinoma [ 22 ]. Another study in the Chinese population examined the contribution of rs3748067 in 352 cervical malignancy patients and 352 healthy controls using the PCR-RFLP method. However, they failed to establish any association between this polymorphism with cervical cancer [ 25 ]. Some other studies also failed to establish the association of rs3748067 variant with cervical carcinoma [ 24 , 26 ].

The latest meta-analysis with rs3748067 variant in the IL-17A gene reported that it was correlated with cervical carcinoma, with T allele carriers depicting an enhanced risk [ 21 ]. Another meta-analysis conducted by Yang and colleagues [ 23 ] reported an elevated susceptibility of cervical carcinoma due to this polymorphism.

In this study, we have also tried to compare the frequency of genotypes of IL-17A rs3748067 between breast and cervical cancer patients. We have found that the frequency of CC homozygotes is greater in breast cancer patients than in cervical cancer patients. The distribution of CT heterozygote and TT mutant homozygotes reveals that the percentage of CT genotypes is higher, but TT genotypes are lower in cervical cancer patients. In addition, the comparison of ORs between breast and cervical cancer patients showed that the ORs were significantly higher for additive model 1 and the over-dominant model in cervical cancer compared to breast cancer.

It is to be mentioned that there are some limitations of the present study, such as the total number of participants included in the study is relatively low. Besides, all sociodemographic and clinicopathologic details of the participants were not possible to collect, which may alter the association. In addition, for this study, we have selected only available SNP from the public electronic database. However, our study has identified the link of IL-17A rs3748067 variant with cervical carcinoma and we are hopeful that our findings will have an impact on further studies that may result in stronger evidence. Besides these, possible interactions between the susceptibility loci and these risk factors should be thoroughly investigated.

This study concludes that rs3748067 polymorphism in the IL-17A gene is associated with cervical cancer, not breast cancer in Bangladeshi patients. However, we suggest studies in the future with a larger sample size.

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request. The details data cannot be shared publicly due to the restriction of the ethical committee.

Abbreviations

Interleukin-17 A

Single nucleotide polymorphism

Myeloid-derived suppressor cells

Signal transducer and activator of transcription

National Institute of Cancer Research and Hospital

Ethylene diamine tetra acetic acid

Tetra-primer amplification refractory mutation system

Polymerase chain reaction

Hardy-Weinberg equilibrium

Confidence intervals

Gene Expression Profiling Interactive Analysis

The Cancer Genome Atlas

Genotype-Tissue Expression

Ferlay J, Colombet M, Soerjomataram I, Parkin DM, Piñeros M, Znaor A, Bray F. Cancer statistics for the year 2020: an overview. Int J Cancer. 2021. https://doi.org/10.1002/ijc.33588 .

Article   PubMed   Google Scholar  

Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Bray F. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209–49.

Aziz MA, Jafrin S, Islam MS, Kabir Y. Interleukins in the development and progression of breast Cancer. In: Interdisciplinary Cancer Research. Cham: Springer; 2022. p. 1–22.

Google Scholar  

Arbyn M, Weiderpass E, Bruni L, de Sanjosé S, Saraiya M, Ferlay J, Bray F. Estimates of incidence and mortality of cervical cancer in 2018: a worldwide analysis. Lancet Glob Health. 2020;8(2):e191-203.

Jafrin S, Aziz MA, Anonna SN, Akter T, Naznin NE, Reza MS, Islam MS. Association of TP53 codon 72 arg > pro polymorphism with breast and lung cancer risk in the south Asian population: a meta-analysis. Asian Pac J Cancer Prev. 2020;21(6):1511–9.

Article   CAS   PubMed   PubMed Central   Google Scholar  

Shah R, Rosso K, Nathanson SD. Pathogenesis, prevention, diagnosis and treatment of breast cancer. World J Clin Oncol. 2014;5(3):283–98.

Article   PubMed   PubMed Central   Google Scholar  

Fakhri N, Chad MA, Lahkim M, Houari A, Dehbi H, Belmouden A, El Kadmiri N. Risk factors for breast cancer in women: an update review. Med Oncol. 2022;39:12–197.

Article   Google Scholar  

Barek MA, Basher MA, Aziz MA, Hossen MS, Jahan N, Afroz N, Begum M, Jafrin S, Uddin MS, Millat MS, Hoque MM, Islam MS. Assessment of the association of CYP1A1 gene polymorphisms with the susceptibility of cervical cancer: a case-control study and meta-analysis. Heliyon. 2023;9(7): e17712.

Muhammad SB, Hassan F, Bhowmik KK, Millat MS, Sarwar MS, Aziz MA, Barek MA, Uddin MS, Ferdous M, Islam MS. Detection of association of IL1β, IL4R, and IL6 gene polymorphisms with cervical cancer in the Bangladeshi women by tetra-primer ARMS-PCR method. Int Immunopharmacol. 2021;90:107131.

Article   CAS   PubMed   Google Scholar  

Ivy SC, Shabnaz S, Shahriar M, Jafrin S, Aka TD, Aziz MA, Islam MS. Association of RAD51 and XRCC2 gene polymorphisms with cervical cancer risk in the Bangladeshi women. Asian Pac J Cancer Prev. 2021;22(7):2099–107.

Nazneen F, Millat MS, Barek MA, Aziz MA, Uddin MS, Jafrin S, Aka TD, Islam MS. Genetic polymorphism of miR-218-2 (rs11134527) in cervical cancer: a case-control study on the Bangladeshi women. MicroRNA. 2021;10(3):219–2246.

Briukhovetska D, Dörr J, Endres S, Libby P, Dinarello CA, Kobold S. Interleukins in cancer: from biology to therapy. Nat Rev Cancer. 2021;21(8):481–99.

Onishi RM, Gaffen SL. Interleukin-17 and its target genes: mechanisms of interleukin-17 function in disease. Immunology. 2010;129(3):311–21.

Krstic J, Obradovic H, Kukolj T, Mojsilovic S, Okic-Dordevic I, Bugarski D, Santibanez JF. An overview of Interleukin-17A and Interleukin-17 receptor a structure, interaction and signaling. Protein Pept Lett. 2015;22(7):570–8.

Chang SH, Mirabolfathinejad SG, Katta H, Cumpian AM, Gong L, Caetano MS, Moghaddam SJ, Dong C. T helper 17 cells play a critical pathogenic role in lung cancer. Proc Natl Acad Sci U S A. 2014;111(15):5664–9.

He D, Li H, Yusuf N, Elmets CA, Li J, Mountz JD, Xu H. IL-17 promotes tumor development through the induction of tumor promoting microenvironments at tumor sites and myeloid-derived suppressor cells. J Immunol. 2010;184(5):2281–8.

Wang L, Yi T, Kortylewski M, Pardoll DM, Zeng D, Yu H. IL-17 can promote tumor growth through an IL-6-Stat3 signaling pathway. J Exp Med. 2009;206(7):1457–64.

Tu JF, Pan HY, Ying XH, Lou J, Ji JS, Zou H. Mast cells comprise the Major of Interleukin 17-Producing cells and predict a poor prognosis in hepatocellular carcinoma. Med (Baltim). 2016;95(13): e3220.

Article   CAS   Google Scholar  

Carvalho DFG, Zanetti BR, Miranda L, Hassumi-Fukasawa MK, Miranda-Camargo F, Crispim JCO, Soares EG. High IL-17 expression is associated with an unfavorable prognosis in thyroid cancer. Oncol Lett. 2017;13(3):1925–31.

Wang L, Jiang Y, Zhang Y, Wang Y, Huang S, Wang Z, Tian B, Yang Y, Jiang W, Pang D. Association analysis of IL-17A and IL-17F polymorphisms in Chinese Han women with breast cancer. PLoS One. 2012;7(3): e34400.

de Moura EL, Dos Santos ACM, da Silva DM, Dos Santos BB, Figueredo DS, Moura AWA, da Silva AF, Tanabe ISB, de Lira Tanabe EL, Lira Neto AB, et al. Association of Polymorphisms in cytokine genes with susceptibility to precancerous lesions and cervical cancer: a systematic review with meta-analysis. Immunol Invest. 2021;50(5):492–526.

Niu AQ, Cao YH, Wang H, Zhang X, Zhu B, Li ZH. Role of IL17A rs2275913 and rs3748067 polymorphisms in the risk cervical cancer. Genet Mol Res. 2017;16(3). https://doi.org/10.4238/gmr16038826 .

Yang S, Li C, Li X, Huang X, Zhao Q, Liu D, Wu S. Relationship of IL-17A and IL-17F genetic variations to cervical cancer risk: a meta-analysis. Biomark Med. 2017;11(5):459–71.

Li L, Tian YL, Lv XM, Yu HF, Xie YY, Wang JD, Shi W. Association analysis of IL-17A and IL-17F polymorphisms in Chinese women with cervical cancer. Genet Mol Res. 2015;14(4):12178–83.

Cong J, Liu R, Wang X, Sheng L, Jiang H, Wang W, Zhang Y, Yang S, Li C. Association between interluekin-17 gene polymorphisms and the risk of cervical cancer in a Chinese population. Int J Clin Exp Pathol. 2015;8(8):9567–73.

PubMed   PubMed Central   Google Scholar  

Lv Q, Zhu D, Zhang J, Yi Y, Yang S, Zhang W. Association between six genetic variants of IL-17A and IL-17F and cervical cancer risk: a case-control study. Tumour Biol. 2015;36(5):3979–84.

Bedoui S, Dallel M, Barbirou M, Stayoussef M, Mokrani A, Mezlini A, Bouhaouala B, Almawi WY, Yacoubi-Loueslati B. Interleukin-17A polymorphisms predict the response and development of tolerance to FOLFOX chemotherapy in colorectal cancer treatment. Cancer Gene Ther. 2020;27(5):311–8.

Bedoui SA, Barbirou M, Stayoussef M, Dallel M, Mokrani A, Makni L, Mezlini A, Bouhaouala B, Yacoubi-Loueslati B, Almawi WY. Association of interleukin-17A polymorphisms with the risk of colorectal cancer: a case-control study. Cytokine. 2018;110:18–23.

Tian J, Liu G, Zuo C, Liu C, He W, Chen H. Genetic polymorphisms and gastric cancer risk: a comprehensive review synopsis from meta-analysis and genome-wide association studies. Cancer Biol Med. 2019;16(2):361–89.

He B, Pan B, Pan Y, Wang X, Zhou L, Sun H, Xu T, Xu X, Liu X, Wang S. Polymorphisms of IL-23R predict survival of gastric cancer patients in a Chinese population. Cytokine. 2019;117:79–83.

He Y, Du Y, Wei S, Shi J, Mei Z, Qian L, Chen Z, Jie Z. IL-17A and IL-17F single nucleotide polymorphisms associated with lung cancer in Chinese population. Clin Respir J. 2017;11(2):230–42.

von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, STROBE Initiative. The strengthening the reporting of Observational studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies. Int J Surg. 2014;12(12):1495–9.

Islam MS, Ahmed MU, Sayeed MS, Maruf AA, Mostofa AG, Hussain SM, Kabir Y, Daly AK, Hasnat A. Lung cancer risk in relation to nicotinic acetylcholine receptor, CYP2A6 and CYP1A1 genotypes in the Bangladeshi population. Clin Chim Acta. 2013;416:11–9.

Aziz MA, Akter T, Hussain MS, Millat MS, Uddin MS, Sajal M, Jafrin S, Aka TD, Akter T, Das C, et al. Association of rs363598 and rs360932 polymorphisms with autism spectrum disorder in the Bangladeshi children. Meta Gene. 2020;25: 100733.

Chandrashekar DS, Bashel B, Balasubramanya SAH, Creighton CJ, Ponce-Rodriguez I, Chakravarthi BVSK, Varambally S. UALCAN: a portal for facilitating tumor subgroup gene expression and survival analyses. Neoplasia. 2017;19(8):649–58.

Fabre J, Giustiniani J, Garbar C, Antonicelli F, Merrouche Y, Bensussan A, Bagot M, Al-Dacak R. Targeting the tumor microenvironment: the protumor effects of IL-17 related to cancer type. Int J Mol Sci. 2016;17(9): 1433.

Karabulut S, Afsar ÇU, Karabulut M, Alış H, Kılıc L, Çikot M, Yasasever CT, Aykan NF. Evaluation of serum Interleukin-17 (IL-17) levels as a diagnostic marker in pancreatic adenocarcinoma. J Gastrointest Cancer. 2016;47(1):47–54.

Karabulut S, Usul Afsar C, Karabulut M, Kilic L, Alis H, Kones O, Bilgin E. Faruk Aykan N. Clinical significance of serum interleukin-17 levels in colorectal cancer patients. J BUON. 2016;21(5):1137–45.

PubMed   Google Scholar  

Li FJ, Cai ZJ, Yang F, Zhang SD, Chen M. Th17 expression and IL-17 levels in laryngeal squamous cell carcinoma patients. Acta Otolaryngol. 2016;136(5):484–90.

Lu Y, Yuan Y. Serum level of interleukin-17 and interleukin-35 as a biomarker for diagnosis of thyroid cancer. J Cancer Res Ther. 2015;11(Suppl 2):C209-211.

Islam MR, Aziz MA, Shahriar M, Islam MS. Polymorphisms in IL-17A gene and susceptibility of colorectal cancer in Bangladeshi population: a case-control analysis. Cancer Control. 2022;29:10732748221143880.

Download references

Acknowledgements

We acknowledge the authority of the NICRH, Dhaka, Bangladesh, for their cooperation and help in the collection and primary storage of blood samples.

Noakhali Science and Technology University (NSTU) Research Cell (ID: NSTU/RC/20/C-84) and National Science and Technology (NST) Fellowship 2020, Ministry of Science and Technology, Bangladesh, have partially funded this research work.

Author information

Authors and affiliations.

Department of Pharmacy, Faculty of Science, Noakhali Science and Technology University, Noakhali, 3814, Bangladesh

Md. Abdul Aziz, Subrina Chowdhury, Sarah Jafrin, Md Abdul Barek, Mohammad Sarowar Uddin, Md. Shalahuddin Millat & Mohammad Safiqul Islam

Laboratory of Pharmacogenomics and Molecular Biology, Department of Pharmacy, Noakhali Science and Technology University, Noakhali, 3814, Bangladesh

Md. Abdul Aziz, Sarah Jafrin, Md Abdul Barek, Mohammad Sarowar Uddin, Md. Shalahuddin Millat & Mohammad Safiqul Islam

Bangladesh Pharmacogenomics Research Network (BdPGRN), Dhaka, 1219, Bangladesh

Md. Abdul Aziz & Mohammad Safiqul Islam

You can also search for this author in PubMed   Google Scholar

Contributions

MAA performed method validation, investigation, software, formal analysis, data curation, visualization, original draft preparation, review, and editing. SC and SJ contributed to formal analysis, review, and editing. MSU contributed to validation, investigation, and data curation. MAB and MSM contributed to writing, reviewing, and editing. MSI contributed to conceptualization, supervision, project administration, funding, software, resources, writing, review, and editing. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Mohammad Safiqul Islam .

Ethics declarations

Ethics approval and consent to participate.

Ethical permissions were taken from the ethics committee of the National Institute of Cancer Research and Hospital (NIRCH) (for breast cancer: NICRH/Ethics/2019/446 and for cervical cancer: NICRH/Ethics/2019/447). In addition, written informed consent was obtained from all subjects before starting the study, and all procedures were conducted in accordance with the Helsinki Declaration.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Supplementary material 1., rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Aziz, M.A., Chowdhury, S., Jafrin, S. et al. Genetic association of Interleukin-17A polymorphism in Bangladeshi patients with breast and cervical cancer: a case-control study with functional analysis. BMC Cancer 24 , 660 (2024). https://doi.org/10.1186/s12885-024-12352-0

Download citation

Received : 28 April 2023

Accepted : 07 May 2024

Published : 30 May 2024

DOI : https://doi.org/10.1186/s12885-024-12352-0

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Interleukin-17A
• Breast cancer
• Cervical cancer
• Association
• Correlation
• Case-control
• Polymorphism

ISSN: 1471-2407

• Submission enquiries: [email protected]
• General enquiries: [email protected]