Pharmaceutical companies often cite human error as the sole cause of a manufacturing deviation or noncompliance issue (1). For medical device manufacturers, identifying and reducing the occurrence of human error is commonly part of the regulatory and overall risk management process (2,3), but for pharmaceutical facilities, the assessment of human error in the workplace is not always so rigorously assessed during deviation investigations. Common corrective actions for human error within a GMP production facility often include retraining or awareness sessions combined with document updates or revisions that include additional document checks and signatures. In most cases, however, there is often no scientific rationale that human error was the actual root cause. The ultimate goal of a root cause investigation within a production facility is to identify the underlying cause of why an incident occurred so that the most effective solution can be identified. Yet corrective actions focused on retraining typically do not fix the true root cause as to why the deviation occurred, and generally will not prevent a reoccurrence of the issue. There may be some merit in retraining programs, but to prevent the same, or similar error occurring again, the retraining should include improved or modified procedures. Simply retraining using existing protocols is unlikely to address the recurring errors. If staff have been trained on the current versions of the procedures, it is difficult to justify, in many cases, why retraining should be included as a preventative action (4–7).
The purpose of a root cause investigation that suggests human error is not to determine what went wrong, but to establish why the actions seemed appropriate at the time and prevent reoccurrence. As such, there is often a regulatory perception that in some cases, staff retraining may be misused as a preventative measure for production incidents, with the perception that there is a lack of understanding of what constitutes human error (1). Most production deviations that occur and are classified as human error are actually due to poorly designed systems, including the structure and content of production documents or Standard Operating Procedures (SOPs). To identify events that are due to human error, it is important to collate metrics and trend the type of errors in the workplace, but also identify if the same error has reoccurred by different production operators, as this provides an indication of an inherent issue in the system design.
Regulatory guidance for pharmaceutical development and manufacture are moving towards a science- and risk-based approach (8). There is now a regulatory expectation for pharmaceutical manufacturers to develop and document scientific rationales for classifying and preventing human error (1). Integrating human error reduction tools within a quality system, however, need not be cumbersome or complicated. Many pharmaceutical companies utilize the Talsico human error reduction methodology, to categorize human error (9) and also to develop effective strategies for prevention (Table 1).
Table 1: Categories of human error that may occur within a production facility.
As part of a continuous improvement program, a manufacturing site could establish procedures that provide guidance on what is classified as human error within the production facility and the associated indicators of such events. A manufacturing site could also consider a simple checklist or questionnaire (1) to assess human error associated with a particular procedure, and include this methodology as part of the deviation process. Alternatively, more elaborate root cause analysis tools such as Ishikawa’s Fishbone analysis (10) can be utilized with a longer term strategy to integrate human error reduction into root cause analysis investigations and CAPA programs. Once the type of human error has been established, it may be possible to evaluate the error hierarchy (Table 2) to define appropriate corrective actions based on strategies to reduce the error from reoccurring. Actions may include specific engineering solutions for error proofing, such as preventing the unlocking of a door unless a switch is in an “off” position, permitting safe access to potentially hazardous areas (gamma irradiation chambers, for example), or using equipment connectors of a specific size or shape to prevent human error. Human error prevention using attention activators within production documents, such as batch records and SOPs has also shown to be effective. Simple attention activators within production documents that have shown to be highly effective in preventing human error include changes in document pattern, color, shape and text format (bold, italic, size, etc.), yet many production documents do not contain any specific formatting that has been included to prevent errors. A common corrective action for human error is the duplication of effort or resource. This often includes multiple checks and signatures (such as within production batch records). Such redundancy is often the most utilized action but is the least effective strategy for human error reduction (Table 2).
Table 2: Strategies for human error prevention (in order of effectiveness).
It is important to exercise caution before deciding that human error is the cause of a production deviation. In some cases, human error may indeed be the cause of an incident, but in many cases human error is a symptom of an underlying problem, rather than the cause. It is clear that the regulators are expecting a shift from simply complying with GMP toward demonstrating that a pharmaceutical organization has an effective pharmaceutical quality system, including procedures to manage human error in the workplace.
1. Poska, R. “Human Error and Retraining.” Journal of GXP compliance 13 (2009): 2–15.
2. Draft Guidance for Industry and Food and Drug Administration Staff – Applying Human Factors and Usability Engineering to Optimize Medical Device Design, U.S. Food and Drug Administration: June 2011 www.fda.gov/medicaldevices/deviceregulationandguidance/guidancedocuments/ucm259748.htm
3. ISO 14971, Medical Devices – Application of risk management to medical devices, International Organization for Standardization: 2007
4. Chalk, S. “Reducing Human Error.” BioPharm International 25, (2012): 58–59. www.biopharminternational.com/biopharm/Quality/Reducing-Human-Error/ArticleStandard/Article/detail/775162
5. Schmitt, S. “The Human Factor.” Pharmaceutical Technology 38 (2014): 82. www.pharmtech.com/pharmtech/GMPs%2FValidation/The-Human-Factor/ArticleStandard/Article/detail/834883
6. Schniepp, S.J. “The Human Error Behind Human Error.” Pharmaceutical Technology 37 (2013) www.pharmtech.com/pharmtech/Insider+Solutions/The-Human-Error-Behind-Human-Error/ArticleStandard/Article/detail/804848
7. Dekker, S. The Field Guide to Understanding Human Error. Burlington, VT: Ashgate Publishing Company, 2006.
8. O’Donnell, K. “Quality Risk Management: Putting GMP Controls First.” PDA Journal of Pharmaceutical Science and Technology 66 (2012): 243-261 9. O’Callaghan, D., Wade, J. Human Error Network Event. (Cork, Ireland, 2009)
10. Ishikawa, K. What Is Total Quality Control? The Japanese Way. Upper Saddle River, NJ: Prentice-Hall, 1985.
Claire Newcombe is the owner and consultant at Applied Biopharm Consultancy, supporting the development and production of biopharmaceuticals. She worked previously as an independent QA consultant at BTG (formerly Protherics UK Ltd) and also within microbiology laboratories at the UK Health Protection Agency, London, UK.
Shanon McKenna is a human factors and usability expert specializing in medical device testing and regulatory compliance. She has over a decade of experience implementing medical technology solutions and holds certifications in business process analysis and systems modeling.
Anthony Newcombe is a Principal Consultant at PAREXEL, advising and consulting on all aspects of pharmaceutical development relating to Biological and Biotechnology Products. He worked previously within Global Industrial Operations at GSK Biologicals and also at Pfizer in the UK and Ireland.
The article was first published in the PDA Letter of October 2014 and is reprinted in our LOGFILE Newsletter by courtesy of PDA.