Is ‘Human Error' The Cause Or The Outcome Of GMP Deviations?

By Amnon Eylath, Broad Spectrum GXP Consulting

In the biopharma industry, “human error” is often selected to describe a deviation for which the immediate apparent cause was the behavior, which could be through action or inaction, of the person(s) performing a task or process. Too often, “retraining” is selected as the preventive action that, hopefully, will prevent that person or persons from repeating the deviation. As anyone with significant experience working in the GMP domain knows, repetition of training may be effective for straightforward and simple work, but not for complex tasks or for activities that require the application of self-awareness, critical thinking, and real-time judgment, all of which are critical in situations that are not nominal to the process, such as, for example:

• The workload is greater than the capacity of staff present.
• The time to complete a task is reduced from the norm.
• The written instructions are complicated, unclear, ambiguous, or have errors.
• Equipment or electronic systems used (e.g., eQMS, ERP, LIMS, equipment controls, etc.) are difficult to operate or understand by someone who may not be a system expert or superuser.
• There are not enough experienced operators or system users assigned to a task or process.

Yes, It’s Complicated

The manufacturing, testing, release, and distribution of medicines and therapeutic products is a series of complex processes. Due to their complexity and the need to attain a very high level of quality at all steps in the process, it is common to observe multiple deviations from procedures, protocols, or analytical methods, or non-conformance to specifications (in-process or release). In most cases, these processes rely heavily on direct human interaction, whether in receipt of raw materials, manufacturing, QC testing, warehousing, or distribution. Fully validated automation of processes does exist, but due to their high cost to install, and lengthy time to implement and debug, automated systems often exist for units of operation but not systemically. Even in the case of full automation, these systems may rely heavily on human activities and decisions, such as the loading of materials, in-process and final material sampling, in-process monitoring, and real-time responses to unplanned process variations, malfunctions, and stoppages.

What Can We Do About It?

If we accept that for most companies there is a continuous human element across all aspects of the manufacturing process, the reduction of deviations caused by human action or inaction in routine or variant situations is critical, due to the potential impact of deviations on product quality, such as late delivery, increased cost of goods, inability to release the product for forward processing or for clinical or commercial use, etc. An additional impact of having many major or critical deviations related to a product, process, or facility would be in the GMP compliance assessment performed by a health authority inspection. A poor assessment would, at a minimum, require immediate time-consuming remediation, could delay product approval, and in extreme situations of many repeat deviations with an impact to product quality, could even prevent product/site approval.

I would like to propose that “human error” is usually the outcome of deficiencies in the documented instructions and procedures, training program, equipment design and operation, ergonomics, process controls, analytical methods, electronic systems, and staffing levels. My approach to this subject came about after I was tasked by the CEO of a major biotechnology company to identify how to reduce human errors in GMP manufacturing at its main site, which included several commercial and clinical manufacturing facilities. One of the errors had even caused a batch of GMP product to be discarded. Others had caused much staff time to be spent on multiple corrective and preventive actions (CAPAs). This company’s concern was compounded by the fact that in the prior year, the number of major and critical deviations, and number of deviations overall, was much lower in these same facilities.

I received close to 90 major deviations (with a likelihood of impacting product quality and/or compliance) and one critical deviation (which had caused the batch loss) to analyze. Most were attributed to manufacturing staff, several to QA staff. My methodology was to review the deviations, interview the QA staff who processed them and oversaw the CAPAs, and interview the manufacturing staff whose actions or inactions had caused the deviations, if available (not all of the people involved were still working at that company by the time of this investigation). I was able to get direct input from the available QA and manufacturing staff, in most cases, or had access to the detailed notes related to each deviation documented.

What surprised me was that when I looked beyond the error performed by the person and identified the true causal factor(s) that contributed to the erroneous action or behavior, it became apparent that, except for two of the deviations analyzed, the main contributing factors that caused a deviation to occur were related to:

• Problematic equipment design impacting ergonomics or no error-proofing built into the operational design.
• Unclear batch record instructions.
• Understaffing of shifts in spite of having  nominal, recommended staffing numbers.
• Loss of highly experienced staff as they were promoted or moved to positions in other plants or sites, etc.

As an example, in the case of the drug substance batch that had to be discarded, the operator had left the cleanroom and did not return until after the maximum time allowed for the processing step for the material. On its face, this was caused by poor behavior by the operator; after all, they had left the suite and returned late. Upon further investigation, it was identified that only two operators were working this shift, even though a prior staffing assessment had recommended four operators per shift.

The operator who returned late to complete the process had not been relieved for hours while working in the cleanroom. They left the production suite to use the lavatories with no one available to cover for them. The new root cause I assigned to this deviation was, “Staffing level failure by the operational management.” The rest of the deviations were assigned new root causes, based on the in-depth analysis. For the two deviations that retained their “human error” root cause, one was a case of purposeful bad behavior by an employee who was leaving the company (events of this type are very rare), and the second was caused by a trained and experienced supervisor, who had a brief lapse of concentration and turned a valve in the wrong direction. In the latter case, one could also make the argument that error-proofing the equipment through color coding, better signage, mechanical or electronic interlocks, etc. could have prevented this type of honest mistake from occurring.

To reference an external source, in the article Frequent Deficiencies In GMP Inspections, Part 1, written by Lea Joos, GMP inspector for the government of Upper Bavaria, she gives an example of “inadequate handling of deviation” as follows.

The deficiency: In the case of the deviation inspected, it was found that the batch number, which had to be manually transferred to the product before production started, was not correctly transferred to the product. The cause was found to be a "human error" in the manual transfer of the batch number during the root cause analysis. An evaluation of possible technical or organizational causes for the "human error" was missing. The evaluation was also missing for the third deviation of this type in the last three months. (Ref.: EU GMP Guide Part I No. 1.4 xiv)

Joos adds that: “Due to the manual transfer, the cause ‘human error’ was obvious: an employee had obviously transferred the batch number incorrectly. However, according to EU GMP Guide Part I No. 1.4 xiv, so-called ‘human error’ should be handled very carefully and should only be recorded as the cause if other technical, process-related, system-related, or organizational causes could be excluded.”

She gives an example of a “Five Whys” root cause analysis exercise that easily identified the true root cause, as the instructions were placed too far from the operator to be read clearly, with the reason being “the instruction documents cannot be placed on the worktable for reasons of hygiene.”

In this example, I would redesignate the root cause of this deviation from “Incorrect manual transfer of the batch number,” apparently a human error, to “Easily readable work instructions were not made available to the operators at this workstation,” with the responsibility lying with the people responsible for designing this setup. Considering that this was a repeating deviation, simple solutions could be implemented that would likely reduce recurrence significantly.

What Are The Challenges In Implementing This Approach Broadly?

Since the industry is used to identifying all deviations that have a direct cause based on human behavior, action, or inaction generically as “human error,” changing the mindset will not be easy. This is in spite of clear communications and regulatory guidelines instructing otherwise. Retraining an employee is much easier, quicker, and less costly than implementing mechanical, ergonomic, visual, communication, documentation, or material changes. The same goes for ensuring that the staff working on GMP processes are at the right number for that task/process, and that they have the appropriate level of education, training, and experience, or are closely supervised by a more experienced person until they do. Based on published regulatory inspections, it can be readily seen that companies are being cited for using human error and retraining as the first response to these types of deviations. Repeat deviations are a clear indication that the initial CAPA was not effective.

In addition, this improved assessment of deviations can be applied to the other domains of GXP activities, such as for GLP, GCP, GDP, and GVP, where procedures, instructions, staff training, complex documentation, records, and electronic systems are expected by the regulations and guidance documents from FDA, Eudralex, MHRA, ICH, etc.

Conclusion

I recommend that all companies involved in GXP activities adopt a built-in process for questioning the true root causes of human error deviations as part of the initial investigation. Simple root cause analysis tools such as the Five Whys example shared in this article can be applied to quickly elucidate what led to the deviation happening. Then, more effective CAPAs can be implemented to lower the deviation rate, improve compliance, reduce costs, and increase employee satisfaction (nobody likes making errors at work). Then, the causes of true human error can also be addressed more effectively.

About the Author:

Amnon Eylath has been working in the biopharma field for over 30 years, covering the complete investigational and commercial product life cycle, from research and development through clinical trials and successful product launch.  He has aided in establishing and managing phase appropriate GXP quality management systems aligning with FDA, EU, MHRA, and global regulations/guidance to ensure patient safety, product safety, and data integrity. He has partnered with FDA and other regulatory agencies since 2007 on clarifying and communicating phase-appropriate cGMP and quality best practices for all phases of investigational medicinal product development, from preclinical to pre-launch, including for cell and gene therapies.