Reducing Human Error on the Manufacturing Floor

For Pharmaceuticals and Biotechnology

As technology advances, human error in manufacturing becomes more and more visible every day. Human error is responsible for more than 80 percent of process deviations in the pharmaceutical and related manufacturing environments. Sadly, little is known about the nature of these events mainly because quality event investigations end where human error investigations should begin.

Human error is about explaining human behavior. Chemical engineers explain product behavior, mechanical engineers explain equipment behavior, industrial engineers explain process behavior, but who explains human behavior?

When we investigate quality events, the focus of those investigations rely on explaining what happened in the process and how the product was affected. A human error usually explains the reason for the occurrence of the deviation; nevertheless, the reason for that error remains unexplained and consequently the corrective and preventive actions fail to address the underlying conditions for that failure. This in turn translates into ineffective action plans that result in non-value added activities, wasting resources and eventually resulting in recurrences and repeated events.


This article is related to the Interactive and Live Demo:
MasterControl Quality Management System (QMS) Overview
To get the full details, please view your free interactive and Live Demo.

Human error is about explaining human behavior. Chemical engineers explain product behavior, mechanical engineers explain equipment behavior, industrial engineers explain process behavior, but who explains human behavior?

Human behavior is complex and just like equipment, product, and process it needs to be analyzed in depth. We would never end an investigation with just "equipment failure." We would explain exactly what the equipment failure was so it could be fixed.

In order to assure effective CAPA's, human error events need to be fully investigated. Actually, regulations require that errors be fully investigated, meaning that you identify the reasons why they occurred. In order to assure we comply with this expectation, we need to understand how human behavior is affected by external variables as well as internal variables.

First we need to understand what a human error is. Human error is defined in many ways. One definition that I like is "any action, performed by a person, which exceeds a system's tolerance." Human error is an error and not an intentional act for harm. Sabotage is not considered a human error, unless the result of the actual intentions is different than was expected. So stating that a human error has occurred does not necessarily mean that is the "human's" fault.

We as humans don't operate in a vacuum. Behaviors are influenced by external as well as internal variables. In manufacturing environments, these variables can be divided into six major categories: procedures, human factors, training, supervision, communication, and the individual itself.

Individuals are certainly responsible for their actions. But before we determine that internal factors like attitude or attention are responsible for the mistake, we as organizations are responsible for eliminating the possibilities of external factor influencing human behavior. Individual performance in manufacturing is proven to be responsible for less than 5 percent of deviations. For example, if an employee fails to notice defects because of lack of appropriate vision, shouldn't the organization make sure visual exams are performed regularly? Even in this example, we can see that a "fit for duty" system is weak.

On the other hand, training is usually used as a corrective action. Although training has proven to be effective for transferring knowledge, skills and abilities (KSA's), it will only work for new employees, new processes or to instruct on changes to existing processes if the employees that will perform the task lack the new KSA's. In fact, training is responsible for less that 10 percent of the deviations related to performance, yet most of the organization's efforts are directed towards less than 10 percent of the actual weaknesses. No wonder CAPA's related to training end up being so ineffective.

What Can Be Done?

The most effective way to control human error is to implement good systems. Systems take care of human factors (any aspect of the workplace or job implementation that makes it more likely for the worker to make an error) as well as external factors. We can start by:

  1. Providing clear, accurate procedures, instructions, and other job aids
  2. Implementing good human factors engineering for control systems, processes, equipment and work environments.
  3. Provide relevant training and practice
  4. Provide appropriate supervision
  5. Assure good communications
  6. Make sure the personnel have all the capabilities needed to succeed in the assigned task.

The Process

Understand the difference between explaining an event and explaining a human error. Once human error has been identified as a cause for the deviation, consider the human error itself as a new event that needs to be explained to assure conditions are identified and fixed.

Past behavior predicts future behavior if changes are not made. We need to perform an assessment of past events and assure the reasons for the error, besides the reasons for the event, are identified. Then we need to categorize these causes (conditions for error) in a systematic/uniform way. This allows you to analyze major contributors and, based on priorities, create an action plan that actually addresses these conditions. For example, if most of your human error events were related to incomplete procedures, then revising them and adding the missing instructions would be an action to consider.

Human error won't be eradicated unless we are able to really identify what is causing humans to err. If eliminating or "fixing" the actual individual eliminates or potentially reduces the probabilities of making that mistake again, then addressing the employee would be effective. If when we challenge ourselves, we can't categorically assure that this will correct the problem; intervening with the individual will only create a liability to the organization and we will end up in the same place we were at the beginning, trying to correct another individual's same mistake.

What do organizations need?

  • A structured human error investigation process
  • Consistent terminology (root causes)
  • Tracking/trending/monitoring system
  • Effectiveness based on root cause recurrence

CAPA's will be effective when we are able to eliminate conditions (causes) that make people deviate from the expected outcome. This is the reason why CAPA effectiveness should be measured by root cause recurrence as well as event recurrence. Most events, although different in nature, share the same causes. Real CAPA effectiveness will be achieved when the number of deviations decreases. not when particular events fail to reoccur.

This way we will not only be more productive but also fair to those that go to work with the intention to do a good job but end up being victims of weak systems.

Dr. Ginette M. Collazo obtained her PhD. in Industrial/Organizational Psychology from the Interamerican University of Puerto Rico. She possesses over 10 years of experience in the pharmaceutical industry with the Technical Training, Organizational Development and Human Reliability fields. Ginette has worked for Bristol Myers Squibb, Johnson & Johnson, Schering Plough and Wyeth. She has completed specialized studies in Human Reliability and has authored various publications on this topic. Having implemented Human Reliability programs in various industries, she understands firsthand what works and what doesn't when trying to change a culture from an error-tolerant environment to an error-free learning organization. Ginette has spoken about this topic in numerous professional conferences, including American Institute for Chemical Engineers, Center for Chemical and Process Safety, American Society for Quality, Interphex and FDAnews.