Skip to content

From Training to Human Reliability: Lessons Learned during the Implementation of a Human Reliability Function in the Pharmaceutical Industry

Pharmaceutical manufacturing sites continuously compete, even within the same company to maintain production volumes and avoid closures. To attain a status of a high performing organization manufacturing sites are implementing strategies and practices like operational excellence and performance reliability. To achieve this goal, innovative approaches to traditional processes need to take place. This paper discusses how by redesigning the training function and by focusing on the expected result (human reliability) instead of the how (training), a reduction of more than 60% of human error events in pharmaceutical manufacturing facilities was achieved.

The Problem: Reduce Human Errors

The ultimate objective of a Training function is to assure that people do things right, people do the right thing, and people do it efficiently. Training, being the process intended for transferring knowledge, skills, and abilities indeed play a vital role in the way people perform. Nevertheless, people’s behaviors are influenced by a lot more. Frequently we are faced with the reality that people, even though they know what to do to either lose weight, lower their cholesterol, or be healthy just don’t do it. Everyday people engage in risky behaviors, even though they are knowledgeable of the possible consequences. Knowledge by itself does not modify behavior.


Organizations want people to be reliable when they are executing day to day operations. Human reliability defines the ultimate goal of any training event. By changing the focus to reliability instead of training, efforts are directed towards the right actions. A human reliability approach focuses on all of those elements that could impact behavior besides the individual himself, like the clarity of instructions, the sequence of actions, clear communication, resources availability, layout designs, and work tools, among others. To ensure flawless execution, organizations could broaden the scope of the training groups from a traditional training function focusing on learning strategies exclusively as a mean to improve performance, to one that addresses other areas that impact behavior as well.


To ensure the correct actions are implemented a diagnosis was performed. This diagnosis consisted of the analysis of failure investigations. A random sample was selected for a 12-month period. These investigations were evaluated considering the following criteria: correct problem defined, correct root causes/causal factors identified, and appropriate immediate, corrective and preventive actions. This step consisted of challenging and validating each of the five elements considered as acceptance criteria. After performing this evaluation, corrections were made to the investigations and events where the criteria were not met.

Once all the correct causal factors/root causes were identified for all events, these were categorized and coded in four iteration levels (as a deductive process from a general to specific view). The first level identified the causal factor type; the second identified the cause category, the third level identified the near cause, and the last level the final cause (s).

Data from six manufacturing processes were considered in the data analysis.


  • More than 80% of failures were related to errors.
  • 8-10% of error events were related to training (knowledge, skills, and abilities).
  • More than 50% were related to procedures, mostly incomplete procedures or confusing procedures.
  • Less than 10% were related to “the individuals” executing the tasks
  • Approximately 30% of events were distributed in other categories like equipment failures, poor administrative management systems, supervision, and communications.

Findings of the investigation process

  • Problem is identifying the event/failure.
  • Identifying one root cause, but failing to identify multiple factors that contribute or cause the failure.
  • Confusion is determining the number of iterations (depth) of the root cause/causes determination process.

Findings in the Correction and Prevention Approach

  • Immediate actions were used in place of CA’s (corrective actions).
  • CA’s (corrective actions) were used in place of PA’s (preventive actions).
  • There were no real preventive actions.
  • Most events used training as a corrective or preventive action for human performance deviations.
  • Trending and tracking mechanisms were non-existent.
  • There were no prediction actions/initiatives.


Human error constitutes a considerable number of failures in manufacturing facilities in the pharmaceutical industry, impacting efficiencies. Training is very effective for preventing errors when there are new employees, new processes and/or changes are made to existing processes. Training has not shown to be effective on failures that are not associated with either a lack of knowledge, skills, and abilities or when the level of awareness needs to be increased on critical tasks. Nevertheless, the training method and contents need to be customized depending on the intention. For skills and abilities, on the job or practical training is appropriate. To increase the level of awareness, people need to understand the consequences of failure and science or reasons for specific requirements of a task. On the other hand, the data indicate that training is responsible for a limited number of deviations or failures. This evidences the fact that, since pharmaceutical manufacturing sites mostly use training to correct or prevent these failures, lots of resources and effort are being placed in a very limited part of the problems, affecting not only the efficiencies, processes, costs and the individuals involved but also the reputation of training professionals that are not being “effective”.

Also, most events are identified as procedure related, either because even though there is much information included in these documents, or the, important information is missing. This can also be considered as a systemic problem, which in turn would be categorized under management systems related to document formats and risk/task reviews.

Another element to be considered is the fact that less than 10% of the events were related to the individual involved. This is particularly important because most organizations consider a human error to be the individual’s fault. This piece of information returns the responsibility of avoiding mistakes to the organization. By designing/ re-designing systems and conditions, failures can be prevented.

Organizational Actions Taken

  • Created an interdisciplinary group that had the necessary knowledge to address human-process interface (hardware, equipment, components, systems and thinking process). These professionals included engineers, psychologists, scientists, teachers and other technical experts.
  • Provided training on human behavior, instructions development, training and investigation processes:
    • Psychology and physiology of human behavior.
    • Human Error prevention techniques.
    • Risk Assessment
    • Procedures design and technical writing.
    • Training design and delivery, and training effectiveness evaluation techniques.
    • Investigation and root causes determination process.
    • Organizational design.
    • Statistics
  • Incorporated Human Reliability experts into the deviation investigation processes.
  • Developed a human error investigation process in which Human Error was treated as an event. Not the cause of failures.
  • Created a corrective action and preventive action (CAPA) effectiveness program.
  • Created a metric mechanism for trending and tracking human performance deviations.
  • Created a cultural change/integration program to assure institutionalization.

Specific Actions Taken

  • Redesigned training courses to ensure training programs capture critical areas.
  • Revised critical procedures to ensure well-written work instructions that take into consideration the way people think and the mental processes that take place when decoding messages.
  • Identified critical tasks and steps to ensure training programs were designed highlighting the criticality and the consequences of each task/step.
  • Designed training that prepared people for managing unexpected situations.
  • Developed programs that considered elements like the level of attention, motivation and situation awareness needed for some tasks.


The 60% reduction in human error events was calculated based on the baseline rate (.047 or 4.7%) before the implementation of the above-mentioned actions. The baseline rate was calculated by dividing the number of human-related events (88) by lots manufactured (manufacturing) and tests performed (laboratories) (1869).