Guest Column | July 17, 2019

4 (More) Microbiological Root Cause Analyses Lessons From Sherlock Holmes

By Paula Peacos, ValSource, Inc.


When a microbial data deviation occurs, microbiologists often struggle to determine the definitive root cause. Even the most experienced microbiologists find the lack of direct evidence and the fact that they are usually dealing with a past event to be major impediments to conducting a successful root cause analysis.

Sir Arthur Conan Doyle’s famous fictional 19th century detective, Sherlock Holmes, was a master of root cause analysis. There was not a case that Holmes could not solve. He achieved such great success by identifying and fully evaluating all available evidence, even the most absurd, and considering nothing else. Part 1 of this two-part article presented some of Holmes’s most famous quotes and demonstrated their applicability to modern day microbiological root cause analyses by prompting an investigator to recognize and fully consider more obscure and often overlooked sources of potential error and deviation.

There is, however, another major factor that can derail an investigation, and that is bias. Bias can take many forms and is often not easily recognized by the investigator. Holmes was keenly aware of its potential impacts and so did not allow bias generated through speculation, random hypothesis, or previous experience to determine or redirect the course of his investigations. Here in Part 2, we will see how applying more of Holmes’ famous observations and investigative principles can uncover hidden bias and greatly increase an investigator’s chance of finding that elusive definitive root cause.

“There is nothing more deceptive than an obvious fact.” – Sherlock Holmes, “The Boscombe Valley Mystery”1

We tend to accept things we believe are true without giving them due consideration. For example, while we need to be able to trust our validations and qualifications, sometimes things go wrong that have nothing to do with the study that was performed. The study is sound, but the situation in which the deviation occurred is different somehow from that which is reflected in the study. While we need to accept known facts as facts, we also need to consider circumstances that can affect them. Failing to consider extenuating circumstances is a form of bias.

Some assumptions that are commonly accepted without further due diligence include the following:

Our sanitization methods are validated and are proven to be effective.

Sanitization methods are effective only if they are executed properly. Could there have been an error in formulation or application? Is there a new member of the cleaning crew? If the procedure requires that no one enter the cleaned area until the floors, etc. are completely dry, are you certain that individuals are not entering the cleaned areas sooner than they should be?

Furthermore, no agent kills everything. If a new organism is recovered or there is an increase in recovery of a known organism, it should be verified that the organism was tested as part of the facility disinfectant efficacy study. It may also be useful to determine if there is a new species or strain of the organism present, as species and strains differ in their resistance levels.

We have a validated overkill cycle, so nothing could have possibly survived.

Sometimes things go wrong that have nothing to do with the study that was performed. For example, consider a validated piece of equipment that is sterilized in place (SIP). If a valve is failing to fully open or close during the sterilization cycle, it is possible that part of that valve may not be exposed to the steam or may not reach sufficient temperature, allowing some of the organisms present to survive. Worse yet, this failure to fully open or close can be intermittent, making detection of the failing part difficult to detect. This becomes an even bigger problem if there are multiple pieces of identical equipment or processing trains and the parts are not dedicated for each unit. Contamination events like these are often investigated as “isolated incidents,” when in fact they are not. Since the issue will not be resolved until that part is discovered and replaced, it is critical to consider such a possibility in the root cause analysis (RCA).

It’s a closed system (or “functionally” closed).

A system is only closed until there is a breach, and breaches can take many forms. If a piece of equipment is cleaned or sterilized in place, the expansion and contraction that occurs during heating and cooling can result in loosening of valves and connections, and minute breaches can develop. During the cooling process, a vacuum is created. If there is a minute breach, a microorganism can be sucked inside the system. Microorganisms only need an opening large enough to accommodate their size; therefore, the breach may not be visible to the naked eye. This scenario is not only applicable to equipment that is SIP, but also items that are removed from an autoclave while still hot. Vibration can also loosen connections and result in a breach. If there is the slightest bit of moisture present, it is possible for a microorganism to follow that path into the interior of the sterilized equipment.

We didn’t detect any contamination at any point upstream in the process.

As it is impossible to test an entire batch for microbiological contamination, representative samples are collected and evaluated. In-process samples are usually extremely small compared to the overall volume of the batch. Complicating matters, microbiological samples are not homogeneous, and improper sampling technique or processing can exacerbate this issue. If a very low level of contamination is present, it may go undetected until the concentration reaches the limit of detection of the assay.

The environmental monitoring (EM) data showed no recoveries.

Environmental monitoring is a critical tool for estimating the microbial populations present in the manufacturing environment. The quality of data generated by an EM program is dependent on many things, including, but not limited to, the design of the program, the number and location of sampling sites, the frequency of sampling, and the ability of the sampling method to recover microorganisms from the environment.

It is estimated that surface sampling methods recover less than 50 percent of what is present on a given surface.2 From a practical standpoint, it is also impossible to sample 100 percent of the surface area of a given room or piece of equipment. Complicating matters, organisms in the manufacturing environment are often severely stressed due to the inhospitable environment and may not grow on the selected microbiological media. (These are known as viable but not culturable, or VBNC.) Finally, an operator can collect a sample from a given site, and immediately afterward, a second operator can touch that site with a contaminated glove. Therefore, it is important to remember that an individual environmental sample represents only a single moment in time and nothing more.

The trending data shows it’s an isolated incident.

Like the EM program, the trending program is only as good as its design and the quality of the data analysis. Is adequate consideration given to sub-alert trends and aberrant data, or is trending performed solely based on excursion rates? Trending solely by excursion rates only demonstrates how many times a single sample yielded a microbial count that exceeded an established alert or action limit. Are overall microbial counts rising in a particular area, or are counts occurring in sampling locations where they had not before? Trending by recovery rate in addition to excursion rates can help to reveal these things.

The operators’ gloves were clean. No growth was observed.

Most manufacturing and testing procedures require operators to frequently sanitize their gloves with an agent such as 70 percent isopropyl alcohol (IPA). Frequent sanitization is intended to remove any contamination present on the gloves so that it is not transferred to other surfaces. However, no evidence of microbial contamination on the gloves at the end of a process does not guarantee that the gloves were contamination-free for the entire duration of the process. Furthermore, usually only the fingertips are sampled, but the palm of the glove may contact critical tools such as forceps.

“Circumstantial evidence is a very tricky thing. It may seem to point very straight to one thing, but if you shift your own point of view a little, you may find it pointing in an equally uncompromising manner to something entirely different.” – Sherlock Holmes, “A Study in Scarlet”3

As mentioned previously, unless you are fortunate enough to observe the cause of the deviation at the time it occurs or have a video recording of the event, the investigator is left with circumstantial evidence to determine the root cause. If a root cause is determined, it is usually based on the sum of that evidence. However, as Holmes so astutely observed, circumstantial evidence can be a tricky thing.

We tend to interpret and accept or dismiss the evidence and draw our conclusions based on our own experience and similar past events. Therefore, the interpretation of the evidence can easily become biased. Conclusions drawn by two investigators from the same piece of evidence can be completely different. It is critical to remember that two deviations that appear to be identical may have entirely different root causes, and, for this reason, investigations are best performed by a team of subject matter experts who individually possess different experiences and viewpoints. In this way, all points can be considered and the correct conclusions drawn.

“It is a capital mistake to theorize before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts.” – Sherlock Holmes, “A Scandal in Bohemia”4

Sometimes we look at a deviation and assume there is no need to investigate further because the root cause appears to be obvious. This is an all-too-common form of bias. When one begins to investigate the finer details, we often find that the “obvious root cause” is completely incorrect.

Many years ago, as a newly hired entry-level bench analyst, I was asked to use a qualified working stock culture to prepare a suspension of Enterococcus faecalis in sterile water. The suspension was to contain 10 to 100 colony-forming units per 0.1 mL for use in growth promotion testing of microbiological media. Although I was a new hire, I had substantial prior experience in preparing this type of suspension.

I carefully prepared the suspension according to procedure, verified the final concentration, and promptly stored the suspension at 2 to 8°C. When the suspension was used about three days later, the count had increased exponentially. Assuming I had somehow made an error in preparation, I repeated the procedure and obtained the same results. I had recorded each step and all associated calculations for both attempts, and the two sets of data were identical.

Because I was a new hire, and this was a routine lab procedure using an existing qualified stock organism, the lab supervisor was certain that I had somehow made an error, even though the documented evidence indicated no error. The supervisor had a second analyst observe as I prepared a third suspension. Again, the same results were obtained, and no error was detected. Finally, the supervisor had the second analyst prepare a fourth suspension. Only when the second analyst’s preparation also grew exponentially did the supervisor consider that perhaps something had changed with the working stock culture that allowed it to rapidly propagate in sterile water at 2 to 8°C. Obtaining a new stock culture eliminated the problem, which never occurred again.

This example clearly demonstrates how hypothesizing in the absence of hard evidence can actually prevent the investigator from identifying the true root cause of a deviation. It was perfectly natural for the lab supervisor to theorize that I had made an error in preparation because I was a new entry-level hire in the lab, and the procedure had been routinely performed by other lab analysts using the same stock culture with no issues. However, the error in the RCA occurred when the supervisor continued to pursue that theory even though the documented evidence clearly suggested otherwise. Had the supervisor accepted the documented evidence as fact instead of discounting it in favor of the “obvious root cause,” I would have been directed to prepare two suspensions, one from the working stock and one from a new stock. This would have clearly revealed the root cause to be an issue with the working stock and would have saved the lab a great deal of wasted time and effort.

The investigator must look at the available evidence and only the available evidence. It is critical not to summarily dismiss or accept any piece of evidence without thoroughly evaluating it within the context in which it occurred.

“It is an old maxim of mine that when you have excluded the impossible, whatever remains, however improbable, must be the truth.” – Sherlock Holmes, “The Beryl Coronet”5

In the end, sometimes the seemingly impossible is indeed possible. The investigator must consider all potential root causes, including the most unlikely, if that is what the evidence suggests. In the previous scenario, the supervisor did not consider the possibility that something had changed with the working stock suspension because although microorganisms can and do grow in sterile water at 2 to 8°C, the growth is usually minimal, as the water provides no nutrients and the cold temperature does not favor the growth of most organisms. This is accepted as microbiological fact and was further supported by the supervisor’s own extensive practical experience. Under those conditions, the organism should not have grown exponentially, but it did. It is also a well-known fact that microorganisms in general have the ability to mutate to survive unfavorable conditions. However, the possibility that the organism might have mutated was initially dismissed because it appeared to be less likely than an occurrence of human error, even though the evidence suggested otherwise.


By simply keeping these quotes in mind when performing microbial data deviation investigations, one can more easily and more accurately determine the definitive root cause. Determining the definitive root cause allows for identification and application of the correct CAPA, which can effectively eliminate or prevent a reoccurrence of the deviation in the future, resulting in overall process improvement and a better product for our patients.

In the end, however, it does not require a detailed RCA to determine why these quotes attributed to a fictional 19th century detective still hold perfectly true more than a century later. In this case, the root cause is indeed truly obvious. Doyle was a genius.


  1. Doyle, Arthur Conan. “The Boscombe Valley Mystery,” The Adventures of Sherlock Holmes. The Complete Sherlock Holmes and Tales of Terror and Mystery, Signature Edition authorized by the Conan Doyle Estate, Ltd., The Complete Works Collection, Digital, 2012
  2. General Chapter 1116, Microbiological Control and Monitoring of Aseptic Processing Environments. USP 42- NF 37, 2019, U.S. Pharmacopoeia, Rockville, MD. Accessed on June 7, 2019 from
  3. Doyle, Arthur Conan. “A Study in Scarlet.” The Complete Sherlock Holmes and Tales of Terror and Mystery, Signature Edition authorized by the Conan Doyle Estate, Ltd., The Complete Works Collection, Digital, 2012
  4. Doyle, Arthur Conan. “A Scandal in Bohemia.” The Adventures of Sherlock Holmes. The Complete Sherlock Holmes and Tales of Terror and Mystery, Signature Edition authorized by the Conan Doyle Estate, Ltd., The Complete Works Collection, Digital, 2012
  5. Doyle, Arthur Conan. “The Beryl Coronet,” The Adventures of Sherlock Holmes. The Complete Sherlock Holmes and Tales of Terror and Mystery, Signature Edition authorized by the Conan Doyle Estate, Ltd., The Complete Works Collection, Digital, 2012

About The Author:

Paula Peacos is a senior consultant with ValSource, Inc. She has over 25 years of industry experience as a microbiologist, working for contract manufacturing organizations as well as small, midsize, and large pharmaceutical organizations. Peacos has extensive experience in aseptic processing, API/drug substance manufacturing, cell therapies, nonsterile production (both clinical and commercial), microbiological laboratory management, and performing compliance audits internationally. She is an experienced trainer and has developed and implemented customized developmental and remedial programs. Peacos has also published articles and delivered presentations at industry meetings on topics such as risk assessment and using microbial recovery rates for trending analysis. You can contact her at