Root Cause Analysis

An accident investigation is the first step in a fact-finding process aimed at avoiding future accidents. It should determine what, why and how an accident happened. Its purpose should not be to blame someone. A good accident investigation will establish the failure modes of the fundamental contributing factors.

Appropriate accident investigations often confirm that many small, less serious accidents occurred earlier as a result of similar system failures. An accident investigation offers the chance to learn a great deal about the fundamental contributing factors present during an accident and thereby increase the opportunity to intervene in the interest of safety. To be useful, accident investigations must be an honest attempt to establish the facts.

There are numerous accident investigation systems in use in the industry. One of the more modern types of accident investigation systems is the root cause analysis, sometimes called the root cause failure analysis. The procedure consists of a set of processes through which the underlying causes of adverse outcomes may be identified. The goal of the investigation is preventing the reoccurrence of such events, according to the United States of America's Joint Commission on Accreditation of Health Care Organizations (JCAHO) (1996).

There are many different processes by which root cause analyses can be performed and the engineering and industrial risk management literature is rife with arguments for and against the different approaches. It is not the purpose of this article to explore those differences.

Root Cause Analysis is designed to systematically evaluate the possible ways that a loss (or series of losses) could have occurred. During the process the investigator collects and arranges factors in such a way as to rule out possibilities and develop a technically feasible loss scenario. The investigator should identify the key causal factors and fundamental contributing factors of the loss scenario. In conclusion the investigator should also produce recommendations to correct the fundamental contributing factors, thus preventing the recurring loss and similar future losses from common contributing factors. This approach may seen to be sufficient to prevent future accidents, but unfortunately the methodology focuses only on a singe cause-effect outcome.

One area of undisputed agreement is the observation that without strong support by upper management, root cause analyses will be performed in a mechanical manner, with the singular purpose of meeting regulatory requirements.

Most real-world events do not follow a simple cause-effect trail. A single factor may have multiple consequences. A combination of factors may bring about a single result, or they may initiate multiple effects. Causes can themselves have causes, and effects can have subsequent downstream effects. The failure mode should also be considered in all of these models.

© 2009 Carl Marx