Root cause

A root cause is an initiating cause of a causal chain which leads to an outcome or effect of interest. Commonly, root cause is used to describe the depth in the causal chain where an intervention could reasonably be implemented to change performance and prevent an undesirable outcome.

The term root cause has been used in professional journals as early as 1905, [ cite journal | year=1905 | date = 18 November 1905 | volume = 166 | title=The Present State of Medical Practice in the Rhondda Valley| journal=The Lancet | pages = 1507 | doi=10.1016/S0140-6736(00)68499-4] but the lack of a widely accepted definition after all this time indicates that there are significantly different interpretations of exactly what constitutes a root cause.

The two biggest differences in viewpoint regard the possibility of an outcome having more than one root cause.

Single cause

The single cause philosophy is based on the belief that there is a single cause for any outcome that, if prevented, would prevent the outcome itself. In this context, the root cause is the cause which dominates over all other contributing factors.

This viewpoint results in the identification of a single root cause that provides a clear direction for preventing an undesired outcome. The subjective criteria used for selection of the root cause from among the contributing factors has been criticized as being arbitrary and inconsistent.

One basis for the argument supporting this as the ‘proper’ interpretation is the decomposition of the words in the phrase – the root cause is the cause at the root of the outcome. While there may be nuances in the meanings of the words, the common usage of the words lead to a straightforward and simple interpretation.

It should be noted however that the notion of a single root cause is the exception, not the rule. Fact|date=July 2008

Multiple causes

The multiple cause philosophy stems from the belief that a root cause can exist for each of the contributing factors that were necessary for a resulting outcome. By preventing any of those necessary causes, the undesired outcome can be prevented.

The result of this philosophy is a branching model that attempts to incorporate all the identified ways that the outcome could be prevented. The inclusive model provides a variety of corrective actions that can potentially break the causal chain.

One basis for the argument supporting this as the ‘proper’ interpretation is the common illustration of the model with the undesired outcome at the top and the causes spreading below like roots spreading from the trunk of a tree.

Application

Effects have causes. The causes may be natural or man-made, active or passive, initiating or permitting, obvious or hidden. Those causes that lead immediately to the effect are often called direct or proximate causes (see proximate causation). The direct causes often result from another set of causes, which could be called intermediate causes, and these may be the result of still other causes. When a chain of cause and effect is followed from a known end-state, back to an origin or starting point, root causes are found. The process used to find root causes is called root cause analysis.

The usual purpose of attempting to find root causes is to solve a problem that has actually occurred, or to prevent a less serious problem from escalating to an unacceptable level (see Near miss (safety), for example). The basic concept is that solving a problem by addressing root causes is ultimately more effective than merely addressing symptoms or direct causes. Consider the following example, where root cause $a$ leads to effect $e$, with a few intervening steps.

:$a\; o\; b\; o\; c\; o\; d\; o\; e$

Assume each of these factors is as described below:

* $e$: car will not start
* $d$: battery is dead
* $c$: alternator does not function
* $b$: alternator is well beyond its designed service life
* $a$: car is not being maintained

The effect, $e$, could be prevented by addressing any of the other factors. For example, attaching jumper cables from another car (addressing factor $d$) will probably allow the problem-car to be started. However, this solution is not likely to provide long-lasting relief from the undesired effect, as factor $c$ will ensure that the car shuts down again in a very short period of time. Addressing factor $c$ by repairing the alternator may solve the problem for a longer period, but factor $b$ will eventually result in another age-related breakdown in the alternator. The alternator could be replaced with a new unit, addressing factor $b$, thus allowing the car to be driven for an extended period of time. However, factor $a$ will eventually ensure that the car breaks down again for some other reason. Clearly, a better solution to the problem (and many other potential problems) is to maintain the car properly, which addresses factor $a$, the root cause.

Note that the preceding example highlights one difficulty with root cause analysis: knowing when to stop. That example could have been carried further to ask why the car wasn't being maintained, and then why the vehicle was designed such that this maintenance was even required. It is often the analysis' frame of reference that determines where the stopping point ought to be. For instance, if the example is viewed from an individual vehicle owner's frame of reference, then factor $a$ may represent a valid stopping point. However, if the frame of reference is moved to the vehicle manufacturer, dealing perhaps with hundreds of thousands of such problems, the proper stopping point may indeed lie in the realm of design.

So as you can see, the root cause is a function of who owns the problem and what corrective action they choose to prevent recurrence. This perspective holds that any root cause is relative and can not be determined until the owner attaches a solution to it. The solution must prevent recurrence, meet the owners goals and objectives, and be within the owner's control to implement.

An issue closely related to solving an existing problem is to foster learning that will embed knowledge (within a person, group, or organization) that may help prevent similar problems from occurring in the future. Such knowledge is often referred to as "lessons-learned". Gaining such knowledge, retaining it, and using it effectively is one of the goals of a learning organization engaged in continuous improvement. [ cite journal | last=Cooke | first=David L. | year=2003 | title=Learning from incidents | journal=Proceedings of the 21st International conference of the System Dynamics Society | url=http://www.systemdynamics.org/conferences/2003/proceed/PAPERS/201.pdf ]

There is little agreement as to the types of conditions that can reasonably be considered root causes. One view holds that, in theory, one would have to return to the Big Bang or point of Creation to find true root causes. An alternate viewpoint is that one need only consider factors within the boundary of the system that exhibits the problem. The former is usually used as one argument against attempts to single out specific factors as root causes, while the latter (or some version of it) is usually proposed as a practical bound within which useful information can be obtained. [ cite book | last=Davies | first=John | coauthors=Alastair Ross, Brendan Wallace and Linda Wright | title=Safety management: A qualitative systems approach | publisher=Taylor and Francis | date=2003 | month=August | location=London | id=0415303710 ]

Practitioners of root cause analysis often define what the phrase "root cause" means for a particular setting and application. The benefits of finding deeper layers of root cause tend to diminish after a certain point. The practical application of root cause analysis therefore often searches only as long as the benefit of answers outweighs the effort of the search.

References

See also

* Root cause analysis
* Causation
* RPR Problem Diagnosis