Of course, it would be easy to focus the blame on the previous mechanic, who at best was grossly negligent and at worst was committing outright fraud.  But what interests me more than fixing the blame is preventing a similar breakdown from happening again.  I certainly won’t use that particular mechanic again, but what prevents the same problem from occurring with a different mechanic?  Even the best mechanic could forget to do part of a job on a particularly bad day.

What I would like is to verify that the work was done, but that is difficult in the case of a timing belt change, since the affected parts are behind a cover and not visible.  Besides, I probably don’t have the experience to tell by sight whether the job was done.  I have a lot more confidence that the second mechanic replaced all of the parts, because he showed me the old parts.  An effective solution for future repairs might be to always ask for the old parts, just so that I can know whether they remembered to perform the entire job.

A group of about 20 high school students were travelling across the U.S. by air for a school-sponsored activity.  The students had been split between two different flights, with 10 students on each flight.  Since there was only one adult in the group, the 10 youngest students were put on the flight with the adult, because the younger students seemed more at risk if their flight were delayed or diverted.

As it happened, the flight for the older students was cancelled, so those students were left to fend for themselves until the next available flight the next day.

Someone who is concerned with risk and reliability will ask, why were the students split into two groups, resulting in half of the students being stranded without an adult?  Did the airline only have enough seats for half of the group on each flight?  It was suggested to me that, as a matter of school policy, large groups were often split in two, so that if one of the planes crashed, then the number of students lost in the tragedy would be reduced.

Would such a policy make sense from a risk management perspective?  The following Cause Map shows the basic root cause analysis for losing all of the students in a plane crash, and how separating the students into two groups prevents (that is, reduces the risk of) losing all of the students.

However, splitting the students into two groups greatly increases the possibility of some students being stranded without an adult, as shown in the Cause Map below.

The missing piece of this risk management puzzle is the probabilities that the causes in the Cause Maps occur.  Risk can be quantified by multiplying the consequence of a failure times its probability.  Over the past year, 1.63% of U.S. plane flights have been cancelled, so if the students are split into two groups, the unescorted group will become stranded at least 1 time in 100.  On the other hand, the probability that a plane flight will crash is on the order of 1 in 10,000,000 and, contrary to what you may think, it is rare that more than a quarter of the passengers die in a crash.

Creating a policy to split up the students is a possible solution to the problem of losing all students in a single crash, but now that the risks are better understood, is it a very effective solution for reducing the overall risk to the students?

I don’t understand how only 1.2% of hospitals can have a death rate higher than the national average (shouldn’t 50% be higher than the average?), but what was most interesting to me was that the article mainly focused on how hospitals in richer, more populated areas tend to have slightly lower death and readmission rates.  I’m sure that the death rate is, in part, determined by causes related to household income, such as the types of food people eat and how much they exercise.  However, a hospital can’t necessarily control how high the quality of life is in its neighborhood.  A root cause analysis of why one hospital’s death rates are higher than another’s should probably focus more on the things that the hospital can control, that is, the hospital’s internal processes.

Want to learn more about healthcare root cause analysis?  See our patient safety blog.

Sometimes there is disagreement about whether the evidence actually supports a particular possible cause. Recent reports from the Flight 447 investigation state that two of the passengers may be linked to terrorist groups. This evidence seems to support the possibility of an in-flight explosion due to terrorism. On the other hand, some have noted that there was a lot of oil found on the surface of the ocean in the vicinity of the crash.   This would suggest an in-flight explosion could not have happened because the oil would have burned up while in the air. This is the point in many root cause analysis investigations where heated arguments begin, and the leader of the investigation might be tempted to just give up.

The key to resolving the argument and making some progress in the investigation might be getting into more detail in the chain of cause-and-effect. In this example, the argument is whether the cause “In-flight explosion” produces the effect “No oil on water.” If this cause-and-effect relationship is accurate, then we should be able to add more causes in between to explain step-by-step how an in-flight explosion results in no oil on the water. Laying out the causes and effects in detail will allow focusing the discussion on what specific cause-and-effect is in doubt and what specific piece of evidence is needed to resolve the disagreement.

The reason for assigning incidents and problems to predefined categories is to collect data on the types of problems that have occurred and establish trends in the numbers of problems occurring in each category.  These trends could highlight systemic issues with the organization’s investigation program or with the organization as a whole.  However, determining which category should be used for a particular problem is not a root cause analysis.

The purpose of a root cause analysis is to determine the best solutions to prevent problems from occurring.  Finding the best solutions requires identifying all of the causes of the issue and getting specific enough about the causes to be able to develop specific, effective solutions.  A root cause analysis cannot stop at causes such as “personnel error,” but must (1) delve into the underlying causes behind the personnel error and (2) ask what else had to happen, besides the personnel error, for the incident to occur.

Once a thorough root cause analysis is completed, then the most appropriate category (or categories) for the incident can be determined based on a complete understanding of all of the causes of the issue.

Similarly, there are many different tools for performing a root cause analysis.  At a fundamental level, a root cause analysis is uncovering “why” something happened by breaking down the issue into its causes and effects.  I like the Cause Mapping method for root cause analysis because it shows how the causes and effects of an issue are connected, without adding confusing terminology or complicated rules.  But just because a Cause Map is clear and uncomplicated doesn’t mean that it lacks the power to unravel the most complex incidents.