Many virtual blows have been exchanged regarding the claim that medical error is the third leading cause of death, but the data is highly variable, and distracts from the goal of reducing medical error and improving patient safety.

If you want to get into a Twitter fistfight, tweet something about medical error being the third leading cause of death in the U.S. Fights will likely erupt between those who regard the estimates as preposterously high, those who see it as an unfair indictment of physicians and hospitals, and those who are patient advocates.

Estimates of Medical Error Mortality

Various papers have been written to estimate mortality due to medical error, and these take a number of different approaches and perspectives. In 1999-2000, “To Err is Human” estimated that 44,000 to 98,000 deaths per year were due to medical error. There was an uproar, and many pundits dismissed it out of hand. Using a weighted average of four previous studies, a meta-analysis in 2013 “A New, Evidence-based Estimate of Patient Harms Associated with Hospital Care” estimated more than 400,000 deaths due to medical error per year. In 2016, “Medical error—the third leading cause of death in the US” estimated 251,454 deaths, based on 35,416,020 hospitalizations.

It is easy to look at the high level of variation between the studies, react to media hyperventilation over the upper bounds of the estimates, and simply denounce all the numbers as, at best, wild guesses, and at worst, an underhanded attack on physicians.

However, and as each of the authors takes pains to explain, the wide margins are primarily due to a very low maturity of measurement, not mischief on the part of the authors. The problem is that because we lack a reliable system that counts all deaths that are in some way due to medical error, we are instead required to rely on estimates.

Counts of Medical Error Mortality

Deaths in medical care are notifiable events, and with the advent of the International Statistical Classification of Diseases and Related Health Problems (ICD), we have a set of metrics to record those deaths. In ICD-10, we can count deaths ranging from Y40-59 “Drugs, medicaments and biological substances causing adverse effects in therapeutic use” to Y83-84 “Surgical and other medical procedures as the cause of abnormal reaction of the patient, or of later complication, without mention of misadventure at the time of the procedure”.

Table 1 shows the results of a query using the Centers for Disease Control and Prevention (CDC) Wonder tool for codes Y40-84. The yearly death count for the gamut of medical error in 1999-2016, ranges from a low of 2,411 (2010), to a high of 3,093 (2016), and a median of 2,540.

Comparing the count to the estimates, it is clear that even the sum of all death counts for the 17 years between 1999 and 2016 barely reaches the lowest annual estimate of the studies cited above.

If the estimates are, in the words of some, “a slander against doctors” for being outrageously high, clearly the counts amount to an outrageously low number. Very few people in the field of medical quality and safety would argue that 2,540 comes anywhere near the reality of annual death toll for medical errors.

What to Do?

The numbers are clearly not trustworthy. Neither the estimates nor the counts instill much confidence, and often lead to a degree of “action paralysis.” People easily become so distracted by the high variation and media outrage, they focus on denouncing the numbers when it might be better to focus on improving safety and quality. I argue that at this point, the specific number is not as relevant as the fact we know that people are dying of medical error and that we have the tools to reduce the number. Regardless of what the true numbers are, I do not think anyone feels that we are anywhere near a good number.

My take is that the best approach to these numbers is not to throw a hissy fit take umbrage that the estimates are unbelievably high, nor that the counts are ridiculously low, but to chalk the upper number and the variance between the studies on the whiteboard as the targets we want to reduce, and then get serious about reducing both measurement variance and the total number of deaths due to medical error.

Using quality-improvement tools such as the “five why’s” and the Ishikawa “fishbone” diagram, we can drill down to find the root causes of errors leading to fatalities. Since there are typically orders of magnitude more near misses than fatalities, it is practical and effective to focus on near misses. Once a root cause of a near miss is identified, we can look for ways to eliminate it in a systematic and progressive manner. We should also look hard at how error is reported, and how to capture medical error as a contributory cause. As we get more precise and accurate numbers, we can revise the estimate on the whiteboard until we trust it, but in the meantime, we would have continued to eliminate causes of near misses and reduce deaths from medical error.

Broaden the Perspective

One effective way to identify causes of near misses, is to make patients part of the quality team. Patients often see healthcare workflow from a different and very intimate perspective, which can enable them to identify near misses that might otherwise go unnoticed. Patient safety could greatly benefit from more input from patients on where near misses occurred.