A report published in May from researchers at Johns Hopkins claims that medical errors are the third leading cause of death in the U.S., behind only heart disease and cancer.
According to the researchers, medical errors account for 251,454 U.S. deaths each year – and they regard this figure as an underestimate.
That’s the sort of finding that makes headlines. Indeed, you might have read about this report in the newspaper or even seen it reported on the evening news.
But as we’ll argue, the methods the researchers used to draw this conclusion are flawed, and that means that the conclusion that medical error is the third leading cause of death is highly questionable.
When a report like this gets broad media coverage, it can foster unwarranted mistrust of medicine, which could prevent people from seeking needed care – a concern to everyone who takes care of patients.
What’s wrong with the methodology?
A medical error can be defined as a decision or action that results in patient harm and that experts agree should have been made differently, given the information available at the time. But applying such a definition in reviewing patient records is fraught with difficulty.
The study’s authors argue that death certificates should be redesigned to recognize that more deaths are attributable to medical error. That’s a reasonable suggestion. But the implication of many media reports that these findings prove hundreds of thousands of people are dying each year due to medical errors is highly problematic.
First, the authors of the Johns Hopkins report did not collect any new data. Instead, they based their conclusions on studies performed by other authors. There is nothing wrong with that in principle.
But in this case, the results are highly misleading because they are based on large extrapolations from very small data sets. The authors based their conclusions on four studies that included a total of only 35 deaths attributable to medical error out of nearly 4,000 hospital admissions. Extrapolating from 35 deaths to a population of 320 million is quite a leap.
In addition, these studies frequently do a poor job of distinguishing between adverse events and errors. They are not the same thing.
An adverse event is defined as any undesirable outcome after a drug or treatment is administered to a patient. Every medical test and therapy – from antibiotics to surgery – is associated with some risk of an adverse outcome. Adverse events can include death, although that is rare. While every adverse outcome is regrettable, it does not prove that an error was made – that based on what was known at the time, a medical professional should have made a different decision or acted in a different way.
Physicians typically cannot know in advance which patients will experience such reactions, so attributing such deaths to error is misleading.
There is another problem with the Hopkins report: two of the four studies it draws on use Medicare data, which generally include patients advanced in years, in relatively poor health and being treated in the hospital. Sad to say, many such patients are at substantially increased risk of death to begin with. Many will die during their hospitalization, no matter how well they are cared for. To attribute such deaths to error is to fail to account for the inevitability of death.
In fact, one of the studies on which the Hopkins report is based even includes a prominent correction factor. The author estimates the number of deaths due to medical error at 210,000. Then, based on the fact that the tools used to identify errors are imperfect, the author chooses to double his estimate of the number of deaths due to error to 420,000.
The sort of medical chart review used in these studies is radically different from caring for patients. The uncertainty and stress associated with caring for the very sickest patients are often invisible to hindsight. Seriously adverse patient outcomes are associated with a greater tendency to blame someone. When a patient has died, we want someone to be responsible, even if every action taken appeared justifiable at the time.
Other research suggests many fewer deaths from medical error
This isn’t the first study to try to assess how often medical errors can lead to death. Other studies paint a very different picture of the number of deaths attributable to error.
In one responding to claims of very high death rates due to medical error, physicians reviewed 111 deaths in Veteran’s Affairs hospitals, attempting to determine whether such deaths were preventable with “optimal care.” VA patients are generally older and sicker than the U.S. population, and thus somewhat comparable to studies based on Medicare data. Also, by using “optimal care,” the study may catch even more deaths than the “medical error” standards, resulting in a tendency to overestimate the number of deaths due to error.
At first, the researchers estimated that 23 percent of deaths could have been prevented. But when they were asked whether patients could have left the hospital alive, this number dropped to 6 percent. Finally, when the additional criterion of “3 months of good cognitive health after discharge” was added, the number dropped to 0.5 percent. Preventable deaths should be viewed in context, and there is a big difference between preventing death and restoring good health.
Applying the rates from the VA study to U.S. hospital admission data, medical error would drop down to number 7 of the top 10 causes of death in the U.S. Applying the additional criterion of three months of good cognitive health, medical error would not even rank in the top 20. Of course, doing so runs the same risks as the Johns Hopkins study; namely, extrapolating from a small study to the entire U.S. population.
To produce a truly balanced account of medicine’s role in causing death, it would be necessary to account not only for the risks but also the benefits of medical care. Many patients with heart disease, cancer and diabetes whose deaths such studies attribute to medical error would not even be alive in the first place without medical treatment, whose benefits vastly outweigh its risks.
Looking at medicine from this point of view, we are fortunate to be living in an era of unsurpassed medical capabilities, when the profession is doing more to promote health and prolong life than at any time in the past.
Perhaps the strongest evidence that such studies overestimate the role of medical error is that the fact that, when causes of death are ranked by authoritative organizations such as the U.S. Centers for Disease Control and Prevention, medical error is not even included in the top ten. Would adding medical error to death certificates change this? We doubt it.
There is no doubt that mistakes occur in medicine every day, and if we take appropriate steps, error rates can be reduced.
But inflated estimates of the number of deaths associated with error do nothing to advance understanding and may in fact make many patients more reluctant to seek care when they need it. A blinkered focus on error, without corresponding accounts of medicine’s benefits, contributes to a distorted understanding of medicine’s role in health and disease.
This piece was coauthored with Jae Hyun Kwon, a student at the Indiana University School of Medicine.