If you’ve seen the headlines, then you know that this open letter to the CDC from three medical students and Martin Makary at Johns Hopkins has gained a lot of media attention. In it, they argue that the CDC should allow doctors to list medical error as a cause of death. So far so good.
Then they argue, through the power of contrivance, that medical error is the third leading cause of death after heart disease and cancer but above COPD (emphasis mine):
We define death due to medical error as death due to 1) an error in judgment, skill, or coordination of care, 2) a diagnostic error, 3) a system defect resulting in death or a failure to rescue a patient from death, or 4) a preventable adverse event. The prevalence of death due to medical error leading to patient deaths has been established in the literature. From studies that analyzed documented health records, we calculated a pooled incidence rate of 251,454 deaths per year.(1) If we project this quantity into the total number of deaths in the year 2013 (2,596,993 deaths), they would account for 9.7% of all deaths in the nation.
Wait, what? All of the medical error data is slightly bullshitty, and doubling down on it to “calculate a pooled incidence rate” for this purpose is no different. Medical errors that occur before death do not necessary cause that death (correlation and causation). Some patients who die “due” to medical error are so sick that the medical error is not the prime (or even secondary) culprit. Perhaps being on death’s door and requiring high level and Herculean care is a risk factor to experiencing medical error. Not all bad surgical outcomes, hospital acquired or postoperative infections are “errors” even if someone labels them as “preventable.”
The accompanying article in BMJ is longer but doesn’t help with the underlying math (emphasis mine):
A literature review by James estimated preventable adverse events using a weighted analysis and described an incidence range of 210 000-400 000 deaths a year associated with medical errors among hospital patients.16 We calculated a mean rate of death from medical error of 251 454 a year using the studies reported since the 1999 IOM report and extrapolating to the total number of US hospital admissions in 2013.
Associated. I have to imagine that this “analysis” is for dramatic effect, to spur popular outrage and official (re)action. It’s like projecting the cancer risk of CT scans based on the atomic bomb survivors exposures, dividing to estimate the risk of a single CT, and then multiplying to get the total risk of all CTs. Sure, it could be right by coincidence, but it’s certainly not rigorous or even definitely based in reality.
The premise that we should actually know if we kill someone is important and makes sense (though telling people to actually put down “medical error” on a death certificate for their patients given the malpractice climate might be a tough sell and means that I imagine in real life it would often be reserved for egregious cases). I personally think that while this sort of presentation may generate discussion, it actually cheapens an important topic within medicine. Medical error is important and we must do more to track it, but tying a number to it in this manner is almost arbitrary.