Counting is not Sesame Street easy in population health science. Case in point: in 1999, a study concluded that medical errors in the United States hospitals were killing 98,000 people a year, the equivalent of two jumbo jet crashes per day. Recently, a new study upped the estimate to 250,000. Never fear — Count von Count is here!
A big part of the counting challenge is deciding who counts in the count and who doesn’t count. In 1999, researchers were counting patients who were considered “high severity” – in regular words, very sick. So, it would be accurate to state that 13.8 percent of severely ill patients will be exposed to a medical error and will subsequently die (whether the error hastened their demise or not).
That’s not what came out of the 1999 study – instead, it stated that 13.8 percent of all hospital patients will be exposed to a fatal medical error. The people studied were on the edge of life and certainly not similar to your average Joe patient having an appendix removed or getting fluids for dehydration. Granted, all hospital care involves some risk of errors but very ill patients are getting more complex care and more of it, increasing the opportunity for errors.
The difference between average Joe hospital patient and severely ill hospital patient is more subtle than apples versus oranges. This explains why there are people in the world with PhD’s in applied biostatistics.
Far short of getting a PhD, you can always use the tried-and-true method of analyzing counts – ask whether the number makes sense. The 98,000 figure did not make sense and the 250,000, even less so. About 700,000 hospitalized patients die every year. If 250,000 is correct, then more than one third of all hospital deaths are from medical errors. This is hard to imagine.
Even if we grant that one-third of patients who die in the hospital had a medical error made, we would also have to say that the error caused them to die. Cause of death is very hard to trace in many cases; assigning responsibility to the medical error is also fraught with difficulty. For example, a cancer patient might get the wrong dose of a medication, unlikely to cause harm. The patient dies a week later; does the medical error have anything to do with his death?
This is not to say that medical errors are not serious or not worth paying attention to. It’s just that medical errors are not nearly as abundant and fatal as the 250,000 count would lead to you believe.
Today’s show brought to you by the number zero, which is the amount of faith we should give breaking-news population health counts until we get the details.