I heard from one of the local ambulance chasers that medical malpractice is the third leading cause of death. Can that many deaths be blamed on doctors and hospitals? My hunch is someone pulled the number in question completely out of their ass. —John

HANG ON. Before we peer into where this idea got pulled out of (and get sued for an unnecessary colonoscopy in the process), let’s get our terms straight. Medical malpractice isn’t what kills you, it’s what your survivors accuse the doctor of when they’re trying to collect damages.

Only about 11,000 such claims are paid out in the U.S. each year, they’re costly to pursue and hard to prove, and nearly 60 percent of total compensation goes to “administrative fees” (read: lawyers), so malpractice suits aren’t even a leading get-rich scheme.

No, the thing you’ll see cited among the top causes of death is medical errors, also known in the literature as “preventable adverse events”: when medical personnel do the wrong thing, or fail to do the right thing, or do the right thing but do it wrong.

This can often take the form of misdiagnosis, or miscommunication between various healthcare providers, or between providers and patient.

And it does happen with some regularity, and patients do die. How many, though? The report that really wound everyone up on this issue was released in 1999 by the Institute of Medicine, titled To Err Is Human. It relied in part on a study of 30,000 records from New York hospitals in 1984, which researchers used to calculate the rate of adverse events per hospitalization (3.7 percent), how many were due to negligence (27.6 percent), and how many led to death (13.6 percent), and then weighted the numbers to estimate figures for the state overall.

What the IOM authors did was to extrapolate these results to the total number of U.S. hospital admissions in 1997, 33.6 million, arriving at a high-end figure of 98,000 deaths and thus enabling the claim that medical error was the fifth leading cause of death for that year. In 2013, a NASA toxicologist turned patients’-rights crusader presented a new report based on more recent hospital data and came up with an even scarier estimate: 400,000-plus deaths due to preventable harm—good for a theoretical third place on the causes-of-death list, right behind cancer.

But even if the raw numbers behind the reports were absolutely correct, assigning all these deaths to medical error doesn’t really make sense. As critics of the 1999 report pointed out, that 13.6 percent of patients who died in the New York study all had life-threatening conditions in the first place, but the authors never establish a baseline rate for how many would have died anyway; they concede that had the adverse events not occurred, the life expectancy for many terminally ill patients wouldn’t have been improved, but don’t work this into their death figures.

Both the 1999 and 2013 reports get some of their data from tertiary hospitals—i.e., where people wind up when their problems are so complicated they’ve already seen two other doctors. Both also analyze a significant number of Medicaid patients, who tend to have a notably high rate of comorbidity (simultaneous multiple ongoing health troubles)—making their risk of death within a month of hospital admittance 40 percent greater than the general population’s. A medical error is still a medical error, but these higher baseline death risks have to be accounted for before drawing major conclusions about causality.

Beyond that, given the byzantine nature of U.S. healthcare logistics and the constant implementation of new technologies, errors may just be part of the game. The theory of “normal accidents”—introduced by Charles Perrow in 1984 and applied to disasters like the meltdown at Three Mile Island and later the Challenger crash—describes serious accidents that occur in complex, high-risk systems as being a result of “multiple failures that are not in a direct operational sequence.”

Basically, the idea is that as small errors occur independently in different areas of the system, they’ll ultimately interact with each other in ways that are more or less impossible to prevent or respond to appropriately. It’s unfortunate, and particularly in the case of medical care often tragic, but that’s the reality when the process is complicated and the stakes are high.

Obviously hospitals still need to minimize mistakes, and evidence suggests a little prevention can go a long way. Two years ago a Milwaukee hospital modified their ID wristbands for not-yet-named newborns to include the mother’s first name as a secondary identifier—rather than “Babygirl Smith” (as it would appear in most of the country’s neonatal ICUs) a kid’s wristband might read “Sarahsgirl Smith.”

The apparent result: a reduction of 36 percent in wrong-patient orders.

Not all fixes are this simple, of course, and if nothing else both these reports effectively bring to light the dangers inherent in our healthcare system.

But that doesn’t mean we’re all doomed every time the steak knife slips and we have to go in for stitches—nor, unfortunately, can we count on that strategy to make our first million.