scorecardresearch Skip to main content

Getting an accurate count on medical errors


Is it too much to ask the US health care system to have a common agreement on what’s a medical mistake and how they should be counted?

In its most recent report, issued this month, the Betsy Lehman Center for Patient Safety tallied nearly 62,000 medical mistakes yearly in the Bay State.

In the state Department of Public Health’s most recent count of “Serious Reportable Events” — meaning any one of 29 preventable mistakes ranging from surgery on the wrong body part to medication errors — 1,267 incidents were reported for 2018.

Did someone miscount? Or were 61,000 mistakes simply deemed not serious?


Those numbers are among many reports and studies offering vastly different estimates of medical mistakes nationwide. Not all these studies count hard numbers of actual events as they occur. Some extrapolate totals from ancillary data. Others may vastly undercount incidents, with a reliance on providers to self-report. Then there are different criteria for different states, and 23 states with no mandated reporting at all.

“There’s no national reporting system worth its salt,” says Arthur Levin, a longtime health care advocate and a contributor to the groundbreaking 1999 report “To Err is Human.”

That report, from the Institute of Medicine (now the National Academy of Medicine) was the first to identify widespread preventable medical mistakes, which it estimated were responsible for nearly 100,000 deaths nationwide. The report sparked several states to begin an annual census of “adverse events” — such as surgery on the wrong body part or items left in patients after surgery — beginning in Minnesota in 2006. Since then, Massachusetts and 25 other states have mandated a count.

Getting the number right is vital: Without accurate statistics, it’s harder to isolate problems and solve them.

Most states show numbers of incidents in the hundreds, not the hundreds of thousands, though different states have different standards for which institutions should report; some cover hospitals only, others also include ambulatory services.


Yet much higher estimates abound. A 2016 study by a Johns Hopkins team suggests errors led to more than 250,000 deaths per year, making mistakes the nation’s third-leading cause of death after heart disease and cancer. Other researchers dispute that, saying, if true, more than 1 in 3 of the 715,000 US hospital deaths per year would be due to an error.

Skepticism should be encouraged, especially on a subject as serious as this. But some criticism comes across as overly defensive, and even the critics should be subject to criticism; for a definition of “error,” one report actually cites Wikipedia.

In a statement, a DPH spokeswoman wrote that its Serious Reportable Event report is “only a subset of all medical errors that occur in the health care setting” reported by hospitals and ambulatory surgical centers.

Betsy Lehman Center officials acknowledge their numbers are far higher than those in the DPH report. The center analyzed insurance claims from 2013 to examine medical codes associated with errors. As a Globe article summarized, those records showed 224 instances of foreign objects left inside patients after surgery, versus just 33 reported to the DPH.

The examination of insurance records, says Barbara Fain, the center’s executive director, is more reliable than the accounts of caregivers or patients immediately following an event. An example is the 1994 death of Betsy Lehman, a Globe health columnist, for whom the center is named, from a massive overdose of chemotherapy that wasn’t uncovered until a review of insurance records several months later.


“That’s why we did the research we did,” says Fain. “Looking at administrative claims data is not the best way of figuring this out, but it’s the only way.”

It also gives a clearer view of the severity of the problem, she says. “The (DPH’s report of) Serious Reportable Events leads people to believe that medical errors aren’t that common.”

Conversely, studies that overestimate the incidence of errors may inure a public bombarded by bad news stories as a problem so large there’s nothing anyone can do about it.

“If you say a million people die unnecessarily, people are going to throw up their hands,” says Levin.

So is there a “Goldilocks zone,” somewhere between hundreds of incidents and hundreds of thousands, that’s easier to accept?

No, because not even a single preventable death is acceptable. There’s no reason why reports on errors can’t be error-free. For better or worse, insurance records rarely leave anything out, and the Betsy Lehman Center’s methodology is the most promising so far.