A new study published in the January issue of Surgery says approximately 4,082 "never events" —i.e., significant and avoidable medical errors, such as accidentally sewing a sponge into the chest cavity, or operating on the wrong patient — occur in American hospitals each year. Which may be blowing things out of proportion if you consider how many surgeries are performed, but still.

The study, authored by a group of Johns Hopkins researchers, extrapolates on the medical malpractice claims that arise from never events. According to its online abstract, the researchers "identified a total of 9,744 paid malpractice settlement and judgments for surgical never events occurring between 1990 and 2010," and, based on "literature rates of surgical adverse events resulting in paid malpractice claims," determined that the number of real "never events" has been dramatically undercounted for the last two decades.

In a weird twist, the study's findings have been miscounted as well. Sarah Kliff over at WonkBlog reports that never events "happen quite frequently, about 500 times a year." (Kliff obtains "about 500" by dividing 9,744 by twenty — the number of years included in the study.) But 9,744 is the number of paid malpractice claims resulting from never events — not the total number of never events that actually occurred. Compare Kliff's statement with the title of the press release she links to: "JOHNS HOPKINS MALPRACTICE STUDY: SURGICAL ‘NEVER EVENTS’ OCCUR AT LEAST 4,000 TIMES PER YEAR".

Likewise, the statistic Kliff cites in her article's headline — "Surgeons left 4,857 objects in patients over the past two decades" — is based on the paid malpractice claims the researchers started with, not the total number of never events they estimate. The actual number of objects left in surgical patients in the past twenty years is probably much, much higher than 4,857.

It's an open question, in any event, whether you should be legitimately frightened about these numbers: