July: When not to go to the hospital

Being admitted to the hospital is never a picnic. But when possible, schedule any therapeutic procedure for some month other than July. At least if you’ll be treated at a teaching hospital. That’s the conclusion of a new analysis that uncovered a cyclical spike across the nation in serious medication errors.

Month to month, over the last several decades, deaths attributable to errors in the amounts and/or types of medicines prescribed or administered at hospitals have varied little. Except during July, when the rate spikes roughly 10 percent, the new study finds. And the likely reason: green doctors. This month-long blip just happens to correspond to when newly minted medical residents are released into teaching hospitals with substantial autonomy to make medical decisions, explain David Phillips and Gwendolyn Barker of the University of California at San Diego.

Owing to their “new resident hypothesis,” they examined death rates for medication errors by county, accounting for the share of teaching hospitals in each. And, they now report, “the greater the concentration of teaching hospitals in a region, the greater the July Effect for . . . [fatal] medication errors.” Indeed, the 10 percent spike disappeared when the new analysis looked only at areas without teaching hospitals.

Phillips, a sociologist who’s recently focused on medical accidents, had been encouraged over and over by physician friends to investigate the so-called “July Effect,” an anecdotally observed peak in hospital deaths. But the few studies that had looked to confirm it failed, Phillips says – probably because they focused on a single hospital, where the sample size was too limited to turn up a small, if strong, trend.

He and Barker, by contrast, probed a national database of more than 62 million death certificates that spanned from 1979 (when hospital status was first recorded in those records) through 2006 (the most recent year for which data were available). They turned up almost a quarter-million deaths that were coded as having not only occurred in a hospital setting, but also been due to medication errors. Both in-patient and out-patient cases were included.

This screening revealed a clear and unambiguous July spike, they report in the Journal of General Internal Medicine, published online ahead of print. It showed up only for medication errors, not other causes of death. And, rather disturbingly, Phillips notes, the July blip “didn’t get smaller when medical residents were required to work fewer hours” – i.e. no more than 80 per week, starting in July 2003.

Analyses of competing causes of death indicate that July increases in fatal medication errors don’t result from a temporary “borrowing” of these deaths, due to miscoding, from other categories. The July Effect also doesn’t reflect a short-term seasonal spike in hospital admissions. The researchers checked that out and found “admissions in July are actually slightly below what you’d expect,” Phillips says.

Owing to the database they used, this pair couldn’t identify whether nonfatal medication errors also peak in early summer, but that’s the presumption.

The solution, Phillips says, should probably be a more gradual transition to autonomy for new doctors. Indeed, he argues, “We provide fresh evidence for reevaluating responsibilities assigned to new residents and increasing the supervision of them.”

And, I asked, wouldn’t it make sense to avoid unnecessary hospitalizations in mid summer? “That’s a good message, too,” Phillips agreed. “In fact,” he noted, “one physician – not knowing that I was studying this – referred to the issue and said to me: ‘If possible, you should probably avoid going in [being hospitalized] during July.’”