Missing from healthcare reform: the autopsy

Part of any real healthcare reform will be improving practices in hospitals, and -- as Obama's proposed commission on comparative effectiveness would do -- identifying what works and what doesn't. Knowing what works and why people get better or not is vital to good medicine. But amid the talk on improving such knowledge as part of healthcare reform, a vital and fairly cheap way to generate some of it -- the autopsy -- is going ignored. This is too bad, as autopsies yield incredibly good information about the quality of both diagnosis and treatment. But they're almost never done.
I explored this 4 years ago in a story for the Times Magazine:

[T]he hospital autopsy is neglected. ...In the 1960's, hospitals in the United States autopsied almost half of all deaths, and the autopsy was familiar to medical students and practitioners alike. The United States now does post-mortems on fewer than 5 percent of hospital deaths, and the procedure is alien to almost every doctor trained in the last 30 years.

This is quite unfortunate, because

nothing reveals error like the autopsy. As [former JAMA editor George] Lundberg noted in a 1998 article, numerous studies over the last century have found that in 25 to 40 percent of cases in which an autopsy is done, it reveals an undiagnosed cause of death. Because of those errors, in 7 to 12 percent of the cases, treatment that might have been lifesaving wasn't prescribed. (In the other cases, the disease might have advanced beyond treatment or there might have been multiple causes of death.) These figures roughly match those found in the first discrepancy studies, done in the early 1910's. ''No improvement!'' Lundberg notes. ''Low-tech autopsy trumps high-tech medicine . . . again and again.''

Lundberg doesn't fantasize that the autopsy can make medicine mistake-free; medicine poses puzzles too various and complex to expect perfection, and indeed error rates run about the same no matter how many autopsies are done. But autopsies can keep doctors from repeating mistakes, and thus advance medicine. Doctors miss things. But without autopsies, they don't know when they've missed something fatal and so are likely to miss it again. They miss the chance to learn from their mistakes. Instead, they bury them. This, Lundberg says, ''is endlessly galling.''
As Lundberg sees it, ''If you want to base your medicine on evidence, if you want to reduce error, if you simply want to know what you are doing, then you should start by evaluating the care given to your sickest patients -- the ones who die.''

When hospitals were using it a lot, the autopsy produced huge rewards, from great advances in cardiac surgery and the identification or detection of problems ranging from SIDS and Legionnaire's disease to HIV, Ebola, and West Nile virus.
As the article notes, several factors have driven the drop in autopsy rates, including the cost ($2000-5000), the time required, doctors' (misplaced) fear of being sued, the lack of connection and familiarity between doctors and patients' families, and a wild overconfidence in high-tech diagnostic tools. ("We get this all the time," one pathologist in my story told me. "The doctor will get our report and call and nsay, 'But there can't have been a lacerated aorta. We did a whole set of scans.' We have to remind him we held the heart in our hands.")
Doctors think they know why patients die. (My own doctor told me this. "We don't do too many autopsies. We generally know why somebody dies.") But about a third of the time, they're wrong.
But the main reason autopsy rates have dropped is that in 1970, the Joint Commission, which accredits hospitals, dropped a requirement that hospitals autopsy at least 20 or 25 percent of deaths to be accredited -- but that requirement was dropped in 1970. Rates by then had dropped to the mid-20s -- and they continued dropping to the low single-digit rates of today.
If we want to improve medicine -- both diagnosis and our knowledge of the actual effects of various treatments -- then we'd do well to refuse accreditation to hospitals that didn't autopsy say, 20 or 30 percent of their deaths.
• Read the whole article at the Times Magazine site.
• You can also listen to a story that NPR's Scott Simon did on the subject just after the Times article appeared, in which he interviewed me and Dr. Pat Lento, head of the autopsy service at Mount Sinai Hospital in NYC.