I read the recent Institute of Medicine (IOM) report on patient safety with dismay, then increasing skepticism and concern. I've reviewed the research from which it concludes that between 44,000 and 98,000 Americans die in hospitals annually due to preventable adverse effects of medical care. I've found that the conclusion is based entirely on adverse event death rates from two studies and an extrapolation of those figures to the 33.4 million annual hospital admissions in the United States.
The first study, from a 1991 issue of the New England Journal of Medicine, estimated that in 30,195 "randomly selected" hospital admissions in New York in 1984, 80 deaths occurred because of negligence. Researchers then applied these percentages to the 2.67 million admissions that occurred in New York state in 1984 and estimated that 6,895 deaths took place because of adverse events. IOM researchers applied the same formula to the 33.6 million admissions in the United States in 1997 to arrive at 86,736, which they rounded up to 98,000, for reasons known only to the authors.
The second study, from a 1999 issue of Inquiry, found that in Utah and Colorado in 1992, death occurred in 6.9% of all preventable adverse events, which occurred in 3.1% of all hospital discharges studied. Applying these researchers' findings yields a national rate of 41,731, which the IOM inexplicably rounded up to 44,000.
I have practiced internal medicine for 24 years, and I have not seen evidence of these statistics. Neither have my partners and colleagues. Why?
First, the selection process of both studies was not random. In another issue of the New England Journal of Medicine, the New York researchers acknowledged that they oversampled births with complications, as well as cases treated by physicians in high-risk specialties like neurosurgery, thoracic surgery, peripherial vascular surgery, cardiac surgery, orthopedics and urology. In short, the researchers said that they deliberately designed their study to target high-risk admissions. You can only speculate to what extent this process inflated the adverse event rate.
The Utah/Colorado study, which found the preventable death rate to be less than half of that found in the New York study, also did not use random selection. Researchers excluded psychiatric, rehabilitation, and drug/alcohol diagnosis-related groups and hospitals that exclusively provide these services. Excluding these lower-risk admissions tended to concentrate adverse events in the admissions that were studied.
The IOM report and the news coverage that followed have assumed that whenever a death occurred in an admission where an adverse event took place, the death was caused by the event. The studies, however, offer no proof of this. Because critically ill patients have more numerous and complex medical interventions, there is a greater chance that an error will be made at some point during the admission. These patients are also more likely to die from their illness.
The IOM report urges Congress to create a new national center for patient safety similar to the Federal Aviation Administration or the Occupational Safety and Health Administration. The center would contain a new national mandatory reporting system for all serious medical errors, and all reports would be open to malpractice attorneys. The stated goal is to reduce errors by 50% within five years.
I submit that the IOM report substantially overstates the extent of medical injury in the U.S. Moreover, the figures in the report already show a greater than 50% reduction in preventable deaths from 1984 to 1992. I would argue that states, hospitals and medical staffs already have extensive record review and quality assurance programs in place and running effectively. If Congress feels a need to do more, a good place to start is to fund a truly random study of adverse events in hospital admissions.