Last Updated: 2003-06-03 17:07:54 -0400 (Reuters Health)

NEW YORK (Reuters Health) - Despite recent improvements in the rate of missed diagnoses, at least eight percent of autopsies still detect potentially serious conditions that were not found by the patients' doctors, according to new findings released Tuesday.

In the report, published in the Journal of the American Medical Association, U.S. researchers estimate that between eight and 24 percent of autopsies currently pick up a condition that the patient's doctor missed, and one that may have been the actual cause of death.

Furthermore, they write that between four and seven percent of autopsies currently performed may detect the most serious type of medical error -- meaning that, if the doctor had properly diagnosed the condition, the patient may have lived. These errors are known as class I errors.

In fact, the authors estimate, among the 850,000 people who die each year in a U.S. hospital, around 34,850 may have survived if an important diagnosis had not been missed.

However, study author Dr. Kaveh G. Shojania of the University of California, San Francisco cautioned that people should not fear that their doctors are always missing life-threatening illnesses.

These errors occur among people who have died, Shojania said, and the current study results say nothing about the vast majority of people who come to the hospital, are treated and released.

"The point here is not that there are huge numbers of patients being misdiagnosed, but it is fair to say that physicians should not be diagnostically complacent -- there is still substantial room for improvement," Shojania told Reuters Health.

To obtain their findings, the researchers reviewed previous research conducted between 1966 and 2002 that investigated autopsy-detected diagnostic errors.

The researchers found that the rate of class I errors decreased by 33 percent every decade, while the rate of major errors -- defined as missed diagnoses of conditions that could have been a cause of death -- dropped by 19 percent every decade.

Shojania said that the decrease in autopsy-detected errors since the 1960s may be a result of improvements in medical technology, enabling more accurate diagnoses.

But despite these improvements, "error rates remain surprisingly high," Shojania noted.

The researcher explained that some experts believe that the high rate of missed diagnoses detected in autopsy stems from the fact that relatively few deaths are investigated by autopsy -- currently, around 1 in 20. Consequently, deaths investigated by autopsy likely represent those that doctors regarded as "diagnostically challenging," the researcher said, and their autopsies are therefore more likely to uncover mistakes.

If this explanation were correct, Shojania reasoned that an increase in the rate of autopsy would include less challenging cases, subsequently causing a decline in the rate of diagnostic errors.

However, the researcher said that some of the studies included in the current report had autopsy rates as high as 100 percent. And while the results suggest that the high rate of errors is, in part, due to the fact that doctors struggled with a diagnosis when the patient was alive, this theory is "not enough to explain away the findings of persistent error discovered at autopsy," Shojania said.

Instead, "the answer is probably a combination of true errors and tough cases," the researcher noted.

Interestingly, many of the missed diagnoses involved well-known conditions, in which the symptoms are very general or perhaps resemble something else, Shojania said. For instance, some diagnostic errors picked up by autopsy include cancer that turns out to be an infection, or an infection that turns out to be cancer.

Shojania added that the current study findings highlight the importance of heeding autopsy results.

"If clinicians paid more attention to autopsy findings, some of today's 'tough cases' might come to recognized as tomorrow's textbook cases," Shojania said.