Tuesday, July 10, 2007

It's not about the Data (again)!

Last night Julie Steenhuysen filed a report for Reuters about a study just published in the Archives of Internal Medicine on the impact of the move to electronic health records. Her lead paragraphs say it all:

Electronic health records -- touted by policymakers as a way to improve the quality of health care -- failed to boost care delivered in routine doctor visits, U.S. researchers said on Monday.

Of 17 measures of quality assessed, electronic health records made no difference in 14 measures, according to a study published in the Archives of Internal Medicine.

The study by researchers at Stanford and Harvard Universities was based on a survey of 1.8 billion physician visits in 2003 and 2004. Electronic health records were used in 18 percent of them.

In two areas, better quality was associated with electronic records, while worse quality was found in one area, they said.

This story was particularly important to me, as a Lipitor user, since that "one area" of "worse quality" was the prescription of statins for high cholesterol; but it is not just my own ox that is being gored. Indeed, my own interpretation of these results is that, when the quality of health care is left in the hands of "policymakers," rather that practicing health care providers, that quality is bound to suffer.

Fortunately, Ms. Steenhuysen gave some indication of what may have influenced those policymakers:

Electronic health records promise to eliminate errors due to bad handwriting and make it easier for doctors to follow a patient's care over time.

Some systems can also flag dangerous drug combinations, or offer advice about tests or drugs the doctor might prescribe.

However, the question of the extent to which factors such as these impact the overall quality of health care practice were left unaddressed, at least by Ms. Steenhuysen if not the Archives of Internal Medicine. Those who seek a better understanding of the issues behind that question need to consult books like Atul Gawande's Better and Jerome Groopman's How Doctors Think, or at least the comprehensive review of Groopman's book that Richard Horton, editor of The Lancet, published in The New York Review. I already wrote about Ellen Goodman's review of Groopman's book when it appeared on Truthdig, but Horton has now given us more of an insider's point of view.

To try to put the Archives study in perspective, I would like to cite a paragraph from Horton's review concerned with what he calls a "deeper fault line within medical practice:"

On average, about 15 percent of a doctor's diagnoses are inaccurate. Groopman directs a well-aimed arrow at a system of medical training that more often than not fails to investigate why these diagnoses are misses. Doctors are rarely taught to ask how an error could have taken place, let alone how it could be avoided in the future. Most are unaware of their mistakes. Even if patients remain unwell, no systematic effort is made to find out where doctors may have gone wrong. Doctors are uncertain about their own uncertainties. (Although for some doctors, such as radiologists, Groopman cites alarming research that shows the worse their performance, the more certain they seem to be that they are right!)

In this context one has to ask whether, even if practicing physicians were more conscientious about error analysis, having electronic records would make a difference. Anyone who took an undergraduate course in databases (if not the more general area of information systems) would know the answer immediately: It depends on what has been recorded electronically! If the decision about record content was left to policymakers more concerned about the risk of applying the data to support a malpractice suit, then it is unlikely that those records would be of much help when error analysis is required. In other words, at a time when public opinion (now under the influence of Michael Moore) about health care just keeps getting lower, technology is there, once again, with a solution to the wrong problem. This leaves the policymakers scratching their head over what went wrong, as oblivious to the possibility that they, themselves, were what went wrong as are those radiologists that Groopman studied!

No comments: