Electronic Medical Records: Hazardous to Your Health?

Here's something I'd like explained, now that I've read amazing findings in this week's New England Journal of Medicine about the pathetic pace at which health information technology, or health IT, is being adopted by hospitals, following a similar NEJM report last summer about foot-dragging physicians: Why are the benefits of health IT and electronic health records, and the fixability of their flaws, apparently taken for granted? More specifically, is it truly worth $30 billion, as the Obama administration proposes, to digitize everybody's individual medical histories, test results, medications, scans, clinical notes from physicians and nurses, and other healthcare detritus and convert them into electronic health records, accessible from anywhere? To enter prescriptions into hospital computers in order to slash the awful toll of patients injured or dead because of the wrong medication, the wrong dose, or a deadly drug interaction?

The new NEJM study found that 1.5 percent of hospitals have equipped all of their major medical services with comprehensive electronic health records. Add 7.6 percent for hospitals with EHRs in as few as one clinical unit. Computerized physician order entry (CPOE) of prescriptions is in use at 17 percent of hospitals. "Abysmally low," said Ashish Jha of the Harvard School of Public Health, the report's lead author, at a press conference on Tuesday. The main reasons identified by the thousands of hospitals that responded to a massive survey: not enough money to buy the systems ($20 million to $100 million), not enough money to maintain them, physician resistance, and unclear return on their investment.

Skepticism about the ability of such systems to fulfill their promise (while not creating risks of their own) was not among the list of reasons that hospitals surveyed were offered. But it's a real question.

Last December, the Joint Commission, which accredits hospitals, issued a "sentinel event alert"—a warning about a significant risk to patient safety, calling for immediate investigation. The commission doesn't put out many of these. There were five in 2008, none in 2007, and three in 2006. They're serious.

The warning was about health IT. Computerized machines that automatically dispense pharmaceuticals, EHRs, and CPOE all pose real risks, noted the alert. "There is a dearth of data," said the commission, on how frequently technology harms rather than helps, but about one fourth of more than 175,000 medication errors registered in 2006 "involved some aspect of computer technology as at least one cause of the error." Mislabeled bar codes, confusing screen displays, and physicians overriding warnings on the screen about a drug were among the examples cited. Bad health IT, or bad use of it, can kill as effectively as low-quality care of any kind.

Nor is it clear that incorporating IT saves money or improves the quality of care. It can accomplish both of these—at large, tightly integrated healthcare systems that can impose a unified approach across many facilities. Veterans Administration hospitals and the Geisinger hospital network come to mind. But most of the U.S. public does not receive medical care from large systems.

The beating of the IT drums has been going on for more than a decade, and it's getting louder. But what is it telling us?

The reason for my several-week absence from the blogosphere was the recent launch of "America's Best Nursing Homes," our latest and by far largest set of health rankings. We'd like to think Best Nursing Homes will make the hunt for a competent, humane source of care for a family member or friend a little less frustrating. We've given nursing homes ratings of one to five stars in various categories, ranked them in tiers, and built in lots of options for custom searches. Geographical location, ratings in different categories, and type of ownership (such as nonprofit with a religious affiliation) are just a few of the choices that can be selected or combined. Is this a useful tool? What more should we do? You tell us.