In a dramatic bid to improve hospital performance, the government plans to post a consumer-friendly comparison of hospitals' heart attack and heart failure death rates on the Internet.

The Centers for Medicare and Medicaid Services (CMS) launched a dry run of the program in December by telling nearly 4,000 hospitals how their 30-day death rates for 2003 compare with the national Medicare death rates of 17.8% for heart attacks and 11.6% for heart failure.

"If we could get the higher-mortality hospitals to achieve the performance of lower-mortality hospitals, we could save probably 10,000 or more lives a year," says Harlan Krumholz of Yale University, who with colleagues at Yale and Harvard developed the statistical methods used in the analysis.

Experts say 30-day death rates offer a more reliable index of a hospital's performance than inpatient deaths and don't reward hospitals that transfer or discharge patients before death.

"The main purpose of all this is to improve quality," says Michael Rapp of CMS. "If I'm running a hospital and see that I fall in a category that's worse than 98% of hospitals, that's going to grab my attention. I'd look to see if we can improve."

Consumers will gain access to the results of the analysis in June, when the agency posts data from July 1, 2005 to June 30, 2006 on a Medicare website called Hospital Compare, hospitalcompare.hhs.gov. Rather than directly posting hospitals' death rates, the site will simply report whether a hospital performs better, worse or on par with the national average.

Hospital Compare now offers a rundown of hospitals' diligence in providing the best therapies for heart disease and other conditions, based on the available scientific evidence.

These so-called performance measures, set out first in the Medicare Modernization Act of 2003, can reassure consumers that doctors at their local hospital are following treatment guidelines, but they don't reveal how patients do after they leave the hospital.

Some hospital administrators, though they support the goal, say the statistics may be misleading.

"It clearly needs to be done, but I'm not sure 30-day mortality is the right measure," says Gary Noskin of Northwestern Memorial Hospital in Chicago. "A patient could have a heart attack (and be treated successfully) and get hit by a bus after he leaves the hospital."

Medicare officials counter that the statistical methods used in the analyses highlight patterns of care, good or bad, not individual cases. The approach was approved by the National Quality Forum, a consortium of professional organizations, businesses, consumer groups, hospital chains and health plans.

"It's very clear that the public is unhappy with the quality of care that's been delivered," says Thomas Balcezak, associate chief of staff in charge of quality at Yale-New Haven Hospital. "This offers a new level of transparency by CMS, a way to show the public how care is being delivered across the U.S., a new opportunity to improve care."