• buy 
• sell $
• about
• My DOTmed 

Hospital report cards might help patients choose a place that best fits their needs, but do the reports actually prompt hospitals to perform better? The answer appears to be a qualified, controversial yes.

Canadian doctors have performed the first randomized, controlled trial on the effect of a hospital learning its "grade" on its practice, and they estimate the experiment, while not prodding huge reforms, might have saved almost 250 lives.

In the decade-in-the-making study published online Wednesday in the Journal of the American Medical Association, the doctors randomly divided 81 hospital corporations in Ontario, Canada into one of two groups: one group, with 42 hospitals, would learn how well its hospital ranked on 18 "process-of-care" quality indicators for heart attack and heart failure; the other, with 39 hospitals, would only learn its rank on these indicators 21 months later, and the doctors would see which group showed greater improvements over a baseline performance.

The guidelines to be assessed were developed by two expert panels which included leading Canadian cardiologists, and the baselines for care quality were established only after nurses visited the hospitals and actually pulled information off patient charts.

The indicators, 12 for heart attacks and six for heart failure, included whether patients got appropriate diagnostic tests, such as echocardiograms, and whether life-saving medication was given in a timely manner.

The doctors say they chose heart conditions because there were well-established, evidence-based guidelines already in place, and because it's well known that not everybody follows them. "There's a big gap between optimal care and actual care," Jack Tu, M.D., Ph.D., a researcher at the Institute for Clinical Evaluative Sciences in Toronto and lead author of the study, tells DOTmed News. "And there's lots of data in regard to what should be done for these patients, as opposed to choosing conditions where there's not a lot of strong evidence," he adds.

Baseline data were gathered from 1999-2001; early feedback hospitals received their "grades" in January 2004, while late feedback ones got them in September 2005. Hospital quality changes on the indicators were assessed at the end of that period.

What they found

Overall, there were no systemic differences in performance between the two groups. However, it wasn't a complete wash: when the doctors broke down the data, interesting trends emerged, as well as significant differences between the groups.

For one, almost three-quarters of hospitals in the early feedback group said they were trying to change heart attack care practices, against slightly less than half of the late feedback group.

But the more important, and dramatic finding, was a 2.5 percent decline in the mean 30-day mortality rates from heart attacks in the early feedback hospitals.

"We think it's probably multifactorial," says Dr. Tu about the decline. "There's not a simple explanation. Every hospital did something slightly different in response to its report card," he says, which could have led to the slight downtick in deaths.

One factor was, in response to early feedback, around 10 to 25 percent of hospitals changed their protocol for delivering anti-blood-clotting drugs to heart attack victims. In the baseline review, Dr. Tu and his colleagues noticed that some hospitals made patients wait to see a specialist, or get transferred to an ICU, before they could get the thrombolytic drugs -- a delay that cost around 10 or 11 minutes, precious time that could be saved by letting the attending ER physician deliver the medicine instead.

Dr. Tu stresses organizational changes, such as instituting standardized protocols and checklists for admitting and discharging patients, might have led to the greatest gains. And these gains are not trivial. "Our estimate is the report card may have saved 250 people in 2004," he says.

From a statistical perspective, Dr. Tu says the findings of the slight improvement in mortality rates in heart attacks are strong, but not perfect. "There's about a one in 20 chance it was [due to] chance, and 19 in 20 it was real," he says. "We chose the most conservative method for analyzing the data. I'm pretty convinced it was the real thing."

Still, he does acknowledge that many have been skeptical of the findings, and believe because the benefits were so modest, they might not be reproduced.

But Dr. Tu counters that though few results were statistically significant, nearly the entire trend was headed one way. "Not a single indicator of our almost 20 [favored] the delayed feedback group," he says. "All our indicators were pointing in one direction."