Like ‘a Lance Armstrong performance,’ Canadian hospitals’ mortality rates may be too good to be true: study

National Post

Tom Blackwell | Nov 13, 2012 12:30 AM ET | Last Updated: Nov 13, 2012 11:06 AM ET

Fotolia Canadian hospitals may have reduced their death rates somewhat, but “gaming” of the numbers is likely a factor, too, says a newly published study. 

 

It was a rare bit of good news about Canada’s health-care system: In the few short years since a federal agency began publishing hospital death rates, the mortality numbers had decreased significantly.

A new study, however, concludes that at least some of that improvement was achieved not by better caring for patients, but by manipulating internal statistics.

The research documented a “dramatic” rise in the number of patients coded by hospitals as palliative — and therefore removed from the publicly released mortality statistics — suggesting an attempt to game the system and artificially shrink the facilities’ death rates.

The [mortality] performance is a Lance Armstrong performance — it’s a bit hard to believe

“Just on the surface, if you looked at some hospitals, the [mortality] performance is a Lance Armstrong performance — it’s a bit hard to believe,” said Dr. Christopher Chong, an internal-medicine specialist at Lakeridge Health in Oshawa, Ont., who led the study. “Mortality is something that is really difficult to actually affect very quickly…. You might want to believe these numbers are true, but they’re just a bit too dramatic and too good.”

In a paper just published in the journal BMJOpen,Dr. Chong, also affiliated with Queen’s University, and his co-authors acknowledged that hospitals could have reduced their death rates somewhat, and that new guidelines for coding palliative-care patients explain some of the increase in those cases. But they argue “gaming” of the numbers is likely a factor, too.

Joe Raedle / Getty ImagesThe most recent Canadian hospital mortality rates, released last December, indicated there had been a significant decrease in the mortality number since 2004 for 53% of the acute-care hospitals surveyed, while almost none had higher rates.

 

Patients are coded as palliative-care — generally meaning their disease is terminal — either during their hospital stay or on discharge. Those cases are not included in mortality statistics, the assumption being the hospital could not have prevented their deaths.

If gaming occurred, it is possible that hospitals deliberately included some patients as palliative who simply did not fit the definition, said Dr. Chong. Or they could have altered their coding practices after the mortality project started to reflect more accurately the number of palliative cases. Even changing for that reason, however, would skew the death ratio, which is meant to be tracked over time.

But the umbrella group for many of the country’s hospitals said Monday it is highly unlikely any of its members had tried to manipulate the figures. Deciding when a patient is palliative is subjective, but coding officials take their work seriously and follow national guidelines, said a spokesperson for the Ontario Hospital Association.

The OHA finds it incredibly hard to believe that any hospital would deliberately abuse the reporting system

“The OHA finds it incredibly hard to believe that any hospital would deliberately abuse the reporting system in order to distort their [mortality] measure,” said the spokesperson, who asked not to be named. “Hospitals are taking great strides and making real efforts in improving patient safety and that is having a positive effect.”

The Canadian Institute for Health Information (CIHI), which produces the figures, does data quality checks every year that include monitoring palliative-care coding, and remains convinced the rate of hospital deaths is truly falling, said Joanna Fillion, a spokeswoman for the agency.

Sholom Glouberman, head of the Canadian Patient Association, said he was not ­surprised by the study’s findings, but suggested the rise in palliative-care numbers may partly reflect the fact hospitals are focusing more on the needs of such patients.

Developed in the U.K. in the 1990s, the awkwardly named “hospital standardized mortality ratios” (HSMR) compares deaths at each given hospital with a national average, with 100 being the baseline. A ratio of 100 suggests no difference from the Canadian average in 2004-05, while below or above indicates death rates higher or lower than the national mean.

Getty Images / ThinkstockThe average rate of patients coded as "palliative" — generally meaning their disease is terminal and the hospital could not have saved them — soared almost 50% between 2004 and 2010, the study found. 

 

The idea is to prod hospitals to change care and lessen the likelihood of patients dying. The first release of figures by the Canadian Institute for Health Information came in 2007, after the project was announced.

The most recent results were released last December, indicating there had been a significant decrease in the mortality number since 2004 for 53% of the acute-care hospitals surveyed, while almost none had higher rates. “This continues to be a good news story,” Kira Leeb, a CHI spokeswoman, told Postmedia News at the time. “It shows a lot of progress.”

Dr. Chong, a Queen’s faculty member, and colleagues at the University of Toronto found that the average mortality ratio fell from 99.6 in April 2004 to 91.06 in March 2010, with most of the decrease coming after the public release of death figures started.

Over the same period, the average rate of patients coded as palliative care soared almost 50%, they found.