CMS (The Centers for Medicare & Medicaid Services)による病院の格付けは死亡率と相関しない
病院機能に関する政府による格付けと死亡率との比較によって，高く格付けされた病院での死亡率と低く格付けされた病院での死亡率において違いはほんどん どないことがわかった．この研究では，研究者がCMSの病院比較ウェブサイトに載せられた3,657の病院を評価した．研究は，病院が心臓発作や心不全・ 肺炎に対し，どれぐらいの頻度で推奨された治療法に応じたかを比較検討したものである・・・
CMS hospital ratings don’t correlate to death rates
A comparison of government ratings of hospital performance and death rates found that death rates were only marginally better at highly rated hospitals compared with lower performing ones.
In the study, researchers evaluated 3,657 hospitals listed on the CMS Hospital Compare Web site. The study compared how often hospitals met recommended treatments for heart attacks, heart failure and pneumonia with each hospital’s mortality rate. Researchers found only small differences in mortality rates for hospitals performing at the 75th percentile on all measures within a category vs. those performing lower than the 25th percentile. For example, for heart failure measures, the absolute mortality reduction ranged from 0.1% for inpatient mortality to 0.2% for one-year mortality.
Researchers found similarly small differences in mortality for heart attack and pneumonia patients--less than 2% for any performance measure. The study was published in the Dec. 13 Journal of the American Medical Association.
The findings indicate that outcomes are influenced by more than clinical measures and practice guidelines, said the study's authors. Whether a hospital uses electronic health records and its staffing levels, for example, may also play a role. There is also some question as to the reliability of the data, they continued. For example, low-performing hospitals may try to game the system by excluding patients who do not receive the processes being measured, and the CMS does not have a mechanism to monitor whether data are complete.
However, the effect of performance on mortality rates would remain mild even accounting for possible problems with data collection, said the authors. They suggested that efforts be made to more tightly link CMS performance measures to outcomes rather than processes of care. An accompanying editorial said that the study’s results raised questions about the appropriateness of using Hospital Compare as the basis of pay-for-performance or consumer hospital selection. In the real world, the editorial noted, multiple clinical variables and patient characteristics affect outcomes.
A CMS official, quoted in the Dec. 12 Washington Post, agreed that more
quality measures are needed but said that researchers would have found
bigger differences between hospitals had they included all 22 measures
used on Hospital Compare.