When ranking hospitals on outcomes for treating acute ischemic stroke in Medicare patients, it might be necessary to take into account the patients' initial stroke severity, researchers found.
A model for predicting 30-day all-cause mortality rates following acute ischemic stroke had significantly better discrimination when the National Institutes of Health (NIH) stroke scale score was included, with the C statistic rising from 0.772 to 0.864 (P<0.001), according to Gregg Fonarow, MD, of the University of California Los Angeles, and colleagues.
Adding that measure of stroke severity - which is not found in administrative claims data - also reclassified a substantial number of hospitals to either better or worse performance categories, the researchers reported in the July 18 issue of the Journal of the American Medical Association.
Although it is believed that adjusting risk models based on claims data is sufficient for distinguishing mortality risk at the hospital level for acute myocardial infarction, heart failure, and pneumonia, that does not appear to be the case for acute ischemic stroke, the researchers said.
"A 30-day mortality model for acute ischemic stroke without adjustment for stroke severity provides lesser discrimination, produces different rankings of hospital performance, and may be biased in favor of hospitals treating less severe strokes than a model with adjustment for stroke severity," they wrote.
Fonarow and colleagues examined data from 127,950 fee-for-service Medicare beneficiaries treated at one of 782 hospitals participating in the Get With the Guidelines-Stroke program from April 2003 through December 2009. All had NIH stroke scale scores available.
The median age of the patients was 80; 57% were female and 86% were white. The mean NIH stroke scale score was 8.23, and the median was 5.
Overall, 14.5% of the patients died from any cause within 30 days, including 5.8% who died during the initial hospitalization.
Including the NIH stroke scale scores in a model created using just administrative claims data significantly improved the prediction of 30-day mortality and resulted in a reshuffling of hospital rankings.
When the hospitals were stratified into the 20% worst-performing, 20% best-performing, and 60% average-performing, about one-quarter (26.3%) moved to a different category after inclusion of stroke severity.
Of the 39 hospitals placed in the top 5% in the model that did not include stroke severity, only 23 remained in that elite group after inclusion of the NIH stroke scale scores. And of the 40 worst hospitals in the model that did not include stroke severity, nearly half (47.5%) were no longer at the bottom after inclusion of the NIH stroke scale scores.
Overall, the addition of stroke severity resulted in a net reclassification improvement of 93.1% and an integrated discrimination improvement of 15% (P<0.001 for both).
"The results of the study ... clearly highlight the importance of incorporating information on stroke severity when conducting health outcomes research in stroke," according to Tobias Kurth, MD, ScD, of the Bordeaux Segalen University in France, and Mitchell Elkind, MD, of Columbia University in New York City.
"Excluding this information," they wrote in an accompanying editorial, "will lead to incorrect ranking of hospital performance by failing to consider that hospitals care for different patient populations."
Fonarow and colleagues acknowledged some limitations of the study, including the possibility that the patients in the study may not be representative of all patients hospitalized for acute stroke, the inclusion of patients in fee-for-service Medicare only, the lack of adjustment for therapies provided, and the lack of information on clinical outcomes aside from mortality.
In addition, they noted that requiring the assessment of the NIH stroke scale in every patient could present some challenges.
"The time and expertise needed to perform even a short standardized stroke severity assessment and ensuring these data are accurately abstracted and entered into the Hospital Compare data collection system are important barriers that will need to be overcome," they wrote.
Primary source: Journal of the American Medical Association
Source reference: Fonarow G, et al "Comparison of 30-day mortality models for profiling hospital performance in acute ischemic stroke with versus without adjustment for stroke severity" JAMA 2012; 308: 257-264.
Additional source: Journal of the American Medical Association