Skip to main content

17-07-2012 | Stroke | Article

Hospital performance at mercy of stroke severity


Free abstract

MedWire News: Accounting for stroke severity has a major impact on the accuracy of models that rank the performance of hospitals in acute stroke care, show findings published in JAMA.

"Considered at the level of the individual patient, this result does not seem surprising: a patient who has a relatively more severe stroke is at increased risk of death," say editorialists Tobias Kurth (INSERM, Bordeaux, France) and Mitchell Elkind (Columbia University College of Physicians and Surgeons, New York, USA).

Nonetheless, they stress the importance of the findings for health outcomes research in stroke. Current models do not include stroke severity, but "excluding this information will lead to incorrect ranking of hospital performance by failing to consider that hospitals care for different patient populations," they say.

In this study, stroke severity was based on the National Institutes of Health Stroke Scale (NIHSS) score in 127,950 Medicare beneficiaries treated in hospitals participating in the Get With The Guidelines (GWTG)-Stroke initiative. The average NIHSS score was 8.23.

In all, 14.5% of patients died within 30 days of stroke onset. The base risk model included 87 variables drawn from claims data and was 77.2% accurate for predicting 30-day mortality, report Gregg Fonarow (University of California, Los Angeles, USA) and team.

But addition of NIHSS score significantly improved this, to 86.4%. Indeed, the NIHSS score on its own distinguished between patients who died or survived by day 30 with 82.2% accuracy.

Adding the NIHSS score to the base risk model had a major impact on hospital rankings, with 57.7% of those classified as having worse than expected stroke mortality being reclassified to "as expected." There were 782 hospitals in the analysis, and adding the NIHSS score changed their rankings by a median of 79 places.

With the base risk model, 39 hospitals were in the top fifth performance percentile. Just 23 of these hospitals remained so after addition of the NIHSS score to the model, and a different 16 hospitals moved into the top category. There was a similar effect on hospitals in the bottom fifth performance percentile.

Kurth and Elkind cautioned that the data come from GWTG-Stroke hospitals, which have a special interest in stroke care. "At other hospitals, stroke severity may not be routinely collected or may be collected less accurately," they say.

They also envisage a scenario in which some hospitals may "systematically overestimate" stroke severity if this becomes an important measure for ranking hospitals.

MedWire ( is an independent clinical news service provided by Springer Healthcare Limited. © Springer Healthcare Ltd; 2012

By Eleanor McDermid