Another Problem With Quality Measures

We all want good quality in health care and to see quality regularly improving.  But many issues have been raised about either the development or application of various quality measures.  An article in the Journal of the American Medical Association examines a measure regarding prevention of venous thromboembolism and finds that surveillance bias undermines the utility of the measure.   (JAMA Article)   Surveillance bias exists when variation in scores or results is due variation in screening or looking for the thing to be measured.  In this example, a hospital that is more aggressive in looking for and identifying potential VTE cases, will likely find more and therefore end up with a worse score on the quality measure.  Paradoxically, a hospital that didn’t look at all might have fewer cases and a better score, but would clearly be delivering worse care to its patients.  The researchers’ analysis included a general quality measure for the hospital, rates of VTE prophylaxis and hospital risk-adjusted VTE rates.  Medicare data was the primary information source.  Overall hospitals have high compliance with VTE prophylaxis, around 94.5%.  As might be expected, hospitals with a higher general quality measure had higher VTE prophylaxis rates.  But those with higher prophylaxis rates also had higher VTE event rates.  There was an unexpected association between supposed higher quality and worse outcomes.  But the higher quality hospitals also had much higher rates of VTE imaging studies, suggesting that they were simply finding more VTEs, not that they actually had more than the lower-quality hospitals.  The results should cause policymakers to stop using this quality measure, as it may actually encourage hospital to provide worse care by not looking for VTEs that they then have to report.