Measuring Hospital Performance Using Mortality Rates: An Alternative to the RAMR

The risk-adjusted mortality rate (RAMR) is used widely by healthcare agencies to evaluate hospital performance. The RAMR is insensitive to case volume and requires a confidence interval for proper interpretation, which results in a hypothesis testing framework. Unfamiliarity with hypothesis testing can lead to erroneous interpretations by the public and other stakeholders. We argue that screening, rather than hypothesis testing, is more defensible. We propose an alternative to the RAMR that is based on sound statistical methodology, easier to understand and can be used in large-scale screening with no additional data requirements.

We use an upper-tail probability to screen for hospitals performing poorly and a lower-tail probability to screen for hospitals performing well. Confidence intervals and hypothesis tests are not needed to compute or interpret our measures. Moreover, unlike the RAMR, our measures are sensitive to the number of cases treated.

To demonstrate our proposed methodology, we obtained data from the New York State Department of Health for 10 Inpatient Quality Indicators (IQIs) for the years 2009-2013. We find strong agreement between the upper tail probability (UTP) and the RAMR, supporting our contention that the UTP is a viable alternative to the RAMR.

We show that our method is simpler to implement than the RAMR and, with no need for a confidence interval, it is easier to interpret. Moreover, it will be available for all hospitals and all diseases/conditions regardless of patient volume.