Hospital standardized mortality ratios

We do not agree with Kaveh Shojania and Alan Forster's assessment of the value of reporting hospital standardized mortality ratios.1 Our view is that the public reporting of hospital standardized mortality ratios in Canada provides a useful and much needed focus on the quality of health care.

The authors criticize the validity of hospital standardized mortality ratios on the basis that they “correlate weakly with other measures of quality of care” and cite as an example one of the findings from a 1987 US study by Dubois and colleagues.2 However, Dubois and colleagues reported in the same study that “detailed reviews by physicians of the records of patients who died during hospitalization revealed a higher rate of preventable deaths in the high [outlier hospitals] than in the low [outlier hospitals].”

In some circumstances, process and outcome measures would be expected to be correlated, but in others they would not, for a number of valid reasons. When these 2 types of measures produce different results, we should not treat the process measures as the gold standard against which a “big dot” (i.e., broad-based) outcome measure like the hospital standardized mortality ratio should be assessed. Both types of measures have strengths and limitations and as such it is important that both be considered when examining the quality of health care within a hospital.

The authors also criticized the precision of the hospital standardized mortality ratio on the basis that “random variation likely accounts for much of the observed differences in mortality among institutions.” In our report of hospital standardized mortality ratios,3 we presented the hospital standardized mortality ratios results and confidence intervals only for large hospitals and regions to minimize the effect of random variation and to inform users of the level of precision associated with a given hospital standardized mortality ratio.

Producing hospital standardized mortality ratios for Canadian hospitals responds to the need for a “big dot” measure of the quality of health care. With an understanding of their limitations and in conjunction with other measures and information, hospital standardized mortality ratios can be used for their intended purposes. Within this context, the hospital standardized mortality ratio is both a valid and useful measure. Our work on developing more and improved quality measures is ongoing.