The ratio of observed-to-expected mortality as a quality of care indicator in non-surgical VA patients.

This study attempts to validate the use of the observed-to-expected (O/E) mortality ratio as an indicator of quality of care. The primary objective is to determine whether medical records of 111 patients who died in Veterans Affairs (VA) hospitals with high overall ratios of observed-to-expected mortality rates show greater evidence that life might have been meaningfully prolonged with more expert care than the records of matched patients who died in VA hospitals with low O/E ratios. Patient matching criteria included: age, diagnosis responsible for length of stay, type of VA hospital, and the mortality probability predicted by logistic regression. Expert physicians blindly and independently reviewed pairs of medical records, assessing comparative care on a symmetrical, nine-alternative visual analog scale. A slight shift in distribution toward better care in low-ratio hospitals was not statistically significant. Results of an additional analysis, not dependent on pairing, showed that preventability of death is more strongly related to physicians' estimates of mortality risk at admission, whether transferred from a nursing home, do-not-resuscitate status, and accuracy of discharge coding than to VA Medical Center O/E ratios.