Background:
Measuring health care quality is increasingly important. Reliably assessing quality requires a review of the care provided and linking the care to a provider. Payers, patients, and regulators have access to quality reporting at the institutional level. There is a movement toward providing measures for individual providers. For hospitalized patients, these measures are usually attributed to the discharging physician. Care of most patients by hospitalists is spread across multiple physicians because of handoffs and the involvement of consultants, so assigning responsibility for outcomes to a single discharging physician raises concerns We sought to determine if quality measures are stable depending on attribution of liability to a given physician as the admitting or discharging physician.
Methods:
We examined quality and efficiency metrics for hospitalists including observed/expected (O/E) length of stay (LOS), O/E cost, O/E mortality, and 30‐day readmission rates. For each hospitalist with at least 125 admissions and discharges in a year, in a single practice, we calculated her or his metrics as the admitting and the discharging physician. We then assigned ranked tertiles for each physician in both roles. We measured change in tertile (0‐2) and absolute change in each metric for every physician when changing from admitting to discharging physician.
Results:
In all metrics, physicians had a tertile change when changing from admitting to discharging physician. Four of 10 physicians changed tertiles in 30‐day readmissions and 2 of those changed by 2 tertites. Eight of 10 physicians changed tertiles in O/E costs, and 2 of those changed by 2 tertiles. The greatest absolute changes for each outcome were: O/E cost, 0.07; O/E LOS, 0.16; O/E mortality, 1.07; and readmission rate, 7.32%.
Conclusions:
For hospitalized patients cared for in a typical hospitalist model, these metrics were unstable depending on the roles used to attribute responsibility. We were primarily interested in percentile change as that is how most quality metrics are reported. Although absolute changes may appear small, these are normalized metrics representing large effects. The largest change in costs for a physician was a $713,034/year reduction as the discharging physician when compared with the costs as the admitting physician, This instability is likely related to attributing liability to a single individual when it is more likely shared among many individuals. These results suggest that current methods of assigning liability to individuals in this setting are inadequate. Alternative approaches to assigning metric attribution could be to consider weighting based on time spent with the patient, sharing attribution with consultants, or weighing liability based on involvement in critical decisions, Although these are all more complex models, they may better assess quality at the provider level.
Author Disclosure:
A. Ansari, none; M. Wall, none; C. Whelan, none.