Correlation between unadjusted and adjusted physician-level estimates of resource use and clinical outcomes in GIM
| Hospital | Length of stay (Corr, p value) | Readmission (Corr, p value) | Mortality (Corr, p value) | Imaging (Corr, p value) |
| All hospitals | 0.96 (<0.001) | 0.96 (<0.001) | 0.81 (<0.001) | 0.96 (<0.001) |
| A | 0.94 (<0.001) | 0.98 (<0.001) | 0.75 (<0.001) | 0.99 (<0.001) |
| B | 0.94 (<0.001) | 0.91 (<0.001) | 0.86 (<0.001) | 0.89 (<0.001) |
| C | 0.87 (<0.001) | 0.96 (<0.001) | 0.92 (<0.001) | 0.92 (<0.001) |
| D | 0.98 (<0.001) | 0.90 (<0.001) | 0.76 (0.001) | 0.96 (<0.001) |
| E | 0.96 (<0.001) | 0.93 (<0.001) | 0.81 (<0.001) | 0.95 (<0.001) |
| F | 0.98 (<0.001) | 0.97 (<0.001) | 0.77 (<0.001) | 0.97 (<0.001) |
| G | 0.98 (<0.001) | 0.94 (<0.001) | 0.84 (<0.001) | 0.98 (<0.001) |
The correlation was calculated between unadjusted and adjusted estimates of each measure. The following patient baseline covariates were used in adjustment: age, sex, Charlson Comorbidity Index score, day of admission, time of admission, Clinical Classifications Software (CCS) classification of discharge diagnosis, hospitalisation within 30 previous days, Laboratory-based Acute Physiology Score and study year. To examine the correlation across all hospitals as a single measure, all physicians were included and we adjusted for hospital fixed effects.
Corr, correlation coefficient; GIM, general internal medicine.