Characteristics of all GIM hospital admissions and categorised across physician quartiles of in-hospital mortality
| Characteristic | All hospital admissions* | Physician mortality quartile 1 | Physician mortality quartile 2 | Physician mortality quartile 3 | Physician mortality quartile 4 | Maximum standardised difference† |
| Admissions, n | 103 085 | 19 781 | 24 298 | 27 568 | 19 786 | |
| Mortality, n (%) | 4474 (4.3) | 761 (3.8) | 1069 (4.4) | 1416 (5.1) | 1228 (6.2) | 0.11 |
| Age, median (IQR) | 73 (56, 84) | 71 (54, 83) | 72 (55, 84) | 73 (56, 84) | 73 (56, 84) | 0.07 |
| Female sex, n (%) | 52 192 (50.6) | 50.6 | 50.3 | 51.0 | 50.2 | 0.02 |
| High Charlson Comorbidity Index score‡, n (%) | 43 910 (42.6) | 42.9 | 41.9 | 42.1 | 43.0 | 0.02 |
| Day of admission (weekday), n (%) | 75 527 (73.3) | 74.1 | 72.5 | 73.3 | 74.6 | 0.05 |
| Time of admission (daytime)§, n (%) | 22 283 (21.6) | 21.2 | 23.7 | 19.6 | 22.4 | 0.10 |
| Hospitalisation in previous 30 days¶, n (%) | 11 091 (11.1) | 11.3 | 11.3 | 11.2 | 11.5 | 0.01 |
| LAPS**, mean (SD) | 20.2 (17.5) | 18.7 (16.7) | 19.3 (16.9) | 20.3 (17.1) | 20.0 (17.2) | 0.10 |
| Discharge diagnoses††, n (%) | ||||||
| Pneumonia | 5602 (5.4) | 1086 (5.5) | 1380 (5.7) | 1531 (5.6) | 1095 (5.5) | 0.01 |
| Urinary tract infection | 5517 (5.4) | 1061 (5.4) | 1190 (4.9) | 1452 (5.3) | 1022 (5.2) | 0.02 |
| Heart failure | 5161 (5.0) | 980 (5.0) | 1335 (5.5) | 1337 (4.8) | 1027 (5.2) | 0.03 |
| COPD | 4634 (4.5) | 876 (4.4) | 1181 (4.9) | 1237 (4.5) | 842 (4.3) | 0.03 |
| Stroke | 3333 (3.2) | 612 (3.1) | 819 (3.4) | 838 (3.0) | 643 (3.2) | 0.02 |
| Gastrointestinal haemorrhage | 3081 (3.0) | 499 (2.5) | 721 (3.0) | 785 (2.8) | 508 (2.6) | 0.03 |
| Delirium, dementia, cognitive disorder | 2940 (2.9) | 510 (2.6) | 695 (2.9) | 763 (2.8) | 520 (2.6) | 0.02 |
| Fluid and electrolyte disorder | 2765 (2.7) | 490 (2.5) | 661 (2.7) | 754 (2.7) | 495 (2.5) | 0.02 |
| Intestinal infection | 2596 (2.4) | 505 (2.6) | 601 (2.5) | 737 (2.7) | 511 (2.6) | 0.01 |
| Diabetes mellitus with complication | 2454 (2.4) | 485 (2.5) | 577 (2.4) | 594 (2.2) | 465 (2.4) | 0.02 |
Physicians were categorised into quartiles within each hospital based on the rate of inpatient mortality and then quartiles were pooled across hospitals
*The data reported for all hospital admissions reflect the entire study cohort (patients admitted or discharged from GIM), whereas the data reported for the physician mortality quartiles reflect only the admissions used in the calculation of mortality (only patients initially admitted to GIM).
†The maximum standardised difference across physician mortality quartiles was calculated using the two quartiles that had the largest standardised difference for each variable (calculated as the mean difference divided by SD).
‡Charlson Comorbidity Index score of 2 or greater was considered high comorbidity.
§Admissions between 08:00 and 17:00 were considered ‘daytime’ admissions.
¶To a GIM ward at a participating hospital.
**Laboratory-based Acute Physiology Score (LAPS) is a validated score to predict inpatient mortality, which ranges from 0 to 256 points, with higher scores indicating a higher risk of mortality.
††The primary discharge diagnosis for each admission was categorised using the Clinical Classifications Software based on International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes.
COPD, chronic obstructive pulmonary disease; GIM, general internal medicine.