Table 3. Adjusted 30-Day Mortality of Hospitalizations Treated by Locum Tenens and Non–Locum Tenens Physicians During 2009-2014, Stratified by Hospital Characteristics.
| Characteristic | Hospitalized Patients Treated by Non–Locum Tenens Physiciansa | Hospitalized Patients Treated by Locum Tenens Physiciansb | Difference, % (95% CI) |
Odds Ratio (95% CI) |
P Valuec | P Value for Interactiond | ||
|---|---|---|---|---|---|---|---|---|
| No. of Hospitalizations | Adjusted 30-d Mortality (95% CI) |
No. of Hospitalizations | Adjusted 30-d Mortality (95% CI) |
|||||
| Overall | 1 780 398 | 8.70 (8.69 to 8.70) |
38 475 | 8.83 (8.52 to 9.14) |
0.14 (−0.18 to 0.45) |
1.02 (0.98 to 1.06) |
.40 | NA |
| Hospital sizee | ||||||||
| Small | 138 741 | 9.12 (9.08 to 9.15) |
6807 | 9.42 (8.67 to 10.18) |
0.31 (−0.49 to 1.10) |
1.04 (0.94 to 1.15) |
.45 | [Reference] |
| Medium | 956 670 | 8.91 (8.90 to 8.91) |
24 574 | 8.92 (8.55 to 9.30) |
0.02 (−0.37 to 0.41) |
1.00 (0.95 to 1.06) |
.92 | .55 |
| Large | 684 987 | 8.31 (8.31 to 8.32) |
7094 | 8.76 (7.99 to 9.54) |
0.45 (−0.34 to 1.23) |
1.07 (0.96 to 1.19) |
.26 | .73 |
| Hospital type | ||||||||
| Public | 200 334 | 9.25 (9.23 to 9.27) |
5204 | 8.72 (7.96 to 9.48) |
−0.53 (−1.31 to 0.25) |
0.93 (0.84 to 1.04) |
.18 | [Reference] |
| For-profit | 272 669 | 8.43 (8.41 to 8.45) |
6442 | 8.52 (7.74 to 9.30) |
0.09 (−0.71 to 0.89) |
1.01 (0.90 to 1.14) |
.83 | .30 |
| Non-profit | 1 307 395 | 8.67 (8.66 to 8.68) |
26 829 | 8.94 (8.57 to 9.32) |
0.28 (−0.11 to 0.66) |
1.04 (0.99 to 1.09) |
.16 | .08 |
| Locum intensity, by tercilef | ||||||||
| Lower | 891 651 | 8.53 (8.53 to 8.53) |
1397 | 11.63 (9.96 to 13.30) |
3.10 (1.43 to 4.77) |
1.47 (1.22 to 1.77) |
<.001 | [Reference] |
| Middle | 541 704 | 8.84 (8.83 to 8.85) |
6537 | 8.72 (8.05 to 9.39) |
−0.12 (−0.80 to 0.55) |
0.98 (0.89 to 1.08) |
.72 | <.001 |
| Upper | 347 043 | 8.89 (8.86 to 8.92) |
30 541 | 8.95 (8.60 to 9.29) |
0.06 (−0.32 to 0.43) |
1.01 (0.96 to 1.06) |
.76 | <.001 |
Locum tenens physicians provided temporary coverage for 4123 non–locum tenens physicians during the study period.
Includes patients treated by 44 520 non–locum tenens physicians.
P value is for comparison of 30-day mortality among hospitalized patients treated by locum tenens and non–locum tenens physicians in each subgroup.
P value for interaction assesses how the difference in 30-day mortality between hospitalized patients treated by locum tenens and non–locum tenens physicians is modified by each category subgroup and was obtained from a formal test of interaction.
Small hospitals are defined as those with fewer than 100 beds; medium hospitals as those with between 100 and 399 beds; and large hospitals as those with 400 or more beds, according to the annual American Hospital Association survey.
Among hospitals with any locum tenens use, we computed the percentage of a hospital’s patients treated by a locum tenens physician, and divided hospitals into terciles along that metric. The lower tercile hospitals involved locum tenens physicians in 0.01% to less than 0.45% of their admissions, the middle tercile in 0.45% to less than 2.5% of admissions, and the upper tercile in at least 2.5% of all admissions. All estimates were adjusted for patient age, sex, race/ethnicity, month of year of admission, day of week of admission, Medicaid eligibility, indicators for 31 Elixhauser conditions, and the admitting Major Diagnostic Category, and hospital fixed effects, with robust standard errors clustered at the hospital level.