Table 2. Bivariable and Multivariable Associations of Baseline Characteristics With Burnout.
| Characteristic | Bivariable Models | Multivariable Modela | ||
|---|---|---|---|---|
| Estimate (95% CI) | P Value | Estimate (95% CI) | P Value | |
| Sex | ||||
| Men | 1 [Reference] | .004 | NA | NA |
| Women | 1.53 (1.15-2.05) | NA | NA | |
| Parenting and domestic tasks weekly, h | ||||
| <22 | 1 [Reference] | .02 | 1 [Reference] | .01 |
| 22-<34 | 1.16 (0.75-1.78) | 1.29 (0.80-1.94) | ||
| 34-<44 | 1.42 (0.93-2.18) | 1.46 (0.94-2.26) | ||
| ≥44 | 1.85 (1.23-2.79) | 1.98 (1.30-3.01) | ||
| Vacation, wk | ||||
| <3 | 1.36 (1.02-1.82) | .04 | 1.43 (1.06-1.93) | .02 |
| ≥3 | 1[Reference] | 1 [Reference] | ||
| Weekly hours spent on patient care | ||||
| <5 | 1 [Reference] | .09 | 1 [Reference] | .04 |
| 5-<10 | 1.27 (0.74-1.72) | 1.13 (0.73-1.74) | ||
| 10-<15 | 1.38 (0.90-2.14) | 1.46 (0.93-2.27) | ||
| ≥15 | 1.71 (1.08-2.70) | 1.83 (1.14-2.92) | ||
| Work Climate Scale score | ||||
| ≤3 | 2.09 (1.29-3.40) | .01 | 2.00 (1.21-3.31) | .02 |
| >3-≤4 | 1.10 (0.81-1.50) | 1.07 (0.78-1.47) | ||
| >4-≤5 | 1 [Reference] | 1 [Reference] | ||
Abbreviations: NA, not applicable; NIH, National Institutes of Health.
Logistic regression model of burnout (defined as a Copenhagen Burnout Inventory work-related burnout scale score of ≥50) as the dependent variable. Independent variables measured at baseline by self-report and considered for inclusion were: sex, age, race/ethnicity, marital status, professional degree, specialty, academic rank, weekly hours spent on parenting and domestic tasks, nightly hours of sleep, presence and number of children, time spent on vacation, typical weekly work hours, typical weekly hours in patient care activities, frequency of K award mentor communications, satisfaction with primary K award mentor, and perceived work climate. Independent variables measured at baseline using public information were grant type, year of grant award, tier of institution at time of K award (based on total extramural NIH funding as previously described), and K-awarding NIH institute tier (based on the total amount of R01 awards granted by the institute). The final model was created after iterative removal of variables with model recalculation until only important covariates remained, as measured by comparing the Akaike information criterion between nested models. The model listed in the table is the final model; sex was not selected for inclusion in the final multivariable model but is included in the table to demonstrate its bivariable association because it was the primary independent variable of interest in the analysis.