Table 3.
Associations Between Intervention and Ratings of Quality of Trainees’ Communication and End-of-Life Carea
| Outcome/ Predictors |
Patients
|
Family
|
Clinicians
|
||||||
|---|---|---|---|---|---|---|---|---|---|
| Trainees Evaluated/ Evaluations, No. |
b (95% CI)b |
P Value |
Trainees Evaluated/ Evaluations, No. |
b (95% CI)b |
P Value |
Trainees Evaluated/ Clinician- Evaluators/ Evaluations, No. |
b (95% CI)b |
P Value |
|
| QOC scorec | 194/1224 | 116/543 | 213/803/2199 | ||||||
|
| |||||||||
| Intervention | 0.38 (−0.14 to 0.90) | .15 | 0.11 (−0.77 to 0.98) | .81 | 0.19 (−0.08 to 0.47) | .17 | |||
|
| |||||||||
| Group | −0.27 (−0.65 to 0.12) | .18 | 0.76 (0.18 to 1.35) | .01 | 0.01 (−0.23 to 0.24) | .94 | |||
|
| |||||||||
| Study period | 0.03 (−0.29 to 0.34) | .86 | 0.07 (−0.48 to 0.62) | .81 | 0.06 (−0.13 to 0.24) | .56 | |||
|
| |||||||||
| Study sited | 0.34 (0.07 to 0.62) | .02 | 0.25 (−0.17 to 0.68) | .25 | 0.08 (−0.16 to 0.33) | .51 | |||
|
| |||||||||
| Stratum | .001 | .52 | .003 | ||||||
|
| |||||||||
| R1 | 0.000 | 0.000 | 0.000 | ||||||
|
| |||||||||
| R2 | 0.52 (0.131 to 0.920) | .009 | −0.29 (−0.79 to 0.22) | .26 | 0.25 (−0.03 to 0.54) | .08 | |||
|
| |||||||||
| R3/fellow | 0.50 (0.14 to 0.86) | .007 | −0.02 (−0.52 to 0.48) | .93 | 0.47 (0.20 to 0.74) | <.001 | |||
|
| |||||||||
| NP | 0.78 (0.46 to 1.10) | <.001 | |||||||
|
| |||||||||
| Overall QOCe | 189/1119 | 114/518 | 211/794/2139 | ||||||
|
| |||||||||
| Intervention | 0.22 (−0.74 to 1.18) | .65 | 0.06 (−1.26 to 1.37) | .94 | 0.14 (−0.16 to 0.42) | .36 | |||
|
| |||||||||
| Group | −0.54 (−1.28 to 0.20) | .15 | 0.46 (−0.54 to 1.46) | .37 | 0.04 (−0.30 to 0.22) | .74 | |||
|
| |||||||||
| Study period | 0.02 (−0.58 to 0.62) | .95 | −0.24 (−1.03 to 0.54) | .54 | 0.08 (−0.12 to 0.28) | .45 | |||
|
| |||||||||
| Study sited | 0.80 (0.28 to 1.32) | .002 | 0.92 (0.21 to 1.64) | .01 | 0.02 (−0.24 to 0.28) | .87 | |||
|
| |||||||||
| Stratum | <.001 | .27 | .009 | ||||||
|
| |||||||||
| R1 | 0.000 | 0.000 | 0.000 | ||||||
|
| |||||||||
| R2 | 1.19 (0.38 to 2.00) | .004 | −0.65 (−1.76 to 0.46) | .25 | 0.29 (−0.03 to 0.62) | .08 | |||
|
| |||||||||
| R3/fellow | 0.79 (0.13 to 1.45) | .02 | 0.30 (−0.61 to 1.20) | .52 | 0.45 (0.15 to 0.75) | .004 | |||
|
| |||||||||
| NP | 1.92 (0.68 to 3.17) | .002 | |||||||
|
| |||||||||
| QEOLCf | 196/1250 | 117/546 | 213/788/2135 | ||||||
|
| |||||||||
| Intervention | 0.25 (−0.27 to 0.77) | .34 | 0.06 (−0.70 to 0.82) | .88 | 0.18 (−0.10 to 0.46) | .20 | |||
|
| |||||||||
| Group | −0.39 (−0.79 to 0.02) | .06 | 0.26 (−0.33 to 0.84) | .38 | −0.01 (−0.25 to 0.23) | .95 | |||
|
| |||||||||
| Study period | 0.10 (−0.23 to 0.43) | .55 | −0.08 (−0.59 to 0.42) | .74 | 0.04 (−0.15 to 0.23) | .68 | |||
|
| |||||||||
| Study sited | 0.38 (0.11 to 0.64) | .006 | 0.44 (−0.02 to 0.90) | .06 | −0.01 (−0.26 to 0.23) | .92 | |||
|
| |||||||||
| Stratum | 0 | .36 | <.001 | ||||||
|
| |||||||||
| R1 | 0.000 | 0.000 | 0.000 | ||||||
|
| |||||||||
| R2 | 0.62 (0.22 to 1.02) | .003 | −0.30 (−0.80 to 0.21) | .25 | 0.32 (0.03 to 0.62) | .03 | |||
|
| |||||||||
| R3/fellow | 0.46 (0.12 to 0.80) | .008 | 0.19 (−0.43 to 0.81) | .55 | 0.55 (0.27 to 0.82) | <.001 | |||
|
| |||||||||
| NP | 0.52 (0.28 to 0.76) | <.001 | |||||||
Abbreviations: NP, nurse practitioner; QEOLC, Quality of End-of-life Care questionnaire; QOC, Quality of Communication questionnaire; R1, first-year resident; R2, second-year resident; R3, third-year resident.
Results with mean scores shown in eTables 5 through 7 in Supplement.
For linear regression models, the b coefficient represents expected number of points change in the observed outcome variable with a 1-point increase in the predictor. For Tobit regression models, it represents the expected number of points change in an uncensored latent variable represented by the observed outcome with a 1-point increase in the predictor.
Estimates were based on robust linear regression. Intercepts (estimated preintervention mean for year-1 residents at University of Washington in the control condition) for the evaluator groups were, respectively, 6.001, 6.445, and 7.343.
0 = University of Washington; 1 = Medical University of South Carolina.
This outcome was censored from above and modeled with Tobit regression. It was modeled as a linear outcome in the cross-classified clinician sample. Intercepts (estimated preintervention mean for first-year residents at University of Washington in the control condition) for the 3 evaluator groups were, respectively, 9.295, 9.221, and 7.502.
For patients and family members, this outcome had a strong ceiling effect and was modeled as censored from above, using Tobit regression. Intercepts (estimated mean value during the preintervention period for first-year residents at University of Washington in the control condition) for the 3 evaluator groups were, respectively, 8.197, 8.244, and 7.472.