Table 3.
Regression coefficient estimates (on log cost-ratio scalea), 95% confidence intervals and p-values from unadjusted and adjusted gamma Generalized Estimating Equation models for cost per day outcome
| Category | Unadjusted Modelsb | Adjusted Modelc | ||
|---|---|---|---|---|
| Coefficient (95% CI) | P-value | Coefficient (95% CI) | P-value | |
| Palliative Care Intensity | ||||
| No involvement | Ref | Ref | ||
| Low involvement | −0.18 (− 0.21: − 0.15) | < 0.001 | − 0.15 (− 0.17: − 0.12) | < 0.001 |
| Medium involvement | − 0.65 (− 0.69: − 0.62) | < 0.001 | − 0.6 (− 0.63: − 0.57) | < 0.001 |
| Age | ||||
| 18–40 | Ref | Ref | ||
| 41–55 | − 0.22 (− 0.27: − 0.18) | < 0.001 | − 0.15 (− 0.18: − 0.11) | < 0.001 |
| 56–70 | −0.32 (− 0.36: − 0.27) | < 0.001 | −0.23 (− 0.27: − 0.19) | < 0.001 |
| 71–85 | −0.45 (− 0.5: − 0.39) | < 0.001 | −0.38 (− 0.42: − 0.34) | < 0.001 |
| 86–104 | −0.64 (− 0.7: − 0.59) | < 0.001 | −0.58 (− 0.62: − 0.54) | < 0.001 |
| Sex | ||||
| F | Ref | Ref | ||
| M | 0.08 (0.07: 0.09) | < 0.001 | 0.04 (0.03: 0.05) | < 0.001 |
| Province | ||||
| ON | Ref | Ref | ||
| AB | −0.02 (−0.19: 0.14) | 0.79 | 0.02 (−0.07: 0.11) | 0.66 |
| BC | 0.02 (−0.1: 0.13) | 0.78 | 0.03 (−0.05: 0.11) | 0.48 |
| MB | −0.02 (− 0.14: 0.11) | 0.81 | 0 (− 0.08: 0.09) | 0.96 |
| NB | −0.02 (− 0.35: 0.31) | 0.91 | − 0.01 (− 0.24: 0.21) | 0.9 |
| NL | 0.05 (−0.1: 0.2) | 0.52 | 0.04 (−0.06: 0.13) | 0.44 |
| NS | −0.19 (− 0.61: 0.24) | 0.39 | − 0.11 (− 0.37: 0.15) | 0.4 |
| PE | 0.48 (0.17: 0.8) | 0.002 | 0.39 (0.09: 0.69) | 0.01 |
| SK | 0.06 (−0.03: 0.14) | 0.23 | 0.07 (0.01: 0.13) | 0.02 |
| Van Walraven Index | ||||
| −14-3 | Ref | Ref | ||
| 4–8 | −0.07 (−0.08: − 0.05) | < 0.001 | − 0.04 (− 0.05: − 0.03) | < 0.001 |
| 9–15 | − 0.1 (− 0.12: − 0.08) | < 0.001 | − 0.09 (− 0.1: − 0.07) | < 0.001 |
| 16+ | − 0.25 (− 0.28: − 0.22) | < 0.001 | − 0.2 (− 0.22: − 0.17) | < 0.001 |
Abbreviations: F Female, AB Alberta, BC British Columbia, MB Manitoba, NB New Brunswick, NL Newfoundland, NS Nova Scotia, ON Ontario, PE Prince Edward Island, SK Saskatchewan
aAs explained in previous work, these “coefficient estimates [are] derived from generalized linear models with logarithmic link functions [and] are interpretable as the logarithm of the relative change in mean cost associated with a one-unit change in the predictor variable” (Austin PC, Ghali WA, Tu JV. A comparison of several regression models for analysing cost of CABG surgery. Stat Med. 2003 Sep 15;22 (17):2799–815. doi: 10.1002/sim.1442. PMID: 12939787)
bThis column displays unadjusted models for each variable
cModel adjusted for age, sex, province, and Van Walraven index