Table 2.
Treatmenta | Unadjusted Results (n=5344)b | Adjusted Results (n=5435)bc | ||||
---|---|---|---|---|---|---|
% | Risk Difference (95% CI) | P Value | % | Risk Difference (95% CI) | P Value | |
Dialysis treatment status | ||||||
No dialysis | 58.8 | Reference | 54.7 | Reference | ||
Acute dialysis | 52.1 | −6.7 (−10.9 to −2.5) | 0.002 | 53.6 | −1.1 (−6.0 to 3.7) | 0.65 |
Maintenance dialysis | 48.2 | −10.6 (−13.5 to −7.6) | <0.001 | 50.8 | −3.9 (−7.5 to −0.4) | 0.03 |
High-intensity treatment | ||||||
2+wks in hospital in last 90 d | ||||||
No | 59.1 | Reference | 55.8 | Reference | ||
Yes | 50.3 | −8.7 (−11.2 to -6.2) | <0.001 | 51.0 | −4.9 (−7.6 to −2.1) | <0.001 |
ICU admission in last 30 d | ||||||
No | 57.4 | Reference | 54.2 | Reference | ||
Yes | 51.6 | −8.9 (−8.7 to −2.7) | <0.001 | 52.3 | −1.9 (−5.1 to 1.3) | 0.25 |
Intensive procedure in last 30 d | ||||||
No | 57.6 | Reference | 54.7 | Reference | ||
Yes | 48.5 | −9.1 (−12.2 to −6.0) | <0.001 | 50.6 | −4.1 (−7.3 to −0.9) | 0.01 |
Death in intensive care unit | ||||||
No | 57.6 | Reference | 55.0 | Reference | ||
Yes | 47.5 | −10.2 (−13.5 to −6.8) | <0.001 | 49.4 | −5.6 (−9.1 to −2.2) | 0.001 |
Palliative and hospice care | ||||||
Palliative care consultation in last 90 d | ||||||
No | 52.7 | Reference | 50.9 | Reference | ||
Yes | 58.3 | 5.6 (3.0 to 8.3) | <0.001 | 57.2 | 6.4 (3.7 to 9.1) | <0.001 |
Hospice services at time of death | ||||||
No | 48.9 | Reference | 48.6 | Reference | ||
Yes | 63.9 | 15.1 (11.9 to 18.2) | <0.001 | 61.5 | 12.8 (9.4 to 16.3) | <0.001 |
Logistic regression with no dialysis as reference group for acute and maintenance dialysis and no receipt as reference group for end-of-life treatment variables, standard errors adjusted for facility-level clustering; presented are the predicted probabilities over the distribution of covariates in the respondent sample; 95% confidence intervals (95% CIs) and P values are for differences in the predicted probabilities.
Dichotomized as “excellent” versus all other responses.
Denominators differ for unadjusted and adjusted models because missing items were imputed in adjusted models only (Supplemental Appendix 2, Supplemental Table 4).
Model adjusted for race, age, sex, next of kin, region, facility complexity, year of death, and individual comorbidities, and weighted for unit nonresponse.