Table 4.

Predictors of kidney transplantation and waitlist-related outcomes, multivariable (competing risk) analysis and resulting subdistribution hazard ratios.

	Kidney transplantation in total cohort		Process toward kidney transplantation
			Waitlisted for transplant		Among waitlisted patients, received transplant
No. of incident events/Total no. of patientsa	415/1152b		656/1152		415/656
Multivariable analysis	HR (CI)	P	HR (CI)	P	HR (CI)	P
Race/ethnicity by current substance use groups c
White, no substance use (referent)	---		---		---
Black, no substance use	0.76 (0.56, 1.02)	0.070	0.73 (0.58, 0.91)	0.005	0.99 (0.74, 1.32)	0.949
Other, no substance use	1.00 (0.70, 1.43)	0.980	0.79 (0.55, 1.13)	0.199	0.96 (0.68, 1.35)	0.804
White, current substance use	0.73 (0.50, 1.06)	0.094	0.55 (0.42, 0.72)	<0.001	1.09 (0.77, 1.53)	0.633
Black, current substance use	0.45 (0.23, 0.85)	0.014	0.32 (0.22, 0.47)	<0.001	0.76 (0.47, 1.25)	0.288
Other, current substance use	0.33 (0.13, 0.84)	0.021	0.60 (0.37, 0.99)	0.049	0.69 (0.26, 1.79)	0.448
Covariates
Age, years	0.97 (0.97, 0.98)	<0.001	0.98 (0.98, 0.99)	<0.001	0.97 (0.97, 0.98)	<0.001
Employment status, unemployed	0.69 (0.55, 0.86)	0.001	0.74 (0.62, 0.88)	0.001	0.68 (0.55, 0.84)	<0.001
Health insurance, public only (vs. any private coverage)	0.75 (0.58, 0.96)	0.022	0.71 (0.59, 0.86)	0.001	0.78 (0.60, 0.99)	0.048
BMI	0.98 (0.96, 0.99)	0.013	0.99 (0.98, 1.00)	0.076	0.98 (0.97, 1.00)	0.056
Hypertension, yes	1.17 (0.93, 1.47)	0.180	1.10 (0.92, 1.31)	0.317	1.13 (0.91, 1.41)	0.272
On dialysis, yes	0.80 (0.65, 0.98)	0.035	0.62 (0.52, 0.73)	<0.001	0.83 (0.68, 1.01)	0.065
Charlson Comorbidity Index,d higher score=worse	0.27 (0.15, 0.50)	<0.001	0.30 (0.19, 0.49)	<0.001	0.39 (0.22, 0.69)	<0.001
Chronic pulmonary disease, yes	0.84 (0.65, 1.09)	0.185	0.83 (0.70, 0.99)	0.034	0.95 (0.73, 1.23)	0.705
History of medical nonadherence, yes	0.82 (0.59, 1.13)	0.226	0.80 (0.63, 1.02)	0.067	0.90 (0.65, 1.24)	0.522
Waitlisted before 2014 implementation of KASe	0.80 (0.77, 0.83)	<0.001	---		0.78 (0.43, 1.40)	0.398
Improvement in model fit over null model: χ2 (df); p	284.7 (15)	<0.001	255.0 (14)	<0.001	124.0 (15)	<0.001

^aFor each outcome, patients were followed until the event of interest or until censoring due to death (competing risk) or other reasons (see Table 1 for numbers of patients by reasons for censoring). Only 11 patients (<1% of all patients; <2% of those waitlisted) were censored because the study observation period ended; they had been waitlisted and were on the waitlist at study’s end). They were followed in the study for median of 8.4 years (IQR, 8.1, 8.8), and had been on the waitlist for median of 6.5 years (IQR, 3.9, 7.9).

^bOf the 415 patients receiving transplants, 134 received living donor transplants. Numbers of outcome events are too small to examine race/ethnicity by substance use groups as predictors separately for living vs. deceased donor transplants.

^cWe chose to compare groups defined by the combination of race/ethnicity and substance use, with the referent group of non-Hispanic white patients, because of the ease of displaying and interpreting specific disparities in the outcomes. An alternative for evaluating our hypothesis (i.e., that the 2 groups of racial/ethnic minority patients who used substances would show particularly great disadvantage on outcomes) is to test a planned contrast using contrast weights to capture the notion of synergistic effects. The statistical tests and p levels associated with evaluating this planned contrast within competing risk models were z=2.99, P=0.003 for the outcome of kidney transplantation; z=4.95, p<.001 for waitlisting, and z=1.34, p=.182 for transplant among waitlisted patients. (Note that decomposing this planned contrast into its component parts, i.e., separately testing main effects for race/ethnicity and substance use, and an interaction effect, would provide only piecemeal evaluation of our hypothesis rather than a focused test of it. Planned contrasts give greater power and precision than piecemeal testing when specific hypotheses such as those pertaining to synergy are proposed.^86,87)

^dLog transformed prior to analysis.

^eIncluded as a time dependent covariate in analysis of time to transplant in full cohort.

Abbreviations: CI, confidence interval; HR, hazard ratio; IQR, Interquartile range; KAS, Kidney Allocation System