Analysis of multivariate longitudinal kidney function outcomes using generalized linear mixed models

Table 1 Bootstrapping on the (a) multivariate models and (b) univariate models: replicate 2,000 sample size 110

Estimate	Outcome	(a) Multivariate models									(b) Univariate models
		SHRI			SHRIS			SPRI^d			SPRIS			RI			RIS
		True beta^a	Mean difference^b	Mean squared difference^b	True beta	Mean difference	Mean squared difference	True beta	Mean difference	Mean squared difference	True beta	Mean difference	Mean squared difference	True beta	Mean difference	Mean squared difference	True beta	Mean difference	Mean squared difference
Intercept	BUN*100^c	3.3899	−0.09021	0.07796	3.3887	−0.10200	0.07740	3.3938	−0.08663	0.08125	3.3895	−0.12931	0.08517	3.3914	−0.08466	0.07908	3.3887	−0.10027	0.08003
	Creatinine*100	0.9405	−0.04489	0.12977	0.9394	−0.06113	0.12942	0.9461	−0.03864	0.13702	0.9422	−0.00693	0.13821	0.9416	−0.04614	0.13064	0.9411	−0.04301	0.13186
	eGFR*100	3.3728	0.13339	0.13729	3.3716	0.14946	0.13749	3.3626	0.11698	0.15495	3.3655	0.05496	0.16161	3.3674	0.11982	0.14741	3.3687	0.06382	0.15037
Time	BUN*10³	−0.0195	0.01402	0.00458	−0.0193	0.01465	0.00458	−0.0186	0.03321	0.00451	−0.01857	0.11809	0.00509	−0.0191	0.02583	0.00456	−0.0188	0.01559	0.00462
	Creatinine*10⁴	−0.0452	−0.46142	0.04717	−0.0451	−0.42203	0.04633	−0.0441	−0.07492	0.04227	−0.04421	−0.52669	0.04300	−0.0452	−0.32036	0.04663	−0.0452	−0.42299	0.04670
	eGFR*10³	0.0532	0.10479	0.00853	0.0534	0.10473	0.00880	0.0503	0.01589	0.00576	0.05049	0.03443	0.00610	0.0518	0.05929	0.00648	0.0518	0.04602	0.00650

^aTrue beta obtained from the application of every model on the renal transplant dataset.
^bMean difference and mean squared difference represent respectively the average deviation and its square of every bootstrap estimate from the true value beta.
^cMean difference and mean squared difference should be multiplied by the number specified next to the outcome name.
^dSPRI model has best results compared to other models: 50% of the estimates have smallest mean difference and mean squared difference under SPRI.

ISSN: 1479-5876