Robust estimation of the effect of an exposure on the change in a continuous outcome

Table 2 Results from the blood glucose study

Method	Linear effect on transformed outcome		Transformation parameter (λ)
Method	Estimate	95% Confidence interval	Estimate	95% Confidence interval
REM^a	–0.054	−0.083, −0.025	0.33	0.28, 0.37
cprobit^b	−0.042	−0.079, −0.005	0.34	0.28, 0.40

^aAdjusted for age and gender by including these variables into the linear predictor
^bImplicitly adjusted for all time-invariant confounders

ISSN: 1471-2288