Table 3.

Model results of the association between a regimen of antifungal cream and the rate of ADL episodes at follow-up time point K among a cohort of lymphedema patients enrolled in a lymphedema management program, Khurda District, Odisha State, India, 2009–2011.

Time (K)	Marginal structural model Ia,b RR (95% CI)	Marginal structural model IIb,c RR (95% CI)	Traditional adjustmentd RR (95% CI)	Crude model RR (95% CI)
3 months	1.08 (0.70, 1.67)	0.94 (0.64, 1.37)	1.12 (0.77, 1.64)	1.76 (1.30, 2.38)
6 months	1.04 (0.60, 1.81)	0.83 (0.45, 1.51)	0.92 (0.47, 1.82)	1.23 (0.89, 1.71)
12 months	0.63 (0.36, 1.09)	0.81 (0.40, 1.60)	0.84 (0.54, 1.30)	1.27 (0.97, 1.67)
18 months	0.77 (0.62, 0.96)	0.55 (0.39, 0.76)	0.86 (0.59, 1.26)	1.07 (0.90, 1.26)
24 months	0.80 (0.63, 1.03)	0.74 (0.54, 1.01)	0.83 (0.62, 1.11)	1.08 (0.88, 1.32)

^aPoisson model using stabilized inverse probability of treatment and inverse probability of censoring weights;

^bAdditionally adjusted for lymphedema status (advanced, stages 4–7 vs. early, stages 1–3) and history of MDA;

^cPoisson model using non-stabilized inverse probability of treatment and inverse probability of censoring weights;

^dAdjusted for presence of inter-digital entry lesions, lymphedema status (advanced, stages 4–7 vs. early, stages 1–3) and history of MDA.