Table 2.

Multiple outcome analysis including both efficacy and tolerability outcomes, Benin, 2005–2008^*

Strategy for handling missing data†	Birth weight information in the event of stillbirth, abortion, or multiple pregnancy
	Not accounting for P value	Accounting for P value
No imputation	0.004	0.005
Random	0.01	0.01
Failure	0.07	0.18

The upper left entry in the table gives the P value of the main analysis.

^*Multiple outcome global comparison of mefloquine (MQ) with sulfadoxine–pyrimethamine (SP): 1) superiority of MQ to SP on at least low birth weight (LBW), placental malaria, and maternal anemia at delivery, 2) non-inferiority of MQ to SP on LBW, placental malaria, maternal anemia, and tolerability. P values according to various strategies for handling missing data and accounting for birth weight information in the 65 stillbirth, spontaneous abortion, or multiple pregnancy cases or not.

^†“No imputation”: missing data were excluded; “Random”: missing data were imputed according to the observed prevalence of the outcome in the cohort, “Failure”: missing data were considered as failures (i.e., LBW, anemia, or placental malaria). Missing data on tolerability were neither imputed nor considered as failures.