Skip to main content

Table 4 Recommendations for analyses in malaria chemoprevention in pregnancy trials based on key safety outcomes from the systematic review

From: Systematic review of statistical methods for safety data in malaria chemoprevention in pregnancy trials

Safety outcome type after treatmentExamplesOptimal statistical method(s)
Time to eventTime to infection/infestation occurrence, time to abortionSurvival methods (e.g. Kaplan–Meier plots, log rank test followed by cox regression model)
Longitudinal continuous outcomesQTc prolongation, haemoglobin, vital signsMixed effects linear/nonlinear models
Recurrent eventsrecurrent hospitalizations, recurrent anaemia, recurrent opportunistic infectionsRecurrent event models (e.g. Poisson model, negative binomial, Andersen Gill model)
Safety outcomes with informative missing data,Treatment switching, Dropout/withdrawal due to AE (i.e. AE-induced non-adherence)Causal inference methods (e.g. mediation analysis)
Multivariate outcomesAbortion, miscarriage Haemoglobin, QTc, alanine aminotransferase levelMultivariate methods: methods for multiple outcomes (e.g. multivariate regression models based on outcome type), methods for repeated outcomes (e.g. multivariate longitudinal analysis, recurrent events models), advanced graphs
Laboratory measurements with temporal variationsVital signs, haemoglobinAdvanced graphical methods, multivariate methods
CountSeizure episodes, dizziness episodesIncidence rate followed by Models for count data (Poisson regression, negative binomial regression)
Outcomes with zero inflated countsSeizure episodes, dizziness episodes, nausea episodesModels for counts (Poisson zero inflated models, negative binomial zero-inflated models)