Epidata

Data were entered onto standardised case record forms, checked against source documents, double entered into Epidata, and analysed using Stata v13 (Stata Corporation, College Station,mTX, USA). Proportional data were compared using chi-squared or Fisher’s exact tests, as appropriate, and continuous data by student’s t (normally distributed data) or Mann–Whitney U (skewed data) tests. The relationship between the fractional fall in Hb on D7 versus baseline and the mg/kg dose of primaquine was assessed by Spearman rho test (skewed data), and with the baseline G6PD enzyme activity by Pearson’s correlation coefficient (transforming the G6PD data to become normally distributed).

Data were collected using pretested, structured questionnaires, doubly entered into Epi data (www.epidata.dk, version 4.4.3.1) and exported for analysis to Stata version 14.0 (StataCorp, College Station, Texas, USA.). Only mother-infant pairs with data at the three time points of follow-up were included in the analysis. Continuous data, if normally distributed, was summarised into means and standard deviations and if skewed, was summarised into medians with their corresponding interquartile ranges. Categorical variables were summarised into frequencies and percentages. The incidence of non-exclusivity of breastfeeding was estimated and its confidence limits calculated using the exact method. Bivariable and multivariable analysis was done using the modified Poisson regression model [23 (link)]. All variables that had a p value < 0.25 at bivariable level and those with biological plausibility were entered into the multivariable model. Variables that were independently associated with non-exclusivity of breastfeeding were determined using the confidence limits.

Data entered into EpiData were exported and analysed using the statistical software Stata v15.1 (StataCorp, College Station, Tx, USA). AWaRe classification (‘Access’, ‘Watch’ and ‘Reserve’ group of antibiotics) was used to group antibiotics for AST.17 Data are summarised using frequencies and percentages. We used bivariable binomial log models to study the association between demographic and clinical factors with demographic and clinical characteristics. These associations are presented as prevalence ratios; P<0.05 was considered statistically significant.

Data were double entered into Epidata, checked, and analyzed using Stata v14 (Stata, College Station, TX). Proportional data between groups were compared using χ ² analysis or the Fisher exact test, as appropriate, and those within groups were analyzed using the Maentel Haenszel test and the exact McNemar significance probability. Continuous data were analyzed by paired (within-group) or unpaired (between-group) t tests for normally distributed data; corresponding nonparametric tests were the Wilcoxon signed-rank test and the Mann-Whitney U test.
We used multiple linear regression (a backward stepwise approach) to determine factors (eg, age, sex, baseline Hb level, G6PD status, Hb type, baseline parasitemia level, and length of illness) associated with the baseline Hb level, the absolute change in the Hb level, the total malaria-attributable change in the Hb level following treatment, and the decrease in RBC count. A linear mixed-effects regression model was used to determine factors associated with changes over time in the Hb level, reticulocytemia, the creatinine level, the LDH level, the haptoglobin level, and the conjugated and unconjugated bilirubin levels. Logistic regression was performed to assess factors associated with recovery of the Hb level and a clinically concerning decrease in the Hb level.

Data were collected with EpiData and analyzed by STATA version 12 (STATA, CA, USA). The count of samples with wild-type and mutant alleles was used to generate the prevalence of the alleles. The chi-square test was used to compare proportions and the statistical significance was defined as a p-value, 0.05.