STUDY SETTING AND DESIGN
The study took place between March and October 2022, in Mopeia, a rural district in Zambezia Province, central Mozambique, which remains highly endemic for malaria despite dual coverage with IRS and ITNs (24). It has a population of approximately 131,000, and covers an area of 7,671km2 (25).
The study was nested within a three-arm cluster-randomised controlled trial, aiming to assess the effect of iMDA on malaria transmission (26). Each cluster consisted of a core area, and a buffer area with a radius of 400 meters around the core. Either oral or written consent was obtained from all participants or their legal guardian. The following interventions were given once per month for three consecutive months:
Ivermectin in humans: a single dose of 400ug/kg in eligible humans;
Ivermectin in humans and livestock: a single dose of 400ug/kg in eligible humans plus 1% injectable ivermectin at a dose of 200 ug/kg to all pigs/cattle in the cluster;
Control: albendazole was given to humans only at a single dose of 400 mg.
All study drugs were delivered door-to-door and drug administration was directly observed by the responsible fieldworker. Within this study, two cohorts were followed up; the safety cohort, and the efficacy cohort. The safety cohort consisted of individuals who met the following criteria: weighing over 15 kg, ability to consent, and a negative pregnancy test for women aged between 13 and 49; and exclusion: known hypersensitivity to ivermectin or albendazole, risk of Loa loa as assessed by travel history, pregnancy, lactating in the first week postpartum, currently participating in another clinical trial, unwilling to consent or adhere to study procedures, severely ill, currently under treatment with inhibitors of CYP3A or P-gp (see Supplementary File 1 for the full list of excluded drugs). The purpose of this cohort was to serve as delivery vehicle for the insecticidal intervention and to monitor the safety of the treatment regime. These participants were followed up at 1,2 and 3 months after the first round of MDA.
The efficacy cohort consisted of children who were ineligible to take ivermectin due to a weight under 15 kg, but that were the primary group affected by malaria in the region; the purpose of this cohort was to assess the impact of the intervention on malaria. Untreated, ineligible participants were followed up at 3 and 6 months after the first round of MDA.
In the analysis, the safety cohort and any participants from the efficacy cohort who had taken ivermectin at least once were categorised as “treated”, and participants from the efficacy cohort who had not taken ivermectin were categorised as “untreated”.
RANDOMISATION AND SAMPLING
During October 2020 to November 2021, a local demography survey took place under a separate protocol, where all households were mapped and the population was enumerated (25). Using this data, clusters were created, and randomized 1:1:1 to each intervention using a computer-generated random sequence. Study investigators were masked from the treatment allocation during the study, however, the study participants were aware of which intervention they received. Following treatment allocation, 13 clusters per arm (39 clusters in total) were randomly selected for follow-up of headlice. Within these clusters, all children enrolled into the efficacy cohort of the main study, and 40 participants enrolled into the safety cohort of the main study were randomly selected for inclusion in this study.
OUTCOMES
The primary outcome was headlice prevalence at three months. Secondary outcomes were headlice prevalence at three months in treated participants, and at 3 and 6 months in untreated participants.
Fieldworkers were trained in headlice detection over a two-day period by JFA and AH through power point presentations, exercises using the tablet, and written assessments. In the field, participant scalps were inspected visually in daylight or under a torchlight, and signs of headlice (live lice or eggs) were recorded. In the main analysis, only participants with live headlice were categorized as having headlice.
RISK FACTOR ANALYSIS VARIABLES
Baseline data on headlice, and data from a demographic survey that took place prior to the start of the trial were used to conduct the risk factor analysis. Household wealth index was calculated using a formula detailed in Xie et al (27). All variables were determined through a pre-designed questionnaire. Bed net use was a binary variable, assessed by asking the participant or their parent/guardian whether they slept under a bed net the previous night. Details of how water sources were categorised are in Supplementary File 2 (28). Open defecation refers to those whose household does not have a sanitation facility available to them.
STATISTICAL ANALYSIS
The sample size of the nested study was calculated to detect a significant difference between the control and the treatment arms in scabies prevalence, which was another outcome that will be reported in a separate publication (Furnival-Adams et al., submitted). Assuming a 7% baseline scabies prevalence in the control group, a cluster size of 40 participants with 13 clusters per arm was required to detect a 60% effect size with 80% power at 5% significance and a kappa of 0.25.
For the risk factor analysis, logistic regression using the lme4 R package (v1.1-26; Bates et al., 2024) was used to conduct bivariate and multivariate analyses in the R programming language, version 4.2.2 (R Foundation for Statistical Computing). In all analyses, odds ratios or adjusted odds ratios with 95% CI were calculated. All variables associated with headlice in bivariate analysis (p < 0.10) were assessed for inclusion in multivariate analysis. Variables that were statistically significantly associated with headlice (p < 0·05) were included in the multivariate model.
To assess the treatment effect, we compared headlice prevalence in the control versus the treatment arms at each time point, using logistic regression model. To account for correlation between clusters, we used generalised estimating equations (GEE) to fit each logistic model of prevalence to the available data. Known risk factors of headlice, age, sex and wealth, were included as covariates in the adjusted models. These data were collected during the census survey that took place prior to the start of the trial. Intervention effects were expressed as adjusted odds ratios (aORs). After analysing the data, we found no statistically significant difference between the two treatment arms (human and human + livestock) existed, and therefore pooled the two intervention arms. Statistical significance was considered when the p-value was < 0.05 using to the Wald test to calculate the p-value.
This study was conducted in accordance with the CONSORT guidelines for reporting randomized controlled trials (Supplementary File 3) (29).