Maternal postpartum deworming and infant milk intake: Secondary outcomes from a trial

2.2. Questionnaire administration

Prior to random allocation, research assistants administered a baseline questionnaire to mothers to collect information on socio‐demographics, deworming history during pregnancy and iron supplementation. At 1‐ and 6‐month postpartum, mothers were administered a questionnaire to collect information on current breastfeeding practices (e.g., exclusive breastfeeding and introduction of complementary foods). Questions on breastfeeding practice were adapted from the World Health Organization (WHO) indicators for assessing infant and young child feeding practices (World Health Organization, 2008).

2.3. Infant milk intake assessment

Water intake from human milk and other sources was estimated in the infants using the dose‐to‐mother deuterium‐oxide turnover technique (Coward, 1984; International Atomic Energy Agency, 2010) at 1‐ and 6‐month following delivery. At each of the 1‐ and 6‐mo assessments, mother–infant pairs were visited in their homes 6 times over a period of 2 weeks. On Day 1, a nonenriched urine sample of at least 3 mL was collected from mothers and their infants to assess baseline levels of deuterium. Following sample collection, mothers were administered a preweighed 10‐g oral dose of deuterium oxide (99.8% purity), accurate to the nearest 0.001 g. Mothers wore an absorbent bib in case of spill. Following administration of deuterium, research assistants added an additional 40 mL of purified water to the vial that contained the deuterium and asked mothers to ingest the remaining liquid. The date and time of deuterium administration were recorded.

Research assistants returned to participants' homes to collect urine samples from the mother on Days 2, 5, 14 and 15 postingestion and to collect urine samples from the infant on Days 2, 4, 5, 14 and 15 postingestion. Urine samples from mothers were collected in presterilized containers, and urine samples from infants were collected using paediatric collection bags. The date and time of urine collection were recorded. Weight and height of mothers were measured at the 1‐ and 6‐month time points using a portable electronic scale accurate to the nearest 0.1 kg and a stadiometer (Seca 213, Seca Corp., Baltimore, MD, USA) accurate to the nearest 0.1 cm. Weight and recumbent length of infants were measured at the 1‐ and 6‐month time points using a digital balance (Seca 354, Seca Corp., Baltimore, MD, USA), accurate to 0.01 kg and a stadiometer (Seca 417, Seca Corp., Baltimore, MD, USA), accurate to 0.1 cm. Urine samples were packed in separate sealed bags to prevent contamination and transported to the laboratory in a cooler on ice. Samples were refrigerated at 4°C until processing.

Upon arrival at the laboratory, all urine samples were aliquoted into sterile 5‐mL cryovials and stored at −20°C. Following specific instructions for sample collection, all samples were shipped on dry ice to the Instituto de Nutrición y Tecnología de los Alimentos, Universidad de Chile (Santiago, Chile) for laboratory assessment. Isotope ratio mass spectrometry was used to measure the ²H enrichment of urine samples. Infant intake of human milk and water from other sources was estimated by fitting deuterium enrichment data to a model for water turnover in mothers and infants, using methods that have been previously described and validated (Coward et al., 1982; Haisma et al., 2003, 2005).

Data collected during home visits (e.g., date and time of sample collection, weight and height/length) were recorded electronically in real time on tablets and reviewed daily for accuracy and completeness. Laboratory management of samples (e.g., labelling and storage) were coordinated using the Laboratory Data Management System (Version 11.1.0.30, Frontier Science, Amherst, NY, USA).

2.4. Stool specimen analysis

Stool specimens were collected from mothers at the 6‐month time point for the determination of STH prevalence and intensity. Specimens were transferred to the laboratory of the local research facility and processed within 24 h of collection using the Kato–Katz technique (World Health Organization, 1998). One microscope slide per specimen was quantitatively examined for helminth ova by trained technologists. STH prevalence was based on the identification of at least one STH helminth egg during microscopic examination. STH intensity, based on parasite‐specific egg counts, was categorized according to WHO cut‐offs for light, moderate and heavy intensity infection (World Health Organization, 2002).

2.5. Power calculations

The study sample size of 200 mother–infant pairs had 80% power to detect a 51‐mL difference in infant milk intake between intervention groups, using a two‐sided Student's t test, an alpha value of .05, a common standard deviation of 130 mL (Albernaz et al., 2003; Haisma et al., 2003; International Atomic Energy Agency, 2010; Limon‐Miro et al., 2017) and an approximate 1:1 ratio of participants allocated to the experimental and control groups. Previous published evidence from Mexico and Brazil suggests that a mean infant milk intake difference of approximately 100 mL would be biologically meaningful (Albernaz et al., 2003; Limon‐Miro et al., 2017). Power calculations were performed using PS Power and Sample Size Calculations Version 3.0 (Copyright © 1997 by Dupont and Plummer).

2.6. Statistical analysis

Baseline characteristics between those lost to follow‐up and their replacements were compared using a chi‐square test for categorical variables and Student's t test for continuous variables. The impact of maternal postpartum deworming on infant milk intake was compared using the intention‐to‐treat (ITT) principle in which participants are analysed according to the groups to which they were randomized.

Crude differences in infant milk intake were compared between groups using Student's t test. Multivariable linear regression models were used to evaluate the impact of maternal postpartum deworming on infant water intake with adjustment for covariables that had been determined to be important a priori from a review of the literature. These variables included maternal age, education level, socio‐economic status (SES) index, gestational age and infant sex. The SES index was created by principal components analysis using the following baseline variables: ownership of a radio and television, type of cooking fuel, type of flooring, water connection to the house and wired electricity in the house, as described previously (Filmer & Pritchett, 2001; Gyorkos et al., 2013; Joseph et al., 2014). The SES index was divided into quartiles for inclusion in multivariable models. Sensitivity analyses, including complete‐case and per‐protocol analyses, were conducted to test the robustness of study findings. Subgroup analysis was also performed among those mothers who reported exclusive breastfeeding.

An exploratory analysis of the associations between infant milk intake and maternal STH prevalence and intensity, both measured at the 6‐month time point, was evaluated using multivariable linear regression with adjustment for possible confounders. A factor was considered as a possible confounder if its association with milk intake at 6 months was significant at the P < .20 level in univariable analysis and if the factor was not known to be on the causal pathway between STH infection and milk intake. Only factors meeting these criteria were included in the final multivariable models to determine the most parsimonious models. STH prevalences and intensities were investigated by species and by intensity category (moderate‐and‐heavy vs. light‐and/or‐uninfected).

Multiple imputation by chained equations (MICE) with 20 imputations was used to impute infant water intake from human milk and nonmilk sources when outcome data for one of the two time points were missing. The MICE model included the following variables collected at baseline: infant sex, gestational age, maternal age, marital status, maternal education level, number of people residing in the household, SES index and intervention group. As one of the objectives of the study was to examine the presence of a differential response to the intervention by infant sex, the initial MICE model also included an interaction term for intervention group and infant sex. However, because no evidence of interaction was found, the interaction term was dropped from the final MICE model.

3.2. Follow‐up

Follow‐up for the 1‐month visit took place between April and August 2014 and follow‐up for the 6‐month visit between August and December 2014. Of the original 199 mother–infant pairs who participated at 1 month, 183 (92.0%) also completed the 6‐month assessment (Figure 1). The main reasons why those who completed the first assessment could not complete the second assessment were relocation to another city or temporarily away from the study area at the time of assessment (n = 10); withdrawal from the study (n = 3); and cessation of breastfeeding (n = 3). Participants who were lost to follow‐up were similar to their replacements in terms of maternal, child and household characteristics. Missing data occurred only due to loss‐to‐follow‐up; all baseline data, breastfeeding status and STH infection results were 100% complete. Infant milk intake was imputed for 15.3% (33/216) of the study.

At 1‐month postpartum, the prevalence of reported exclusive breastfeeding, and combined exclusive and predominant breastfeeding, was 48.2% (95% CI: 41.3%, 55.2%) and 92.5% (95% CI: 87.8%, 95.4%), respectively. At 6‐month postpartum, the prevalence of reported exclusive breastfeeding, and combined exclusive and predominant breastfeeding, was 1.0% (95% CI: 0.2%, 4.0%) and 10.0% (95% CI: 6.5%, 15.0%), respectively. There were no statistically significant differences between intervention groups with respect to reported breastfeeding status at either time point.

3.3. Effect of maternal postpartum deworming on infant milk intake

Mean infant milk intake at 1‐month postpartum was 756 ± 16 and 774 ± 18 mL day⁻¹ in the albendazole and placebo groups, respectively (Table 2). Mean infant milk intake at 6‐month postpartum was 903 ± 16 and 908 ± 18 mL day⁻¹ in the albendazole and placebo groups, respectively. There were no statistically significant differences in mean infant milk intake between intervention groups at 1‐ or 6‐month postpartum. Similar findings were obtained from complete‐case and per‐protocol analyses and in subgroup analyses restricted to mothers who reported exclusive breastfeeding (data not shown).

We would like to acknowledge our ongoing research collaboration with our colleagues in Peru, without whom this research would not have been possible. We would also like to thank Gabriela Salazar and her team at the Instituto de Nutrición y Tecnología de los Alimentos for their expertise and laboratory analyses for the deuterium oxide technique. Lastly, we thank Dr. Sonia Semenic for her insightful comments on the manuscript. Funding for this study was provided to TWG by the Bill and Melinda Gates Foundation (OPP1066903) and Fonds de Recherche du Québec‐Santé (Research Institute Grant) and to LSM by the International Development Research Centre (106690‐99906075‐052) and the Michael Smith Foreign Study Supplement, Canadian Institutes of Health Research (124574). The funders had no role in the design, analysis or writing of this article.