Effect of adding azithromycin to the antimalarials used for seasonal malaria chemoprevention on the nutritional status of African children

Mass administration of azithromycin has reduced mortality in children in sub‐Saharan Africa but its mode of action is not well characterised. A recent trial found that azithromycin given alongside seasonal malaria chemoprevention was not associated with a reduction in mortality or hospital admissions in young children. We investigated the effect of azithromycin on the nutritional status of children enrolled in this study.


Introduction
Mass drug administration (MDA) with azithromycin (AZ) is a highly effective approach to the control of trachoma. Reductions in the incidence of skin, gastrointestinal and respiratory infections have been recorded after mass distribution with AZ, and reductions in childhood mortality have been found to varying degrees. Among Ethiopian children aged 1-9 years, a 49% reduction in all-cause mortality was found following a single dose of AZ administered yearly, biannually or quarterly, a reduction sustained during the 26-month follow-up period [1,2]. The MORDOR (Mortality Reduction after Oral Azithromycin) trial conducted in Malawi, Niger and Tanzania randomised children under 5 years of age to receive twiceyearly mass distributions of oral AZ for four years. Relative to children receiving placebo, an overall reduction in mortality of 13.5% (95% confidence interval (CI): 6.7, 740 19.8) was recorded. The impact on mortality was higher in Niger, an area with intense and seasonal malaria transmission [18.1% (95% confidence interval (CI): 10.0, 25.5)], than in Malawi or Tanzania [3].
The mechanism through which mass distribution of AZ reduces mortality is not understood [4]. Improved nutritional status among children receiving AZ is a possible underlying mechanism through which mortality might be reduced. Azithromycin is an antibiotic widely used to treat bacterial infections. Diarrhoeal disease and lower respiratory tract infections are associated with growth faltering in many low-income countries [5][6][7]. It is therefore plausible that mass distribution with AZ improves nutritional status by reducing these infections.
Previous research has shown improved nutritional status in livestock after the administration of AZ regardless of infection status [8]. The administration of AZ to pregnant women improved birthweight outcomes of neonates in Malawi [9]: mean child length and development scores were consistently higher and mortality lower in the group receiving AZ plus monthly sulfadoxine-pyrimethamine (SP) than in the group receiving monthly SP alone.
Similar beneficial results have been observed in children following antibiotic administration, but inconsistently. A meta-analysis of randomised controlled trials (RCTs) in low-and middle-income countries reported that antibiotic use increased height (0.04 cm/month) and weight (23.8 g/ month). However, these trials all involved distribution of antibiotics other than AZ [8]. Two trials in Niger found no difference in prevalence of wasting, stunting, being underweight, or mid-upper arm circumference (MUAC) between communities that received mass administration of AZ and those that did not [10,11]. Children of mothers in The Gambia who received AZ during labour were less likely to be malnourished or underweight at the age of 11-13 months than children of mothers who received a placebo [12]. Also in The Gambia, a comparison of height and weight in children who had been enrolled in a trachoma trial found no evidence that three annual rounds of MDA with AZ improved anthropometric indices compared to a single round of AZ, in spite of high treatment coverage [13].
In settings of seasonal malaria transmission, a highly effective approach to malaria control involves the administration of the antimalarials sulfadoxine-pyrimethamine and amodiaquine (SP-AQ) to children once a month for three or four months during the season of high malaria transmission, an intervention known as seasonal malaria chemoprevention (SMC). SMC prevents clinical malaria and severe malaria by about 70%À80% under trial conditions, and there is growing evidence of an impact on mortality from large-scale evaluations [14,15]. In 2017, 15.7 million children were included in SMC programmes in 12 countries (Burkina Faso, Cameroon, Chad, Gambia, Ghana, Guinea, Guinea-Bissau, Mali, Niger, Nigeria, Senegal and Togo) [16].
To assess the effect of adding AZ to SMC, a randomised, double-blind, placebo-controlled trial was conducted in Hound e district of Burkina Faso and in the Bougouni district of Mali [17]. The addition of AZ to SMC did not result in a lower incidence of the primary trial outcomedeath or hospital admissions not due to trauma or surgery. However, gastrointestinal, upper respiratory tract and skin infections, and non-malarial febrile illnesses occurred less frequently among children who received AZ instead of placebo [17]. Given the broadspectrum activity of AZ, and its impact on these causes of morbidity, we investigated whether regular administration of AZ would lead to improvements in nutritional status.

Study centres
The centres where this trial was conducted in Burkina Faso and Mali are predominantly rural with the majority of the population engaged in farming and/or petty trading [17]. The rainy season lasts from July to October, followed by a long dry season ( Figure S1). The entire population of Burkina Faso (19.2 million) and 91% of the population of Mali (16.9 million) are classified as living in an area of high malaria transmission (>1 case per 1000 population per year) [16]. Demographic and Health Surveys (2010 Burkina Faso, 2012-2013 Mali) estimate the prevalence of stunting in children at 34.6% in Burkina Faso and 38.3% in Mali, and the prevalence of wasting as 15.5% in Burkina Faso and 12.7% in Mali [18,19].

Enrolment and randomisation
A household census was conducted in June 2014, and children of either sex who were 3 to 59 months of age on 1 August 2014 were eligible for enrolment in the trial. Randomisation to the AZ or placebo groups was conducted at the household level (to avoid the potential effect of within-household transmission of infection). Drug combinations were administered in four cycles of three days, at monthly intervals, for three successive malaria transmission seasons (2014-2016) ( Figure S2). Infants 3-11 months of age received 250 mg of sulfadoxine and 12.5 mg of pyrimethamine plus 75 mg of amodiaquine on day 1 and 75 mg of amodiaquine on days 2 and 3 (Guilin Pharmaceutical, Shanghai, China). In addition, they were randomly assigned to receive either 100 mg of AZ or matching placebo each month on days 1, 2 and 3. Children 1-4 years of age received double these doses.

Nutrition data
Clinical surveys were conducted approximately four to six weeks after the last dose of each SMC round had been given in 2014, 2015 and 2016 (i.e. between four and five months after the first course) ( Figure S2). Two thousand children were randomly selected by an independent statistician from the randomisation list each year (stratified on randomisation by group and age) in each country for each survey. During these surveys, weight, height and MUAC were measured by trained nurses supervised by physicians, and a blood sample was obtained for measurement of haemoglobin (Hb). Children were weighed using a weighing scale (Seca©), a tray was used for small children or the difference was calculated when weighing mother and baby relative to mother alone. Weight was recorded to the nearest 0.1 kg. Recumbent length measurements were taken for children under 2 years and standing height measurements for children above 2 years using a stadiometer (Seca©). Length/ height measurements were taken twice and the average recorded. MUAC was measured on the left arm to the nearest 0.1 cm with a MUAC tape. Hb concentration was measured using a HemoCueâ. In August 2015, anthropometric data were also collected prior to the first SMC distribution (MUAC obtained for all study participants), and prior to the second SMC distribution in August 2016, weight, height and MUAC data were collected from the children selected for the full anthropometric survey at the end of the malaria transmission season that year ( Figure S2).

Statistical analyses
Individual height-for-age (HAZ), weight-for-age (WAZ) and weight-for-height (WHZ) Z-scores were calculated in Stata (Version 15, College Station, TX, USA), using the WHO 2007 reference populations [20]. MUAC-for-age Z-scores were calculated in Stata based on the WHO 2006 reference values.
Binary outcomes stunted, underweight, wasted, low MUAC-for-age were defined as individuals with corresponding Z-scores below -2 (HAZ, WAZ, WHZ, MUAC-for-age, respectively). Nutritional outcomes were defined as severely stunted/underweight/wasted or very low MUAC-for-age when corresponding Z-scores were less than -3. Moderate anaemia was classified as Hb concentration <10g/dL and severe anaemia as an Hb concentration <7g/dL [21].
Each binary outcome was compared between treatment arms (SMC + placebo or SMC + AZ) using Poisson regression to obtain prevalence ratios accounting for clustering at the household level. The primary analysis was an intention-to-treat analysis performed on the pooled data from both countries (controlling for country). Effect modification between country and study arm was investigated using the likelihood ratio test.
Country-specific analyses were performed for Burkina Faso and Mali separately, along with a per-protocol analysis (including only children who had received at least three cycles of SMC in the current intervention year). Linear regression was used to compare the continuous outcomes (mean Z-scores and haemoglobin) between study arms.
In 2015 and 2016, when pre-and post-SMC anthropometric data were available, linear regression was used to compare difference in means over time between AZ and placebo groups.
Finally, to explore potential differences in the effect of AZ by age, pre-specified sub-group analyses were conducted including only children less than 12 months at administration of the first dose of SMC in each year and children 13-24 months at administration of the first dose of SMC in each year.

Study population
In July 2014, 19 578 children from 9618 households were enrolled across both countries and randomly assigned to receive either SMC plus AZ (9735 children) or SMC plus placebo (9843 children) on up to four occasions each year for three years. The two study arms were well matched with regard to baseline characteristics such as age, sex, use of long-lasting insecticidal net (LLIN), distance to the nearest health facility and hospital. The prevalence of rapid diagnostic test (RDT) confirmed malaria at the first SMC cycle ranged between 5.1% and 6.4% over the three years and was similar between trial arms. The percentage of children who received all four monthly courses of the assigned regimen (SMC + placebo or SMC + AZ) was 75.0% in the first year, 66.3% in the second year, and 64.6% in the third year. The percentage of children who received at least three monthly courses was 92.8% in the first year, 86.8% in the second year, and 84.3% in the third year [17].

Surveys
At the end of the malaria transmission season survey in 2014, we measured the height, weight, MUAC and Hb concentration of 4009 children. After the transmission season in 2015 and 2016, measurements were taken from 3976 and 3975 children, respectively. Among children selected for the end-of-season surveys, MUAC of 3837 children was also measured before the malaria transmission season in 2015, and pre-SMC MUAC, weight and height data of 3826 children were collected in 2016.

Nutrition status (regardless of trial arm)
The prevalence of stunting (HAZ < À2) remained consistently high in both countries over the study period, ranging from 24

Effect of AZ on nutrition outcomes
In both study sites, the prevalence of children with stunting, wasting, underweight or low MUAC was very similar between the two randomisation groups in all three years (Figures 2 and 3). The prevalence ratio (prevalence in SMC + AZ group relative to SMC + placebo group) for the primary comparison of pooled data from both countries at the 2016 post-SMC survey was 0.92 (95% CI 0.83, 1.02) for stunting, 1.01 (95% CI 0.90, 1.13) for being underweight, and 0.94 (95% CI 0.81, 1.08) for being wasted ( Table 1). The lowest MUAC-for-age prevalence ratio was 0.79 (95% CI 0.62, 1.01) in 2016 and for anaemia, 0.93 (0.83, 1.05) in 2014 (Table 1).
Comparison of pre-and post-SMC anthropometric data for Burkina Faso and Mali pooled in 2015 (MUACfor-age only) and 2016 (HAZ, WAZ, WHZ, MUAC-forage) did not provide evidence for a change in nutrition indicators between surveys (Tables S4 and S5). Analysis of a sub-sample of data for children less than one year of age at the time of administration of the first yearly dose of SMC similarly revealed no difference between study arms ( Figures S3-S5), as did analysis of children 13-24 months of age at the time of administration of the first yearly dose of SMC ( Figures S6-S8).
Results of the per-protocol analysis were consistent with those from the intention-to-treat analysis; the addition of AZ to SMC did not improve nutrition outcomes when compared to SMC and placebo (Tables S2 and S3).

Discussion
Despite the fact that the addition of AZ to SMC significantly reduced the incidence of acute respiratory and gastrointestinal infections and non-malarial febrile illnesses [17], up to four monthly courses of directly observed AZ did not improve the nutritional status of children in the study sites in Burkina Faso or Mali when distributed alongside the SMC regimen. A large majority of nutritional indicators did not differ significantly in prevalence between AZ and placebo groups in either country at any of the end-of-season surveys in the three-year trial. In the few instances where a difference in prevalence of a particular indicator was found in one of the study sites, the difference was not replicated on other occasions in either site and is thus likely be a chance finding when considered in the context of the large number of comparisons undertaken.
The largest difference in mortality between AZ and placebo treatment groups in children in the MORDOR trial conducted in Malawi, Niger and Tanzania, which administered AZ up to four times at six-month intervals, was found among children in the youngest age group (1-5 months of age) (24.9%, 95% CI 10.6-37.0) [3]. Due to the lower age limit for SMC of three months, there were very few children in this age range at the end of the season, and consequently, it was not possible to evaluate impact on nutrition in this age group. However, when restricting analysis to children < 12 months of age we did not find an effect of AZ on nutritional status ( Figures S3-S5), nor among children 13-24 months, coinciding with the period post-weaning when nutritional status might be expected to be especially poor ( Figures S6-S8).
Differences were seen in the direction of change of nutrition status over the study period between Burkina Faso and Mali. All five key nutritional indicators (stunted, underweight, wasted, low MUAC-for-age, anaemic) increased in prevalence over the study period in Burkina Faso while they decreased in prevalence in Mali, with the exception of the prevalence of low MUAC-forage which also increased in Mali. These patterns are difficult to explain, especially changes in the prevalence of wasting in both countries in 2014 and 2016, and do not correlate with differential rainfall between the countries ( Figure S1) nor with differences in the incidence of gastrointestinal illness (Figures S9-S10). The prevalence of stunting was lower in both study sites than reported in the national DHS, possibly due to 'healthy participant effect' and/or the shorter hunger season in these districts than in many other parts of Burkina Faso or Mali. Administration of AZ with SMC reduced the incidence of acute respiratory and gastrointestinal infections and so it might have been expected to have a beneficial impact on nutrition. However, a large majority of the episodes of diarrhoea detected in this study were acute episodes which are likely to have had only a transitory effect and are not likely to have led to a detectable effect in a single survey conducted at the end of rainy season up to four months after an acute episode had been noted. In addition, poor nutrition rather than infection is the predominant cause of malnutrition, including stunting, in young children in both Burkina Faso and Mali.
A possible explanation for the absence of a benefit on nutritional status from administration of AZ is that antibiotics alter the ecology of the intestinal microbiome which may hinder nutrient supply and allow overgrowth of species with potential pathogenicity, possibly counterbalancing any beneficial effects [22][23][24]. Although the delivery of four monthly courses of AZ provides a period of sustained protection, it is possible that the timeframe was too short to observe major improvements in nutritional status.
The high level of coverage achieved with SMC, involving frequent contacts between young children and care givers, provides an opportunity for additional interventions aimed at improving nutrition. Which additional interventions might prove effective is uncertain. In our study, community workers administering SMC were instructed on the importance of referring children with severe malnutrition to a local nutritional rehabilitation centre and this may have saved some lives. However, it did not have any impact on the prevalence of clinical malnutrition at the population level [17]. In Nigeria, overall nutritional status among children aged 6-24 months did not improve in children who received nutritional supplementation alongside SMC compared to children receiving SMC alone, although the study was small and not randomised [25].
An important characteristic of this study was its large sample size, allowing small differences in nutritional status between randomisation groups to be detected. However, it had a number of weaknesses including the infrequent measurement of nutritional status and the lack of a control group not receiving SMC. Exploration of the role of potential confounders such as socio-economic status would also have been valuable but was not possible in a trial of this size. A further limitation is the generalisability of results beyond the setting of the study and SMC distribution regimen. Through partaking in the study, the children in the placebo arm are likely to have received a higher level of care than the standard in each of the study sites, thus potentially diluting effects the intervention may have had. However, given the large number of indicators measured and a lack of any consistent differences between AZ and placebo, the overall interpretation is likely to be robust.
In conclusion, in these two areas of intense seasonal malaria transmission in Burkina Faso and Mali, where nutritional status is poor and fluctuates seasonally, the addition of AZ to the antimalarials used for SMC did not result in improved nutritional status.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Figure S1. Monthly rainfall (mm) for Burkina Faso and Mali. Figure S2. Schematic of data collection activities. Figure S3. Distribution of nutrition outcomes between study arms in children younger than one year of age (2014). Figure S4. Distribution of nutrition outcomes between study arms in children younger than one year of age (2015). Figure S5. Distribution of nutrition outcomes between study arms in children younger than one year of age (2016). Figure S6. Distribution of nutrition outcomes between study arms in children 13-24 months of age (2014). Figure S7. Distribution of nutrition outcomes between study arms in children 13-24 months of age (2015). Figure S8. Distribution of nutrition outcomes between study arms in children 13-24 months of age (2016). Figure S9. Incidence of gastro-intestinal illnesses in Burkina Faso over the study period. Figure S10. Incidence of gastro-intestinal illnesses in Mali over the study period. Table S1. Test of interaction between country and treatment arm. Table S2. Per protocol analysis Burkina Faso. Table S3. Per protocol analysis Mali. Table S4. Comparison of Pre-and Post-SMC MUAC and MUAC-for-age data (2015), pooled data from both countries. Table S5. Comparison of Pre-and Post-SMC data MUAC-for-age, WAZ, HAZ, WHZ (2016), pooled data from both countries.