Effectiveness of a monthly schedule of followup for the treatment of uncomplicated severe acute malnutrition in Sokoto, Nigeria: A cluster randomized crossover trial

Background Community-based: management of severe acute malnutrition : (SAM) involves weekly or biweekly outpatient clinic visits for clinical surveillance and distribution of therapeutic foods. Distance to outpatient clinics and high opportunity costs for caregivers can represent major barriers to access. Reducing the frequency of outpatient visits while providing training to caregivers to recognize clinical danger signs at home between outpatient visits may increase acceptability, coverage, and public health impact of SAM treatment. We investigated the effectiveness of monthly clinic visits compared to the standard weekly follow-up in the outpatient treatment of uncomplicated SAM in northwestern Nigeria. Methods and findings We conducted a cluster randomized crossover trial to test the noninferiority of nutritional recovery in children with uncomplicated SAM receiving monthly follow-up compared to the standard weekly schedule. From January 2018 to November 2019, 3,945 children aged 6 to 59 months were enrolled at 10 health centers (5 assigned to monthly follow-up and 5 assigned to weekly follow-up) in Sokoto, Nigeria. In total, 96% of children (n = 1,976 in the monthly follow-up group and 1,802 in the weekly follow-up group) were followed until program discharge, and 91% (n = 1,873 in the monthly follow-up group and 1,721 in the weekly follow-up group) were followed to 3 months postdischarge. The mean age at admission was 15.8 months (standard deviation [SD] 7.1), 2,097/3,945 (53.2%) were girls, and the mean midupper arm circumference (MUAC) at admission was 105.8 mm (SD 6.0). In a modified intention-to-treat analysis, the primary outcome of nutritional recovery, defined as having MUAC ≥125 mm on 2 consecutive visits, was analyzed using generalized linear models, with generalized estimating equations to account for clustering. Nutritional recovery was lower in the monthly follow-up group compared to the weekly group (1,036/1,976, 52.4% versus 1,059/1,802, 58.8%; risk difference: −6.8%), and noninferiority was not demonstrated (lower bound of the confidence interval [CI] was −11.5%, lower than the noninferiority margin of 10%). The proportion of children defaulting was lower in the monthly group than in the weekly group (109/1,976, 5.5% versus 151/1,802, 8.4%, p = 0.03). Three months postdischarge, children in the monthly group were less likely to relapse compared to those in the weekly group (58/976, 5.9% versus 78/1,005,:7.8%, p = 0.03), but cumulative mortality at 3 months postdischarge was higher in the monthly group (159/1,873, 8.5% versus 106/1,721, 6.2%, p < 0.001). Study results may depend on context-specific factors including baseline level of care and the clinical status of children presenting to health centers, and, thus, generalizability of these results may be limited. We conducted a cluster randomized crossover trial of a monthly schedule of clinic visit follow-up in the treatment of uncomplicated SAM among children aged 6 to 59 months with a noninferiority design (ClinicalTrials.gov ID: {"type":"clinical-trial","attrs":{"text":"NCT03140904","term_id":"NCT03140904"}}NCT03140904). The study protocol and the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) Checklist are provided as S1 Protocol and S1 STROBE Checklist. The study was conducted in 10 outpatient clinics in the Binji and Wamako local government areas (LGA) in Sokoto state in northwestern Nigeria. This setting is a rural area largely representative of the Sahel region with endemic malnutrition and seasonal peaks of acute malnutrition during the lean season prior to harvest. While national SAM prevalence in Nigeria is 2% [10], northwestern Nigeria experiences the highest rates of malnutrition in the country, with reported SAM prevalence of 7.9% in children <5 years old in Sokoto state in 2018 [10]. CMAM was introduced in northern Nigeria in 2009, and by 2014, achieved coverage of 37% of children eligible for care [11]. The program includes clinic-based care provided by trained health care workers. SAM treatment was provided free of charge in all 10 study sites by the Sokoto State Ministry of Health (SMOH), with support from UNICEF. The study population included children newly admitted for the treatment of uncomplicated SAM at 1 of 10 outpatient clinics between January 2018 and November 2019. Eligibility criteria for treatment of uncomplicated SAM according to local program criteria were minimum weight of 3.5 kg, midupper arm circumference (MUAC) <115 mm and/or grades 1 and 2 edema, absence of current illness requiring inpatient care, and age 6 to 59 months. Additional inclusion criteria were residence in the catchment area of one of the study clinics and written informed consent of the parent or legal guardian. Cases of relapse, where the child was previously successfully treated, discharged as cured and returned with a new episode of acute malnutrition within 2 months of discharge, were eligible for admission, as were those transferred from inpatient care to outpatient care to continue treatment. Returned defaulters who were absent for 3 consecutive visits before recovery and returned to continue treatment and children transferred from another outpatient site were not eligible. Other exclusion criteria for the study included reported history of allergy to peanuts and any other condition which, in the judgment of the field investigator, would interfere with or serve as a contraindication to protocol adherence or the ability to give informed consent. The study compared 2 schedules of follow-up: The proposed monthly schedule of follow-up involved an extension of the period between which the child received a physical assessment by a trained health worker from 1 week to 1 month. During this time, caregivers were asked to monitor the clinical and anthropometric status of the child at home and return to the outpatient clinic for medical attention upon development of any clinical sign of concern (e.g., diarrhea, vomiting, fever, lethargy, lack of appetite, edema, cough, difficulty breathing, or convulsions). To support caregivers in these tasks, culturally appropriate educational tools and messages, including materials for facility-based instruction (e.g., posters and pictorial flip charts), as well as for home-based use (e.g., pocket pictorial cards), were developed with Médecins Sans Frontières France and the Laboratoire d’Etudes et de Recherche sur les Dynamiques Sociales et le Développement Local (LASDEL) in 2014. The tools covered the key topics of (a) home-based MUAC measurement; (b) clinical surveillance of key danger signs or symptoms; and (c) appropriate storage and use of the monthly therapeutic ration. The feasibility and acceptability of these tools were tested in a pilot study in Madarounfa, Niger [12,13]. Caregiver understanding of clinical danger signs increased following training and agreement between MUAC measurements was high between nurses and caregivers. All children admitted for outpatient treatment at a participating study clinic received the same schedule of follow-up (e.g., weekly or monthly clinic visits). Due to lower than expected enrollment in the first 11 months of the study, a 12-month extension of the enrollment period, together with a crossover design to increase study power and efficiency [14], was implemented. On December 17, 2018, the crossover was executed in the field: Outpatient clinics that had been administering monthly visits began to administer weekly visits to all newly admitted children and vice versa. Intervention assignment among children already under follow-up did not change at the time of crossover, and children were analyzed as the group their sites were randomized to at the time of their enrolment. As crossover occurred at the site level, staff ensured children already enrolled completed treatment until discharge under their site’s initial treatment and administered the new crossover treatment only to newly enrolled children. There was no washout period at the site level. At each follow-up visit at the health clinic, a physical exam and anthropometric surveillance were performed, and a therapeutic ration of ready-to-use therapeutic food (RUTF) was distributed. Study staff carried out anthropometric measurements with the use of standardized methods and calibrated instruments [15]. Child height (recumbent length if <85 cm) was measured to the nearest 0.1 cm using a wooden measurement board. Weight was measured to the nearest 0.1 kg using electronic seca scales. Standard medical care was provided as per the national protocol for the treatment of acute malnutrition [16]. As per study protocol, all children received an unannounced home visit within 2 weeks of admission by community health workers to assess the safety of the program and a 3-month postdischarge clinic visit to assess longer-term outcomes. At the home safety visit, study staff assessed incident morbidities, measured MUAC, and counted the number of used RUTF sachets in the household as a proxy to assess appropriate use of the RUTF rations early in treatment. Children with clinical danger signs identified at the home safety visit were referred to the health center for care, and fees for transportation to the hospital, medicine, and testing during eventual inpatient care were taken in charge by the study. To measure RUTF adherence, caregivers were instructed to return to each scheduled clinic visit with used and unused RUTF sachets distributed at the previous visit. A child was defined as compliant if, at the end of treatment, their caregiver returned used at least 80% of sachets provided before the final clinic visit of treatment. The unit of randomization was the SMOH-supported outpatient clinic. The 10 health clinics in Binji and Wamako LGAs were stratified by size of admission (6 sites with ≥500 admissions per year and 4 sites with <500 admissions per year), and centers within each stratum were randomized in a 1:1 ratio to one of the 2 schedules of follow-up. Randomization assignment was made by lottery, in which a local representative selected the name of one of the 2 interventions from an opaque plastic jar. Due to the nature of the intervention, neither the investigators nor participants were blinded. The primary effectiveness outcome was nutritional recovery, defined in both groups as being free from medical complications, having MUAC ≥125 mm on 2 consecutive clinic visits, and no edema. Secondary effectiveness outcomes assessed at program discharge were nonresponse (defined as not meeting the definition of nutritional recovery by 12 weeks from admission); default (defined as 3 missed scheduled clinic visits in the weekly follow-up group and 1 missed scheduled clinic visit in the monthly follow-up group); hospitalization (defined as all-cause, due to weight/edema changes, and/or due to clinical complication, and not including events that ended in death); death from any cause; and among recovered children, weight gain (g/kg/day) at week 4 and week 8 from admission and at program discharge and length of stay from admission to program discharge. At each scheduled clinic visit, incident morbidities assessed by study staff or caregiver report included edema, diarrhea, vomiting, fever, cough, lack of appetite, high temperature, tachypnea, dehydration, pallor, and superficial skin infection at the clinic visit or in the previous week. To assess the early safety of the program at the 2-week home visit, morbidities and change in MUAC (mm/day) from the admission visit to the home visit were recorded. In addition, the RUTF deviance was calculated as the difference between the number of used RUTF sachets observed at home compared to the number expected based on the time since admission and the child’s prescribed dose, as described previously [12]. At 3 months postdischarge, longer-term outcomes assessed were all-cause hospitalization, all-cause cumulative mortality, and relapse after treatment recovery (defined as a maternal report of admission to any therapeutic feeding program within 3 months of discharge from the index admission or MUAC <115 mm or edema observed at the time of the postdischarge visit). Given an observed probability of nutritional recovery in the weekly follow-up group of 0.55, intracluster correlation of 0.01, between-cluster between-period correlation of 0, significance level of 0.05 using the 1-sided Score test, and allowing for 10% loss to follow-up, a sample size of 175 children per outpatient clinic in each period (1,750 per group) was sufficient to achieve >90% power to detect a noninferiority margin difference between the group proportions of −0.10 [17]. The noninferiority margin of 10% was chosen to represent a difference in recovery with programmatic and policy importance within this context. The analysis of the primary outcome, nutritional recovery, was a noninferiority analysis. If the lower bound of the observed risk difference confidence interval (CI) was greater than the noninferiority margin of −0.10, there would be evidence that monthly follow-up was noninferior to weekly follow-up. To estimate risk difference for nutritional recovery, we used a binomial regression with an identity link. For each of the other program outcomes, morbidity at the home safety visit, RUTF adherence, and all postdischarge outcomes, we used log-binomial regression to estimate relative risk and associated 95% CIs comparing intervention groups. Weight gain and length of stay among recovered children, as well as change in MUAC observed at the safety visit, were analyzed using linear regression. We analyzed the incidence of morbidities at clinic visits using Poisson regression, with the number of visits as an offset term. For every analysis, we used generalized estimating equations with clustering by site within each crossover period [18], choosing a full model accounting for clustering by site between crossover periods where appropriate [19]. We defined the lean season as the period of increased admission before the harvest period (July to October) and included an interaction between intervention group and season to assess differences in program effectiveness and RUTF adherence by season. Finally, we assessed differences in RUTF adherence by whether there were ±2 children under 5 years living in the household by including an interaction term. All analyses were performed on the modified intention-to-treat basis, which excluded individuals with protocol violations (n = 72) and those who withdrew consent during the study (n = 99). Data entry was performed using EpiInfo version 7.2.2.2, and sample size calculations and analyses were performed using SAS version 9.3 (SAS Institute, Cary, North Carolina, United States of America) and R, version 4.0.2. The study was approved by the research ethics committee of Harvard T.H. Chan School of Public Health and the SMOH, Nigeria and was conducted in accordance with the Declaration of Helsinki. An independent data and safety monitoring board reviewed study progress and safety events. Caregivers provided written informed consent before admission and were made aware of their ability to withdraw from the study at any time.