Changes in growth, anaemia, and iron deficiency among children aged 6–23 months in two districts in Nepal that were part of the post‐pilot scale‐up of an integrated infant and young child feeding and micronutrient powder intervention

Abstract There is limited research on integrated infant and young child feeding (IYCF) and micronutrient powders (MNPs) programmes operating at scale, despite widespread implementation. This study uses cross‐sectional baseline (n = 2,542) and endline (n = 2,578) surveys representative of children 6–23 months in two districts in Nepal that were part of a post‐pilot scale‐up of a IYCF–MNP programme. Multivariable log‐binomial regression models were used to estimate prevalence ratios (PRs) for stunting (length‐for‐age z‐score <−2), wasting (weight‐for‐length z‐score <−2), underweight (weight‐for‐age z‐score <−2), anaemia (altitude‐adjusted haemoglobin <110 μg/L), moderate or severe anaemia (altitude‐adjusted haemoglobin <100 g/L), iron deficiency (inflammation‐adjusted ferritin <12 μg/L), and iron deficiency anaemia (iron deficiency + anaemia [IDA]) at endline versus baseline and also to compare children in the endline survey based on frequency of mothers' interactions with female community health volunteers (FCHVs; >1× per month or monthly vs. <1× per month) and MNP coverage (1 or ≥2 distributions vs. none among children 12–23 months). Endline children were significantly less likely to be stunted than baseline children in both districts (multivariable‐adjusted PR [95% CI]: 0.77 [0.69, 0.85], P < 0.001 and 0.82 [0.75, 0.91], P < 0.001 in Kapilvastu and Achham, respectively); however, only Achham had significantly lower prevalences of underweight, moderate/severe anaemia, iron deficiency, and IDA at endline. At endline, 53.5% and 71.4% of children had tried MNP in Kapilvastu and Achham districts, respectively, consuming an average of 24 sachets from the last distribution. Frequent maternal–FCHV interactions were associated with a reduced risk of stunting and underweight at endline, whereas repeat MNP coverage was associated with reduced risk of anaemia and IDA. Future research using experimental designs should verify the potential of integrated IYCF–MNP programmes to improve children's nutritional status.


| INTRODUCTION
Undernutrition in early life-particularly from conception to 2 years of age-increases the risk of premature death and illness in childhood and also has lifelong consequences for child development (Black et al., 2013). Micronutrient deficiencies, often called "hidden hunger" because individuals with these conditions may not feel or appear hungry or deficient, are particularly prevalent globally-affecting over a third of people worldwide (Micronutrient Initiative, 2009). Among children aged 6-59 months in low-and middle-income countries, 43% suffer from anaemia (Stevens et al., 2013), an estimated quarter of which is due to iron deficiency (Petry et al., 2016). Iron deficiency anaemia has been shown to impair child development, particularly the motor development of infants and young children, and also has negative consequences for cognition and schooling as children age (Black et al., 2013). Interventions designed to address multiple micronutrient deficiencies have gained particular attention because several deficiencies often cluster within the same individuals and communities (Allen, Peerson, & Olney, 2009;Christian & Tielsch, 2012;Neufeld & Ramakrishnan, 2011). Micronutrient powders (MNPs)-which come in single-dose, light-weight, shelf-stable sachets-can be mixed with a variety of semi-solid foods to increase the availability of vitamins and minerals in children's diets (Zlotkin et al., 2005). Meta-analyses of randomized controlled trials have shown that regular MNP consumption can reduce the risk of anaemia in children aged 6-23 months by a quarter and can cut the prevalence of iron deficiency in half (De-Regil, Suchdev, Vist, Walleser, & Peña-Rosas, 2013;Salam, MacPhail, Das, & Bhutta, 2013); however, evidence from large-scale programmes using MNPs is limited (Hirve et al., 2013;Jefferds et al., 2015;Menon et al., 2007;Nyhus Dhillon et al., 2017;Rah et al., 2012;Serdula et al., 2013;Suchdev et al., 2012;Vossenaar et al., 2017) (Jefferds et al., 2015). After a feasibility study assessed the acceptability of MNP and developed key messages and strategies for implementing an integrated IYCF-MNP programme in Nepal, a pilot programme was implemented in 2010-2011 in six districts (Jefferds et al., 2015). In 2012, the MoH and UNICEF began scaling up the integrated IYCF-MNP programme, which as of 2016, had reached 25 of the country's 75 districts. In this study, we assess changes in the nutritional status (growth, anaemia, and iron deficiency) of children 6-23 months in two districts in Nepal that were part of a post-pilot scale-up of an integrated IYCF-MNP programme.

| SUBJECTS AND METHODS
In Nepal, MNP has been locally branded as Baal Vita, locally translated as "vitamins for children." Each sachet contains 15 micronutrients including 10 mg of elemental iron, 4.1 mg of zinc, 17 μg of selenium, 0.56 mg of copper, 90 μg of iodine, 400 μg of vitamin A, 5 μg of vitamin D3, 90 μg of folic acid, 6 mg of niacin, 0.9 μg of vitamin B12, 30 mg of vitamin C, and 0.5 mg of vitamins E, B1, B2, and B6 (WFP/WHO/UNICEF, 2007). The integrated IYCF-MNP intervention supports improved IYCF practices as well as the distribution of 60 sachets of MNP free of charge to all children aged 6-23 months every 6 months through local health facilities and through female community health volunteers (FCHVs). The programme used a cascade training approach with a manual based on the UNICEF (2012) community-based IYCF training tools with additional information on the appropriate storage, distribution, and utilization of MNP adapted by the MoH for the Nepal context to train health workers and FCHVs. Health workers and FCHVs were in turn expected to counsel mothers during routine health visits and through community outreach. In accordance with the national FCHV strategy, FCHVs (who volunteer their time) were expected to hold monthly mothers group meetings where they would counsel mothers using the IYCF-MNP counselling cards, flipcharts, and cooking demonstrations and were also encouraged (but not required) to conduct home visits or individual counselling if they had the time to do so. FCHVs and health workers were also responsible for distributing 60 sachets of MNP to mothers every 6 months; mothers were instructed to feed their child one sachet per day, with an expected 4-month gap after finishing the 60 sachets before the next

Key messages
• In two districts in Nepal that were part of a post-pilot scale-up of an integrated infant and young child feeding-micronutrient powder (IYCF-MNP) programme, the prevalence of stunting significantly declined from baseline to endline, likely due to secular improvements in socio-economic status as well as the co-location of multiple nutrition and health programmes.
• Repeat MNP coverage was independently associated with a reduced risk of anaemia and iron deficiency anaemia after adjustment for socio-economic and programme-related variables. The district with higher MNP coverage also experienced a significant reduction in the prevalence of anaemia, moderate/severe anaemia, and iron deficiency anaemia from baseline to endline.
distribution. The intervention also included updated counselling aids (e.g., IYCF flipcharts including MNP content, new posters, pamphlets, and radio jingles) re-enforcing several key IYCF messages, with a focus on the timely introduction of complementary foods at 6 months of age; improving the energy density and nutritional quality of local complementary foods (including training on how to make home-made superflour); information about MNP including the appropriate preparation and serving of food with MNP; and age appropriate minimum meal frequency, dietary diversity, and active feeding strategies. The IYCF content also emphasized early breastfeeding initiation, breast milk exclusivity for the first 6 months, and continued breastfeeding after 6 months of age for at least 2 years or longer, as well as sanitation messages related to safe water, handwashing, nail clipping, use of latrines, and appropriate disposal of child faeces. 18 months before the endline survey; thus, all children in the endline survey (who were aged 6-23 months) had been exposed to the full programme (without MNP stock-outs) for 18 months or since they were eligible (as of age 6 months). This provided sufficient time for the oldest participants in the endline survey (children 18-23 months) to complete all three of their distribution cycles (60 MNP sachets every 6 months) after the relaunch but before the endline survey. Notably, in 2014, Suaahara, a multidistrict U.S. Agency for International Development-funded programme, also initiated programme activities in Kapilvastu and Achham (Cunningham et al., 2017). Suaahara targeted mothers and children in the first 1,000 days with an integrated nutrition, health services, family planning, WASH, and agriculture/homestead food production programme. The Suaahara programme provided health workers and FCHVs with additional IYCF trainings; thus, many FCHVs and health workers received trainings from both the Suaahara and integrated IYCF-MNP programmes with overlapping IYCF messages (though Suaahara did not include an MNP component). Both programmes also supported IYCF key messages with mass media, and both programmes supported health worker and FCHV group meetings, home visits, and food demonstrations to strengthen IYCF practices among mothers of children under 23 months.
2.1 | Survey design, sampling, and data collection Prior to expanding to nine additional districts in 2013, a programme evaluation was designed for Achham and Kapilvastu districts (in the hills and plains ecological zones, respectively). Cross-sectional baseline and endline surveys representative of children aged 6-23 months in the two districts were implemented prior to programme initiation in December to February 2012/2013 and then again in January to February 2016. During consultations on the evaluation design, the stakeholders of the integrated IYCF-MNP programme decided that it would not be practical to include "control" districts in the impact evaluation given the rapidly changing landscape of nutrition programmes in Nepal and the likelihood that additional nutrition interventions would be initiated in other districts between the baseline and endline surveys. The baseline survey sample size was calculated based on an estimated change in anaemia prevalence from 65% (Nepal Demographic and Health Survey, 2011) to 50% and iron deficiency from 50% to 40%. The results of the baseline survey data were used to estimate sample sizes for the endline survey. The assumed design effect for anaemia was 3.2 and 2.0 for iron deficiency. With an assumed household response rate of 95%, an individual response rate of 90%, and an overall response rate of 85.5%, we estimated the need for total sample size of 2,662 children 6 to 23 months in the two districts. Function to allow for the tarred weighing of young children held by the mother measured weight to the 0.01 kg. Trained laboratory technicians collected venous blood specimens from children and then assessed haemoglobin, malaria, and Helicobacter pylori (H. pylori) infections at the household using the HemoCue® Hb-301 photometer (Angelholm, Sweden), the malaria antigen (HRP2/pLDH) combination rapid diagnostic kit for Plasmodium falciparum and Plasmodium vivax, and the H. pylori QuickVue rapid test kit, respectively. A disposable pipette transferred a drop of blood from the blood collection tube onto a piece of parafilm for assessing haemoglobin, malaria, and H. pylori.
Although malaria is endemic in parts of Nepal, particularly during the hot season from April to October, at the time of our baseline and endline assessments (all from December to February), there was only one positive case of malaria at baseline and three at endline and no positive cases of H. pylori. The HemoCue® Hb-301 photometer is self-calibrating, but laboratory technicians also performed additional quality control procedures at the beginning of each day using three levels of liquid controls (low, normal, and high; Eurotrol, Burlington, MA, USA). The laboratory technicians placed the blood specimens into a cold box containing frozen gel packs and a thermometer. At a central portable laboratory established in the field in each cluster, the specimens were centrifuged on the same day (within 1-2 hr after collection), transferred into cryovials, and then stored in portable freezers.
At the end of each day, the processed specimens were transferred to the freezers in the District Public Health Offices for storage until the end of data collection. At the end of the survey, all specimens from the District Public Health Offices were then transferred to the National Public Health Laboratory for storage at −86°centigrade.
The specimens from National Public Health Laboratory were sent to the VitMin Laboratory (Willstaett, Germany) for the testing of ferritin, C-reactive protein, and alpha-1 acid glycoprotein presented in this analysis using an in-house sandwich enzyme-linked immunosorbent assay technique (Erhardt, Estes, Pfeiffer, Biesalski, & Craft, 2004).
Length-for-age, weight-for-length, and weight-for-age z-scores (LAZ, WLZ, and WAZ) were calculated using the WHO (2006) growth standards ((WHO Multicentre Growth Reference Study Group & de Onis, 2006). Each child's age was back-calculated by subtracting the child's birth date from the interview date. Stunting, wasting, and underweight were defined as <−2 standard deviations for LAZ, WLZ, and WAZ, respectively. In accordance with WHO recommendations, all extreme LAZ (<−6 or >6), WLZ (<−5 or >5), and WAZ (<−6 or >5) values were set to missing (WHO, 2006). As additional quality checks of length and weight data, we examined the per cent of missing data for age, sex, length, and weight; end-digit preference for length and weight values; and SD for LAZ, WLZ, and WAZ. Each child's haemoglobin was adjusted for altitude based on the global positioning system coordinates of the household (WHO, 2011). Anaemia was defined as altitude-adjusted haemoglobin <110 g/L. Moderate or severe anaemia was defined as altitude-adjusted haemoglobin <100 g/L; children with moderate and severe anaemia were combined into a single category because only six and seven children in the baseline and endline surveys were classified as severely anaemic (haemoglobin <70 g/L). Serum ferritin was adjusted for inflammation (C-reactive protein and alpha-1-acid glycoprotein) using the Biomarkers Reflecting Inflammation and Nutritional Determinants of Anemia linear regression technique with internal data-driven reference levels (Suchdev et al., 2016). Iron deficiency was defined as inflammation-adjusted serum ferritin <12 μg/L. Iron deficiency anaemia was defined as iron deficiency plus anaemia.

| Ethical approval
Ethical approval was obtained from the Nepal Health Research Council for both the baseline and endline surveys. For each survey, interviewers described the purpose, procedures, risks, and benefits of the study and allowed mothers to ask questions before inviting mothers and children to participate in the survey. Mothers or other legal guardians then provided written informed consent to enrol themselves and their children in the study. If the mothers or guardians were illiterate, then a witness signature was obtained.

| Statistical methods
Descriptive statistics-frequencies with proportions and means with standard deviations-were used to compare baseline and endline sociodemographic, nutrition, water, and sanitation and programme characteristics by district. Multivariable log-binomial regression models were used to estimate the prevalence ratios for binary outcomes comparing children in the endline survey to those in the baseline survey (Spiegelman & Hertzmark, 2005), whereas continuous outcomes were compared using linear regression models; all models account for clustering using an exchangeable correlation structure.
Covariates for multivariable models were selected a priori based on a review of the literature and include child's sex, age, caste, maternal education, household food insecurity level, and household asset tertile developed from a principal component analysis based on household ownership of electricity, radio, television, mobile, refrigerator, table, chair, bed, sofa, watch, computer, fan, traditional grain miller, and bicycle (Vyas & Kumaranayake, 2006). In order to explore the large reduction in stunting from baseline to endline in the two districts, the point and interval estimates of the per cent of the change from baseline to endline that was mediated by intermediate variables, such as sociodemographic characteristics, water, sanitation, hygiene, health, and IYCF indicators, were also estimated using log-binomial regression models.
Because some programme exposure variables, such as MNP coverage and the frequency of maternal-FCHV interactions, were only available in the endline survey, log-binomial models were used to assess the relationship between programme exposure variables and children's nutritional status in the endline survey only. In a previous analysis, we found that key IYCF practices (including feeding the child the minimum dietary diversity, meal frequency, and acceptable diet) improved in the programme area from baseline to endline and that several IYCF practices were significantly more prevalent among mothers who received IYCF counselling from an FCHV and interacted with their FCHV at least twice per month compared with mothers who did not receive FCHV IYCF counselling or interacted with the FCHV less than once per month (Locks et al., 2018). Because mother-FCHV interactions may have contributed to children's nutritional status through pathways external to discussing IYCF only (such as discussing WASH or child health more generally), in this analysis, we compare children's nutritional status solely based on the frequency with which their mothers interacted with their FCHVs for the child: frequent (>once per month) and intermediate (once per month) compared with infrequent interactions (<once per month). Because very few mothers reported "never" interacting with their FCHV, mothers who reported that they never interacted with their FCHV or did so less than once per month were collapsed into a single reference group. We also compared children's nutritional status based on the number of times the mothers received MNP for the child (≥2 MNP distributions or a single The two districts were first modelled separately; however, because trends across the two districts were similar, the districts were combined, and district was added as a covariate. Because analyses assessing the relationship between programme indicators and health outcomes were not intended to provide population-based estimates, sample weights were not applied to these analyses; all models account for clustering using an exchangeable correlation structure. All analyses were conducted in SAS 9.4 (Cary, NC, USA).

| RESULTS
In all, 2,543 and 2,577 children participated in the baseline and endline surveys, respectively, reflecting response rates of 97% and 96% of targeted children (   In analyses of the influence of demographic, health, and programme covariates on the observed changes in stunting from baseline to endline, we found that 43.2% (P = 0.002) and 36.8% (P = 0.01) of the change in the prevalence of stunting from baseline to endline in Kapilvastu and Achham, respectively, could be explained by sociodemographic characteristics and water, sanitation, health, and IYCF indicators that were measured in the baseline and endline surveys (  Prevalence ratios and corresponding 95% confidence intervals and P values obtained from generalized estimating equations using a log link and binomial distribution, accounting for correlated errors within clusters using an exchangeable correlation structure. When the log-binomial model did not converge (in the models for stunting), a Poisson distribution was used. For continuous variables, difference of means and corresponding 95% confidence interval and P values were obtained from generalized estimating equations using the identity link and normal distribution accounting for correlated errors within clusters using an exchangeable correlation structure.
b All models adjust for child's sex, age and caste, maternal education, household food insecurity level and asset tertile, source of drinking water, type of toilet, whether anyone in the household participated in the "Open Defecation Free" campaign, and whether the child received deworming medication and/or vitamin A supplementation in the previous 6 months. c Stunting, wasting, and underweight defined as <−2SD for length-for-age, weight-for-length, and weight-for-age z-scores, respectively (WHO, 2006).  Water, sanitation, and health variables are indicators that were not targeted by the IYCF-MNP intervention. They include source of drinking water (piped water, tube well, or unimproved water source), type of toilet (flush toilet, improved pit latrine, unimproved pit latrine vs. no toilet), whether the household participated in the "Open Defecation Free" campaign, and whether the child was dewormed and/or received vitamin A in the previous 6 months. c The IYCF and hygiene behaviours that were added last to the full multivariable model include all practice indicators that were targeted by the IYCF-MNP programme and were measured at baseline and endline. These include when the mother reports introducing complementary foods (before, after, or at 6 months), frequency of breastfeeding in the previous day, frequency of solid foods in the previous day, number of food groups consumed in the previous day, whether the mother engaged in any responsive feeding activity (singing, talking to or eye contact with the child while feeding) in the previous day, and whether the child consumed locally made superflour in the previous day. It also includes hygiene indicators that were targeted by the IYCF-MNP programme key messages including whether there was observed soap and water at the household handwashing station, whether the mother reports using soap the current or previous day, whether she reports washing her hands with soap before preparing food or feeding the child in the previous or current day, whether she reports using soap after cleaning the child's stool in the previous or current day, and whether the household has a nail clipper.
characteristics, frequency and location of mother-FCHV interactions for the child, and whether the child received minimum acceptable diet in the previous day (Table 5).

| DISCUSSION
In this analysis of baseline and endline surveys that were representative of children aged 6-23 months in two districts in Nepal that were  Prevalence ratios and corresponding 95% confidence intervals and P values obtained generalized estimating equations using a log link and binomial distribution, accounting for correlated errors within clusters using an exchangeable correlation structure. When the log-binomial model did not converge (in the model for stunting), a Poisson distribution was used. For continuous variables, difference of means and corresponding 95% confidence interval and P values were obtained from generalized estimating equations with the identity link and normal distribution accounting for correlated errors within clusters using an exchangeable correlation structure. Because we did not intend to perform population-based estimates, sample weights were not applied to the analysis.
b Adjusted models control for district, child's sex, age and caste, maternal education, household food insecurity level and asset tertile, and the location of mother-FCHV interactions for the child (home visits only, visits elsewhere in the community only, or both). c Stunting, wasting, and underweight defined as <−2SD for length-for-age, weight-for-length, and weight-for-age z-scores, respectively (WHO, 2006). d Anaemia defined as haemoglobin <110 g/L. Moderate or severe anaemia defined as haemoglobin <100 g/L. Haemoglobin was adjusted for altitude (WHO, 2011). N for anaemia and haemoglobin is 720, 679, and 858 for FCHV <1× per month, FCHV once per month, and FCHV >once per month groups, respectively. e Adjusted for inflammation using the Biomarkers Reflecting Inflammation and Nutritional Determinants of Anemia linear regression technique (Burke et al., 2017;Suchdev et al., 2016). f Iron deficiency defined as serum ferritin <12 μg/L. N for serum ferritin and iron deficiency is 709, 667, and 824 for FCHV <1× per month, FCHV once per month, and FCHV >once per month groups, respectively.
g Iron deficiency anaemia defined as serum ferritin <12 μg/L and haemoglobin <110 g/L. 2013); however, optimal IYCF practices-particularly complementary feeding of diverse, nutrient-rich foods like animal-source foods-have been shown to improve paediatric anthropometry (Dewey & Adu-Afarwuah, 2008). We found that mothers who interacted with an FCHV at least twice per month were significantly less likely to have children who were stunted or underweight. These findings are consistent with a previous analysis from this programme that showed that several key IYCF practices (including minimum dietary diversity and minimum acceptable diet) improved significantly from baseline to endline in both districts and that mothers who received IYCF counselling from an FCHV and interacted with their FCHV at least twice per month were significantly more likely to engage in several key IYCF practices (Locks et al., 2018). Taken together, the evidence indicates that the co-location of the integrated IYCF-MNP programme and the Suaahara programmes, both of which substantially invested in strengthening the capacities of FCHVs in child nutrition and IYCF, combined with secular trends in underlying socio-economic status, likely explains the observed improvements in child stunting and underweight. Notably in our analysis, mothers who saw their FCHV monthly were significantly less likely to have a child who was wasted compared with mothers who saw their FCHVs less than once per month, but frequent (>once per month) interactions were not associated with a reduced prevalence of wasting. It is possible that in this cross-sectional analysis, there is reverse causation and that mothers of children who recently became wasted also recently began visiting their FCHV more frequently than the average monthly interaction.
In mediation analyses, we were able to attribute 43.2% and 36.8% of the decline in the prevalence of stunting to changes in sociodemographic characteristics and water, sanitation, hygiene (WASH), health, and IYCF indicators that were measured in our baseline and endline surveys. When the IYCF and hygiene behaviours (that were targeted by both the integrated IYCF-MNP and Suaahara programmes) were added to the multivariable model that already Analysis is restricted to children 12-23 months because children under 12 months were not expected to receive more than one MNP distribution.
b Prevalence ratios and corresponding 95% confidence intervals and P values obtained from log-binomial regression models accounting for correlated errors within clusters using an exchangeable correlation structure. When the log-binomial model did not converge (in the model for stunting), a Poisson distribution was used. For continuous variables, difference of means and corresponding 95% confidence interval and P values were obtained from linear regression models accounting for correlated errors within clusters using an exchangeable correlation structure. Because we did not intend to perform populationbased estimates, sample weights were not applied to the analysis. Multivariable models control for district, child's sex, age and caste, maternal education, household food insecurity level and asset tertile, mother-female community health volunteer interaction for the children frequency (<once per month, monthly, and >once per month) and location (home visits, visits elsewhere in the community, or both), and whether the child received the minimum acceptable diet in the previous day. c Stunting, wasting, and underweight defined as <−2SD for length-for-age, weight-for-length, and weight-for-age z-scores, respectively (WHO, 2006 (Reerink et al., 2017).
We also found that the prevalence of iron deficiency, iron deficiency anaemia, and moderate-severe anaemia reduced significantly in Achham district from baseline to endline. 2016) will be essential for improving future programmes using MNP (Tumilowicz, Schnefke, Neufeld, & Pelto, 2017).
Limitations of our study include the cross-sectional nature of the baseline and endline surveys and the lack of comparison group; it is thus possible that some of the observed changes in nutritional status between baseline and endline are due to secular trends not related to the IYCF-MNP programme. In the endline analyses assessing the association of frequency of mother-FCHV interactions and MNP coverage with children's nutritional status, we are also limited by the lack of randomization of the exposures, and thus, underlying differences between mothers who saw their FCHV frequently versus infrequently, or among children who received MNP versus those who did not, may ultimately be confounding the associations between programme exposure and nutrition indicators. We did, however, adjust for several demographic covariates, including maternal literacy and socio-economic variables, which did not substantially change our findings. Last, some data, including IYCF practices and experiences with FCHV, are maternal report and subject to various biases including socially desirability and time recall; though notably in the Nepal pilot IYCF-MNP programme, maternal recall of the number of sachets consumed and the directly observed difference (based on subtracting the number of observed unopened sachets from the number received) were highly correlated (Ng'eno et al., 2017).

| CONCLUSION
In two districts in Nepal that were part of a post-pilot scale-up of an integrated IYCF-MNP programme, the prevalence of stunting significantly declined from baseline to endline in both districts, likely due to secular improvements in socio-economic status as well as the colocation of multiple nutrition and health programmes. Repeat MNP coverage was independently associated with a reduced risk of anaemia and iron deficiency anaemia after adjustment for socio-economic and programme-related variables. Achham, the district with higher MNP coverage, also experienced a significant reduction in the prevalence of anaemia, moderate/severe anaemia, and iron deficiency anaemia from baseline to endline. As Nepal, and other national programmes, continue to scale-up integrated IYCF-MNP programmes, embedded implementation research on how to achieve and sustain high MNP coverage and adherence will be essential for maximizing the benefit of integrated IYCF-MNP programmes.

CONFLICTS OF INTEREST
The authors declare that they have no conflicts of interest.

AUTHOR CONTRIBUTIONS
MEJ, PD, ZM, DW designed the evaluation; PD, RP, SC, NP, NJ and PL oversaw program implementation; PD, RP, SC, NP, NJ, PL, MEJ, ZM, and DW provided training, supervision and support of data collectors, and provided interpretation and approval of analyses; LML analyzed data; and LML, AG, MEJ wrote the paper and had primary responsibility for final content. All authors read and approved the final.