Serum zinc status is a matter of concern among children and non-pregnant women in a nationwide survey of Nepal

Nationally representative population data on zinc status in Nepal is lacking at present. This study analyzed data from the recent Nepal National Micronutrient status survey 2016 to determine the prevalence of zinc deficiency and associated risk factors among children aged 6–59 months (n = 1462) and non-pregnant women aged 15–49 years (n = 1923). Venous blood was collected from the participants to measure micronutrients such as zinc, markers of anemia, RBP (vitamin A), and markers of inflammation. Stool samples were collected to assess soil-transmitted helminths and Helicobacter pylori infection. Socio-demographic, household, and other relevant factors were collected by a structured questionnaire. Serum zinc concentration was measured by Microwave Plasma Atomic Emission Spectrometry, and zinc deficiency was defined according to the International Zinc Nutrition Consultative Group’s guidelines. Logistic regression was used to examine the predictors of zinc deficiency among the participants. The overall zinc deficiency in children was 22.9%, while it was higher in non-pregnant women (24.7%). The prevalence of anemia among zinc-deficient children was higher (21.3%) than the zinc non-deficit children (18.7%). The prevalence of anemia was 18% among zinc-deficient non-pregnant women compared to 22% non-deficit non-pregnant women. Predictors associated with zinc deficiency among the study children were living in rural areas (AOR = 2.25, 95% CI, [1.13, 4.49]), the occurrence of diarrhea during the two weeks preceding the survey (AOR = 1.57, 95% CI, [1.07, 2.30]), lowest household wealth quintile (AOR = 0.48, 95% CI, [0.25, 0.92]) and lower vitamin A status (AOR = 0.49, 95% CI, [0.28, 0.85]. The predictors associated with zinc deficiency among non-pregnant women were: being underweight (AOR = 1.55, 95% CI, [1.12, 2.15]), fever occurrence during two weeks preceding the survey (AOR = 1.43, 95% CI, [1.04, 1.98]), H. pylori in the stool (AOR = 1.33, 95% CI, [1.04, 1.71]), lowest household wealth quintile (AOR = 0.62, 95% CI,[0.40, 0.94]) and being at risk of folate deficiency (AOR = 0.58, 95% CI,[0.36, 0.94]). We conclude that community-level intervention programs focused on rural children and women to prevent diarrhea, improve nutrition counseling, and provide economic opportunities in rural communities may help to lower zinc deficiency and other micronutrient deficiencies in the Nepalese population. We believe that intervention programs to address zinc deficiency should not be isolated. Instead, integrated approaches are beneficial to improve overall micronutrient status, such as encouraging dietary diversity, providing livelihood opportunities to the unemployed, micronutrient supplementation to vulnerable populations, and consumption of zinc-rich animal-based foods.


Discussion
This paper details findings from the Nepal National Micronutrient Status Survey 2016 (NNMSS2016), focusing on zinc deficiency in children aged 6-59 months and NPW aged 15-49 years. The prevalence of zinc deficiency in our study was 22.7% in children and 24.7% in NPW. Zinc deficiency is considered a public health concern when the prevalence crosses 20% 36 . Based on our findings, zinc deficiency in Nepal is a public health problem, and we suggest longitudinal or interventional studies targeting the risk groups in different ecological regions across the country. A few cross-sectional studies conducted among school-age children, women of reproductive age, and pregnant women have also shown a high prevalence of zinc deficiency in Nepal 34,37,42 . Zinc deficiency is not only a public health problem in developing countries like Nepal; recent data from some developed countries such as Japan and New Zealand have also indicated report a higher burden of zinc deficiency 43,44 . We didn't find nationally representative data from India on zinc deficiency, a neighboring country of Nepal that shares similar sociocultural and food practices 45 . However, several studies from various regions of India report that the prevalence of zinc deficiency was high (43.8%) in children aged 6-60 months and about 53% in non-pregnant women 46,47 .
The estimation of many nutritional markers such as ferritin, zinc, total protein, etc., are affected by inflammation. This is because inflammation may alter the synthesis of their carrier proteins or cause translocation of the metals from the blood into the liver. So, measurement of these micronutrients, when inflammation is present in the body, results in under or overestimation of these micronutrients 48 . One study reported 31% of Nepalese children 6-8 years of age had high levels of α-acid glycoprotein (AGP) and C-reactive protein (CRP), prominent markers of inflammation 49 . Also, unpublished data from a baseline survey in 2012 in Nepal showed a high percentage of children (6-23 months of age) with sub-clinical inflammation. Therefore, the micronutrient concentration of the children in this study has been adjusted based on the AGP and CRP values, which minimizes bias 50 .
Our findings show that children residing in rural areas were at a higher risk of zinc deficiency compared with those from urban areas. An African study (Burkina Faso) also showed a high prevalence of zinc deficiency in children residing in rural areas 51 . The location of residence (rural vs. urban) is a non-modifiable predictor of zinc status. The increased risk of zinc deficiency in children residing in rural areas might reflect the limited consumption of food products from animal sources compared with those living in urban areas. Studies show that low dietary intake is the major cause of micronutrient deficiencies in developing countries. The practice of consuming mostly plant-based diets as a staple in rural communities in developing countries might contribute to increased zinc deficiency in children 52,53 . Meat products are quite expensive, and poor people cannot afford to buy them. Animal products such as meat and oysters are a good source of zinc 54 , but these are very limited and often hard to get in rural communities in Nepal. Plant-based diets, typical in Nepal, contain a high amount of phytates which are potent inhibitors of zinc absorption in the intestines 35 . A low dietary diversity reflecting low consumption of nutrients is also associated with low serum zinc status in pregnant women 55 . However, our study did not assess food consumption patterns in the participants, which could have further highlighted the contribution of dietary intake in maintaining serum zinc levels.
Many international organizations are helping the GoN to address maternal and infant nutrition problems. The Suaahara II (2016-2023) project funded by USAID has been helping GoN to reduce malnutrition in women and infants. They are working through a multi-sectoral approach to combat malnutrition such as promoting water, sanitation, and hygiene (WASH) practice, encouraging households to grow and produce micronutrientrich foods in their kitchen garden, poultry farming, access to diverse and nutritious foods to children, and radio program to raise awareness on cooking nutritious foods 28 . These interventions ultimately help to improve the micronutrient status of the population. www.nature.com/scientificreports/ As our findings also implicate lower wealth quintile (reflecting the lower economic status of the household) associated with more prone to zinc deficiency, reducing poverty would address micronutrient deficiency. Heifer International is an organization currently working in Nepal to address poverty reduction by promoting smallscale livestock enterprises (such as goat and dairy) in the communities 56 . They encourage the consumption of meat products, which might aid in zinc intake and improve micronutrient status at the household level.
The occurrence of diarrhoea in children during the two weeks preceding the survey was associated with a higher risk of zinc deficiency. A study conducted among Nepalese children aged 6-35 months reported a significant association between dysentery and zinc status 50 . Diarrhoea or dysentery leads to loss of body fluids, and zinc can be excreted in the stool 57 . Studies have suggested zinc supplementation improves gastrointestinal mucosal integrity and promotes the immune system, thereby potentially reducing the severity and duration of diarrhoea 58 . Based on our findings, nutritional awareness, improving the dietary pattern to include meat products, Table 1. Socio-demographic and health characteristics of children aged 6 to 59 months by the status of zinc deficiency, Nepal National Micronutrient Status Survey, Nepal, 2016. Ns are unweighted. Values presented are mean(95%CI) or percent(95%CI). All estimates account for weighting and complex sampling design. AGP ɑ-1 acid glycoprotein, CI confidence interval, CRP C-reactive protein, RBC red blood cell, RBP retinolbinding protein. a Zinc deficiency defined as serum zinc < 65.0 μg/ dL for nonfasted, morning (i. e. before 12 pm) samples and < 57.0 μg/ dL for non-fasted, afternoon (i. e. after 12 p.m.) samples, inflammation adjusted 38 . b Hemoglobin adjusted for altitude and smoking 39 . c Anemia defined as altitude-and smokingadjusted Hb < 12.0 g/dL 39 . d Soil-transmitted helminths including hookworm, Trichuris trichura, and Ascaris lumbricodes. e Biomarker was regression-adjusted to a pooled country reference to adjust for inflammation, using CRP and AGP (ferritin) or AGP only 40 . f Iron deficiency defined as inflammation-adjusted serum ferritin < 15.0 μg/L 39 . g Vitamin A deficiency was defined as RBP < 0.64 μmol/ L. h Folate cut off based on the risk of megaloblastic anemia defined as RBC folate < 305.0 nmol/L 41 .  provision of health/medical services, and livelihood programs coupled with nutritional counseling for people living in rural areas might help improve zinc status as well as overall nutritional status. The body mass index is a modifiable predictor of zinc status in NPW. As our data show that being underweight heightens the odds of zinc deficiency, maintaining body weight, fitness, and consumption of adequate nutrients might help prevent zinc deficiency among those NPW. Infection with H. pylori was associated with a higher likelihood of zinc deficiency, as did fever during the two weeks preceding the survey. H. pylori infection is more commonly discussed in the context of anemia 59,60 , and evidence linking H. pylori with zinc status is scarce. A study in dyspepsia patients showed an association of H. pylori-induced gastric inflammation with reduced zinc concentration in gastric tissues. H. pylori infection can induce increased reactive oxygen isotypes resulting in oxidative stress and zinc deficiency, which further exacerbates inflammation 61 . Another notion is that inflammation caused by H. pylori in the stomach might trigger systemic acute phase responses that could translocate trace metals from blood to the liver resulting in the deficiency of metals, including zinc 62 . There is evidence that H. pylori infection can transmit within a family through personal contact, and proper hygiene maintenance may prevent its transmission and spread 63 . The relationship between fever and zinc deficiency is less clear; however, a study found that children with dengue and zinc deficiency had a slightly longer duration of fever and hospital stay as compared to children with normal zinc levels 64 . The drop in zinc level in the blood during fever may be due to higher hepatic synthesis of zinc-binding acute phase proteins, including metallothioneins 65 , and translocation of metals from circulation to liver 62 .

Socio-demographic and health characteristics
Studies have suggested a possible interaction between stool helminths, serum zinc, and vitamin A status in infants. The complex interaction between host micronutrient status and parasitic infection may involve the impact of parasites on host nutritional imbalance as well as low immune status due to deficiency of micronutrients such as zinc and vitamin A [66][67][68] . We didn't find any association between STH and zinc status in our population. However, some studies show the interaction of micronutrients with parasite infection. For example, Curtale et al. showed vitamin A deficiency was associated with the presence of Ascaris lumbricoides eggs, and vitamin A supplementation resulted in a reduction in Ascaris eggs 68 . Similarly, low zinc status was also associated with increased parasitic survival and increased parasitic load in school children from Kenya 69 .
Our study observed that good economic status (rich vs. poor) and risk of folate deficiency were protective against zinc deficiency in NPW. Economic status is a non-modifiable factor. However, through proper economic support and livelihood opportunities from the government, people with low socio-economic status can gradually increase their living standards. Income growth will enable them to buy nutritious foods that may help to www.nature.com/scientificreports/ improve their overall nutrition uptake. Low folate level as a protective factor for zinc deficiency in our study is quite intriguing, as many studies suggest no significant interaction between folate and zinc in the context of intestinal zinc. Evidence of the interaction between folate and zinc is less clear 70 . However, folate could impair intestinal absorption of zinc through the formation of insoluble chelate compounds 71 . This may be the reason behind the reduction of zinc deficiency when there is associated folate deficiency. Thus, future studies, preferably interventional in design, should investigate the role of folate in maintaining zinc status in NPW. Low zinc levels in a high proportion of children aged 6-59 months and NPW aged 15-49 years in our study suggest that zinc deficiency is a public health concern in Nepal. Our study identified several modifiable predictors of zinc deficiency, such as body mass index, diarrhoea, and fever occurrence. Appropriate intervention, as discussed above, could improve zinc status as well as the overall nutritional status of children and women. Interventional studies with zinc supplements, particularly in vulnerable groups, are warranted to verify the relationship between zinc status and its observed risk factors.
Strengths and limitations of the study. The NNMSS 2016 is a nationally representative sample of all ecological regions and socio-demographic groups, which is its biggest strength. This is the first study from Nepal using nationally representative data to assess the prevalence and predictors of zinc deficiency. This study has some limitations as well. Perhaps the biggest limitation of our study is its cross-sectional design; we could not ascertain causality between the predictors and zinc deficiency. This study does not include samples from adult males and respondents aged 49 years and above. We were also unable to determine the contribution of dietary zinc on serum zinc levels because of a lack of dietary information.

Conclusion
Our study suggests that zinc deficiency and other micronutrient deficiencies are prevalent in Nepalese children and NPW, especially those from rural communities and low socio-economic status. We also assume that micronutrients may interact with one another, resulting in multiple micronutrient deficiencies. Our findings indicate that the government should implement policies and programs to reduce poverty, generate employment opportunities targeted to people living in rural areas, the poor, those underweight, and improve both the zinc status and overall nutritional health of the population.

Materials and methods
Study setting. We used cross-sectional data from a nationally representative survey, the Nepal National Micronutrient Status Survey 2016 (NNMSS-2016). So, this paper is a part of the larger NNMSS-2016 data. The GoN has been implementing public health programs to improve the nutritional status of vulnerable groups of the population such as children, adolescents, non-pregnant and pregnant women. Nutrition-specific programs include the iron-deficiency anemia control program, the iodine deficiency disorder elimination program, vitamin A deficiency control programs, and programs on improving maternal and child health. The NNMSS was conceptualized to obtain nationally representative data on micronutrient status and associated factors and determine priority process and outcome indicators for national supplementation and fortification/intervention programs.

Study participants. The study participants included children aged 6-59 months and NPW aged 15-49 years.
Out of the total sample size of 1728 for the children, 19 of them either refused to participate or were not available at home after three attempts. Thus, finally, 1709 children completed the survey. Serum zinc data was only available from 1462 children. Therefore, the final sample size for the analysis was 1462 for children.
Similarly, the calculated sample size for women was 2160, of which 16 women either refused to survey or were not available at home, and finally, 2144 of them completed the survey. The serum zinc data was only available from 1923 of these women. Thus, accordingly, the final sample size in the analysis was 1923 for women.
Sampling design. The stratified multistage cluster sampling was adopted for the survey. Clusters (wards) were used as primary sampling units (PSUs). The minimum cluster size was considered as 100 households, and clusters were merged with the adjacent wards if the cluster size was less than 100. If any cluster had more than 300 households, it was divided into segments of approximately 100 households. The average number of households in a rural ward is 104, while the number reaches up to 800 households in an urban ward. Here in our survey, wards were taken as clusters. Therefore, the minimum cluster size was considered 100 households.
In the first stage of sampling, six clusters from each stratum in the Mountain and 15 clusters from each stratum in the Hill and Terai were selected using probability proportional to size (PPS) by cluster population size. The majority of the population lives in Terai (50%), and Hill (43%), and only 7% lives in Mountain. Therefore, the stratification by ecological zones was done to ensure reasonable estimates for the Mountain. However, the same number of clusters were selected from Hill and Terai, although they had slightly different populations. Altogether 180 clusters were selected (Mountain = 30, Hill = 75 and Terai = 75) and a total of 36 clusters from each developmental region. The distribution of clusters across developmental regions and ecological zones is presented in the www.nature.com/scientificreports/ In the second stage, the data collection team created the maps of the selected clusters with the help of the ward officers a few days prior to data collection. Numbering was done for each household in each cluster. The survey staff then selected 24 households from each cluster using a systematic sampling method. Thus, a total of 4320 households (180 × 24) were invited for the survey. Finally, eligible individuals (children aged 6-59 months and NPW aged 15-49 years) from the household were selected for the interview and sample collection.

Data and biological samples collection.
Out of the calculated sample size of 4320 households, only 4309 households completed the survey, and 11 households refused to participate in the survey or were not available at the time of the survey (please refer to the section "study participants" for the number of children and NPW participants). Data collection took place between April 2016 to June 2016. A total of 14 teams of 13 members each included a field supervisor, enumerators, and phlebotomist, and laboratory technicians who collected all data from the participants. Trained phlebotomists and pathologists collected blood (in EDTA and plain vials), urine and stool samples from the participants at their houses. Blood samples were non-fasted samples as fasting was not possible in the survey design. The survey staff collected stool samples within 24 h of the interview. The samples were shipped via cold chain to Kathmandu, where they were stored at − 80 °C at the National Public Health Laboratory (NPHL). The blood samples for hemoglobin and malaria; and stool samples for testing soiltransmitted helminths (STH) were analysed in the field immediately after collection. The personnel responsible for collecting all the biological samples were laboratory technicians. Pathologists were involved in the testing of STH in the stools. All other samples, including blood and urine were transported to National Public Health Laboratory, Kathmandu. From NPHL, the blood samples were sent to different national and international laboratories for the laboratory analysis of micronutrient status (zinc, iron, ferritin, sTfR, retinol-binding protein, and Table 4. Predictors of zinc deficiency among NPW aged 15-49 years, Nepal National Micronutrient Status Survey, Nepal, 2016. Estimates are unadjusted odds ratios and adjusted odds ratios with 95% confidence intervals from logistic regression models, accounting for weighting and complex sampling design. a Biomarker was regression-adjusted to a pooled country reference to adjust for inflammation, using CRP and AGP (ferritin) or AGP only 40 . b Folate cut off based on the risk of megaloblastic anemia defined as RBC folate < 305.0 nmol/L 41 .

Potential predictors
Unadjusted odds ratio (95% CI) Adjusted odds ratio (95% CI) P www.nature.com/scientificreports/ folate) and inflammation measures (C-reactive protein and Alpha-1 acid glycoprotein). We corrected for inflammation in children but not in women. The BRINDA project, which analyzed nutritional data from 13 nationally representative surveys, showed that inflammation was correlated with serum zinc in children, but the correlation of inflammation with serum zinc in women was weak and inconclusive 72 . This is why we have not adjusted for inflammation in our samples from women. Altitude adjustment was made for the hemoglobin levels. Data on socio-demographic characteristics, participation in national nutrition programs and other interventions, recent micronutrient supplementation (zinc, iron, folic acid, vitamin A, multiple micronutrient supplementation or powders), two-week recall of fever, cough, and diarrhoea, and other relevant information were collected using a structured questionnaire.

Age group, (%)
The household wealth quintile was determined based on the world bank report authored by Gwatkin et al. 73 . The wealth quintile was constructed using the first principle component of household assets and the types of materials used for flooring, roofing and external walls, sources of drinking water, and possession of sanitation facilities. A weight (factor score) was generated for each asset through principal component analysis. The resulting scores were standardized in relation to a normal distribution with a mean of zero and a standard deviation of one. Each household was then assigned a score for each asset, and the scores were summed up for each household. The sample was finally divided into quintiles from lowest to highest (five). A single asset index was developed for each household.
Body mass index (BMI) was calculated for the anthropometric measurement of children and women. The wealth status was calculated based on household assets, and principal component analysis was used to calculate the wealth index, and further wealth quintile was categorized using a ranked technique to translate it to the ordinal scale. For detailed methodology used, please refer to the NNMSS 2016 report 32  When zinc is introduced into nitrogen produced by microwave energy at high temperature, the zinc atoms get excited and, when return to a lower energy state, generates emission spectra at 213.85 nm. The intensity of the emitted spectra is proportional to the concentration of zinc and measured using a calibration curve. Initially, the protein content of the serum is precipitated by adding concentrated nitric acid. Then serum is centrifuged, and diluted supernatant with free zinc is injected into the MP-AES by the auto-sampler. In each run, a calibration curve of zinc standards from 0 to 300 ppb (µg/L) is read along with samples and QC. A sample is run in duplicate every four samples, and a blank, a 100 ppb standard, and serum sample(internal quality control) are placed every 15 samples. A BIORAD serum chemistry quality control, Liquid Assayed Multiqual in levels 1,2 and 3 were used for the quality control of zinc. The blank and 100 ppb standards are prepared from the zinc standard solutions (0 and 1 ppm). All the standards and controls are treated as if they were samples.
Each control sera was diluted according to the manufacturer's protocol, and duplicate samples were processed in each run. Within each run, samples of the pools of the sera obtained in the laboratory were included to determine their concentration and verify the repeatability of the assay. The CV of the sera pool measurement was 12.37%. The INCAP laboratory routinely tests quality control (QC) sera developed by the National Institute of Standards and Technology (NIST). All the results of the NIST control samples were within the acceptable range in each run. The inter-assay coefficient of variation (CV) was < 10% for zinc.
Serum zinc values obtained from the laboratory were corrected for inflammation in children but not in women. Zinc deficiency was defined using a cut-off as described by International Zinc Nutrition Consultative Group 38 as follows: For children 6-59 months: morning, non-fasting: < 65 μg/dL or afternoon, non-fasting: < 57 μg/dL. For non-pregnant women: morning, non-fasting: < 66 or afternoon, non-fasting < 59 μg/dL depending on the time of day: morning (until noon), non-fasting: 66 μg/dL; afternoon, non-fasting: 59 μg/dL.
Blood samples for measuring hemoglobin and malaria were tested in the field immediately after collection. Hemoglobin was measured by a photometric method using HEMOCUE Hb 301 analyser 76 . Following criteria was used for defining anemia 77 , children 6-59 months: < 11.0 g/dL; Non-pregnant women 15-49 years: < 12.0 g/dL. A number of factors can cause anemia, such as malaria infection, Helicobacter pylori infection, or soil-transmitted helminths (STH). A malaria test in the blood was done using a rapid diagnostic test kit (malaria antigen combo RTK). The STH was also assessed in the field by microscopic examination of the stool samples for the presence of parasite eggs (Trichuris trichiura, Ascaris lumbricoides, and Hookworms) following the Kato Katz technique 78 . www.nature.com/scientificreports/ Helicobacter pylori infection causes gastric acidity, leading to reduced vitamin B12 absorption, which can also cause anemia. Therefore, the collected stool samples were used for Helicobacter pylori antigen detection using an enzyme-linked immunoassay (ELISA) test kit on a MAGO clinical analyzer at Siddhi Policlinic Laboratory Kathmandu, Nepal. For the validity of each analytical test, positive and negative controls were used where the absorbance was at least 0.8 OD units and less than 0.09 OD units, respectively. The laboratory runs its own bench QCs (low, medium, and high) with each run. The inter-assay variation (CV) was < 10% for each analytical run. The INCAP also analyzed the serum samples for retinol and modified-relative dose-response ratio (MRDR) measurement using High-Performance Liquid Chromatography (HPLC). The INCAP lab tests bench control materials distributed in each specimen plate. Each run processes three levels (low, medium, and high) of bench QCs. Each run was accepted based on the following criteria: > 50% internal standard recovery; sufficient peak separation between retinol and MRDR peaks; MRDR ratio between 0.01-0.07; MRDR ratio below 0.05 when retinol ratio is below 30 μg/dL; and MRDR ratio above 0.03 when retinol ratio is above 30 μg/dL. The inter-assay CV was < 10% for MRDR and retinol, which is acceptable for survey specimens.
The samples for the measurement of ferritin, C-reactive protein (CRP), Retinol Binding Protein (RBP), and Alpha-1 acid glycoprotein (AGP) were performed at The VitMin Lab, Willstaett, Germany, using sandwiched ELISA technique 79 . The CRP and AGP are measured to assess inflammation, and micronutrient parameters are adjusted according to the inflammation status. The antibodies used for this measurement are as follows: Capture antibodies: ferritin (Code A0133, Dako), RBP (Code A0040, Dako), CRP (Code A0073, Dako Denmark). Detection antibodies: antiferritin-horseradish peroxidase (HRP) (Code P0145, Dako), anti-RBP-HRP (Code P0304, Dako), anti-CRP-HRP (Code P0227, Dako). A Bio-Rad serum control sample (Liquicheck, Bio-Rad) was used as a standard for the calibration curves. The lab routinely tests a single QC pool in 10 different wells randomly distributed in each 384-well plate. The inter-assay CV for ferritin, CRP, RBP, and AGP were 3.5%, 5.4%. 4.2%, 4.6% and 4.9% respectively. A CV of about 10% is acceptable in the ELISA technique.
Serum retinol measured by HPLC is the WHO recommended indicator for vitamin A status. The full sample collected from the survey measured retinol-binding protein (RBP) using a sandwich ELISA at the VitMin Laboratory in Germany 80 . RBP values should be calibrated to serum retinol to determine the cut-offs. Serum retinol was measured from a subsample of 200 children (6-59 months) and 100 non-NPW. A maximum of 2.5% of the retinol values was removed after applying the outlier test (CLSI EP9-A2 and CLSI EP9-A2). The R 2 in the subsample of children and NPW were 0.74 and 0.78, respectively. Then a linear regression was used to determine RBP cut-off equivalent to retinol < 0.70 µmol/L 81 .
Samples for measuring RBC folate were sent to The Peking University, Institute of Reproductive and Child Health Laboratory (Beijing, China). RBC folate concentration in RBC hemolysate was measured by a gold standard microbiological method following the protocol of O'Broin and Kelleher 82 . The laboratory also tested QC materials (low, medium, and high) distributed in each 96-well plate. Each run also contained one blind QC replicated in 22 wells throughout the plate. The inter-assay variation (CV) was 4.1% for folate. The inter-assay CV was < 10% in the microbiological assay.
External quality control of the laboratory analysis All laboratories that analyzed this study samples participated in CDC external quality assurance (EQA) programs. The EQA program also includes Vitamin A Laboratory and External Quality Assurance (VITAL-EQA) that includes QA for ferritin, vitamin A (retinol and RBP), CRP, and RBC folate. The precision for the measurement of ferritin, RBP, and CRP concentration was > 90-95% with < 0.5% bias for ferritin and CRP and < 4.0% bias for RBP. The precision and bias for folate measurement were > 90% and < 4.0%, respectively. Statistical analysis. All analyses were performed using Stata 15 (StataCorp LLC, Texas, USA). The reported values were weighted by sample weights in order to obtain national estimates. Logistic regression was used to assess unadjusted and adjusted odds ratios (AOR) and considered the sampling design wherein ecological zones were used as strata and 'wards' as a cluster. Only significant associations observed in bivariate logistic regression were included in the multivariate model. Multicollinearity of the covariates was checked using variance inflation factors (VIFs). The VIFs for all covariates that were included in the logistic regression analysis were less than 2.0. P < 0.05 was considered to be statistically significant.
Ethics statement. Ethical approval to conduct the survey was taken from the Ethical Review Board (ERB) of Nepal Health Research Council (NHRC) (Reg. No. 201/2015). The research was conducted in accordance with the National Ethical Guidelines for Health Research in Nepal 2019 and the Declaration of Helsinki. All adult participants gave informed consent before they were included in the study. Informed consent for the children was taken from their parents/or legal guardians.