Factors associated with anaemia in a nationally representative sample of nonpregnant women of reproductive age in Nepal

Abstract We used cross‐sectional data from the 2016 Nepal National Micronutrient Status Survey to evaluate factors associated with anaemia among a nationally representative sample of nonpregnant women 15– 49 years (n = 1, 918). Haemoglobin, biomarkers of iron status and other micronutrients, infection, inflammation, and blood disorders were assessed from venous blood. Soil‐transmitted helminth and Helicobacter pylori infections were assessed from stool. Sociodemographic, household, and health characteristics and diet were ascertained by interview. We conducted bivariate analyses between candidate predictors and anaemia (haemoglobin <12.0 g/ dL, altitude‐ and smoking‐adjusted). Candidate predictors that were significant in bivariate models (P < 0.05) were included in the multivariable logistic regression model, accounting for complex sampling design. Anaemia prevalence was 20.2% (95% confidence interval [CI] [17.6, 22.8]). Associated with reduced anaemia odds were living in the Mountain and Hill ecological zones relative to the Terai (adjusted odds ratio [AOR] 0.35, 95% CI [0.21, 0.60] and AOR 0.41, 95% CI [0.29, 0.59], respectively), recent cough (AOR 0.56, 95% CI [0.38, 0.82]), hormonal contraceptive use (AOR 0.58; 95% CI [0.38, 0.88]), ln ferritin (micrograms per litre; AOR 0.43, 95% CI [0.35, 0.54]), and ln retinol binding protein (micrograms per litre; AOR 0.20, 95% CI [0.11, 0.37]). Residing in a house with an earth floor (AOR 1.74, 95% CI [1.18, 2.56]), glucose‐6‐ phosphate dehydrogenase deficiency (AOR 2.44, 95% CI [1.66, 3.60]), and haemoglobinopathies (AOR 6.15, 95% CI [3.09, 12.26]) were associated with increased anaemia odds. Interventions that improve micronutrient status, ensure access to hormonal birth control, and replace dirt floors to reduce infection risk might help reduce anaemia in this population.

control, and replace dirt floors to reduce infection risk might help reduce anaemia in this population.

K E Y W O R D S
anaemia, maternal nutrition, micronutrient status, nepal 1 | INTRODUCTION Worldwide, anaemia affects an estimated 29% of nonpregnant women of reproductive age (Stevens et al., 2013) and is thought to contribute to 115, 000 maternal deaths annually (Ezzati, Lopez, Rodgers, & Murray, 2004). In South, East, and South-East Asia, approximately a quarter of the anaemia burden is thought to be due to iron deficiency (Petry et al., 2016); however, additional factors contribute to anaemia through underproduction or excessive loss of red blood cells. Deficiency in micronutrients other than iron, infection, inflammation, blood disorders, and blood loss from worm infection, menses, or other causes directly contribute to anaemia (WHO, 2017a) whereas intermediate causes, such as dietary intake, are influenced by food security, access to health services, and sociodemographic characteristics.
Understanding the context-specific factors associated with anaemia is key to developing effective, evidence-based public health programmes and policies. Despite national-level initiatives to reduce iron deficiency and anaemia, anaemia prevalence in Nepal has The physiology of anaemia is relatively well understood globally; however, less is known about context-specific determinants of anaemia among women in Nepal. Previous studies of the risk factors for anaemia among nonpregnant women of reproductive age in Nepal have evaluated iron status, reproductive history, and sociodemographic characteristics (Chandyo et al., 2007;Chandyo et al., 2016;Gautam, Min, Kim, & Jeong, 2019) but have not evaluated many known potential causes, including deficiencies in micronutrients other than iron, infection, inflammation, and blood disorders.
To address this important knowledge gap, the Nepal National Micronutrient Status Survey (NNMSS) collected data on potential causes of anaemia to inform programmatic decision-making (Ministry of Health and Population [Nepal], 2018). The NNMSS is comprehensive, nationally representative survey that collected data on potential causes of anaemia-including multiple biomarkers which are rarely included in large-scale surveys due to logistical complications and cost.
The objective of these analyses was to identify factors associated with anaemia among nonpregnant women of reproductive age 15-49 years in Nepal. Blood disorders including αand β-thalassemia, sickle cell, haemoglobin E, and glucose-6-phosphate dehydrogenase deficiency (G6PD) were analysed using complete blood count, high performance liquid chromatography, DNA analysis, and PCR (Access Bio Korea Inc.

| Biomarkers of nutritional status
To correct for the role of inflammation on biomarkers of iron status, we regression-adjusted ferritin and sTfR to a pooled country reference using CRP and AGP (ferritin) or AGP only (sTfR; Namaste et al., 2017). Iron deficiency was defined as adjusted ferritin was defined as RBP <0.64 μmol/ L. The population-specific RBP cutpoint equivalent to serum retinol <0.70 μmol/ L was calculated by regressing RBP on retinol in a subsample of 100 WRA for whom serum retinol was assessed using high performance liquid chromatography from the same blood draw as RBP (WHO, 1996)

| Infection and inflammation
We categorized blood disorders into two groups: (a) haemoglobinopathies (thalassemias, hemoblogin E, and sickle cell) and (b)

| Dietary intake
We included consumption of flesh, organ, or blood-based foods, legumes, green leafy vegetables, vitamin A-rich fruits and vegetables, and tea (an iron inhibitor) the day preceding the survey as binary variables (yes/no). Minimum dietary diversity was defined as intake from five or more of 10 main FAO food groups the day preceding the survey (FAO and FHI 360, 2016). Pica was defined as any consumption of clay, earth, termite mounds, ice, uncooked rice, or starch during the 7 days preceding the survey.

| Reproductive and other health variables
We included lactation status, giving birth during the 5 years preceding the survey, hormonal contraceptive use, intake of any micronutrient supplements (multivitamin, iron, iron-folic acid, vitamin A, and/or zinc) the week preceding the survey, and receipt of deworming tablets during the 6 months preceding the survey as binary variables (yes/no). We created a household wealth score using principal components analysis of housing characteristics and assets. We then divided wealth into tertiles. Improved water source was defined as piped water, tube well borehole, protected well or spring, stone tap, rainwater, or bottled water (WHO and UNICEF, 2017). We defined severe household food insecurity as households who often cut back on meal size or number of meals and/or ever experienced any of the three most severe conditions (there no food to eat of any kind in the household because of lack of resources to get food; any household member goes to sleep at night hungry because there was not enough food; and any household member goes a whole day and night without eating anything because there was not enough food; Ballard et al., 2011).

| Sociodemographic variables
Improved water source, open defecation, earth floor, and severe household food insecurity were included as binary variables (yes/no).

| Statistical methods
We evaluated differences in sociodemographic and health characteristics by anaemia status using Rao-Scott chi square tests and linear contrast tests for categorical and continuous variables, respectively.
We used Rao-Scott chi square tests rather than Pearson to allow for complex sampling design correction.
We conducted bivariate analyses between candidate predictors and anaemia status. We tested variables with multiple categories as a group. Non-normally distributed variables were log transformed.
Where candidate predictors had P < 0.05 in bivariate models, we included them in the multivariable logistic regression model. To identity collinearity, we used eigenvalues <0.01 and conditionality index >30.
We conducted all analyses in SAS v.9.4 (SAS Institute Inc., Cary, North Carolina). All analyses were weighted and accounted for complex sample design. We set statistical significance a priori at two-sided P < .05.

| RESULTS
In total, 20.  (Chandyo et al., 2007). We were unable to estimate total iron intake; however, 69.7% of women reported consuming organ, flesh, or blood-based foods the day preceding the survey, suggesting that the majority of women consume iron-rich food sources. Reported intake of micronutrient supplements, however, including iron tablets/ syrups and multiple micronutrient supplements, was low (6.5%).
Despite <1% of WRA having vitamin A deficiency, vitamin A status was associated with anaemia in our study. Vitamin A is essential to mobilize of iron stores for erythropoiesis, and for immune function (WHO, 2017a). Although we were unable to estimate total vitamin A intake, reported consumption of food sources high in vitamin A during the day preceding the survey did not vary by anaemia status. Policies or programmes to support frequent physiological intakes of vitamin A or pro-retinol carotenoids through low-dose supplements, fortification, or improved diets could improve vitamin A status and potentially reduce anaemia (Mason et al., 2011).
k Haemoglobinopathies include ɑ-and β-thalassemia, haemoglobin E, and sickle cell. l Receiving deworming during the 6 months preceding the survey. m Biomarker was regression-adjusted to a pooled country reference to adjust for inflammation, using CRP and AGP (ferritin) or AGP only (sTfR; Namaste et al., 2017). n Iron deficiency defined as inflammation-adjusted serum ferritin <15.0 μg/ L (WHO, 2017a). o Vitamin A deficiency was defined as RBP <0.64 μmol/ L. The population-specific RBP cut-point equivalent to serum retinol <0.70 μmol/ L was calculated by regressing RBP on retinol in a subsample of 100 WRA for whom serum retinol was assessed using HPLC from the same blood draw as RBP (WHO, 1996). p Folate cutoff based on the risk of megaloblastic anaemia defined as RBC folate <305.0 nmol/ L (Institute of Medicine 1998). q Zinc deficiency defined as serum zinc <66.0 μg/ dL for nonfasted, morning (i. e. before 12 pm) samples and < 59.0 μg/ dL for nonfasted, afternoon (i. e. after 12 p.m.) samples (IZiNCG 2012). r Minimum dietary diversity defined as intake from ≥5 of the 10 main food groups (grains, legumes, nuts, dairy, flesh foods, eggs, green leafy vegetables, vitamin A-rich fruits and vegetables, other fruits, and other vegetables) the day preceding the survey based on Food and Agriculture Organization recommendations for minimum dietary diversity for women (FAO and FHI 360, 2016). s Micronutrient supplement intake includes: multivitamin, iron-folic acid tablets, iron tablets, and/or zinc tablets consumed the 7 days preceding the survey. among WRA in Nepal (Chandyo et al., 2007), and a study using data from the 2016 DHS reported that women 15-49 years who were currently using hormonal contraception had 37% lower odds of anaemia (Gautam et al., 2019).
Residing in a house with a dirt floor was associated with 1.74 times higher odds of anaemia relative to residing in a house with another floor type. An evaluation of the Mexican government program Piso Firme found that replacing dirt floors with concrete cement floors was associated with an 81% reduction in anaemia prevalence T A B L E 2 Multivariable binomial logistic regression predicting anaemia among nonpregnant women 15-49 Years, Nepal National Micronutrient Status Survey, Nepal, 2016 (n = 1, 918) Unadjusted odds ratio (95% CI) Adjusted odds ratio (95% CI) P Note. Estimates are unadjusted odds ratios and adjusted odds ratios with 95% confidence intervals from logistic regression models, accounting for weighting and complex sampling design. Anaemia was defined as altitude-and smoking-adjusted Hb <12.0 g/ dL (WHO 2017). Abbreviations: AGP, ɑ-1-acid glycoprotein; BMI, body mass index; CI, confidence interval; CRP, C-reactive protein; G6PD, glucose-6-phosphate dehydrogenase deficiency; RBP, retinol binding protein. a Recent cough defined as self-report of cough during the 2 weeks preceding the survey. b Biomarker was regression-adjusted to a pooled country reference to adjust for inflammation, using CRP and AGP (Namaste et al., 2017). c Other Terai ethnicities include Terai/ Madhesi ethnicities not including Terai/ Madhesi Brahmin/ Chettri. d Other blood disorders ɑand βthalassemia, haemoglobin E, sickle cell, and other blood disorders not including G6PD.
among children (Cattaneo, Galiani, Gertler, Martinez, & Titiunik, 2009). Dirt floors can expose household members to faecal matter, worms, protozoa, and other parasites (WHO, 2017a), increasing prevalence of infection (Benjamin-Chung et al., 2015). Infection and inflammation can cause both micronutrient malnutrition and anaemia (Balarajan et al., 2011). Although we directly measured STH infection and biomarkers of inflammation, these indicators were not associated with anaemia in bivariate models. However, dirt floors may represent an infection not otherwise captured. It is also possible that dirt floors represent overall living conditions; however, household socioeconomic status was not associated with anaemia in bivariate models.
Although infection is a known risk factor for anaemia, recent cough was inversely associated with anaemia. Recent cough might be a proxy for household air pollution. Smoke exposure is associated with increased haemoglobin because RBC production increases to compensate for chronically low blood oxygen concentrations (Nordenberg, Yip, & Binkin, 1990). Exposure to biomass smoke was associated with higher prevalence of respiratory symptoms in Nepal (Kurmi et al., 2014). In our study, women who reported recent cough had a higher prevalence of cooking with biomass fuel relative to women without recent cough (Table S2). Thus, recent cough may be a proxy for exposure to smoke from burning biomass for cooking fuel.
We identified nonmodifiable factors associated with anaemia among women. WRA residing in the Mountain or Hill ecological zones had lower odds of anaemia relative to women residing in the Terai ecological zone. Chronic exposure to arsenic via contaminated groundwater due to the geology of the Terai might explain persistently high burden of anaemia in this zone. Populations in the Terai are exposed to arsenic concentrations above the upper limit of drinking water per the World Health Organization (>10 μg/ L; Pokharel, Bhandari, and Viraraghavan, 2009;WHO, 2017b). Arsenic can depress haem metabolism (Hernandez-Zavala et al., 1996) and increase erythrocyte haemolysis (Mahmud, Foller, & Lang, 2008). A study among women in Bangladesh reported a positive association between arsenic exposure and anaemia (Heck et al., 2008). Because vitamin B 12 and folate are required to metabolize inorganic arsenic, deficiencies in these micronutrients could further contribute to anaemia (Gamble et al., 2005). Future research might explore arsenic exposure and anaemia in the Terai. Because ethnicity and blood disorders were included in multivariable models, regional differences in anaemia are unlikely due to these factors; however, other cultural, dietary, or other factors might also help explain anaemia in the Terai.
G6PD and haemoglobinopathies had the strongest associations with anaemia in this population. WRA with G6PD had more than double the odds of anaemia, and WRA with haemoglobinopathies had more than 6.1 times higher odds of anaemia relative to women without these conditions. While inherited disorders are nonmodifiable, our findings have may implications for frontline health workers and program planning. Because exposure to some foods and commonly prescribed antibiotics, antimalarials, and anthelmintics can induce acute haemolysis among people with G6PD (Beutler, 2008), frontline health workers could be trained about the prevalence among the population and contraindications for these drugs (WHO, 1989

| Strengths and Limitations
To our knowledge, this analysis is the first to examine causes of anaemia among WRA in Nepal using comprehensive, nationally representative data on multiple potential causes of anaemia-many of which are rarely included in large-scale surveys in low-income and middle-income countries. Due to the cross-sectional study design, we were unable to establish causality between candidate predictors and anaemia status. The NNMSS did not collect data on all micronutrients for which deficiency could lead to anaemia. Although plasma vitamin B 12 was measured, we excluded it from these analyses due to data quality. Dietary recall questions were limited in scope, which might explain the lack of findings for any diet-related indicators and anaemia.

| CONCLUSION
Our analysis suggests a combination of effectively implemented strat-

CONFLICTS OF INTEREST
The authors do not report any conflicts of interest.