Prevalence of and factors associated with antenatal care seeking and adherence to recommended iron-folic acid supplementation among pregnant women in Zinder, Niger

The World Health Organization recommends iron and folic acid (IFA) supplementation for pregnant women. The high prevalence of anaemia among pregnant women in Niger warrants better understanding of the utilization of antenatal care (ANC) and IFA. We aimed to assess the prevalence of and factors associated with ANC coverage and adherence to IFA recommendation among pregnant women. Pregnant women (n = 923) from 64 randomly selected villages within the catchment area of 12 health centres were interviewed during a baseline household survey in Zinder, Niger. ANC and IFA coverage were 60.1% and 43.6%, respectively. Only 71.7% of women who attended ANC received IFA. Of the 401 women who reportedly received any IFA supplements, 99.3% had attended any ANC during their current pregnancy and 68.6% reported adherence to recommended IFA supplementation (i.e., consumed IFA every day in the previous week). Women with gestational age ≥27 weeks were more likely to have attended ANC than women with gestational age <27 weeks (85.9% vs. 27.5%, odds ratio [OR]: 21.81, 95% confidence interval [CI]: 13.81, 34.45). Women who reportedly received husbands' advice about attending ANC were more likely to attend ANC (OR: 1.48, 95% CI [1.03, 2.11]) and adhere to IFA recommendations (OR: 1.80, 95% CI [1.04, 3.13]) compared to those who did not receive any advice. ANC attendance is crucial to ensure distribution of IFA supplementation among pregnant women in Zinder. Interventions to improve ANC and IFA adherence will require promotion of early ANC, ensure availability of IFA at ANC, and involve husbands in ANC. This study was a community‐based cross‐sectional survey of pregnant women carried out in two health districts in the Zinder region of Niger using a continuous enrollment schedule over a period of 12 months to account for seasonality (Mar 2014–Feb 2015). The survey was part of the baseline evaluation of the Niger Maternal Nutrition (NiMaNu) Project, which was registered with the U.S. National Institutes of Health (http://www.ClinicalTrials.gov, http://clinicaltrials.gov/ct2/show/{"type":"clinical-trial","attrs":{"text":"NCT01832688","term_id":"NCT01832688"}}NCT01832688?term={"type":"clinical-trial","attrs":{"text":"NCT01832688","term_id":"NCT01832688"}}NCT01832688&rank=1). Households with potentially eligible pregnant women within the catchment areas of 12 integrated health centres (CSIs) were identified for the baseline survey employing a multistage clustered sampling design. Briefly, 12 CSIs (seven from Mirriah Health District and five from Zinder Health District) were initially selected by convenience sampling and were randomized to the order of participation. Within the catchment area of each CSI, five to eight villages were selected per CSI resulting in a total of 64 villages. The village in which the sampled CSI was located (CSI containing village) was automatically included in the study. One village with a health post (CS) was randomly selected per CSI (CS containing village) among all CS containing villages in the catchment area of that particular CSI. The remaining 3–6 villages per CSI were randomly selected from among the remaining villages (not containing any CSI or CS) after stratification by population size and distance to the health facility. To explore the impact of seasonality, we targeted enrollment of ~77 pregnant women per CSI to enroll of 923 women over 12 months. Pregnant women were identified and interviewed using the random walk method (United Nations, 2008), with a starting point randomly selected for each village (market, primary school, or mosque). Particpant enrollment continued until the desired sample size was reached (16–20 women per village) or until there were no more eligible pregnant women in the village. Women of any age were eligible to participate in the survey if they reported being currently pregnant, had resided in the village for the six previous months, and planned to stay for at least two additional months. Those who presented with a severe illness warranting immediate hospital referral or were unable to provide consent due to mental disabilities were excluded from the study. All eligible pregnant women were interviewed through household visits using pretested structured questionnaires. Interviews were conducted by eight trained local female interviewers in Hausa, the predominant local language. Data were collected on sociodemographic characteristics of the households, anthropometric measurements, household food insecurity, dietary diversity, and knowledge, attitude, and practices of the women about ANC and IFA. To obtain village level information on village characteristics, infrastructures and services (Table S1), health centre personnel, and/or heads of villages were interviewed. In cases where respondents were not able to provide all village level information, we asked the interviewee to refer us to additional village representatives with relevant knowledge. The study protocol and consenting procedures were approved by the National Consultative Ethical Committee of Niger and the Institutional Review Board of the University of California, Davis, USA. In case the pregnant woman was illiterate, an impartial witness was present during the consent process, who confirmed that the information in the consent document was accurately explained, and that consent was freely given. Informed consent was documented with a signature or a fingerprint. We constructed two primary dichotomous outcome variables. The first outcome was utilization of ANC, and the second outcome was adherence to recommended IFA supplementation. ANC use was defined as whether the woman participated in any ANC visit to any health facilities including CSI or CS during her current pregnancy regardless of her estimated gestational age. Adherence to recommended IFA supplementation was defined as whether the woman consumed IFA supplement daily in the previous 7 days if she had reported receiving any IFA supplement during her current pregnancy. The independent variables examined are listed in the Table S1, and some additional explanations are provided below. A weighted average of gestational age was created in which reported measures for last menstrual period (LMP; LMP by months, LMP by lunar cycle, and LMP by proximity to a religious or cultural event), time elapsed since first quickening was felt and two fundal height measures together received each one third of the weight, as described in more detail elsewhere (Hess & Ouédraogo, 2016). We defined first trimester: <13 weeks, second trimester: ≥13 to <27 weeks, and third trimester: ≥27 weeks. Low mid‐upper‐arm circumference was defined as <23 cm (Ververs, Antierens, Sackl, Staderini, & Captier, 2013). Women's dietary diversity was constructed from a food frequency questionnaire based on a woman's food intake in the past 24 hours. After an initial open recall, during which women were asked to list all foods consumed in the past 24 hours, a list‐based method was used to ask the woman if she had consumed any foods during the past 24 hours from each of the food groups (FAO & FHI 360, 2016). Consuming ≥5 food groups (out of 10) in the past 24 hours was considered as meeting the minimum dietary diversity for women (MDD‐W) (FAO & FHI 360, 2016). Socio‐economic status of the household was estimated using three proxy indicators. Housing quality index was constructed based on whether the household had an improved roof, improved walls, or improved floors. It was a continuous scale from 0 to 1 where 1 was the highest possible score corresponding to the best housing quality. A household asset index was based on ownership of a set of assets including a radio, a mobile phone, a battery, a bicycle, a moped/motorcycle, and a car/truck. The household livestock index was a standardized composite measure of household small animals including goats, sheep, and poultry including a conversion factor for feed requirements (FAO, 2003). Household food insecurity was assessed using the Household Food Insecurity Access Scale of the Food and Nutrition Technical Assistance/USAID (Coates, Swindale, & Bilinsky, 2007). A household was categorized as food secure, or mildly, moderately or severely food insecure. Data analyses were performed using Stata version 12.1 (Stata Corporation, College Station, TX, USA). A statistical analysis plan was created prior to analysis (Hess & Ouédraogo, 2016). For our two outcomes (utilization of ANC and adherence to recommended IFA supplementation), we considered two sample sizes. All 923 pregnant women who were interviewed during the household visits were included in the ANC coverage analyses. Women who received IFA (n = 401) provided information on adherence to recommended IFA supplementation (Figure 1). Flow chart of the pregnant women included in the ANC and IFA analyses. ANC = antenatal care; IFA = iron‐folic acid supplement Descriptive analyses were used to assess the characteristics of women and villages. If a characteristic contained a homogenous response and less than 3% of respondents were in any category, we attempted to combine categories in logical ways to have each category contain at least 3% of respondents. The prevalence of outcome by categories of independent variables was measured taking into account the cluster design nature of the survey. To test for multicollinearity among independent variables, we ran collinearity diagnostics to calculate variance inflation factor and tolerance for each of the variables. We used variance inflation factor ≥2 (or equivalently, tolerances ≤0.50) as cut‐off values to minimize the possibility of multicollinearity within the variables. Bivariate logistic regression analyses were carried out to evaluate the unadjusted associations between dependent variables and each of the independent variables. All these bivariate models were adjusted to assess whether any of the significant associations in the bivariate models are attributable to variation in seasonality (i.e, time of year interview was conducted), estimated pregnancy trimester, or household socio‐economic status. For the purpose of analyses, we classified the seasons as (a) dry hot (Mar–May), (b) wet or lean (Jun–Sep), and (c) dry cool, harvest season (Oct–Feb; USAID & FEWS Net, 2011). Multivariable logistic regression models were constructed to identify the factors associated with ANC attendance and adherence to recommended IFA supplementation when controlling for the effect of other covariates. The decision of which covariates to include was based on public health significance and plausibility instead of any statistical criteria. In order to identify the women at risk of inadequate utilization of ANC, pregnancy‐related covariates that are meaningful and targetable in public health practices were included in the multivariable model. All models were adjusted for women's education as a variable of particular interest and relevance, because education has been found to be less associated with ANC use in sub‐Saharan Africa than in other regions of Africa and Asia (WHO & UNICEF, 2003). All models were constructed using mixed effect logistic regression procedure where CSI was considered as cluster nested within the health district as a fixed effect and village nested within CSI as a random effect. Results were considered significant at p < .05 and the final models present the results as odds ratios (ORs) with 95% confidence intervals (95% CI).