Timing of first antenatal care contact, its associated factors and state-level analysis in Nigeria: a cross-sectional assessment of compliance with the WHO guidelines

OBJECTIVES: To assess the compliance of WHO guidelines on the timeliness of antenatal care (ANC) initiation in Nigeria and its associated factors and to provide subcountry analysis of disparities in the timing of the first ANC in Nigeria. DESIGN: Cross-sectional. SETTING: Nationally representative data of most recent pregnancies between 2013 and 2018 in Nigeria. PARTICIPANTS: Women with pregnancies within 5 years before the study. PRIMARY AND SECONDARY OUTCOME MEASURES: The outcome variable was the trimesters of the first ANC contact. Data were analysed using descriptive statistics, bivariable and multivariable multinomial logistic regression at 5% significance level. RESULTS: Of all the 21 785 respondents, 75% had at least one ANC contact during their most recent pregnancies within the five years preceding the data collection. Among which 24% and 63% started in the first and second trimester, respectively. The proportion who started ANC in the first trimester was highest in Benue (44.5%), Lagos (41.4%) and Nasarawa (39.3%) and lowest in Zamfara (7.6%), Kano (7.4%) and Sokoto (4.8%). Respondents aged 40-49 years were 65% (adjusted relative risk ratio (aRRR: 1.65, 95 % CI: 1.10 to 2.45) more likely to initiate ANC during the first trimester of pregnancy relative to those aged 15-19 years. Although insignificant, women who participate in their healthcare utilisation were 4% (aRRR: 1.04, 95 % CI: 0.90 to 1.20) times more likely to have early initiation of ANC. Other significant factors were respondents’ and spousal educational attainment, household wealth quintiles, region of residence, ethnicity, religion and birth order. CONCLUSIONS: Only a quarter of pregnant women, initiated ANC contact during the first trimester with wider disparities across the states in Nigeria and across the background characteristics of the pregnant women. There are needs to enhance women’s autonomy in healthcare utilisation. Concerted efforts on awareness creation and empowerment for women by all stakeholders in maternal and child healthcare are antidotes for early ANC contact initiation. The study setting is Nigeria. Nigeria is divided into 36 states and the Federal Capital Territory for administration purposes as shown in figure 1. The states are further grouped into six regions. The states are are made up of local government areas (LGAs), and each LGA is divided into local administrative units. The LGAs are subdivided into convenient areas, for election purposes, called census enumeration areas (EAs). Map of Nigeria showing the 36 states, the federal capital territory, by the geopolitical zones. We analysed the data collected among women of reproductive age in Nigeria. We used secondary data from the 2018 Nigeria Demographic Health Survey (NDHS). The NDHS is one in the series of surveys conducted by Inner City Fund (ICF) Macro International, Calverton, Maryland, USA, in conjunction with the Nigeria National Population Commission (NPC).32 The DHS data are cross-sectional in design and nationally representative household surveys. DHSs are conducted every 5 years in low-income and middle-income countries. Two-stage sampling procedures were used for the 2018 NDHS survey. The sampling frame is the Population and Housing Census of the Federal Republic of Nigeria which was conducted in 2006 by the Nigeria NPC. The primary sampling unit referred to as a cluster are the EAs in the census frame. In each state, samples of EAs were selected independently in a two-stage selection. At the first stage, 38 EAs were selected with probability proportional to EA size in each state. At the second stage, 30 households were selected in every selected EAs using equal probability of systematic sampling. All eligible women of reproductive age (15–49 years) in all the selected households were interviewed. Sampling weights were applied in the analyses to account for the differences in response rates and population sizes of the states. A total of 41 821 women aged 15–49 years were interviewed in the 2018 NDHS.32 All eligible respondents were asked if they had any pregnancy or birth within 5 years preceding the survey. Those who answered in affirmative to haven had at least a birth within the preceding 5 years were asked questions on the number of ANC contacts made, the onset of the ANC visits, and the ANC provider etc for the pregnancy starting from the most recent. Our analysis is based on the information on the most recent pregnancies of each of the respondents. A total of 21 785 women provided relevant information. Of these 21 785 women, 16 448 (75.5%) attended ANC and were thus included in the final analysis. The outcome variable is the timing of the first ANC visit among women. The time of the first ANC was reported in months by the mother and was grouped as first trimester (early initiation), second trimester (late initiation) and third trimester (very late initiation). We grouped the States in Nigeria into two: below 85% or ‘greater than or equal to’ 85% global ‘no antenatal contact’ prevalence.1 33 The states’ performances regarding the proportion of women who initiated ANC visits in the first trimester during the most recent pregnancy is presented, with the states grouped into having 0%–33%, 34%–67% and >67% early ANC visits. Based on existing literature,23 34–36 the independent variables used in this study are maternal age (15–19, 20–24, 25–29, 30–39, 40–49 years), educational attainment (no education, primary, secondary and higher), spouse educational attainment (no education, primary, secondary and higher), employment status (currently employed vs unemployed), spouse employment status (currently employed vs unemployed), access to media (at least one of radio, television, newspaper or not), household wealth tertile (low, middle and high), women’s autonomy using who decides respondents healthcare utilisation (respondent alone, respondent/spouse and spouse alone). Other included independent variables are birth interval (firstborn, <36 months and ≥36 months), birth order (1, 2, 3, 4 and 5+), number of children ever born (none, 1–2, 3–4, 4+), current marital status (currently married or living together, divorced/separated/widowed, never married), place of residence (rural vs urban), religion (Islam, Christian, others) and ethnicity (Hausa/Fulani, Igbo, Yoruba and others). Family mobility (had stayed less than 5 years at residence or not), wanted child when became pregnant (wanted then, wanted later, or wanted not more), household headship (male vs female), health insurance coverage (yes vs no), acceptance of wife-beating (yes vs no). We also assessed four community-level factors in the descriptive analysis. The communities are synonymus to the EAs. The four factors are the community poverty rate (high or low), community unemployment rate (high vs low), community illiteracy rate (high vs low) and community media access rate (high vs low). We computed the neighbourhood socioeconomic status disadvantage as a composite score using principal component analysis and grouped into lowest, middle and highest categories. It is the proportion of respondents within each community with no media access, who are illiterates, who are poor and who are unemployed. The ‘xtile’ function in Stata V.16 was used to categorise the already provided wealth index scores (V.191) in the DHS data into three tertiles. Descriptive statistics, bivariable and multivariable multinomial logistic regression were used. The ‘SVY’ command for survey data in Stata V.16 was used to adjust for the study design used and the sample weights. Frequency tables showing percentages were used to describe the distribution of study respondents’ characteristics and we cross-classified the outcome variables by the respondents’ characteristics (table 1). Graphs and maps were produced using Microsoft Office 365 Excel and PowerPoint editable maps, respectively. Association between respondents’ characteristics and timing of first ANC visit *Significant χ2 at 0.05. ANC, antenatal care; SES, socioeconomic status. We used the ‘multinomial logistic’ command in Stata to implement bivariable and multivariable regression models. It is a procedure for estimating the risk ratio (RR) of factors associated with the outcome variables. Variables that were significant at p<0.20 were included in the multivariable model.31 The multinomial logistic regression model computes the maximum-likelihood estimates of the probability of success of an event. In the binary logistic regression, the ORs are computed as the ratio of odds of success divided by the odds of failure, the ith coefficient is ϕi=exp(bi) with SE siϕ=ϕisi, where si is the SE of bi estimated using the logit function. Assuming that the predicted index of the jth observation is defined as Xib. The predicted probability of a positive outcome is Pj(yj≠0|Xj)=exp(Xib)1+exp(Xib)(1). Whereas, multinomial logistic regression is used for the categorical dependent variable, with three or more categories. If y has three outcomes 1, 2 and 3 whereby ‘3’ is not necessarily greater than ‘1’ or ‘2’. Then, multinomial logistic regression is useful in modelling the nominal outcome variables.37 The relative probability of any of the levels, say y=2 to the base outcome, say y=1 is p(y=2)p(y=1)=eXβ(2) which is the relative risk ratio (RRR). Assuming that X and βk(2) are vectors equal to (x1+x2+…xi…..+xk) and (β1(2)+β2(2)+…+βi(2)+…βk(2)), the ratio of the relative risk (RR) for a one-unit change in xt is eβ1(2)x1+….+eβi(2)x(i+1)+……+eβk(2)xkeβ1(2)x1+….+eβi(2)xi+……+eβk(2)xk=eXβ(2)(2) Thus, the exponential of the coefficient is the RRR for a one-unit change in the corresponding variable. This is easily interpreted as the ratio of the probability of choosing one outcome category divided by the probability of choosing the baseline category.37 38 Patients and the public will be involved in the dissemination plan.