Coverage-level and predictors of maternity continuum of care in Nigeria: implications for maternal, newborn and child health programming

The study is a secondary analysis of 2018 Nigerian Demographic and Health Survey (NDHS) data. NDHS is a cross-sectional population-based and nationally representative survey routinely collected in five years’ intervals in Nigeria. Nigeria is administratively grouped into six geopolitical zones (Northcentral, Northeast, Northwest, Southeast, Southsouth and Southwest) with an average of 6 states per geo-political zone and the federal capital territory (FCT) as the administrative headquarter [14]. Each state is further divided into local government areas that serve as the lowest and the closest administrative cadre of government for the people. The 36 states and FCT are shown in the study area map in Fig. 1.

The sampling frame of the 2018 nationally representative NDHS was obtained from the list of rural and urban enumeration areas collated by the National Population and Housing Census (NPHC) in Nigeria. A two-stage stratified random sampling design was used in the 2018 NDHS, where 1400 enumeration areas consisting of 820 rural and 580 urban strata were selected using probability proportional to size at the first sampling stage. Hence the difference in the number of urban and rural strata. Equal probability systematic sampling was then used to select the same number of households (30 households per enumeration area) in the second sampling stage. A total of 41,821 (22,658 in rural and 19,163 in urban) women participants were interviewed in the cross-sectional survey that achieve a 99% response rate [14]. 21,447 women who had at least one ANC visit and whose information were at least non-missing in one of the maternity CoC pathway made up the weighted sample size of the study. The survey also collected information on women’s demographics, socioeconomic and health-related characteristics that includes the key measures of the maternity continuum of care (ANC, SBA and PNC) investigated in this study.

Outcomes of interest in this study are the maternity continuum of care received during pregnancy (ANC), childbirth (use of SBA) and post-delivery (PNC). A postpartum woman is regarded to have completed the three gamut of care if she received the recommended 4 or more ANC contacts in a healthcare facility during pregnancy, move on to utilize SBA i.e., delivery assisted by at least a doctor, nurse or midwive and subsequently received postnatal checkup within the first 48 h after childbirth [14]. The combined outcome was based on the WHO recommendation of at least 4 ANC visits and the use of SBA at birth, especially in low-resource settings of the lower-middle-income countries [11, 12]. We measured PNC within the first two days after birth which has been reported in the 2018 NDHS due to most maternal morbidity and mortality that occur at the time and therefore highlighted PNC (within two days) as an important measure in the maternity CoC model [33]. We avoided the adaptation of the recently recommended 8 ANC contacts since the DHS framework was designed on a minimum of 4 ANC visits as the optimal number of ANC visit and also; because the strategy to implement the 8 ANC visits was recently devised in the orientation package for healthcare providers in Nigeria after most of the respondents have had the indexed childbirth [14, 35, 36]. The outcome variable was obtained from the combination of responses to the following questions:

Three sets of dichotomous variables were extracted, such that; a positive response to question ‘1’ is 4 or more ANC and negative response is ANC visit less than 4 (0, 1, 2, 3), response to question ‘2’ that delivery was assisted by doctor/nurse/midwife is a positive response and otherwise a negative response and similarly positive response to question ‘3’ is ‘Yes’ and ‘No’ is the negative response. The sequence of maternity continuum of care was drawn from the combination of positive responses. Hence, positive response to; question 1 indicate ANC (4 +) visits, question 2 indicate ANC (4 +) visits and SBA use and question 3 indicate maternity CoC completion in this study i.e., when ANC (4 +), SBA and PNC were all received.

Independent variables included in this study were based on similar factors considered by previous studies that investigated the maternity continuum of care [3, 5, 30‐33, 37]. This can be defined under the broad categories as; socio-demographic characteristics, maternal health and birth factors, quality of pregnancy care received, economic status and physical and autonomy factors [13, 38, 39].

Descriptive statistics of the background characteristics and outcomes were reported in frequency and percentages. Missing data were reported for at least 1% of the observation and otherwise negligible i.e., less than 1%. Three sequences of maternity CoC model defined under the space that; postpartum women received at least 4 ANC visits during pregnancy was coded as 1 and 0 otherwise – model 1, continued from ANC (4 +) to use SBA at childbirth was coded as 1 and 0 otherwise – model 2 and completed the three key CoC which is from ANC (4 +) to SBA and to PNC after childbirth was equally coded as 1 and 0 otherwise – model 3 were fitted.

Bivariate and Multivariate analysis that includes all the significant variables retained in the stepwise regression (final models 1, 2 and 3) were performed to determine the likelihood and significance of each of the predictor variables and the combined set of the predictors respectively. The respective unadjusted and adjusted odds ratio were reported for the binary logistic regression analysis of models 1 and 2 while the unadjusted and adjusted e(form) or exp(b) equivalent of the odds ratio was reported in the multivariable complementary log–log analysis of model 3. Data were weighted with the women’s sample weight indices included in the NDHS data and the svyset command was used to adjust for unequal group/population sizes due to the complex survey design. Bivariable and multivariable statistical analysis were performed at 10% and 5% level of significance (95% confidence level) respectively, using Stata version 16.0 (Stata Corp, Texas, USA). Variable (Union type) that causes multicollinearity (variance inflation factor > 5) was subsequently removed from the multivariate analysis.

The multiple binary logistic regression and the complementary log–log modeled the odds of optimal ANC uptake and continuation to the use of SBA and PNC as a binary response [P(\({Y}_{i}=0\)), P(\({Y}_{i}=1)]\) [40, 41]. The multiple logistic model which equates the function of the odds to a linear combination of the regression terms and the predictors is generally expressed as:

where: \(\mathrm{ln}\left(\frac{P}{1-P}\right)\) is the log odds (P is the probability of success and 1-P is the failure probability).

\({\beta }_{0}\) is the logistic regression constant.

\({\beta }_{1}+\dots + {\beta }_{p}\) are the px1 vector of regression coefficient or estimates of the multiple predictors.

\({X}_{i1}+\dots +{X}_{ip}\) are the nxp matrix of explanatory variables predicting the log odds in the model.

When the probability of success “P” is very large or very small (less than 10%) leading to asymmetrical S-shape compared to the symmetric logistic curve [42], the use of the complementary log–log model becomes more appropriate (accurate) as it’s in rare CoC outcome. The complementary-log–log model is generally stated as:

where log{-log[1-π(x)]} is the complementary log–log transformation with binary response (0, 1).

Figure 2 shows the pathway of the continuum of care model (from pregnancy to delivery and to postpartum), based on the key maternity health service received at each stage. Antenatal care assessed maternal health service received in pregnancy, skilled birth attendant utilization at delivery and postnatal care at postpartum. Among the 21,447 pregnancies reported in NDHS 2018, 75.1% (16,114) received antenatal care at least once. Antenatal care was optimal in this study when a postpartum woman received at least 4 contacts (n = 12,362, P = 57.6%) which made up pregnant women’s inclusion in the first stage of the CoC model (Model-1).

About 37.4% (8023) of Pregnant women continued to use skilled birth attendants at childbirth. Implying that 20.2% (4339) who had delivery performed by unskilled births attendants after receiving optimal antenatal care drop out at the second stage of the CoC model. Continuation to use of postnatal care service at the third stage after receiving optimal ANC and SBA indicates completion of the CoC by only 6.5% (1388) while 30.9% (6635) dropped out from postnatal care service after receiving optimal ANC and SBA. Dropout from any of the CoC models at any stage along the pathway will lead to incomplete receipt of the essential maternal health service across pregnancy to the puerperium continuum. The CoC model pathway is shown in Fig. 2.

Table 1 shows the percentage distribution of women who had at least one birth in the last 5 years preceding the survey by background characteristics. About 25% of the women respondents were in the early maternal age (15–24 years), while 27.5% were in the late maternal age (35–49 years). Nine of ten (91.4%) were married, 2.8% were cohabiting while 3.5% are either divorced or widowed. 45.2% and 34.1% of women and their partner has no education while only 8.5% and 14.2% of women and their partners have completed higher education respectively. 61.0% of respondents resides in the urban area and, are not exposed to mass media and only 28.1% belonged to the rich wealth quintile. Most (35.6%) of the women respondents were from the northwest geopolitical zone compared to the few (9.0%) from the south-south (Table 1).

About 88% of the women wanted the pregnancy and about 50% have had at least four births. Only 2.2% were covered by health insurance while about 48.5 and 28.2% reported big problems in getting money for medical help and in reaching medical facilities respectively (Table 1). Most (56.2%) of women’s healthcare decisions were made by the partner while only 8.8% of the women made their healthcare decision alone. Nurse/Midwives were the providers of ANC for 56.8% of the women while 26.4% either had no-one or utilized TBA (Table 1). Only nearly 18% had first ANC in 1^st trimester. 70.6, 65.8 and 64.8% of women had their blood pressure measured, blood and urine sample taken at ANC respectively. 69 and 53.1% of the women took iron folic acid and at least 2 dose of tetanus vaccine during pregnancy respectively. 40.4% of the women delivered at a hospital and 3.1% of them were through Caesarian mode. 51.2% of the women recently delivered a male child while 48.8% delivered a female child. About 2.8% of the women delivered a child that is very small in size while 8.9% delivered a very large child (Table 1).

Figure 3 shows that only 6.5% of the women population completed the maternity continuum of care (received optimal ANC, continue to use SBA at childbirth and received PNC service after delivery) while 93.5% of the women had an incomplete maternity continuum of care (at least one of the optimal ANC, SBA and PNC service was not received). 37.4% received optimal ANC and SBA, 57.6% received Optimal ANC only and 75.1% received at least one ANC (Fig. 3).

The result of backward stepwise regression in Table 2 highlighted selected variables associated with women uptake of optimal ANC (4 +) (model 1), continuation to use of SBA (model 2) and the use of PNC (model 3). In model 1, marital status (-2logL = 2.049, p = 0.152) and Employment type (-2logL = 0.855, p = 0.355), and other variables whose inclusion led to insignificant change in -2logL were removed from the full model 1 while the remaining variables whose removal led to significant (p < 0.10) change in -2logL were retained (Table 2). Similarly, institutional delivery (-2logL = 2341.697, p = 0.000) along with other variables led to a significant (p < 0.10) change in -2logL and were therefore retained in model 2. Also, the removal of women and partner education, women healthcare decider, provider of ANC, time of first ANC and others that led to a significant change in -2logL (Table 2) were subsequently retained in the final model 3 (CoC completion).

It is not improbable that the study suffers from responder bias since the data quality depends on respondents’ ability to recall events in the last five years preceding the survey. The study investigated the association between women’s characteristics and maternity CoC and does not infer that these factors are causes of maternity CoC due to the cross-sectional design of the data. Therefore, interpretation should be limited to the association. The application of secondary data posed the difficulty of data incompleteness and restricted the authors to the choice of the available set of independent factors assessed in the survey, which was minimized by the analysis of a weighted sample of women with at least one birth in the last five years preceding the survey and the automated model selection approach. Non-availability of variables or data to assess the type of healthcare facility or women first place of care which experience could be the reason for dropout or incomplete CoC was also a limitation. However, the study strength can be observed from the application of a nationally representative sample which increases the study generalizability. The fact that we adjust for complex survey design based on the sample weighting, clustering and stratification improves the reliability of the study findings and accuracy therein. Furthermore, this is the first study that assessed the maternity CoC completion in Nigeria using the well-known and utilized WHO standards on a minimum of 4 ANC, SBA by doctor/nurse/midwife and first PNC within the first 48 h and therefore presents an opportunity to strategize towards transitioning into the newly recommended minimum of 8 ANC contacts and achieving the SDG-3 in Nigeria.

We infer from the study findings that, coverage of the three essentials maternity continuum of care (ANC, SBA and PNC) in Nigeria is below the 90% recommendation, which will halt the attainment of the 2030 SDG on improving childhood health in Nigeria. A centralized strategy that will improve MNCH program practice is however required to unify national programs and breach the coverage gap. Governmental and nongovernmental agencies need to be steadfast in providing improved support for sensitization programs around early ANC initiation and an optimal number of ANC visits (the minimum of 4 and transition into 8). Improved ANC packages that strengthen women in; pregnancy care, healthcare decision power and educational awareness for childbirth preparedness are recommended to improve pregnancy outcomes. Capacity building for pregnant women to improve the use of SBA at delivery and the least utilized PNC at post-delivery is also vital to optimize mother and child survival. Contextual research investigating the reason for dropout and non-compliance with the WHO recommendations of the maternity continuum of care is required to better provide intervention strategy to improve on the low completion coverage.