Socioeconomic inequalities in maternal healthcare utilization in Zambia: 1997-2014

Background Exploring the trends and socioeconomic inequalities in the use of maternal healthcare utilisation between 1997 and 2014. Methods Data were analyzed using descriptive and multivariate regression methods. Results Women in the higher wealth quintiles (Q4 and Q5) generally had higher prevalence of using health facility delivery and postnatal care services compared with those in the lower wealth quintiles (Q1 and Q2), whereas the prevalence of timely and adequate ANC visit was comparatively higher among those in the lower wealth quintiles. Conclusion Findings indicated important sociodemographic inequalities in using maternal healthcare services, addressing which may help promote the utilisation of these services.


Introduction
Zambia is a land-locked country in Southern Africa with around two-thirds of the total population living below international poverty line [1]. Despite being one of the fastest growing economies in Africa, a large proportion of the Zambian population is still facing significant challenges in meeting the basic amenities such as proper housing, access to clean water, and sanitation and health care.
The demography is characterised by widespread poverty, food insecurity and malnutrition, and a fragile healthcare system which are contributing to high burden of maternal and child mortality in the country. Historically, Zambia has a high maternal mortality ratio [2,3]  The causes of low maternal healthcare services uptake are numerous and multifaceted, and broadly classified into behavioral, cultural, economic, and sociodemographic factors at individual level and remoteness of health facility, inadequate infrastructure and skilled human resource for healthcare at community level [9,10,11,12,13]. A large number of studies has been conducted on Demographic 3 and Health Survey data that adopted the Andersen and Newman Behavioural Model to identify the factors associated with poor use of ANC, facility delivery and postnatal services [14,15]. To date, there is no such study or report available for Zambia, and thus the factors behind the low utilisation of maternal healthcare services are not yet fully understood. Although the a few subnational studies have been conducted [12, 16,17], there remains a lack of comprehensive research data showing the full spectrum of maternal healthcare services use such as timing and adequacy of antenatal care, health facility delivery and postnatal care. Therefore, the present study was conducted to explore the factors associated with these four components of maternal healthcare using nationally-representative data from Zambia Demographic and Health Surveys (DHS) conducted since 1996-1997. DHS provides information on a wide range of demographic and socioeconomic factors which were selected systematically to fit within the scope of the Andersen and Newman Behavioural Model. Findings of the present study are expected to enhance the understanding of the enabling and predisposing factors associated with the uptake of maternal healthcare services in Zambia as well as in the neighbouring countries with similar economic and sociocultural environment.

Data source
Data for this study were collected from the last four rounds of Zambia Demographic and health survey (ZDHS 1997(ZDHS , 2002(ZDHS , 2007(ZDHS , 2014

Patient and public involvement
We did not directly involve Zambia women in study planning.

Outcome measures
The outcome variables of interest were: 1) timing of first antenatal care, 2) frequency of antenatal care, 3) facility delivery, and 4) postnatal care for the latest childbirth occurring within the last five years from the survey. All of these items are assessed based on the self-report of the participant.
Timing of first antenatal care was classified as timely (if within the first trimester) and late (if beyond the first trimester) [19]. Frequency of antenatal care visits was defined as adequate and inadequate as per World Health Organization recommendation [20]. Facility delivery was defined as childbirths occurring at any healthcare institution, and home delivery if otherwise [21]. Use of postnatal care (Yes/No) was assessed by asking whether or not the respondent underwent any health checkup by a health professional after delivery. This study considered a postnatal check-up within 48 hours after birth as a potential maternal healthcare indicator as per the WHO.

Explanatory variables
Selection was explanatory variables was guided by Andersen's behavioral model of health service utilisation which postulates that healthcare utilisation is a function of three major factors: 1) predisposing factors, 2) enabling factors and 3) need factors [15]. For this study, the data were secondary and hence the selection of the explanatory variables in line with the behavioral model was not possible. To this effect, the three sets of factors were met by using proxy indicators. Predisposing 5 factors were indicated by the sociodemographic variables including: age (15-19, 20-

Data analysis
Data were analyzed with Stata version 14. Datasets were merged and cleaned by applying the inclusion criteria: experience of at least 1 childbirth in the preceding 5 years. As the surveys used cluster sampling techniques, all analyses were adjusted for this by using the svy command [24]. This command uses the information on sampling weight, strata, and primary sampling unit provided with the datasets. Following that, the dataset was accounted for the cluster sampling design, sampling strata and weight by using complex survey mode. Sample characteristics were described as percentages. Prevalence of antenatal care, facility delivery and postnatal care was presented as bar charts. Following that, multinomial logistic regression models were used to estimate variations in the odds of receiving the maternal healthcare services. Results of four outcome variables were presented in four separate tables, each divided into three subsamples: overall, urban and rural. Model fit statistics were run after the regression analysis using the variance inflation factor (VIF) command. No multi-collinearity was detected as VIF values were below 10 for all the models. All tests were twotailed and was considered significant at alpha value of 5%.

Results
In  (Table A). In short, majority of the women were aged between 20-24 years, rural residents, followers of Christianity, of Bemba ethnicity, had more than two children, lived in maleheaded households, had primary level education, and had outdoor employment.    Figure 2 shows the prevalence of maternal healthcare services across age groups. The prevalence of timely ANC visit, health facility delivery, and postnatal care were higher among those aged between 20 and 34 years than the adolescent and those aged above 34 years.  Women in the higher wealth quintiles (Q4 and Q5) generally had higher prevalence of using health facility delivery and postnatal care services compared with those in the lower wealth quintiles (Q1 and Q2), whereas the prevalence of timely and adequate ANC visit was comparatively higher among those in the lower wealth quintiles.

Multivariate analyses
Predictors of timely initiation, at least one and adequate use of antenatal care were shown in      Interestingly, husband's educational status also appeared to be an important predictor adequate ANC, facility delivery, and PNC services. These findings imply the need for continued effort to improve women's education status and health communication through mass media channels to promote the use of maternal healthcare services in the country.
Significant economic (wealth quintile), geographic (urban/rural) and cultural differences (religion/ethnicity) were also observed in the odds of utilising the maternal healthcare services.
Women in the rural areas had higher odds of making early ANC contact, but lower odds of using facility delivery and PNC services. Having an employment didn't any strong impact in particular, however, higher parity showed a negative association with taking the services. Interestingly, the odds of wealth quintile were not as pronounced as educational status, despite the fact that these two indicators usually show similar effect in terms of healthcare utilisation. The present findings showed that women from higher wealth status households generally had higher odds of having used adequate ANC, facility delivery and PNC services. For PNC, the effects are particularly higher among urban women. The underlying mechanisms behind these differential effects of the different predictors are 11 hard to pinpoint by quantitative methods, and therefore calls for more comprehensive and qualitative studies. Nonetheless, the findings offer plenty of insight into the diverse sociodemographic factors that influence the use of maternal healthcare services in Zambia. Policy efforts to improve maternal healthcare service utilisation should therefore focus on developing innovative interventions strategies to address the sociocultural barriers to accessing care. Evidence-driven social determinants based approach by incorporating the specific needs of subpopulations may facilitate a more equitable provision of these essential and life-saving healthcare services to the vulnerable communities.

GHO | Key Country Indicators | Zambia -key indicators. WHO
http://apps.who.int/gho/data/node.cco.ki-ZMB?lang=en.       Maternal healthcare utilisation by type Figure 1 shows that prevalence of women making the first ANC visit within the first trimester has increased from 11.8% in 1996-97 to 23.9% to 2013-14, whereas that of making at least four ANC visit has decreased from 72.3 to 55.3% Maternal healthcare utilization by age groups Figure 2 shows the prevalence of maternal healthcare services across age groups. The prevalence of timely ANC visit, health facility delivery, and postnatal care were higher among those aged between 20 and 34 years than the adolescent and those aged above 34 years.

Figure 3
Maternal healthcare utilisation by educational level Figure 3 shows the prevalence of maternal healthcare services according to educational level. The prevalence of timely ANC visit, health facility delivery, and postnatal care were higher among those aged between 20 and 34 years than the adolescent and those aged above 34 years. Women who had no education had the lowest prevalence of making timely and adequate ANC visits, facility delivery and postnatal care, while those had primary level education had the highest prevalence of using these services.
32 Figure 4 Maternal healthcare utilisation by household wealth quintile Figure 4 shows the prevalence of maternal healthcare services according to household wealth status. Women in the higher wealth quintiles (Q4 and Q5) generally had higher prevalence of using health facility delivery and postnatal care services compared with those in the lower wealth quintiles (Q1 and Q2), whereas the prevalence of timely and adequate ANC visit was comparatively higher among those in the lower wealth quintiles.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download. Appendix.docx