Date Published: June 20, 2019
Publisher: Public Library of Science
Author(s): Linnea A. Zimmerman, Suzanne O. Bell, Qingfeng Li, Antonia Morzenti, Philip Anglewicz, Amy O. Tsui, Joshua Amo-Adjei.
The importance of the family planning service environment and community-level factors on contraceptive use has long been studied. Few studies, however, have been able to link individual and health facility data from surveys that are nationally representative, concurrently fielded, and geographically linked. Data from Performance Monitoring and Accountability 2020 address these limitations. To assess the relative influences of the service delivery environment and community, household, and individual factors on a woman’s likelihood of using a modern contraceptive in five geographically and culturally diverse sub-Saharan African countries. Nationally representative, cross-sectional data from PMA2020 were linked at the household and service delivery level. Country-specific and pooled multilevel multinomial logistic models, comparing non-users, short- and long-acting method users were used. The variables elected for inclusion in our multivariate analyses were guided by the conceptual framework to profile the different levels of influences on individual use of modern contraception. Average marginal effects were calculated to improve interpretability. We find that the effect of contextual factors varies widely but that being visited by a health worker who spoke about family planning in the past 12 months was consistently and positively associated with individual use of short-acting and long-acting contraception. Characteristics of the nearest health facility did not generally exercise their own independent influences on a woman’s use of contraception, except in the case of Burkina Faso, where the average distance between individuals and the nearest family planning provider was significantly greater than other countries. Inclusion of country fixed effects in the pooled models and the relevance of covariates at different levels in the country-specific models demonstrate that there is significant variation across countries in how community, individual, and service delivery environment factors influence contraceptive use and method choice. Context must be taken into account when designing family planning programs.
The importance of the family planning service environment and community-level factors on fertility and contraceptive use for low- and middle-income countries (LMICs) has long been studied [1–5]. This body of research has highlighted the influence of service environment factors (e.g. proximity, quality, and facility type), on a range of female contraceptive outcomes, including current practice, choice of method, and duration of use. Community-level measures of socioeconomic development (e.g. mean education level, mean household wealth, and religious composition) and contextual factors (e.g. gender norms, spouse communication, and social capital) have been found to be associated with modern method use  and consistent use . The knowledge base that has accumulated over the past decades has grown in rigor with the application of advanced statistical methods for model estimation, in particular hierarchical linear modeling or multi-level analysis (MLA) to address the nested data structure of individuals within households and households within communities. MLA has enabled the assessment of cross-level influences of higher-level observed factors on lower level individual outcomes and of unobserved variation at multiple levels. In addition, improvements in computational efficiency have facilitated the estimation of multilevel factors’ influences after adjusting for complex survey design.
Table 1 summarizes the sample characteristics for females across the five countries while Table 2 presents the characteristics of the SDP sample. The SDP characteristics are shown in two ways; first, those of the entire sample in each PMA round, and second, as the distribution by woman; that is, the percentage of women whose nearest health facility met each characteristic.
The aim of this analysis was to examine the multilevel relationship between community norms, demand generation efforts, factors in the contraceptive supply environment, and individual characteristics with the probability of short-acting and long-acting modern contraceptive use in five sub-Saharan African countries. By evaluating the effect of variables in each of these domains, we sought to show empirically where resource investments into health and development policy and program efforts may enable and raise the probability of contraceptive use.
Use of modern contraception in the five sub-Saharan African study countries has been trending upward in recent years, suggesting that demand is increasingly satisfied either through individual motivation (Bongaarts 2017) or facilitated by improved systems of contraceptive delivery. This analysis, enhanced by multilevel modeling of linked household, female, and SDP data, shows that access and service provision characteristics of proximal health facilities can influence contraceptive use decisions when choice of provider is limited, independent of community norms and individual factors. Additionally, dissatisfaction with contraceptive use at the community level can significantly affect use patterns and should be considered when designing policies and programs. Investments to decrease distance to services, improve satisfaction with methods, and increase health worker outreach are likely to increase modern method use.