How Communities Shape Unmet Need for Modern Contraception: An Analysis of 44 Low-and Middle-Income Countries

Metheny, Nicholas S.

Background:

Access to modern contraception and its resulting decline in fertility is associated with reduced poverty, improved health outcomes for women, and gains in female empowerment. However, fewer than 40% of women in the least-developed countries are currently using any method of modern contraception. As a result, fertility rates are higher in Low and Middle Income Countries (LMICs) than in developed countries and declining rates of fertility in the second half of the 20^th century have slowed or stalled in many LMICs- especially those in sub-Saharan Africa. Despite worldwide efforts to increase access to modern, pharmacological methods of family planning, less than half of the total demand for family planning was met by modern methods in 2015 in 54 of the world’s poorest countries. Unmet Need, a measure of the disparity between a partnered, fecund woman’s desire for modern family planning and her use of contraception, is a useful indicator in tracking progress towards universal access to family planning. Unmet need consists of two components- unmet need for spacing and unmet need for limiting. Having an unmet need for limiting occurs when a woman wants to increase the interval between her births but is not using modern contraception, while the latter is present when a woman wants to stop having children altogether, but is not using modern contraception. While the individual-level drivers of unmet need are well documented, there is little data on how community-level characteristics influence this important indicator. To explore this, we created a comprehensive dataset of parous women from the 44 Demographic and Health Surveys (DHS) published since 2010 (n=528,101). We analyzed the influence of 13 community-level variables on unmet need as a whole and for its two components, unmet need for spacing and unmet need for limiting, as outcome variables. We then repeated the analysis, stratifying countries by low, medium, and high total fertility rate (TFR) to determine if the effect of community-level variables change with TFR. This is the first multi-country analysis that examines how community-level factors shape unmet need in resource-constrained settings.

Methods:

Data: This analysis utilized data from the women’s questionnaire of all 44 Demographic and Health Surveys collected in or after 2010. The initial sample for this study contained all women ages 15-49 from the 44 countries studied (n=721, 539). Nulliparous women do not answer DHS questions related to fertility and childbirth, 193,438 childless respondents were removed from the sample, resulting in a final sample of 528,101 pregnant, postpartum-amenorrheic, and/or parous women across five WHO regions. Three outcome variables were considered: total unmet need, unmet need for spacing, and unmet need for limiting. All are binary variables coded one for total unmet need, unmet need for limiting, or unmet need for spacing. Community-level variables were categorized into four domains: 1) community demographics and fertility norms consisting of five variables: the mean age at marriage for women in the community, mean age at first intercourse for women in the community, mean age at first birth for women in the community, mean ideal of number of children each woman would have in the community, and gender composition of the children in the community; 2) community economic prosperity, measured by the mean household wealth index score for each PSU; 3) community gender norms and inequities, measured by the mean community violence justification index score, mean community decision-making autonomy score, proportion of women in the community with at least a primary education, proportion of men in the community with at least a primary education, and ratio of men to women employed in the community; and 4) health literacy and media exposure, measured by the mean community HIV knowledge index score and mean exposure to family planning media messages in the community.

Analysis: A multilevel modeling approach was used to account for the hierarchical nature of DHS data and allow for the observation of community-level influences on unmet need for family planning. The PSU was included as the only random effect term. This allowed the intercept to vary across communities and provided a measure of the extent to which the odds of reporting unmet need, unmet need for spacing, and unmet need for limiting varied between PSUs. After controlling for individual and household-level factors known from the literature to influence contraceptive use, the 13 community-level covariates were added. Identical models for unmet need, unmet need for spacing, and unmet need for limiting were fitted using STATA 14.

To assess whether the role of community-level effects on unmet need varies with fertility, the countries were then stratified by total fertility rate (TFR) and divided into Low, Medium, and High countries using UN data of TFR. Separate random effects logistic regression models were fitted for total unmet need, unmet need for limiting and unmet need for spacing in in each group of country (low, medium and high fertility). In total, 12 random-effects multiple logistic regression models were produced to examine the effect of community-level variables on unmet need for family planning in the study sample.

Results:

Results indicate that unmet need is significantly influenced by community-level variables in all three models (Unmet Need: SE=0.008, σ_µ=0.148, PSU random intercept= 0.13, (95% CI: 0.08-0.21); p<0.000; Spacing: SE= 0.008, σ_µ=0.140, PSU random intercept= 0.10, (95% CI: 0.05-0.19); p=<.000; Limiting: SE= 0.010, σ_µ=151, PSU random intercept= 0.01, (95% CI: 0.00-0.02); p<0.000). These models measure the variation in unmet need, unmet need for spacing, and unmet need for limiting between communities which is not explained by any of the included fixed effects. While the models themselves are significant, they indicate substantial unobserved heterogeneity in the determination of unmet need in the study sample. The effect size and directionality of community variables in all four domains changed significantly by TFR. For example, while residing in a community with a higher average age of cohabitation was associated with reduced odds of having total unmet need and unmet need for limiting in the 44-nation sample, it was associated with increased odds of total unmet need and unmet need for spacing in high fertility countries (unmet need: OR: 1.12, (95% CI: 1.02-1.23), p<0.05; spacing: OR: 1.13, (95% CI: 1.01-1.27), p<0.05). Conversely, communities with higher boy-girl ratios were associated with reporting increased odds of unmet need and unmet need for limiting in the larger sample, but was not significant in any of the stratified models. For economic prosperity, women in the 44-nation sample who resided in wealthier-than-average communities reported more unmet need for limiting, increased community wealth was associated with reduced odds of unmet need, unmet need for spacing and unmet need for limiting in low fertility countries (unmet need: OR: 0.86, (95% CI: 0.81-0.92), p<0.000); spacing: OR: 0.85, (95% CI: 0.78-0.93), p<0.000; limiting: OR: 0.88, (95% CI: 0.82-0.94), p<0.000) and with reduced total unmet need in high fertility countries (OR: 0.88, (95% CI: 0.78-0.98), p<0.005). Concerning the domain Gender Norms and Inequalities, women in communities of higher primary education attainment reported less unmet need of all kinds in the 44-nation sample (unmet need: OR: 0.29, (95% CI: 0.24-0.36), p<0.000; spacing: OR: 0.73, (95% CI: 0.56-0.99), p<0.05; limiting: OR: 0.19, (95% CI: 0.15-0.25), p<0.000), but this was associated with reporting more unmet need for spacing in medium fertility countries (OR: 1.83, 95% CI: 0.15-2.92), p<0.05). HIV knowledge score, nested in Community Health Knowledge and Media Exposure, showed that living in communities with greater average knowledge of HIV was associated with reporting more total unmet need and unmet need for limiting (unmet need: OR: 1.10 (95% CI: 1.05-1.16), p<0.000; limiting: OR: 1.12, (95% CI: 1.05-1.19), p<0.000). However, higher average scores on the HIV knowledge index were associated with reporting less total unmet need in low fertility countries (OR: 0.91, (95% CI: 0.84-0.99), p<0.05).

Discussion:

This is the first large, multi-country study of the community’s effect on unmet need in LMICs. Our analysis of 13 community-level variables shows that the community has a significant impact on a woman’s access to modern contraception. Further, it highlights that combining these indicators into four broad domains is a valid way to view their collective effects on unmet need. The change in directionality of these indicators across different contexts (defined by levels of TFR) suggests that research studies, interventions, and policies should be tailored to the community domains which have the largest effect for the target population. Nonetheless, additional studies are required to explicate the mechanisms underlying the relationship between these community-level effects and unmet need, unmet need for spacing, and unmet need for limiting. This information is critical to accelerating progress towards universal access to family planning and reproductive health services in Low and Middle Income Countries.