Determinants of fertility in Rwanda in the context of a fertility transition: a secondary analysis of the 2010 Demographic and Health Survey

Population growth and high fertility rates in resource poor settings can be a challenge for both the society and individuals. A growing population can affect the well-being of that population in terms of socioeconomic development, environmental sustainability, and resource supply [1]. Resource poor countries with population growth are challenged to create jobs for a budding workforce while their governments lack resources to meet increasing demand for services and infrastructure [2]. The effect of high fertility is also challenging for individuals. When many children are born to one mother, there is an economic burden on her household and an increased chance of her family entering into poverty [3]. In families that do not have enough resources for education, food, and health care, children - especially girls - may be forced to drop out of school and to marry early [2]. High fertility also increases the risk that a child is born prematurely or with low birth weight [4] and becomes stunted as she grows [5], and premature birth increases maternal health risks [6].

The “demographic transition” describes a widely observed phenomena whereby a population transitions from high levels of mortality and fertility to low levels of mortality and fertility [7]. This transition is typified by an initial drop in child morality due to improved infrastructure, health system developments, and socioeconomic improvements followed by a decrease in fertility rates years later. While countries in sub-Saharan Africa have been slow to enter and pass through the fertility transition, Rwanda has been an exception [8]. Rapid improvements to the health system, infrastructure, and social programs over the last decade have launched Rwanda into a rapid fertility transition. Between 2005 and 2010, the mortality rate among children under five was halved from 152 to 76 deaths per 1000 live births, marking one of the fastest improvements in child mortality in human history [9]. The average fertility rate in Rwanda dropped from 6.1 births per woman to 4.6 births per woman in 2010 after the percent of women using a modern method of contraception increased from 17% in 2005 to 52% in 2010 [10]. In other East African countries the fertility decline is still low: in Uganda fertility declined between 2006 and 2011, from 6.7 children per woman to 6.2 children [11]; in Kenya fertility dropped from 4.9 in 200 to 4.6 in 2008 [12] while in Tanzania the fertility was 5.7 in 2004–05 and slightly declined to 5.4 births per woman in 2010 [13].

This drop in fertility and uptake of contraception in Rwanda coincides with a major shift in attitudes by government officials about family planning as it related to economic development policies. Faced with the reality that Rwanda has the highest population density of any country in Africa (416 persons per square kilometer with an annual population growth rate of 2.6% [14]), smaller families and limited population growth became priorities for individual well-being as well as national progress. Officials subsequently launched widespread campaigns to shift public attitudes toward acceptance of small families with an informal goal of bringing the total fertility rate to less than 4 children per woman [15]. Following the implementation of mandatory free primary education, and in response to the rising cost of living, the government has performed sensitization campaigns to encourage couples to have only as many children as the family can afford to feed, educate, and care for. This has been reinforced at the community-level by community health workers and community leaders in monthly “community works” meetings. Despite this major shift in fertility, there are many families in Rwanda still having large families; more than 20% of women between 15 and 49 currently have had five or more births [16].

Rwanda is in an important demographic transition that is setting a course for the country’s economic development and bucking trends in a region of slow fertility transition. Understanding predictors of fertility can support the development of policies and interventions that both support families to achieve their desired fertility and inform government economic policies and infrastructure development plans. Results may also inform fertility policies and programming elsewhere in sub-Saharan Africa. This article examines some of the determinants of fertility rates in Rwanda, looking separately at women who have ever been married/cohabitating and women who have never been married.

This analysis is based on the data collected from the 2010 Rwanda Demographic and Health Survey (RDHS). The RDHS is a nationally representative two-stage cluster sample designed to provide population and health indicators at national, province and district levels. Villages were the primary sampling unit (PSU) with 413 rural PSUs and 79 urban PSUs sampled, stratified by district. Twenty-six households were sampled per PSU [16]. Women provided informed consent before participating in the survey; we were granted permission by the MEASUREDHS Project to use these de-identified data for this analysis.

A total of 13,671 women age 15 to 49 years participated in the 2010 RDHS. Women who never had sex were excluded from the analysis. Among women who reported ever having sex, the analyses were stratified by the 8,309 women currently or ever married/cohabitating and the 1,220 women who had never married.

The main outcome, level of fertility, is defined as the total number of children ever born to women in the childbearing period (15–49 years). The predictor variables were mainly the proximate determinants of fertility - current marital status, age at first cohabitation, age at first sexual intercourse, age at first birth, use of contraceptives - and socio-demographic variables that may predict fertility in Rwanda. Twenty variables in the 2010 RDHS were identified as potential predictors of fertility.

We developed a multivariable Poisson regression model with a natural logarithmic link function to assess associations between predictor variables and fertility rates. The offset term in the Poisson model was set to natural log of current age of the woman. The modeling development involved two stages. Potential predictors were identified in bivariate analyses. Variables that were differentially distributed across ever married or never married women with different numbers of children were retained based on a chi-squared test (p < 0.05). We tested remaining variables for collinearity at the r > 0.5 level, and excluded the variable that was more weakly associated with fertility. Multivariable models were built using backward stepwise regression considering ever married and never married women separately. Variables that were significantly associated with fertility at the 95% confidence level were retained, and all models were adjusted for age, province, and urban/rural residence. Adjusted incidence rate ratios (IRR) with 95% confidence intervals (CI) are reported to assess the association of fertility with the demographic characteristics (age, number of siblings, number of unions), socioeconomic characteristics (education, wealth, religion, residence, types of earnings), geography (province), fertility behavior and desires (age at first sex, unmet need for family planning, ideal number of children) and, among ever married women, husband’s desire for children. All analyses were performed in Stata v12 with svyset statements to apply sampling weights, and account for clustering and stratification.

The bivariate results for determinants of fertility and the number of children ever born is presented for women who were ever married/cohabitated in Table 1, and for women who have never married in Table 2. Five percent of ever married/cohabitated woman had no children, 45.7% had 1 to 3 children, and 49.3% had more than 3 children. All potential predictors considered in the bivariate analyses were significant for this group. For never married women, 49.6% had no children, 48.3% had 1 to 3 children, and 2.1% had more than three children. All considered covariates were significant in the bivariate analyses except urban/rural residence and religion (p = 0.101 and p = 0.386, respectively).

The results of the multivariable analysis are presented in Table 3. These results showed lower fertility among women with more education and with greater household wealth. In ever married/cohabitated women, more education was progressively associated with fewer children; the IRR was 0.96 (95% CI: 0.94-0.98), 0.90 (95% CI: 0.87-0.94) and 0.66 (95% CI: 0.61-0.73) respectively for women with primary, secondary and higher education levels compared to women with no education. In never married women, only women with more than a secondary education had fewer children than women with no education (IRR = 0.22; 95% CI: 0.09-0.52). Household wealth had a stronger effect on limiting fertility among never married women (IRR = 0.58, 95% CI: 0.46-0.63) than ever married/cohabitated women (IRR = 0.93, 95% CI: 0.90-0.96). Working status was associated with higher fertility among never married women; women employed for cash (IRR = 1.34; 95% CI: 1.06-1.70) or in-kind compensation (IRR = 1.38; 95% CI: 1.11-1.71) had more children than unemployed women.

Age of sexual debut was strongly associated with fertility rate; women who were 25 years or older at first sex had less than half the fertility rate as women whose first sex was before age 19 in both ever married/cohabitated women (IRR = 0.49, 95% CI: 0.47, 0.52) and never married women (IRR = 0.43, 95% CI: 0.34, 0.56). In ever married/cohabitated women, having an unmet need for family planning was associated with higher levels of fertility (IRR = 1.07, 95% CI: 1.05-1.09), while unmet need was associated with lower fertility in never married women (IRR = 0.69, 95% CI: 0.51, 0.94).

Several additional factors were associated with fertility in ever married/cohabitated women. In these women, their ideal number of children was associated with fertility level; fertility was 1.22 times higher (95% CI: 1.18,1.25) among women who wanted 5 or more children compared to women who wanted 0 to 2 children. Women whose partners wanted a different number of children had higher fertility than women who wanted the same number of children as their partners (husband wanted more children IRR = 1.04, 95% CI: 1.01-1.07, husband wanted fewer children IRR = 1.04, 95% CI: 1.02-1.07). Ever married/cohabitated women who have been in more than one union had 0.87 times the fertility rate of women who had been in only one union (95% CI: 0.85, 0.90). Women from large families with 8 or more siblings had slightly higher fertility than women from average size families with 5 to 7 siblings (IRR = 1.03, 95% CI: 1.01, 1.05), and women with 4 or fewer siblings had similar fertility as women from average size families.

Although we cannot draw causal conclusions from these results, our study suggests several risk factors for high fertility including having an unmet need for family planning, early sexual debut, limited access to education and economic opportunity for women, valuing larger families over small families, and married couples disagreeing about desired number of children. We suggest several strategies for continuing to reduce fertility, particularly delaying sexual debut though increased educational and economic opportunity for women, increased access to reproductive health knowledge and services in schools and through public campaigns, and involving men in family planning programs and campaigns.

Rwanda is unique in Africa for its sharp declines in desired and actual fertility in recent years. The unique circumstances that led to this decline may provide lessons for other countries in the region. However, despite this fertility decline, the fertility rate remains above the national target of 3 children per woman. The government’s strong and coordinated position to sensitize the public about the benefits of smaller families, promote women’s empowerment through education, and encourage delayed sexual debut through late marriage may have all played roles in the steep decline in fertility. Based on these results we recommend additional programs to improve access to reproductive health services, particularly to young, unmarried women, and involve men in family planning decision making.