Association between skilled maternal healthcare and postpartum contraceptive use in Ethiopia

The Ethiopian Demographic and Health Survey (EDHS) was first implemented in Ethiopia in 2000 and a total of four national surveys have been conducted as of 2016. The present analysis uses data from the latest two rounds of the EDHS, collected in 2011 and 2016. The EDHS is a cross-sectional survey conducted every five to six years to measure the demographic and health-related characteristics of the population [26]. The EDHS 2011 survey was conducted from December 2010 to June 2011 and includes interviews with a total of 16,515 women of reproductive age [27]. The EDHS 2016 survey was conducted from January to June 2016 and includes a sample of 15,683 women of reproductive age [21]. A stratified, two-stage cluster sampling procedure was used to recruit the nationally representative sample in both surveys. The 11 administrative regions found in Ethiopia were initially stratified into urban and rural areas. Each stratum was subdivided into districts (woredas), and woredas were then subdivided into kebeles (the smallest administrative unit in Ethiopia). Finally each kebele was subdivided into Enumeration Areas (EAs) that served as sampling clusters. A total of 624 EAs were included in EDHS 2011 and 645 EAs were included in EDHS 2016. In the present analysis, we included married/in sexual union, non-pregnant women who gave birth in the 12 months prior to data collection. As a result, a total of 2545 married/in sexual union women in EDHS 2011 and 2386 married/in sexual union women in EDHS 2016 were included in our final analysis (Fig. 1). The analysis was restricted to married/in sexual union women to allow cross-country comparisons of contraceptive prevalence rate (CPR), which is often calculated only for married women and those in sexual union. Additionally, we excluded women who were not married or in sexual union because they are not considered at risk for pregnancy and therefore do not require methods of pregnancy prevention, the main outcome of interest in this analysis.

In this analysis the dependent variable is a binary variable for postpartum contraceptive use. Postpartum contraceptive use is defined as any self-reported use of a modern contraceptive method (modern methods reported in this sample included: female sterilization, implant, intrauterine device (IUD), injectable, oral contraceptive, emergency contraceptive, lactational amenorrhea method (LAM) or condom) at the time of data collection. Those women not reporting any modern method use and those reporting traditional method use (periodic abstinence or withdrawal) were classified as non-users.

We examined the association between postpartum contraceptive use and four key maternal healthcare services, all coded as binary variables: institutional delivery, presence of a skilled birth attendant, skilled antenatal care, and skilled postnatal care. Institutional delivery was classified based on the place of delivery. A woman was defined as having an institutional delivery if she had given birth (the most recent time) at a hospital, health centre, or any private facility; otherwise women were classified as having a home delivery. Skilled attendance at birth was based on who attended the delivery irrespective of the place of delivery. Those women whose most recent birth was attended by a professional (doctor, midwife or nurse, health officer) possessing midwifery skills were classified as having skilled attendance at birth. Skilled antenatal care included any woman receiving one or more episodes of antenatal care by a skilled provider. Skilled postnatal care was based on health check-ups by skilled professionals (doctor, midwife or nurse, health officer) for women who delivered in a health facility or at home within the first six weeks of the participant’s most recent delivery.

Demographic, reproductive health and other factors were also examined as potential covariates in this analysis. Demographic variables include age, rural versus urban residence, region, religion, participant’s education, partner’s education, employment status, wealth, and parity. Reproductive health and other factors include birth order, concordant family planning preferences, fertility preferences, women’s decision-making, contraceptive knowledge (the ability to mention at least one method), any visit from a family planning worker in the last 12 months, and exposure to family planning media messages (newspaper, radio, television, or mobile SMS text) in the last few months. Women’s decision making was defined by whether a woman reports that she can make decisions by herself or jointly with her partner about the following: decisions about health service utilisation, decisions about household purchases, and decisions to visit family or relatives. Specific categories for these variables are shown in Tables 1, 2, and 3.

This analysis includes nearly 5000 married/in sexual union women of reproductive age with a birth in the 12 months preceding the survey; 2545 of these women participated in the 2011 EDHS while the remaining 2386 were participants in the 2016 survey. As seen in Table 1, about half of all women (47% in 2011 and 50% in 2016) were between the ages of 25 and 34, with a median age of 28 years. The majority of the respondents (87% in 2011 and 88% in 2016) were rural residents. Oromo and Amhara ethnic groups were the predominant groups in both rounds of the EDHS. With regard to the level of education, the proportion of women who did not have any formal education decreased from 67% in 2011 to 59% in 2016 and the proportion of women who completed secondary education more than doubled between surveys (4% in 2011 to 9% in 2016). Likewise, the proportion of participants whose partners completed secondary education nearly doubled between surveys from 9 to 15%. About 53% of these postpartum women were not working outside the home. When looking at the household wealth, 46% of participants in both surveys were in the poorest group, with no change between surveys in terms of the distribution of wealth.

As shown in Table 2, in both surveys, more than a third of women had between one and two children, more than a fourth had three to four children, and nearly a third had five or more children. For the majority of participants (80 to 85%), their most recent birth was not their first birth, although first-time deliveries increased by about 5 % when comparing the 2011 and 2016 samples. Few changes were seen between surveys in terms of family size concordance, with just about 40% of couples agreeing on ideal family size and approximately one in four women reporting that their partner wants a bigger family than they do. In both surveys, fertility preferences remained approximately stable, with only about six to 8 % of women reporting a desire for another child soon; about one third reported not wanting any more children and more than half reported wishing to delay their next pregnancy. Ability to mention at least one contraceptive method remained nearly universal in both survey rounds and exposure to family planning media messages stayed stable at around 25%. Notably, the percentage of women reporting some degree of decision-making power increased from 50 to 68%. Also of note, the percentage of women who reported a visit by a family planning worker in the preceding 12 months doubled between surveys from 16 to 33%.

In the last five years, the prevalence of contraceptive use increased by 53% from 15% in 2011 to 23% in 2016 (Table 3). In terms of the contraceptive method mix in both surveys, injectable contraception was the predominant method (Fig. 2). While there were little to no changes in the contribution to the national method mix by pills, female sterilization, condoms, and LAM, there was a notable change in the contribution of implants (10% in 2011 vs 16% in 2016) and IUD (1% in 2011 vs 4% in 2016) (Fig. 2).

With respect to maternal healthcare, while only about one in ten women reported delivery in a health facility in 2011, more than a third of our 2016 sample reported delivery in a health facility. Use of public facilities became nearly universal within our 2016 sample, as compared to private facility use, among those women not delivering at home. Additionally we observed a threefold increase in the use of skilled assistance during delivery and a twofold increase in the number of women receiving skilled antenatal care, when comparing the two survey samples. We also observed modest increases in the number of women with four or more antenatal visits and skilled postnatal care (Table 3).

In our 2011 sample of postpartum women, two demographic characteristics were strongly associated with postpartum family planning. Women in the highest wealth category had nearly three times the odds of postpartum contraceptive use compared to women in the lowest wealth category. In addition to wealth, education was strongly associated with postpartum contraceptive use in the 2011 sample; those with at least a primary education were nearly twice as likely to use contraception, compared to women with no education. In our 2016 sample of postpartum women, education remains associated with contraceptive use but the magnitude of the association between wealth and family planning use is greatly reduced and the result is only marginally significant (Tables 4 and 5).

In the 2011 sample of postpartum women, the strongest association with postpartum contraceptive use was seen among women who reported a desire to avoid future pregnancy. Women who wanted no more children had six times the odds of postpartum contraceptive use compared to women who would like to have a child soon. We did not find any relationship between women with a desire to delay pregnancy and contraceptive use in our 2011 sample of women. In the 2016 data, our sample of women also demonstrated a strong likelihood to use contraception in the postpartum period if they preferred no more children; in contrast with the 2011 data however, our 2016 sample of women were also significantly more likely to use contraception if they had a desire to delay a future pregnancy (AOR: 4.38, 95% CI: 1.46 – 13.18) (Tables 4 and 5).

Regarding use of skilled maternal healthcare, no statistically significant associations were found in the 2011 sample of women. Women who delivered in a health facility or with a skilled attendant were no more likely than women with home births or those without skilled attendance to use postpartum family planning. Similarly, women with antenatal care visits were no more likely to use postpartum family planning than those without such visits. We found only a marginally significant relationship between postnatal care and postpartum contraceptive use in the 2011 data. In contrast, delivery within a health facility as well as obtaining skilled antenatal care are both significantly associated with postpartum contraceptive use in the 2016 sample, with adjusted odds ratios of 1.71 (95% CI: 1.12-2.62) and 2.41 (95% CI: 1.41-4.10) respectively (Tables 4 and 5).