Individual and facility-level factors associated with women’s receipt of immediate postpartum family planning counseling in Ethiopia: results from national surveys of women and health facilities

Background: Immediate postpartum family planning (IPPFP) helps prevent unintended and closely spaced pregnancies. Despite Ethiopia’s rising facility-based delivery rate and supportive IPPFP policies, the prevalence of postpartum contraceptive use remains low, with little known about disparities in access to IPPFP counseling. We sought to understand if women’s receipt of IPPFP counseling varied by individual and facility characteristics. Methods: We used weighted linked household and facility data from the national Performance Monitoring for Action Ethiopia (PMA-Ethiopia) study. Altogether, 936 women 5–9 weeks postpartum who delivered at a government facility were matched to the nearest facility offering labor and delivery care, corresponding to the facility type in which each woman reported delivering (n = 224 facilities). We explored women’s receipt of IPPFP counseling and individual and facility-level characteristics utilizing descriptive statistics. The relationship between women’s receipt of IPPFP counseling and individual and facility factors were assessed through multivariate, multilevel models. Results: Approximately one-quarter of postpartum women received IPPFP counseling (27%) and most women delivered government health centers (59%). Nearly all facilities provided IPPFP services (94%); most had short- and long-acting methods available (71 and 87%, respectively) and no recent stockouts (60%). Multivariate analyses revealed significant disparities in IPPFP counseling with lower odds of counseling among primiparous women, those who delivered vaginally, and women who did not receive delivery care from a doctor or health officer (all p < 0.05). Having never used contraception was marginally associated with lower odds of receiving IPPFP counseling (p < 0.10). IPPFP counseling did not differ by age, residence, method availability, or facility type, after adjusting for other individual and facility factors. Conclusion: Despite relatively widespread availability of IPPFP services in Ethiopia, receipt of IPPFP counseling remains low. Our results highlight important gaps in IPPFP care, particularly among first-time mothers, women who have never used contraception, women who delivered vaginally, and those who did not receive delivery care from a doctor or health officer. As facility births continue to rise in Ethiopia, health systems and providers must ensure that equitable, high-quality IPPFP services are offered to all women. Our analysis was guided by a conceptual framework based informed by the literature. Figure 1 depicts factors hypothesized to influence women’s receipt of IPPFP counseling. The first level identifies women’s individual sociodemographic characteristics, such as age, parity, and residence. The second level includes women’s individual reproductive characteristics; for example, prior use of contraception, receipt of ANC, and mode of delivery. The second level addresses characteristics reflecting IPPFP service availability at the facility-level; for example, facility type, the monthly ratio of deliveries to available providers, and the availability of contraceptive methods. Drawing upon the literature, we hypothesize that these individual and facility factors individually and collectively inform whether women receive IPPFP counseling. Conceptual framework informing study of immediate postpartum family planning counseling in Ethiopia This cross-sectional analysis uses data from PMA-Ethiopia, a collaborative study between Addis Ababa University (AAU), the Ethiopian FMOH, and Johns Hopkins Bloomberg School of Public Health (JHSPH). PMA-Ethiopia generates timely cross-sectional and longitudinal data on reproductive, maternal, and newborn health indicators to inform national and regional government priorities and policies. PMA-Ethiopia’s survey follows pregnant or recently postpartum women through 1 year postpartum and is conducted in six regions, collectively representing 90% of Ethiopia’s population [37]. PMA-Ethiopia conducted a full census in 217 enumeration areas (EA) across six regions and listed 36,614 households between October and November 2019. All women aged 15–49 in these households were screened for participation in the survey and were eligible if they were pregnant during the survey or delivered within 9 weeks preceding the survey. The baseline interview, conducted by resident enumerators, collected information about women’s sociodemographic characteristics, receipt of ANC, prior use of family planning, pregnancy intentions, and gestational age. Information about labor and delivery was collected at the follow-up interview; interview questions asked about pregnancy complications, delivery experiences, and postpartum experiences and behaviors, including whether women received IPPFP services. Women who were already 5–9 weeks postpartum at enrollment were asked the labor and delivery questions during the baseline interview. Informed consent to participate was obtained at the screening and prior to enrollment by all participants and was provided orally. Altogether, 2860 pregnant or recently postpartum women were enrolled in the survey. Women enrolled during pregnancy completed a baseline interview and a follow-up interview 6 weeks postpartum, with a follow-up rate of 88.9% (n = 2537) [37]. PMA-Ethiopia also conducts annual cross-sectional assessments of public and private facilities located in EAs used for the household survey. The Service Delivery Point (SDP) survey, herein referred to as the facility survey, collected data between September and November 2019 on facility characteristics and quality of care indicators for reproductive, maternal, and newborn health services, including family planning. Our analysis uses 1) individual-level data collected from postpartum women who provided information about their delivery and postpartum experiences at the baseline or follow-up interviews; and 2) service availability data from the facility survey. All PMA-Ethiopia surveys received ethical approval from both the JHSPH [00009391] and AAU Institutional Review Boards [075/13/SPH] [37]. Additional information on the PMA Ethiopia survey design, including weighting and informed consent procedures, is described elsewhere [37, 38]. Figure 2 presents the selection process for the analytic sample of postpartum women. Among the 2860 pregnant or recently postpartum women in our survey, we excluded pregnant women who had not yet delivered (n = 1180), those who had delivered but had not yet completed their follow-up interview (n = 19), those who were not within 5–9 weeks postpartum (n = 27), and women missing data on time since delivery (n = 7). Among the remaining 1627 eligible women with completed interviews, we excluded women who delivered at home, at a health post, or at other facilities unequipped to offer IPPFP services (n = 615), as information about IPPFP counseling and services was measured exclusively among women who delivered at a facility where family planning services could be provided immediately postpartum (hospital, health center). Due to the small number of women who delivered at private, faith-based, or non-governmental health facilities (only 3% of eligible women), we further restricted our sample to exclude women who did not deliver at a government facility (n = 51) and those with incomplete data on outcomes and exposures (n = 25). Our final analytic sample included 936 postpartum women who delivered in a government facility. Analytic sample – postpartum women As we did not collect the name of the facility where each woman delivered, we matched each woman with the nearest facility that matched the facility type each woman reported attending for labor and delivery care. For example, if a woman reported delivering at a health center, we identified the nearest health center based on straight-line geodetic distance. Due to a lack of complete information on road networks, the point distance between each woman’s household and the nearest matching facility was used as a proxy for physical access. We identified 224 facilities as the closest matching facility type to women in our analytic sample; the characteristics of these facilities were utilized in our analysis. Supplemental Fig. S1 presents further information about the facility analytic sample selection. The primary dependent variable, receipt of IPPFP counseling, was assessed by asking women “Before you left the facility, did a provider talk to you about a family planning method?” We also assessed the proportion of women who received both IPPFP counseling and a contraceptive method, which was ascertained by asking women if they received a method before they left the facility, and if so, which method. To assess individual-level correlates of IPPFP counseling, we measured a number of sociodemographic and reproductive health covariates from the household survey [37]. Sociodemographic variables included age, residence (urban/rural), region, education, religion, marital status, wealth, and parity. Reproductive health variables included number of ANC visits, receipt of PPFP counseling during ANC, mode of delivery (vaginal/Cesarean), prior use of contraception, pregnancy intention, type of delivery attendant, and experience of obstetric complications. We hypothesized that women who experienced more complicated deliveries because of obstetric complications would be less likely to receive IPPFP counseling due to the emergent nature of the delivery and competing healthcare priorities. Obstetric complications measured were self-reported severe bleeding, membrane leak or rupture without labor pain (> 24 h), membrane leak or rupture (at < 9 months gestation), malpresentation or malposition, prolonged labor (> 12 h), and convulsions or fits. These were assessed as any complications vs. no complications. Facility-level characteristics included availability of IPPFP services, ascertained by asking the facility in-charge “Is immediate postpartum family planning provided at this facility?”. Additional facility characteristics included short- and long-acting method availability, recent stockouts, ratio of monthly deliveries to providers, and presence of family planning guidelines. We defined method availability by whether each of five non-barrier family planning methods appropriate for IPPFP were in-stock and observed on the day of the survey. These included long-acting methods; specifically, implants and non-hormonal intrauterine devices (IUDs), and the following short-acting methods: progesterone-based pills, progestin-based injectables, and emergency contraception [14, 39]. The proportion of women who reported using exclusive breastfeeding as their primary contraceptive method was calculated among women who answered “No” to the question, “Do you plan to use a method of family planning, other than breastfeeding, within a year of giving birth?” Recent method stockout was defined by whether any method was reported out-of-stock at any time in the past 3 months, and if so, which method (short-acting, long-acting, or both). We also examined the ratio of monthly deliveries to providers, as a proxy for caseload volume, as a categorical variable and generated tertiles based on facility distributions. Finally, we explored the presence of national family planning guidelines in the delivery room of each facility. We used descriptive statistics to examine the distribution of women’s sociodemographic and reproductive characteristics (Table 1) and receipt of IPPFP methods, among women who received IPPFP counseling (Table 2). We described facility characteristics in two ways; first by the distribution of characteristics of the 224 linked facilities and second, by the percentage of women who delivered in a facility of each type (Table 3). We assessed the bivariate distributions of IPPFP counseling and each covariate at both levels (Individual, Table ​Table1;1; Facility, Table ​Table3).3). We used multivariate, multilevel models to estimate differences in women’s odds of receiving IPPFP counseling by individual and facility-level characteristics, adjusting for clustering of women within the facilities. Our final adjusted model included individual and facility-level covariates driven by theory and conceptual relevance, while also accounting for established confounders of women’s receipt of reproductive health services (i.e. age and urban/rural residence) (Table 4) [40, 41]. We examined model fit by analyzing model fit statistics (i.e., Aikake’s Information Criterion (AIC) values) and assessed collinearity at 0.6 using a correlation matrix for all analytic variables. We excluded wealth from the multivariate analysis due to its high correlation with urban/rural residence (r = 0.749) and religion due to small cell sizes, limiting statistical power. Statistical significance for the adjusted multivariate analysis was set to p < 0.05. All analyses were weighted to reflect the national population of pregnant and postpartum women in Ethiopia and accounted for the complex survey design [42]. All analyses were conducted in StataSE, Version 16 [43]. Sociodemographic and reproductive characteristics of postpartum Ethiopian women who delivered at a government health facility (n = 936)b aAmong women who received at least 1 ANC visit bAll values were weighted to account for the complex survey design Receipt of IPPFP methods or referrals among women who received IPPFP counseling (n = 251a) aUnweighted bWeighted cAmong women who received IPPFP counseling, but did not receive a method or a referral Facility-level (n = 224) and woman-level (n = 936) descriptive characteristics of linked health facilities LARC Long-acting and reversible contraception, SA Short-acting contraception *Represents the design-based chi-squared test between each facility-level characteristic and government facility type (health center/hospital) **Represents the weighted bivariate logistic regression between each woman-level facility characteristic and women’s receipt of IPPFP counseling Adjusted odds of receiving immediate postpartum family planning counseling by women’s individual and facility-level characteristics (n = 936) AOR Adjusted Odds Ratio, LARC Long-acting and reversible contraception, SA Short-acting contraception *p < 0.10, **p < 0.05