Determinants of postpartum uterine atony in urban South Ethiopia: a community-based unmatched nested case–control study

In this study, a community-based unmatched nested case–control study design was applied. The cases and controls originated from a community-based prospective cohort study that was conducted among pregnant women in five urban settings: namely Hossana, Shone, Gimbichu, Jajura, and Homecho, which are located in Hadiya Zone, South Ethiopia. Hadiya Zone is located 232 km from the capital city, Addis Ababa. In the zone, there are 1 general hospital, 3 primary hospitals, 62 health centers, and 311 health posts that provide health services for the community (Hadiya Zone Health Bureau report, unpublished).

Study participants involved in this study were subsets of a community-based prospective cohort study that was conducted to assess the effects of inter-pregnancy intervals on pregnancy outcomes [20]. For the study, a cohort of pregnant women who were at the end of their first trimester of confirmed pregnancy (after 12 weeks of gestation) were enrolled via house-to-house visits, from July 8, 2019, to March 30, 2020. Women who were pregnant at the time of recruitment, who were able to recall the date of last childbirth, who had a live birth during the most recent childbirth, who had no recent stillbirth, and who had no recent abortion were included. In the study, a total of 2578 pregnant women were enrolled and followed until September 30, 2020 [20].

For this particular study, all participants who had experienced uterine atony during a follow-up time (July 08, 2019 to September 30, 2020) were considered as cases and those who had no uterine atony were considered as controls.

In nested case–control studies, the sample size depends on the number of observed cases from a cohort study and their corresponding controls. Accordingly, a total of 93 cases were observed during the follow-up. Uterine atony is a relatively rare condition. To increase statistical power (precision), a 1:4 ratio was considered. Therefore, the sample size becomes 465 (93 cases and 372 controls). For every uterine atony case, four controls were randomly selected from the frame of the cohort (the risk set) by using a random number generator in open-Epi software. The sample frame was prepared for each study site separately so that for a given case, the corresponding four controls were selected from where the case was obtained (Fig. 1).

Independent variables were: Socio-demographic, economic and reproductive characteristics, which are measured as follows: The age of a woman is measured in years by asking how old she is and when she was born. Age at first childbirth measures at what age she has given her first birth, irrespective of the outcomes of the birth, and is reported in completed years. Parity is measured by asking a woman the number of times she gives birth, irrespective of the birth outcomes. The response was reported in numbers, and categorized as prim-para (if she has given birth only once) or multipara (if she has had more than one birth). The inter-pregnancy interval is measured by asking women about the dates of their most recent childbirth and their last menstrual period. Then it was computed by subtracting the date of recent childbirth from the date of the last menstrual period (LMP). For women who had difficulty recalling the date of LMP, ultrasound was used to estimate gestational age at the hospitals. LMP was computed by subtracting the duration of gestation, and then the value of the inter-pregnancy interval was calculated. The inter-pregnancy interval was categorized as < 24 months (short) and 24–60 months (normal), which is based on the World Health Organization recommendation for birth spacing [21]. Pregnancy intention refers to whether the woman had a plan for pregnancy at the time of conception. If a woman had a plan, the response was categorized as ‘intended; if not, it was categorized as 'unintended'. Unintended pregnancy includes both mistimed and totally unwanted pregnancies [22]. Pre-eclampsia was measured as the presence of pregnancy-related high blood pressure (≥ 140/90 mmHg) and protein in the urine, after 20 weeks of gestation [23], and was taken from the clients’ chart. If it occurs, it is reported as "present," otherwise it is reported as "absent." Premature rupture of membranes was measured as a rupture of membranes before the onset of labor and is categorized as ‘present’ or ‘absent [24]. The duration of labor is the time that a woman stays in labor. If the true labor exceeds 12 h, irrespective of the stages, then it is categorized as ‘prolonged labor, otherwise ‘not prolonged' [25]. The progress of labor refers to whether labor progresses in its natural course or is augmented or induced using utero-tonic drugs. If labor was induced or augmented, then it was classified as ‘assisted’, otherwise 'normal'. The mode of delivery is whether a woman gives birth spontaneously vaginally, assisted vaginally with instruments (forceps and vacuum), or via cesarean section. Birth characteristics refer to the number of babies born at the time of delivery (single or multiple). Gestational age refers to the number of weeks elapsed from the date of a woman’s last menstrual period until the date of delivery. If it was < 37 completed weeks, it was categorized as ‘preterm’, if it was from 37–41 weeks, including the 37^th week, it was categorized as ‘term’, and if it exceeded 41 weeks, it was categorized as 'post-term' [26]. Birth weight is the weight of the newborn at birth, measured in grams. It was categorized as < 4500gms and ≥ 4500gms (macrosomia) [27].

The questionnaire was prepared in English from existing related literature (published articles and Ethiopia Demographic and Health Surveys) based on the study objectives [28, 29]. The English version was translated to Amharic by two native speakers of the Amharic language (one was public health and the other was English language and literature in their professions). Then back translation to English was done by another two individuals who could speak English (again, one was from public health and the other from English language and literature). The questionnaire was pre-tested on 50 pregnant women in Durame town, where the actual study population is culturally related.

Baseline data about sociodemographic and reproductive variables were collected at the household level during enrolment via face-to-face interviews. Ten trained midwives collected data and five public health professionals made supervisions. The data collectors at each health facility were assigned and the list of participants was given for each of them to collect outcome data. The outcome (uterine atony) and other clinical data were collected from clients’ charts, during the time of delivery before discharge [20].

Data were entered in Epi-data version 3.1 software and exported to R version 4.2.2 software for the analysis. For continuous variables, mean and standard deviation were calculated. For categorical variables, frequencies and percentages using cross-tabulation were calculated. For missing data, a complete case analysis approach was applied. Interactions for possible effect modifications were checked for the predictor variables. To identify determinants of uterine atony, an unconditional binary logistic regression analysis was done. All independent variables that showed a significant association with uterine atony in the binary unconditional logistic regression model at P < 0.20 were included in the multivariable unconditional logistic regression model. Then predictor variables with confidence intervals for odds ratios that did not include 1 and P < 0.05 were declared statistically significant determinants of uterine atony. To measure the impact of the determinants, attributable fractions (AF) and population attributable fractions (PAF) were estimated using the adjusted odds ratios (AOR) and percent of cases exposed as follows: AF = [(AOR-1)/AOR]*100; PAF = AF*% of cases exposed. Percent of cases exposed = the number of exposed individuals among those who had an outcome divided by the total number of individuals with the outcome multiplied by 100 [30, 31].

Two days training was given for data collectors and supervisors on the concept of the questionnaire and how to approach the participants ethically. A pre-test was conducted in Durame town, which has a similar socio-cultural context to the study setting. Supervisors checked the data collection process closely. To minimize selection bias, we used community-based recruitment to include pregnant women during the study period using predefined eligibility criteria. Additionally, ultrasound was used for those women who had difficulty remembering the date of their last menstrual period due to different reasons, such as contraceptive use and breastfeeding. Epi-data was used to control errors during data entry. The data were explored to check for outliers and missing values.

A total of 2578 pregnant women from the end of first trimester were followed-up until delivery. Of these, 29 (1%) were lost to follow-up, and their pregnancy outcomes could not be ascertained. The pregnancy outcomes were ascertained for the remaining 2549 study participants. One woman had a spontaneous abortion before 28 weeks of gestation. Hence she was not followed-up any more. The final analysis was done for the remaining 2548 study participants. Of them, 93 developed uterine atony, yielding the incidence of uterine atony 3.7% (Additional Fig. 1).

In this study, data were missing for age (n = 3) and age at first childbirth (n = 1). Missing data were not related to the outcome (uterine atony) since the missed variables were collected at baseline before the outcome status was ascertained.

The mean age of both cases and controls was 27.01 ± 3.06 years and 27.05 ± 3.43 years, respectively. Nearly similar proportions of cases (80.6%) and controls (82.3%) attended formal education (1–12 grade or above). Regarding religion, 93.5% of cases and 89.2% of controls were protestant. About 93.5% of cases and 90.6% of controls were Hadiya in ethnicity. About 60.2% of the cases and 59.9% of the controls had given birth two and more times (multiparous). Higher proportion of cases (71%) had a short inter-pregnancy interval than the controls (55.1%). Likewise, a higher proportion of cases (24.7%) had prolonged labor than the controls (8.6%) (Table 1).

In the binary unconditional logistic regression model, five variables: the inter-pregnancy interval, duration of labor, birth characteristics, gestational age at birth, and birth weight, were associated with uterine atony at P < 0.20. When these variables were fitted in the multivariable unconditional logistic regression model, three variables: the inter-pregnancy interval, duration of labor, and birth characteristics, were found to be associated with uterine atony with a 95% CI at P < 0.05.

In this study, women with a short inter-pregnancy interval (< 24 months) were twice (AOR = 2.13, 95% CI: 1.26, 3.61) more likely to experience uterine atony than women with an inter-pregnancy interval of 24–60 months. This means that among women who had a short inter-pregnancy interval, about 53% of uterine atony was attributed to the short inter-pregnancy interval (AF = 53.05%, 95% CI: 20.63%, 72.29%). In the study population, about 38% of uterine atony was attributed to the short inter-pregnancy interval (PAF = 37.61%, 95% CI: 14.63%, 51.25%). Women who had a prolonged labor were twice (AOR = 2.35, 95% CI: 1.15, 4.83) more likely to experience uterine atony than those who had no prolonged labor. Among women who had prolonged labor, about 57% of uterine atony was attributed to the prolonged labor (AF = 57.44%, 95% CI: 13.04%, 79.29%). In the study population, about 14% of uterine atony was attributed to prolonged labor (PAF = 14.18%, 95% CI: 3.22%, 19.58%). Women who have given multiple birth (twins) were three times (AOR = 3.46, 95% CI: 1.25, 9.56) more likely to experience uterine atony than those with singleton birth. Among women who had multiple birth (twins), about 71% of uterine atony was attributed to multiple birth (AF = 71.09%, 95% CI: 20%, 89.53%). In the study population, about 6% of uterine atony was attributed to multiple birth (PAF = 6.11%, 95% CI: 1.72%, 7.69%) (Table 2).