Predictors of Intrapartum Stillbirths among Women Delivering at Mulago Hospital, Kampala, Uganda

2.1. Study Variables

This study’s independent variables were socio-demographic characteristics, medical, obstetric, and family history that were compared among cases and controls. The main exposure variable was the proportion of participants with birth asphyxia among the cases and controls.

2.2. Study Design and Setting

This was an unmatched case-control study conducted from October 29, 2018 to October 30, 2019. Cases were women who had given birth to fresh stillbirths, while controls were women who had given birth to live babies. This study was conducted at Mulago hospital. Mulago Hospital is a National Referral hospital for Uganda and a teaching hospital for Makerere University College of Health Sciences. It is located in the central region within Kampala, which is both a commercial and capital city. Kampala has a total population of five million by day and about one and half million at night.15 Mulago hospital conducts about 32,000 deliveries per year. Most (60 %) patients are admitted as referrals from the surrounding health facilities (Kiswa HCIII, Naguru Hospital, Kisenyi HCIV and Kawala HCIII, Komamboga HC III and Kisugu HC III). On average, 50 intrapartum stillbirths occur per month to approximately 600 stillbirths per year, according to the 2016 departmental records.

2.3. Data Collection Technique and Tools

Data collection was conducted by four midwives. They were trained by the Principal Investigator about the purpose of the study and how to conduct the interviews using the data collection tool before the data collection process began. We used interviewer-administered questionnaires to collect the data. They had both open- and closed-ended questions. The questionnaires were in English and Luganda (which is a local language spoken by the women in the area of study). The women who understood English were given questionnaires written in English and the women who understand the local language were given questionnaires written in the local language. Women who did not understand any of these languages were excluded from the study.

The information collected included the socio-demographic characteristics, medical and family history, present and past obstetric performances. Socio-demographic characteristics included information on age, marital status, educational level, and distance from the hospital. Medical history included the history of medical diseases (e.g. hypertension, diabetes mellitus, renal diseases, malaria in the current pregnancy, anemia, urinary tract infection, and HIV serostatus) and obstetric history (e.g. parity, antenatal clinic attendance, obstetric complications, partogram use, fetal heart monitoring, mode of delivery, babies’ weight and sex.

The research assistants measured the blood pressure of women and collected information on their HIV status. Routine counseling and testing for HIV are done for all women who deliver at Mulago hospital. HIV status of all the women was extracted from the clinical notes. Participants were identified by study identification numbers on the case report forms. All the case report forms were kept in individual files in a secure filing cabinet in the study office. Additional records were kept in the clinical and laboratory record books at the study office in the labor ward.

2.4. Eligibility Criteria

The study included women aged 15-49. Cases were women who had delivered a stillborn baby at the gestational age of 28 weeks or more or weighed 1000g or more. Women were excluded as cases if they delivered a stillbirth in which fetal death occurred before the onset of labor, or if they delivered before arrival to the hospital, or if they delivered a baby with congenital abnormalities. Controls were women who delivered live babies after 28 weeks of pregnancy or more. Women were excluded if they delivered before arrival to the hospital or delivered a baby with congenital abnormalities.

2.5. Sample Size Estimation and Sampling Technique

The sample size was calculated using a formula for comparing two proportions.16 Using a proportion of asphyxia of 14.8% among the controls and 26% among the cases17 as was found in a study in Western Uganda, a sample size of 474 women (158 cases and 316 controls, in the ratio of 2 controls per case) was sufficient with 95% confidence level, power of 80% and odds ratio of worth detecting at least 2.

Research Assistants from the labor ward identified women with intrapartum stillbirth delivery. The Principal Investigator confirmed the diagnosis of intrapartum stillbirth. The women who had stillbirths were managed using the standard hospital protocol of managing stillbirth which involved determining the cause of death and helping women cope with bereavement, counseling, and support. The women were given information about the study and they were requested to participate. Consent was administered to the women who accepted to join the study. Recruitment of these participants was done consecutively until the sample size was achieved. For each identified case, two women with live birth were immediately identified by systematic sampling from the register of all the live births in the last 24 hours. The Principal Investigator checked the women for eligibility. They were given information about the study by the Research Assistants. If they accepted to join, then informed consent was obtained.

2.6. Data Management and Statistical Methods

The data collected was cleaned, coded, and double entered using EpiData version 3.1.18 We exported the data to Stata software version 14.0 (Stata Corp: College Station, TX, USA) for cleaning and analysis.19

Continuous variables were compared between groups using the independent sample t-test. Categorical variables between groups were compared using the Chi-squared test. Continuous variables were categorized and compared between groups as above.

Bivariable analysis was done to establish the strength of association between the maternal socio-demographic characteristics, medical and obstetric history, and the risk of stillbirth delivery.

Factors with a p-value of 0.2 or less at bivariate analysis and factors known a priori to be associated with stillbirth were entered into a multivariable logistic regression model and adjusted. The backward elimination method was used until a stable model was obtained. Results are reported as adjusted odds ratios with the corresponding 95% confidence intervals.

2.7. Ethical Approval

The study was approved by the Makerere University School of Medicine Research and Ethics Committee, the Mulago Hospital Research and Ethics Committee, and the National Council for Science and Technology in Uganda. The participants gave written informed consent. Only the participants’ identification numbers were used and no names were entered in the database.

3.1. Socio-demographic and Medical Characteristics of the Cases and Controls

A total of 474 women were included in the analysis, 158 as cases with an intrapartum stillbirth and 316 as controls without an intrapartum stillbirth. The cases were more likely to stay more than five kilometers from the hospital (61% Vs. 43%), P-value <0.001; to be referred (70% Vs. 41%), P-value <0.001 and have suffered trauma during pregnancy (2.5% Vs. 0.3%), P-value 0.020 than the controls. Other study characteristics are presented in table 1.

3.2. Factors Associated with Intrapartum Stillbirths at Bivariate Analysis

The factors associated with intrapartum stillbirth delivery were the distance from Mulago hospital, being referred by a health worker from a lower health facility to Mulago hospital, antepartum hemorrhage, malpresentation, prolonged labor, mode of delivery, and trauma during pregnancy (Table 2).

3.3. Predictors of Intrapartum Stillbirths

After controlling for other factors, referral to a hospital was associated with 2.4 times the risk of delivering stillbirth compared with women who were not referred. Women who developed antepartum hemorrhage were 6.2 times more likely to deliver stillbirth compared to women who did not have an antepartum hemorrhage. Women with malpresentation were associated with 5.5 risks of delivering stillbirth compared to women who did not have malpresentation. Women who had prolonged labor were 5.1 times at increased risk of delivering stillbirth compared to women who did not have prolonged labor. Women delivered by cesarean section were 7.7 times more likely to deliver a stillbirth than women who delivered vaginally. Finally, women who had trauma during pregnancy were 26 times more likely to deliver stillbirth than women who did not experience trauma in pregnancy (Table 3).

4.1 Limitations

This was a facility-based study that may not reflect the actual picture in the community. However, it has illustrated how the first, second, and third delays contribute to stillbirth delivery.