This was a hypothesis-generating study to identify factors associated with stillbirth which could be amenable to modification to reduce stillbirth. We were able to determine factors associated with stillbirth in Zimbabwe in a sample of adequate size. Some of these factors were then validated in a sample of consecutive births taken from the birth register. This approach has enable us to identify some risk factors that have been reported previously in low-resource settings, such as lack of antenatal care, increasing maternal age, and presence of maternal medical complications. It has also identified risk factors which have not been widely reported in LMICs, including previous stillbirth and seasonality. This study has also demonstrated that the effect of some risk factors varied between different studies and locations (such as the effect of HIV infection).
Strengths and limitations of this study
This study was strengthened by data collection from a large cohort of women by clinically trained staff who were familiar with local documentation and practice which enabled detailed information to be obtained. However, data acquisition could not keep pace with the number of births which could have introduced selection bias although notably the demographics characteristics of both cohorts were similar. Data collected were reliant upon information recorded in the antenatal and intrapartum case notes which was sometimes missing; the presence of data may be negatively influenced by disclosure of conditions which may be stigmatising (e.g. syphilis, previous stillbirth). The inclusion of local maternity staff in the study enabled nuanced discussions about the relationship between confounding factors which informed the directed acyclic graph (DAG) and the subsequent multivariable analysis.
The use of the DAG also enabled the researchers to identify limitations in our model. For example, our theoretical DAG identified socioeconomic status and cigarette smoking as potential confounding factors, but only limited data were available for these variables which meant they could not be accounted for in our model which may have introduced bias into our results.
Regional Context
The factors associated with stillbirth here are in agreement with those described by Aminu et al. in a systematic review of 142 studies of risk factors for stillbirth in LMICs, 49 of which were from Africa.4 Due to variations employed in source studies and risk factors studied formal meta-analysis was not possible. In their narrative synthesis, the authors identified lack of antenatal care and previous stillbirth, which were amongst the largest independent effects in our study.4 Notably, these effects are not unique to LMICs. Systematic reviews and meta-analysis of observational studies demonstrate a relationship between stillbirth and increasing maternal age, with the greatest effects seen in women ≥40 years of age.11 A meta-analysis of studies from HICs found that women who have a history of stillbirth were more likely to have a stillbirth in a subsequent pregnancy.12 The fact that these associations are present irrespective of setting suggests that they have their origins in human biology. However, their effect sizes vary which indicates that other local sociodemographic characteristics may moderate this increased risk. For example, the effect size of having a previous stillbirth in Zimbabwe was greater than reported in HICs. This may be because women with a previous stillbirth were more likely to live in a rural location and have less access to antenatal care.
Comparison with prior studies of stillbirth and perinatal death in Zimbabwe reveals variation in the stillbirth rate between different regions and time-periods ranging from 1.7 to 6.1%.13-16 This could be attributable to various factors including changes in economic prosperity over time or differences in the urban/rural mix of population between sites. These studies report variation in factors associated with stillbirth. Crowther reviewed 53,665 births in Harare in 1983, of which 1,204 were stillbirths, 17.0% of stillbirths were macerated of unknown cause, 14.0% were attributed to intrapartum asphyxia, 5.4% were associated with hypertensive disorders of pregnancy and 8.1% with antepartum haemorrhage.14 Aiken reviewed 466 stillbirths at Mpilo Hospital from 1989-1990, describing causes of stillbirth were congenital syphilis (21.7%), birth asphyxia (23.8%), unexplained stillbirths (21.5%), congenital malformations (7.3%), pregnancy-induced hypertension (9.9%) and placental abruption (8.8%).13 A review of women with HIV infection conducted at the same time found 15% of women who had a stillbirth were HIV positive; HIV mothers had more stillbirths associated with syphilis and congenital infection.17 Feresu et al. examined 985 stillbirths and 17,174 live births in Harare in 1997-1998 using a threshold of 20 weeks’ gestation and 500g to define stillbirth; this study found maternal age ≥35 years, rural location and women who were unbooked for antenatal care were associated with stillbirth, in this population nulliparity was protective.15 A population-based survey conducted in 2006-7 undertook verbal autopsy in 11 areas, one of which was near Bulawayo; 1296 stillbirths from 45,023 live births were reviewed. This report identified that maternal disease was thought to be causal in 7.8% of stillbirths.16 Lastly, an interview study of 103 cases and 206 controls conducted in 2009 in Mashonaland found lack of education, labour complications, home birth, HIV infection and low birthweight (<2.5kg) were associated with stillbirth.18 This region has a high proportion of people from the Apostolic church who have little engagement with medical services. These studies agree that poor access to maternity care, medical complications (particularly hypertension) and labour complications are important risk factors for stillbirth in Zimbabwe as they have been consistently observed over time.
Importantly, we observed a lower proportion of intrapartum stillbirth than reported previously which may have been due to a focus on reducing term stillbirth at Mpilo hospital in 2017;19 the levels reported in our study were consistent with these data from the preceding year in the maternity unit under study. We also did not observe an association between HIV-positive status and stillbirth reported in earlier studies.17,18 This may be because testing and antiretroviral therapy to reduce vertical transmission are embedded within contemporary maternity care in Zimbabwe.
A novel association with stillbirth in Zimbabwe described in this study was the association of stillbirth and birth in the cold or hot season compared to the mild season. A systematic review of 32 studies found that pregnancy length and birth outcomes were altered, particularly in summer and winter.20 The four included studies that examined stillbirth found higher rates of stillbirth in winter (cold) from studies in the Northern Hemisphere and with summer (hot) in Australia.20 One subsequent study examining the effects of seasonality in Nepal found the peak incidence of stillbirth was in January (cold). Thus, our findings appear consistent with other studies from the literature.21 Further research is required to better understand whether this association is independent, or whether it is mediated by behavioural changes e.g. use of indoor stoves/fuel, difficulty accessing maternity care or alterations in diet.
Clinical Implications
The strong association between prior stillbirth and subsequent stillbirth in our study population was particularly striking. Of the women who had a stillbirth 39/46 (85%) had a previous stillbirth. This percentage was higher than a study from Malawi which found 62.7% of mothers who experienced a perinatal death had previously had a perinatal death.22 Four other studies, three of which were conducted in Africa (Ghana, Nigeria and Zambia), found an association between stillbirth and a history of stillbirth. The crude effect sizes ranged from 1.94 – 5.7.23-25 The consistency of this observation suggests that it is robust and likely to occur in different LMICs. Recurrent stillbirth is particularly significant given the stigma and taboo that surround stillbirth in many societies; beliefs that stillbirth is the result of a curse or errant maternal behaviour are likely to be reinforced if a mother experiences recurrent deaths.26 However, women who have a history of stillbirth may represent a group for whom care can be modified. After a stillbirth has occurred, women could be counselled about the importance of attending for antenatal care from an earlier time point in a subsequent pregnancy. Regular attendance in maternity services could ensure adequate screening for syphilis and hypertensive disorders of pregnancy, and potentially administration of prophylactic aspirin which reduces the risk of perinatal death.27,28
Our findings, and those of earlier studies, emphasise the importance of antenatal care as unbooked women have a significantly increased risk of stillbirth as well as increased risk of maternal and neonatal mortality. Since 2018, antenatal care in Zimbabwe has been free at the point of care, removing one barrier to accessing maternity care. However, additional services may require additional payment. Improvements in intrapartum care, including increased skilled birth attendants and access to Caesarean section, may have reduced the proportion of intrapartum stillbirths in our study population. Access to evidence-based interventions in antenatal and intrapartum care should continue to be prioritised, as their implementation will reduce stillbirths, neonatal and maternal deaths achieving a triple return on investment.26