Risk factors for stillbirths: how much can a responsive health system prevent?

Stillbirth is an unfavourable outcome of pregnancy, which is still prevalent in many countries despite remarkable efforts to improve the care of pregnant women. Around 2.6 million third trimester stillbirths occur worldwide, out of which nearly 98% are reported from low and middle income countries (LMIC) [1]. The stillbirth rate (SBR) in India is 4/1000 births [2]. These rates are likely to be underestimates, as data on stillbirths are difficult to capture [2, 3].

Several risk factors for stillbirths have so far been documented. Maternal factors, such as advanced maternal age, teenage pregnancies, maternal nutritional status, history of prior pregnancy losses, complicated pregnancies [4] and multiple pregnancies increase the risk of stillbirths. Poor socio-economic conditions have also been found to be associated with stillbirths [5‐9]. In LMICs, quality of health care is an important risk factor for stillbirth, alongside other factors. Lack of access to antenatal care services, quality of care during childbirth and delayed caesarean sections are common risk factors [10].

In India, most data on stillbirths are from hospital-based studies. Population-based data are not collected routinely and those collected are subjected to gross under-reporting and selection bias [11]. Bihar is a high-priority Indian state with a significant burden of maternal and child mortality. The routine health data collected through the Health Management Information System (HMIS) collates information from health facilities across the country. In 2015, the HMIS reported an SBR of approximately 2% for institutional deliveries in the state [12]. However, the Sample Registration System (SRS), a geographically representative population-based system of collecting data on vital statistics, reported the SBR in Bihar as 1 per 1000 births in 2015 [2].

Studies conducted in LMICs suggest that a sizeable proportion of stillbirths may have occurred during the intra-partum period, indicating that improved quality of care during childbirth and caesarean section can help prevent these stillbirths [13]. Moreover, there are anecdotal reports from certain parts of India on the administration of unwarranted use of oxytocics to augment labour, even in home deliveries [14, 15]. Studies have shown that oxytocin is used in most instances while some also administer misoprostol [16, 17]. A small pilot study conducted by the same group of authors in a neighbouring district of the same state found that nearly 20% of the women received more than two doses of such medicines to augment labour before reaching the health facilities for delivery.

For this study, the objective was to investigate the association of potential risk factors including various socio-demographic and health system factors, with special reference to the practice of administering two or more doses of oxytocic medicines to augment labour, with stillbirths in two districts of Bihar.

This study was conducted in two UNICEF priority districts of Bihar -Gaya and Purnea- between July and August 2015 using a case-control study design. In these two districts, 23 blocks were chosen (16 from Gaya and 7 from Purnea) based on the number of stillbirths reported in HMIS from facilities. In order to minimize the extent of recall bias, the enrolment of participants was limited to the previous six calendar months.

The interviewers screened a total of 421 cases of stillbirths prior to the interviews, from both districts based on the line lists available so as to meet the sample size. Cases were confirmed as stillbirths (based on signs of life immediately after birth) during the screening; it was observed that 21 (5%) listed stillbirths were actually early neonatal deaths, falsely reported as stillbirths. These were excluded from the study. Thus, data were collected from 400 confirmed stillbirths and 800 controls.

In the present population-based case-control study, untreated hypertension during pregnancy, longer than 10 min taken to attend to pregnant women when they reach the facilities for delivery and presence of any complication during labour that warrants treatment emerged as significant modifiable health system risk factors for stillbirths. The other risk factors included increased maternal age, preterm births, primigravidae, exposure to active or passive smoking, and history of injury or physical violence during pregnancy.

Stillbirths is a sensitive marker of the strength of the health system [21]. In Indian settings, stillbirths have customarily been under-reported, given the lack of consistency in the definitions and lack of demand for data surrounding stillbirths [22]. Additionally, early neonatal deaths are falsely reported as stillbirths; in a study conducted elsewhere in India it was found to be 15% [18]. Irrespective of settings, the largest proportions of stillbirths are antepartum deaths [23]. In India, the current reporting system for stillbirths does not include their categorization as fresh or macerated. As a result, it is difficult to differentiate the antepartum stillbirths from intrapartum stillbirths. Moreover, due to the retrospective nature of our study design, it was difficult to elicit whether the stillbirths were antepartum or intrapartum, hence they were considered as one group in this study.

Identification of high-risk pregnancies during antenatal period can facilitate timely initiation of therapy and referral to the appropriate facility to prevent or minimise antepartum deaths [24]. In our study, the coverage of antenatal care was less than 60% even among live births, thereby plummeting the chances of early identification and management of high-risk pregnancies. It is suggested that four antenatal visits may not positively impact the outcome of pregnancy until the coverage is at least 60% [21]. Women with high blood pressure in pregnancy were almost three times more likely to have stillbirths. In our study, 52% of women with high blood pressure among cases reported having consumed antihypertensive medications while it was 62% among controls, although compliance to advice and medication would be a concern in both groups. Similar findings have been reported from other studies [25, 26] suggesting that early diagnosis and adequate medical intervention for hypertensive disorders can help reduce the risk of stillbirths.

Exposure to active or passive smoking led to an almost 2-fold risk of having a stillborn baby in our study, similar to existing evidence that maternal smoking was associated with a 47% increase in the odds of stillbirth [27]. Mothers of stillbirths were more likely (p = 0.002) to have been the victims of physical violence during pregnancy, which was similar to other studies [28, 29]. Since this information is quite sensitive in nature, there is a likelihood of under-reporting [30]. This may have resulted in a non-differential misclassification of the exposure, meaning that the actual strength of association might be more than what is reported in this study.

The optimal caesarean section (CS) rate to reduce the proportion of intra-partum stillbirths lie between 5 and 10% [31]. Evidence from LMICs shows that with every percentage rise in CS rate from 0 to 8%, intra-partum stillbirth drops by 1.6 per 1000 births [31]. The rate of CS in our study was <1%, far below the optimal rates. Studies have also demonstrated that good quality obstetric care, and not mere provision of CS services, is critical [32]. In a recent analysis, the maximum benefit was noted from appropriate care during labour and childbirth, including complications (70% of all stillbirths averted), followed by enhanced antenatal care focussing on detection of complications (21%) [33]. The proportions of instrumental deliveries among the cases and controls in our study were 6% and 1.3% respectively. There may be instances in which instruments might have been applied to expedite the delivery in a confirmed case of stillbirth. Hence, it is difficult to interpret the association between instrumental deliveries and stillbirths.

In our study, mothers who delivered prematurely were 4.5 times more likely to have a stillborn child, as compared to full-term deliveries. Preterm births and low birth weight babies increase the risk of stillbirths manifold [34]. A study highlighted that spontaneous preterm birth or preterm ruptured membranes were the key factors in 41% of stillbirths less than 28 gestational weeks [35]. This is in sharp contrast to our study where more than 75% stillbirths occurred between 28 and 37 weeks of gestation, although we took 24 weeks as the cut-off to identify stillbirths. However, it is not known how many stillbirths could have been averted through public health interventions like administration of antibiotics for preterm, premature rupture of membranes and antenatal corticosteroids for preterm labour [36].

In our study, administration of two or more doses of oxytocics to augment labour was associated with stillbirths in the unadjusted analyses. More than 35% mothers of stillborn children and 25% mothers of live born babies reported the use of such medications before reaching the facilities. Previous studies from India have shown that unwarranted use of oxytocics increases the risk of stillbirth, similar to our finding when considered independently [14, 16, 17, 37]. Oxytocics are usually administered only for a valid indication in facilities with provision for C-Sections [38]. Use of such oxytocics only to hasten the process of delivery without a valid indication and medical supervision may be detrimental. Nevertheless, rational use of oxytocics in health facilities along with regular intrapartum monitoring can positively favour obstetric outcomes.

Substandard obstetric care contributes to 20–30% of all stillbirths [35]. Reaching a facility within 30 min, initiation of management within 10 min of reaching the facility and timely management of complications during labour could have prevented 17%, 37% and 20% of stillbirths respectively, according to our analyses. It is comparable to the estimated reduction in stillbirths by 25% with 60% coverage of basic interventions for skilled obstetric care [36]. This is however true only for prevention of intra-partum stillbirths. In our study, nearly 21% of mothers of cases and 48% mothers among controls admitted that they felt the last fetal movement during labour implying that more cases could have been antepartum stillbirths.

The study had its own set of limitations. Birth weight is an important parameter, since management of fetal growth restriction can avert 20% of antepartum and intra-partum stillbirths [36]. However, weighing of stillborn babies is not yet practiced in India [18]. Stillbirth reporting has improved through routine HMIS, however recording of all birth outcomes using comparable definitions, ante-partum versus intra-partum stillbirths and birth weight are weak links within the system [29, 34]. This was observed in our study as well. Increased and improved data availability can provide a solution for targeted interventions where those would be most effective [1]. Also, since the cases in our study were chosen from lists available at healthcare facilities, those women who had a stillbirth at home and who never visited the facilities for any post-partum complications were missed out completely, since they would not be registered in the HMIS. This was a major limitation in a situation where one-third to one-fourth of the deliveries take place at home. In the absence of reliable data, we were not certain if home-delivered stillbirths are different from facility-based stillbirths. Home-based stillbirths may be less likely to exhibit complications compared to facility stillbirths. There is a likelihood of bias resulting due to differential misclassification, which can change the direction of the observed association. The sample size was calculated considering administration of more than two doses of oxytocics to augment labour as the principal exposure. For the remaining factors, exploratory analysis was conducted and hence may not have been adequately powered to establish an association.

Besides these, our inability to differentiate between antepartum and intrapartum stillbirths remains a major challenge. Since we are focussing on health system related factors, it would have been useful to analyse the two groups separately. Paucity of this information might have led to undermining of the actual effect. From the reported information on the instrumental deliveries, it is difficult to ascertain if instruments were used more often to expel stillborn babies or these were used to deliver live babies who were later declared as stillborns when the babies were found dead. Approximate time to reach the facilities and time to initiate management after reaching the facilities were self-reported by mothers/family members. This could have resulted in recall bias; we tried to minimize this bias by keeping the interval between the event and the interview to less than 6 months. However, this might have affected both the groups leading to non-differential misclassification and hence the actual strength of association could be more than what is inferred from this study.

Since autopsies or other tests were not performed on stillborn babies, it is difficult to establish any causation. This poses a hindrance to understanding the cause of death, and increases the proportion of unexplained deaths [35]. Use of a validated verbal autopsy tool to collect data could address this lacuna to some extent. The tool has reasonable validity in identifying and discriminating between causes of stillbirth [19, 39, 40]. Any history of medical complication during pregnancy in our study was corroborated from available records. In our setting, availability of medical records was a challenge, thus resulting in probable reporting bias. However, our study focussed on risk factors and not causes of stillbirths. Besides, this study though representative of the state, focussed mainly on rural populations.

Despite these limitations, the study highlights the importance of basic and advanced antenatal and intra-natal care for averting stillbirths. Conducted on a large sample and in household settings, the study captured existing factors related to obstetric care. This work is one of the few exploratory studies which have used routine health information system data to identify stillbirths from the community in this setting. Yet another strength of this study was the ability to relate the health-system related risk factors with stillbirths.

To conclude, a strong health system with a good coverage of basic antenatal care and timely institution of emergency obstetric services can avert a significant proportion of stillbirths in similar settings. A good quality antenatal care package can improve compliance, and detect complications early during pregnancy for timely action. Certain unwarranted complications that may occur during the last trimester of pregnancy can also be adequately addressed with appropriate measures taken as soon as women reach the health facility. Stillbirths are adverse pregnancy outcomes, most of which can be averted by a pro-active and strong health system through appropriate obstetric practices. Judicious use of oxytocics and other medications during labour can improve pregnancy outcomes. Additionally, mechanisms to improve routine surveillance of stillbirths to detect contributory factors and the reporting of stillbirths including information on birth weight and fresh vs. macerated stillbirth would contribute to strengthening the health information system.