Deferred and referred deliveries contribute to stillbirths in the Indian state of Bihar: results from a population-based survey of all births

The Indian government adopted a target of < 10 stillbirths per 1000 births by 2030, the first ever national stillbirth prevention target, in 2014 as part of the India Newborn Action Plan (INAP) [1]. This is an important step in the recognition of stillbirth as a public health concern as India has the largest burden of stillbirths globally with an estimated 590,000 stillbirths in 2015 [2].

We have previously reported a stillbirth rate (SBR) of 21.2 (95% CI 19.7–22.6) per 1000 births for the Indian state of Bihar in 2012–2013, which were the first state-wide robust population-based estimates of stillbirth from India given the significant underreporting of stillbirths in the Sample Registration System and the Demographic Health Surveys [1, 3, 4]. We reported on the risk factors for stillbirths and also highlighted the missed opportunities in the health system to reduce stillbirths [3]. However, as no comparative data were available on livebirths in the previous survey, hence, the acute need remained to better understand risk factors for stillbirths among all births that could further assist with reduction of stillbirths [3]. We now address this gap by reporting detailed epidemiology of stillbirths from a population-based survey of all births wherein the risk factors were documented for each birth irrespective of the birth outcome from Bihar state. This Indian state, with a population of 104 million in year 2016, is the third most populous state, with 11% of it being urban [5].

A total of 23,940 births in year 2016 were identified during the enumeration from 182,486 households (96.2% participation) covering a population of 945,216. Three hundred and thirty-eight stillbirths were identified giving an estimated SBR of 15.4 (95% CI 13.2–17.9) per 1000 births for the state. The SBR for boys was 15.7 (95% CI 12.7–19.2) and for girls was 14.6 (95% CI 11.5–18.5) per 1000 births.

A significantly higher proportion of births was stillborn (8.6%, p < 0.001) among women (175, 0.9%) for whom the delivery was deferred (Table 1). Of these women, 143 (81.7%) mentioned that they were sent back home as the health provider informed them that “there was still time for delivery”. The proportion of deferred deliveries (Additional file 1) was significantly higher in the private facilities (2.1%, p < 0.001) as compared with public facilities or home deliveries, and in breech position (2.8%) as compared with normal position deliveries (0.8%, p < 0.001).

Birth weight measurement was reported only for 40 (14.5%) of stillbirths (Table 1), and women did not know if birth weight was taken in 219 (84.6%) stillbirths. Considering the size of baby at birth, 42.2% the stillborns were reported to be of an average size, 37.5% as small/very small, and 10.2% as larger than usual in size. For the 703 (3.5%) births where the mother was informed during pregnancy that baby was not growing adequately (Table 1), 41 (5.8%) were < 2.0 kg, 474 (67.4%) were ≥ 2.0 kg and 105 (14.9%) were not weighed at birth.

A total of 827 (5.9%) among the 14,106 women who had a facility delivery were considered as “referred” (Table 1). These referral births were significantly more likely to be stillborn (5.6%, p < 0.001) as compared with non-referred births (0.9%). Considering these referral births, the prior facility was public facility for the majority (687, 83.1%) followed by private facility (115, 13.9%), 16 (1.9%) had gone to both types of facilities, and 9 (1.1%) elsewhere. A varied referral pattern was seen for livebirths and stillbirths (Fig. 1). A significantly (p < 0.001) higher proportion of referrals between private facilities was seen for stillbirths (38.9%) than for livebirths (8.4%); however, a significantly higher proportion of referrals from private to public facility was seen for livebirths (29.9%) than for stillbirths (10%). The major reasons for moving between facilities for delivery were (not mutually exclusive) the health provider asked to go to another facility (64.6%), the health provider refused treatment (23.3%), long wait at the prior facility (14%), and the woman did not like services at the prior facility (11.6%). These reasons were similar for livebirths and stillbirths except that among those with stillbirths, 8.7% mentioned “the health provider had refused to deliver dead baby” as the reason to move to another facility. Considering all births (Additional file 3), nurses delivered 45.1%, untrained Dai 20.1%, and doctors 15.9% of the babies, with the proportion of stillbirths significantly higher in births delivered by others (p < 0.001; untrained and unskilled people). Among the facility births (Additional file 3), the proportion of deliveries by doctors was higher in the referred than non-referred births (55.5% vs 19.9%) with proportion of stillbirths significantly higher in the former (6.54%; p < 0.001). Similar results were seen for deliveries by a nurse at a facility based on referral (p < 0.001; Additional file 3). When considering only facility deliveries and inclusion of referral deliveries as a risk factor in the sequential logistic regression model (Additional file 4), the referred deliveries were 3.32 times more likely to result in a stillbirth. The referred deliveries were explored further to understand the role of breech position of the baby, and push/ forceful pull performed during delivery by the health provider by place of delivery. Significantly high odds of stillbirth were seen with referred deliveries that were at public facilities and with breech position of the baby (Additional file 5).

Open image in new window

Figure 2 summarises the classification process for the identification of antepartum and intrapartum deaths. For the 239 (86.9%) stillborn for whom data both on the baby’s movement felt by mother before delivery and skin maceration were available, 128 (53.6%) were considered as antepartum (5.6, 95% CI 4.3–7.2 antepartum stillbirths per 1000 births), 96 (40.2%) as intrapartum deaths (4.5, 95% CI 3.3–6.1 intrapartum stillbirths per 1000 births, and the rest (6.2%) could not be classified. The reported mismatch between the baby’s movement and appearance was more pronounced for antepartum deaths wherein fresh appearance was reported for 60.2% of the cases (Fig. 2). The distribution of stillbirths with gestation period of 7 months was 14.1% and 6.2%, with 8 months was 21.9% and 8.3%, and with > 8 months was 64.1% and 85.4% among antepartum and intrapartum stillbirths, respectively. Boys accounted for 56.3%, 55.2%, and 58.8% of antepartum, intrapartum, and non-classifiable stillbirths, respectively (p = 0.915).

Open image in new window

To our knowledge, this study is the first large-scale state-wide population-based assessment of all births to facilitate understanding of epidemiology of stillbirths by type in India. The stillbirth rate was 15.4 per 1000 births for Bihar state with no statistically significant difference between boys and girls. In addition to demographic and clinical risk factors for antepartum and intrapartum stillbirths, this study has documented the increased risk of stillbirths in deferred and referred deliveries highlighting aspects of health care around labour and delivery that need attention in addition to improved skills of health providers to reduce stillbirths.

The smallest babies, in terms of gestational age and birthweight, are known to be at the highest risk of stillbirth [18, 19]. In this study, gestation period of ≤ 8 months was the strongest risk factor for both antepartum and intrapartum stillbirths, and particularly high for antepartum stillbirths. A total of 2.9% of all babies and 29.1% of stillborn babies had gestation period of ≤ 8 months in our study. As the pregnancy length was captured in months in our study, it is not possible for us to comment on whether the babies with gestation period of ≤ 8 months were very or moderately pre-term [9]. Pre-term labour is considered to be a syndrome initiated by multiple mechanisms, and more understanding in classification of pre-term and of its risk factors is needed [20, 21, 22]. Our data do not allow us to distinguish between pre-term labour as an increased risk for stillbirth or if a condition that caused stillbirth precipitated the early delivery, but these highlight that in addition to improving data on gestational age, differentiating spontaneous and medically induced pre-term births is of relevance to INAP to reduce adverse birth outcomes with rising caesarean section rate in India [23].

Antepartum deaths accounted for half of the stillbirths in this population, as reported previously and elsewhere [3, 18]. The women with deferred births were fivefold more likely to have an antepartum stillbirth. The major reason for them to be sent back from the health facility was that the health provider thought there was still time for delivery. Also, women who reported spontaneous labour and foul smelling discharge were significantly more likely to have antepartum stillbirth. We did not document why these women thought they were ready to deliver when they first visited the health facility for delivery. We did not document the frequency of foetal movements for all births or whether these were checked by the health provider before deferring the delivery; only the absence of foetal movements in stillborn was documented. It is well established that reduced foetal movement is associated with increased risk of poor pregnancy outcomes such as foetal growth restriction and stillbirth [24, 25, 26] and that maternal perception of decreased foetal movements is associated with adverse pregnancy outcomes, including stillbirth [26, 27]. Also, a sudden increase in vigorous foetal movements indicating foetal compromise possibly due to an hypoxic-ischaemic insult is also reported [26]. Monitoring of foetal movements at every antenatal care visit and teaching the pregnant women how to monitor foetal movements are recommended in the Indian health worker guidelines for delivery care [28]; however, not much is known about implementation of these in practice. To strengthen surveillance mechanism to prevent stillbirths in INAP [1], further research is needed to understand what methods to monitor are appropriate, what the women are informed about the foetal movements by health providers [29, 30, 31, 32], the process of foetal heart rate monitoring in medical records at health facilities, and on what basis the health providers defer a delivery.

Foetal growth restriction is known to be associated with antepartum stillbirth [33, 34, 35]. Poor foetal growth and baby small-for-gestational age at birth were associated with antepartum stillbirth in the INTERGROWTH-21 Project [16]. We did not find a significant association of stillbirth with women being informed that the baby was not growing adequately inside the womb. We are unable to comment on birthweight as most of the stillborn babies were not weighed at birth as also reported by us previously [3]. However, of importance to note in this study is that only 23.7% of the livebirths were not weighed as compared with 84.6% of the stillbirths. Furthermore, the size of the baby in this study did not necessarily corroborate with birthweight as nearly 70% of the babies who were reported to be not growing adequately inside the womb weighed ≥ 2 kg at birth. Although the Indian government guidelines on the newborn care facilities include weighing of the newborn at birth [36], it is clear that those stillborn are not weighed. Also, weighing newborn is currently not included in the health worker guidelines for delivery care [28]. As the INAP envisages strengthening surveillance mechanism for stillbirths [1], it is imperative for these guidelines explicitly state that every newborn should be weighed irrespective of the baby’s live status at birth.

Pregnancies with no ANC were associated with a higher risk of antepartum stillbirths and with borderline significance for overall stillbirths. ANC is meant to deliver maternal interventions to reduce adverse outcomes by preventing, identifying, or treating infections, pregnancy-induced conditions, and undernutrition [33, 37, 38]. Syphilis and malaria are among the largest contributors to stillbirths of infectious causes in the low-income countries [39, 40, 41], whereas placental conditions, pregnancy-induced hypertension, and congenital anomalies contribute to the antepartum deaths globally [40, 42]. Considering all births in our population, syphilis, malaria, and pregnancy-induced hypertension were reported only by 0.4%, 2.3%, and 2.6% women. However, based on our experience, it is important to note here that negative reporting does not necessarily mean that these conditions did not exist. It includes a variety—women who were checked and informed that they did not have a given condition, those who were checked but not informed whether they had or did not have the condition, women who were not checked for these conditions, and women who did not know if they were checked for these conditions. Among the 81.2% women who reported at least one ANC visit in this study, 48%, 46.9%, and 31.1% were not asked to give blood sample, give urine sample, or have their blood pressure checked even once, respectively. All these point to inadequacy of antenatal care, and significant reduction in antepartum stillbirths is unlikely in India unless INAP can ensure improved quality of ANC services to deliver maternal interventions in addition to improved coverage.

We found evidence of sub-optimal quality of care around labour and delivery as babies presenting in breech position, pregnancies with multiple foetuses, and deliveries with “push and forceful pull” performed during the delivery by the health provider had a significantly higher odds of stillbirth, both for antepartum and intrapartum stillbirths. Furthermore, antepartum stillbirth was more likely to be delivered vaginally. Emergency obstetric care including caesarean section is highly effective in reducing stillbirths [33, 43, 44]; however, as we have previously highlighted, some health providers in obstructed deliveries and/or prolonged labour continue to wait for a vaginal delivery instead of opting for a timely caesarean section that could possibly reduce the chance of stillbirth in some cases [3]. Better quality of care during labour and childbirth in deliveries with complications is likely to result in the highest number of stillbirths averted [37]. With intrapartum stillbirths accounting for 40% of all stillbirths in this population, improving quality of care during labour and childbirth as proposed in the INAP should facilitate notable reduction in intrapartum deaths [1].

Place of delivery was an important risk factor in determining stillbirth as an outcome. Though the proportion of home deliveries in this population has declined significantly by 20.5% over the last 5 years [7], 30% of the women still delivered at home with these deliveries having significantly higher odds of stillbirths. As the home deliveries conducted by untrained birth attendants (such as untrained Dai, family member, and friends) had a higher risk of stillbirth as compared with a skilled birth attendant, it is important to understand birth preparedness in addition to encouraging safe facility deliveries [45, 46, 47]. The private sector facilities also had a higher odds of stillbirth than public sector facilities, antepartum in particular. As highlighted by this study, this finding needs to be interpreted within the context of referral of deliveries from one facility to another, both within and between public and private facilities [48]. In general irrespective of the place of delivery, the high odds of stillbirth for deliveries with breech position of the baby and those that included push/forceful pull of the baby by the health provider highlight the need for skill-building of health providers to not only identify and manage complications, but also to do timely referral [1, 28, 49]. Majority of the deliveries with breech position of the baby (75.4%) were referred by one health provider to another in this study. Also, pregnancies with multiple foetuses had significantly higher odds of stillbirth than singleton pregnancies [16]. We did not document the time of death of the babies, but we have previously highlighted in this population the reluctance of health providers to deliver a dead baby which also results in referral for delivering a dead baby [3]. Poor quality obstetric referral services, higher odds of adverse birth outcomes in referral deliveries, and low competence of health staff providing emergency obstetric care in India have been documented previously [48, 49, 50, 51, 52]. INAP recognises the shortage of appropriately trained human resources as a major bottleneck in improving quality of care in addition to poor quality of care, and recommends effective referral mechanisms for complicated deliveries [1]. As indicated by these findings, improved monitoring and evaluation practices are necessary along with better understanding of how referrals work to inform implementation of effective referral mechanisms [48, 53, 54]. With private sector dominating provision of emergency obstetric care in India [55, 56], INAP needs to encourage partnership with private sector for improved quality of care in addition to enforcing regulations on private sector [1, 57].

The finding of greater risk of stillbirths with higher maternal age and in first born has been reported from many developing countries [18, 58, 59]. Boy babies are reported to at a 10% higher risk of stillbirths than girl babies [60], and 62% of stillborn were boys in our previous assessment which included only stillbirths [3]. Though boys accounted for 56.4% of all stillbirths in this study, we did not find the SBR to be different for boys and girls when considering all births. It is important to note that the odds of stillbirth were significantly higher for boys in logistic regression until model 4 and then lost statistical significance in the last model on including delivery-related risk factors. This finding needs to be explored further. Solid cooking fuel use had a slight but significant association with intrapartum stillbirth, and such an association with stillbirth has been reported recently from rural communities including India [61].

There are some limitations to the study findings. As is the case with surveys, the findings should be interpreted within the context of recall bias of the respondent. The gestational age was captured in months instead of weeks as the pregnancy length in India is reported in months. The last menstrual period forms the basis for most and is considered a reliable estimate for measuring gestational age in both developing and developed country settings [62, 63]. To classify stillbirths as antepartum or intrapartum, a previously used cut-off of within 8 h since the last felt baby’s movement was considered [3, 10, 11, 12], and we preferred the baby’s movement over the description of stillborn baby because appearance is reported to be a less accurate proxy for death-to-delivery interval [11, 13, 14]. If we had given preference to stillborn baby’s description over movement, 51 (18.5%) cases would then be classified as antepartum and 96 (34.9%) cases as intrapartum deaths. INAP should encourage documentation of accurate time of delivery and that of the baby’s demise where possible in medical records at health facilities to differentiate antepartum and intrapartum stillbirths, and should establish mortality audit to address breakdowns in clinical care or processes to strengthen stillbirth surveillance [64].

There are several strengths of this study. Some previously available population level reports on stillbirths have been either been on stillbirths only, case-control in nature, small sample size, and/or from a select city/area [40, 65, 66]. This study’s strength is the large-scale state-wide representative data on all births that allows for detailed understanding of epidemiology of stillbirths by type. We aimed to generate robust estimates for all births and stillbirths by documenting all in/out migration among the reproductive age women who had a pregnancy outcome in the period of interest that provides an appropriate denominator for the stillbirth estimation. We strengthened the numerator for these estimates by confirming stillbirth at three points in time from enumeration through the analysis by confirming signs of life, which is different from the DHS surveys which do not confirm signs of life from the respondent when documenting a stillbirth [1, 3]. We used a hierarchical approach to the risk factor analysis that gave importance to distal determinants of stillbirths, and have presented results in a manner that allows for better understanding of association of various risk factors of interest [16].

On comparing the SBR in this study with that in 2012 [3], a reduction of 27.4% (95% CI 12.9–41.8) is seen in Bihar over 4 years. As we do not have comparative data on the coverage of services for these years on all births in 2012 survey, it is difficult for us to comment on the possible reasons for this reduction. However, comparing data on livebirths in the same population from this study and a previous survey in 2011 [7], there has been a significant increase in coverage of antenatal care and public sector delivery (10.2% and 18.8%, respectively; z test p < 0.001) and decrease in home deliveries (20.5%; z test p < 0.001) in this population. Improvements in intrapartum practices through nurse mentoring are also documented in the Bihar Technical Support Programme [67]. These could have possibly played a role in stillbirth reduction in this population.