The period around childbirth is the one having the greatest lifetime risk of death.1 Of the 2.6 million stillbirths which take place in a year across the world, half are reported to be intrapartum.2 Almost 2.5 million babies die annually in the first month after birth, one-third of them during the first 24 hours (day 1).3 Over 95% of the stillbirths and neonatal deaths occur in low- and middle-income countries.2,4 In India, 15%–87% of the stillbirths are reported to be intrapartum5–7 and day-1 deaths comprise nearly one third of all the neonatal deaths.6–8 Most intrapartum deaths are associated with asphyxiating complications of labor and delivery.9 Birth asphyxia is the cause of about one fourth of neonatal deaths globally,10 also in India.6–8,11
Injections with drugs like oxytocin, also called oxytocics or uterotonics, are used for the induction and augmentation of labor. When given to a women before the birth of her baby, they are known to reduce placental perfusion and its oxygen supply12,13 and may increase the risk of stillbirth, birth asphyxia, and neonatal encephalopathy.13–16 Antenatal oxytocin administration requires close monitoring of fetal heart rate and uterine contractions, and should only be done in facilities where emergency cesarean section is available.17,18 In many developing countries, including India, there is widespread use of oxytocics before delivery even in settings where such monitoring and emergency obstetric care is not available.19–23 There is, to our knowledge, no study that has adequately estimated the full extent to which antenatal administration of uterotonics increases the risk of intrapartum stillbirth and early neonatal death.
We conducted a large randomized trial in 18 population clusters to evaluate the Integrated Management of Neonatal and Childhood Illnesses strategy in Haryana, India.24,25 While reviewing the verbal autopsies of stillbirth and neonatal deaths, we observed that more than half of the women, including those who delivered at home, received injections to initiate or augment labor, that is, uterotonic drugs. Hypothesizing that such injections could increase the risk of death around birth, we undertook a case–control analysis to estimate the extent to which antenatal uterotonic administration was associated with stillbirth or day-1 death. Our large sample size of 2076 cases and 532 controls enabled us to estimate separately the strength of such association for babies born at home, in private, and in government facilities. Additionally, we measured the association between such administration and the risk of intrapartum stillbirths and deaths on day 1, day 2, or day 3, and from day 4 to 7.
We conducted a case–control analysis of data obtained during an already completed cluster-randomized trial in the state of Haryana in North India.24,25 Our trial field staff visited all households in the study area every month to identify new pregnancies and inquire about the outcome of previously identified pregnancies. They also visited mothers of live-born babies on day 29 to ascertain whether the babies were alive. An independent team visited families with a stillbirth or neonatal death within 5 weeks and administered a standardized and validated verbal autopsy tool26,27 consisting of an open narrative and a structured questionnaire.28 We began data collection in January 2008 and completed it in March 2011.
Selection of Cases and Controls
We selected as cases the intrapartum stillbirths and live-born babies who died within 24 hours of delivery, that is, those who experienced an early death. A stillbirth was recorded when the family reported that, after birth, the baby did not breathe, cry, or move, even slightly. Our controls were neonates who died after day 7 (8–28 days), that is, those who experienced a late neonatal death, because information on administration of uterotonics was reported for stillbirths and neonatal deaths but not for babies who survived beyond the neonatal period. Although we would have preferred to include as controls all infants who survived the first day of life, that would have necessitated a prospective cohort design to capture uterotonic injections administered to induce or augment labor to all women who participated in the trial. Not only was the current study conceived after the trial was completed; it would also have been ethically unjustifiable to do such a prospective study without arresting inappropriate injections with uterotonics, thereby abolishing the exposure. Because administration of uterotonics during labor is unlikely to increase the risk of death in babies beyond the first week of life,7,11 we consider this proxy sampling29 strategy to yield controls with an appropriate exposure distribution.
We defined exposure as administration of an injectable uterotonic to the woman before delivery, that is, when she reported having been given an injection to increase the labor pains, or an injection after which pains increased, or if following an injection the baby was delivered within 2 hours. Healthcare providers who were reported to have administered these injections in public or private facilities and during home deliveries unanimously reported the injectable uterotonic to be oxytocin. This is consistent with other studies reporting oxytocin being the most commonly used uterotonic in India.30,31
We described the case, control, and exposure definitions in an a priori formulated analysis plan. For both cases and controls, the trial data also provided information on other potential confounders and effect measure modifiers. We defined a family as poor if it belonged to the lowest three quintiles using the methods employed by the Indian National Family Health Survey.32 Delivery was considered to be conducted by a skilled birth attendant when the mother reported that it was performed by a qualified doctor, nurse, or auxiliary nurse midwife. We defined a baby as preterm if the mother reported the baby to be born before its due date. Government health facilities included subcenters, primary health centers, and hospitals; private health facilities included clinics, small nursing homes, and hospitals.
An obstetrician and a pediatrician, neither of whom was aware of our hypothesis, independently reviewed the verbal autopsies for stillbirth. They coded the underlying cause and timing of stillbirth using the Causes of Death and Associated Conditions classification.33 They considered a stillbirth to be intrapartum when, after the onset of labor, a mother reported loss of fetal movements or the health worker informed her of the absence of heart sounds, and there were no signs of maceration. It was considered antepartum when loss of movements or fetal heart sounds was reported to have occurred before the onset of labor, and maceration may have been present. When sufficient information was not available regarding fetal heart sounds, fetal movements, and maceration, they categorized the timing as indeterminate. The two experts extracted information on administration of uterotonics from the narrative as given by the mother or caregiver. We compared the timing of stillbirth and administration of such drugs as assigned by the two experts. S.B.M. independently coded any disagreements pertaining to timing of stillbirth and administration of uterotonics. They were deemed resolved when her coding matched that of one of the experts. If the coding by these three experts did not match, we categorized the timing of stillbirth or administration of oxytocics as unknown. S.B.M. was unaware of the timing of stillbirth when she assigned such administration and vice versa.
Equally unaware of our hypothesis, two pediatricians independently assigned the cause of neonatal death and a third pediatrician reviewed any disagreements. They classified a day-1 death as being caused by birth asphyxia if the baby did not cry or breathe immediately after birth or was given assistance to breathe.27 For neonates who died later than 24 hours of being born, in addition to the above features, the presence of at least one of the following signs starting on or before day 3 had to be present for the death to be ascribed to asphyxia: having convulsions, being limp or flaccid, unresponsive, or unconscious.27 A pair of supervisors who were nonmedical graduates with over 5 years’ experience in field research extracted information on administration of oxytocics from the narrative, using the same definition as for stillbirth. S.B.M. reviewed any disagreements. The reviewers and S.B.M. were kept unaware of the assigned cause of death.
We conducted statistical analyses using Stata version 15.1 (StataCorp LLC, TX). We estimated odds ratios (ORs) for the association between uterotonic injection administered before or during delivery and early versus late death using logistic regression, and included the vce(cluster) option in Stata to account for dependencies between individuals within each of the 18 population clusters.
Oxytocin injections are only administered by care providers, and potentially confounding conditions and complications with causal associations to this exposure must have a pathway through the care provider—either through care-seeking behavior of the affected woman, or through the care provider’s response to the observed condition or complication. The presence of a skilled birth attendant, and place of delivery (at home, in a private or in a public health facility), therefore, represents the minimal sufficient adjustment set for our data and were, based on the corresponding directed acyclic graph (DAG; Figure 1), included as potential confounders in the final logistic regression model, which provided the adjusted OR (aOR) for uterotonic administration on early versus late death. The estimates from further analyses were adjusted for these two covariates except for that stratified on the place of delivery, where we adjusted for the presence of a skilled birth attendant. To explore any residual confounding, in which an unobserved quality of care providers would open a confounding pathway for clinical conditions and complications, we also designed an alternative DAG depicting the set of potential confounders that could be observed before or during delivery, and that could drive either care seeking behavior or response by care providers (eFigure in the eAppendix; http://links.lww.com/EDE/B681).
We calculated the percent population attributable risk (%PAR) from the aOR and the proportion of cases whose mothers had received such oxytocics (p) with the formula %PAR = 100 × p × [(aOR − 1)/aOR].34 In addition to our predefined intention of estimating the association between oxytocics and early death separately in the homes, in government, and in private birth facilities, we compared the strength of the association between administration of oxytocics and early death in male versus in female babies because boys are at a higher risk of asphyxia death.35 Likewise, we compared the aOR for oxytocic injections and early death between babies of nulliparous versus of multiparous women because there are indications that the former experience a higher risk of birth asphyxia and even asphyxia death.36 To quantify biologic interactions,29 we estimated the relative excess risks due to interaction37 with their 95% confidence intervals (95% CIs) using Stata’s icp command.
In a separate explanatory analysis of possible mechanisms by which uterotonic injections could exert their effect, while maintaining late deaths as controls, we separately estimated the aORs for intrapartum stillbirth, for death occurring during the first two and during the subsequent 22 hours of life, and during the subsequent 6 days. Furthermore, we used a generalized linear regression model of the binomial family with a log link, again taking the design effect of the population clusters into account, to estimate the extent to which antenatal administration of uterotonics predicted asphyxia. We then used logistic regression to estimate the extent to which asphyxia predicted early neonatal death in live-born babies. We concluded this analysis by including this possible mediator in our logistic regression model for the association between oxytocic injection and subsequent death in the first day of life. We then quantified any mediation by asphyxia in the causal pathway from oxytocics to death, as described previously.38
Ethics committees from Society for Applied Studies (SAS), Delhi and WHO approved the trial.24 All participants provided informed consent.
We recruited 77,587 pregnant women into the Integrated Management of Neonatal and Childhood Illnesses trial and determined the pregnancy outcomes in 67,348 (87%).24 There were 1499 stillbirths and 60,702 live births. We recorded the vital status on day 29 in 60,480 (99.6%) live-born neonates and recorded 2570 deaths. There were 60,938 (90%) single pregnancies, 507 of the women (1%) carried twins, and 9 (0.01%) carried triplets. We included early deaths, that is, the 838 (56%) intrapartum stillbirths and 1238 (48%) day-1 deaths, as cases, and the 532 (21%) late neonatal deaths as controls (Figure 2).
Birth asphyxia was the underlying cause in 712 (58%) of the 1238 day-1 deaths and this proportion declined progressively from birth to the end of the neonatal period (Table 1). Antenatal administration of oxytocics was reported in 74% of 1997 early deaths and in 62% of 517 late deaths, yielding an unadjusted OR of 1.8 (eTable in the eAppendix; http://links.lww.com/EDE/B681). Similar proportions were observed in 16 of the 18 population clusters (Figure 3).
In crude analyses, several other factors that occur before delivery predicted early as compared with late death: nulliparity, preterm delivery, and when the baby was a boy (eTable in the eAppendix; http://links.lww.com/EDE/B681). Some conditions that coincide with labor were associated with a higher risk of early versus late death, including malpresentation and fever on the day delivery began. Delivery by a skilled attendant and in a private facility was also associated with a higher risk. In the 1639 babies who did not have any of the conditions associated with early death, that is, normally formed singletons not born prematurely, with a vertex presentation and by a mother without fever, administration of oxytocics strongly predicted early death, with an OR of 2.1 (95% CI = 1.6, 2.7).
When we in our final model included as covariates the minimal sufficient set of potential confounders, that is, skilled birth attendant and place of delivery, the aOR of oxytocics for predicting early death was 1.7 (95% CI = 1.4, 2.1). The model emanating from the explorative DAG, which depicted additional potential confounders (eFigure in the eAppendix; http://links.lww.com/EDE/B681), yielded an aOR for oxytocics of 1.7, only 2% higher than that of the estimate obtained from our final model. Referring to the change-in-point-estimate approach proposed by Greenland,39 we therefore used the minimal sufficient set of potential confounders in further analyses. The aOR for oxytocics of our final model translates to a PAR of 31% (95% CI = 22%, 38%).
Although the administration of oxytocics was associated with a substantially increased risk of early death in home deliveries and in private facilities, we found no such association in government facilities (Table 2). Uterotonic injection was even more predictive of early death among boys than girls and among babies of nulliparous as compared to of multiparous women (Table 2). Use of uterotonics predicted intrapartum stillbirth more strongly than death among live born babies; this association waned during the first 24 hours of life and was negligible in the subsequent 6 days of the first week (Table 3).
Administration of oxytocics predicted asphyxia in the first week of life with an adjusted relative risk of 1.6 (95% CI = 1.3, 1.9) (not tabulated). Asphyxia was strongly predictive of death in the first 2 hours of life with an aOR of 76 (95% CI = 44, 131), with a substantial association maintained in the subsequent 22 hours (aOR 31 [95% CI = 18, 55]) and in the remaining 6 days of the first week of life (aOR 9.4 [95% CI = 6.0, 15]). Including asphyxia in the final model of early versus late death during the first day of life reduced the aOR from 1.5 (95% CI = 1.2, 1.8) to 0.96 (95% CI = 0.72, 1.3), translating to what appears to be a complete mediation by asphyxia for the association between oxytocic administration and day-1 deaths and quantified to 1.5/0.96 or 1.5 (95% CI = 1.3, 1.9).
In our population-based case–control study, we found that the use of antenatal oxytocics was associated with a substantially increased risk of intrapartum stillbirth and day-1 death. The oxytocics-related risk of death declined with passage of time during the first week, which was paralleled by a declining proportion of asphyxia-related deaths. It is known that boys are more likely to die from asphyxia35 and therefore of interest that antenatal administration of oxytocics predicted early death among boys more strongly than among girls. Considering that nulliparous women experience a higher risk of having a baby die with asphyxia than those who have previously given birth, it is also intriguing that in our study, uterotonic injection predicted early death even more strongly in babies of nulliparous as compared to in babies of multiparous women. The temporal relationship between oxytocic injection and death, the strength of the association and the biological plausibility, which was strengthened by asphyxia seemingly mediating the entire effect of oxytocics on day-1 death, strongly suggests a causal relationship. Such administration was reported in 62% of home deliveries, and 81% of the deliveries in health facilities, which is consistent with findings of other community-19–22 and hospital-based studies12,14,15,22,31 in South Asia. With oxytocin being recommended for the prevention of postpartum hemorrhage in all deliveries, concerns have been raised about the improved availability of this life-saving drug, because it could simultaneously increase its misuse for acceleration of labor.12
The PAR of 31% in our study, more than twice that for late stillbirth of the very large MOMA study in Africa,23 indicates that 637 of our 2076 intrapartum stillbirths and day-1 deaths were attributed to antepartum administration of uterotonics. We estimated the number of intrapartum stillbirths and day-1 deaths in India that could be attributed to such administration, referring to reports1,2 of the number of day-1 deaths and stillbirths, and to the current study for the proportion of stillbirths that were intrapartum. Of the expected 600,000 annual intrapartum stillbirths and day-1 deaths in India, if the use of oxytocics before delivery is as common in the country as a whole as it was in our study area, 184,000 (95% CI = 131,000, 228,000) deaths can be averted if oxytocics are used appropriately. This estimate is, however, fraught with uncertainty beyond what is captured in its 95% CI. The available figures for intrapartum stillbirths and day-1 deaths in India1,2 are themselves estimates, the former modeled based on predictor variables, the latter on vital registration and Demographic and Health Survey data. That said, there is little doubt that deaths on the first day of life contribute to approximately one third of the neonatal deaths in most low- and middle-income countries, including in India, and that intrapartum stillbirths, albeit often unrecorded, are even more common.3,5,7 Furthermore, our own data, although collected for the purpose of a very large population-based cluster-randomized trial in North India, may not be representative of the entire country. Given the above considerations and also other reports of widespread and misinformed use of oxytocics, our findings strongly suggest that full adherence to guidelines on the use of oxytocics, in India alone, would save more than twice the number of babies that die in stillbirth across all high-income countries.2
Somewhat surprisingly, although administration of uterotonics at home and in private health facilities predicted early death, we did not find such an association for babies born in government health facilities. This may be a chance finding, but could also be explained by health workers in government facilities using somewhat lower doses, administrating the drug only after being instructed to do so by a qualified physician, appropriate fetal monitoring, recording progress of labor, and ensuring that cesarean section is available and performed when needed.
Strengths and Limitations of the Study
The strength of our study is its community base, large sample size, low attrition, and minimal missing data. The study, however, has limitations. Our proxy-sampled controls were babies who died in the late neonatal period rather than all infants who survived beyond day 1. Based on current knowledge, however, we have no reasons to believe that administration of uterotonics increases the risk of death to a measurable degree in babies beyond the first week of life,7,11 so we trust that our proxy sampling approach29 would include controls with an exposure distribution reflecting that of babies who survived even beyond the neonatal period. Our data speaks to this. Thus, we found no overall association between uterotonic drug injection and neonatal death from day 2 to 7. We therefore do not expect that this strategy would substantially underestimate our observed effect size.
Case–control studies such as ours may be fraught with biases. Thus, it is possible that mothers of neonates who die would better recollect injections given during labor than those with babies who were alive. However, our proxy-sampling approach could actually reduce such information bias, because it is less likely that such recall would differ between those who experience an early death as compared to those whose baby died after its first week but before it crossed its fourth week of life. Moreover, the fact that the entire association between uterotonic administration and early neonatal death was driven by babies with asphyxia, an entity mothers in labor would be unlikely to identify or link to injections, indicates that such differential exposure misclassification is less likely. Extracting the information on oxytocics administration from the verbal autopsies is prone to misclassification that is nondifferential, because our experts were not aware of our hypothesis. Such misclassification would tend to bias the aOR toward 1.
We cannot rule out that some of the injections identified by our experts as oxytocics were actually antibiotics to treat infections. However, such misclassification is unlikely to have biased our aOR estimate. Thus, adding fever to our final model or removing it from the model derived from the explorative DAG presented in the eAppendix; http://links.lww.com/EDE/B681 left the ORs virtually unchanged. Moreover, injectable antibiotics could have been given to prevent infections in the case of prolonged rupture of membranes, prolonged labor, or in the presence of foul-smelling amniotic fluid. However, these variables did not predict early versus late death (eTable in the eAppendix; http://links.lww.com/EDE/B681), and forcing them into our final model or removing them from our alternative model left the ORs nearly unchanged. We also considered the possibility that prolonged labor or prolonged rupture of membranes could elicit the administration of oxytocics and independently lead to early death. However, this is unlikely because in those without prolonged rupture of membranes or prolonged labor, such administration was as strongly predictive of early death as in our entire sample.
Our experts are likely to have misclassified some antepartum stillbirths as being intrapartum and vice versa, a misclassification that is likely to be nondifferential, and which could result in an underestimated aOR.
Data collection for the study ended in 2010, and it is possible that antenatal administration of oxytocics has become better informed by guidelines in subsequent years. In 2016, we therefore went back to the providers who were reported to have administered oxytocics in the verbal autopsies; regrettably more than three quarters of the respondents reported oxytocin use for increasing labor pains to be as common as it was previously.
This is to our knowledge the first population-based study to estimate the extent to which inappropriate antenatal administration of uterotonics contributes to high intrapartum stillbirth and early neonatal death risk. The MOMA study that took place across six countries in West Africa measured risk factors for late stillbirth,23 and perinatal death40 at the community level. The late stillbirths are likely to have included antepartum stillbirths, and perinatal deaths are likely to have also included deaths from prematurity and sepsis, conditions not likely to have been related to uterotonic injections. In rural Maharashtra in India, the use of oxytocics to augment labor was associated with a high risk of severe asphyxia and “fresh stillbirths,”20 which we interpret as being mainly intrapartum. In a study in Uttar Pradesh, India, Jeffery et al.21 found that use of oxytocin before birth was associated with an increased risk of stillbirth and neonatal death. Both these small Indian studies included only home births, and the investigators did not seem to have considered adjusting for possible confounders. Studies have shown an increased risk of neonatal encephalopathy,14 birth asphyxia,16 need for neonatal resuscitation, and stillbirth13 with antenatal oxytocin administration. All these studies were, however, hospital-based and their findings may not apply to the community level, as women giving birth in hospitals are likely to have a higher risk of complications, especially in low-income settings.41
Our study is strongly consistent with the hypothesis that use of antenatal oxytocics, when to a large extent given on inappropriate indications and without adequate access to emergency obstetric care, substantially increases the risk of intrapartum stillbirth and day-1 deaths. These findings are sobering when seen in light of other studies showing the extent of inappropriate oxytocin use during facility as well as home deliveries. Ongoing studies as well as systematic reviews will further strengthen the evidence for this important public health problem.
We are grateful to the participants for their cooperation throughout the study. We thank the Government of Haryana, the Civil Surgeons, participating health and ICDS officers, and health workers of districts Faridabad and Palwal. We are thankful to the pediatricians and obstetricians who reviewed the stillbirth and neonatal verbal autopsies, and to Paul Murrel for developing a new application for the gridGraphics package. We thank Kenneth J. Rothman for his guidance with respect to epidemiologic theory and terminology and our mentor, the late Prof. Maharaj K. Bhan, as well as Tarun Choudhary, Håkon K. Gjessing, and Hans Steinsland for fruitful discussions on the statistical analyses. We also acknowledge the core support provided to SAS by the Department of Maternal, Newborn, Child and Adolescent Health, World Health Organization (Geneva).
1. Oza S, Cousens S, Lawn JE. Estimation of daily risk of neonatal death, including the day of birth, in 186 countries in 2013: a vital-registration and modelling-based study. Lancet Glob Health. 2014;2:e635–e644.
2. Lawn JE, Blencowe H, Waiswa P, et al.; Lancet Ending Preventable Stillbirths Series study group; Lancet Stillbirth Epidemiology investigator group. Stillbirths: rates, risk factors, and acceleration towards 2030. Lancet. 2016;387:587–603.
3. United Nations Inter-agency Group for Child Mortality Estimation (UNIGME). Levels & Trends in Child Mortality: Report 2018, Estimates developed by the United Nations Inter-agency Group for Child Mortality Estimation. 2018.New York: United Nations Children’s Fund.
4. Oestergaard MZ, Inoue M, Yoshida S, et al.; United Nations Inter-Agency Group for Child Mortality Estimation and the Child Health Epidemiology Reference Group. Neonatal mortality levels for 193 countries in 2009 with trends since 1990: a systematic analysis of progress, projections, and priorities. PLoS Med. 2011;8:e1001080.
5. McClure EM, Wright LL, Goldenberg RL, et al.; NICHD FIRST BREATH Study Group. The global network: a prospective study of stillbirths in developing countries. Am J Obstet Gynecol. 2007;197:247.e1–247.e5.
6. Bapat U, Alcock G, More NS, Das S, Joshi W, Osrin D. Stillbirths and newborn deaths in slum settlements in Mumbai, India: a prospective verbal autopsy study. BMC Pregnancy Childbirth. 2012;12:39.
7. Baqui AH, Darmstadt GL, Williams EK, et al. Rates, timing and causes of neonatal deaths in rural India: implications for neonatal health programmes. Bull World Health Organ. 2006;84:706–713.
8. Indian Council of Medical Research Young Infant Study Group. Age profile of neonatal deaths. Indian Pediatr. 2008;45:991–994.
9. Aminu M, Bar-Zeev S, White S, Mathai M, van den Broek N. Understanding cause of stillbirth: a prospective observational multi-country study from sub-Saharan Africa. BMC Pregnancy Childbirth. 2019;19:470.
10. Lawn JE, Blencowe H, Oza S, et al.; Lancet Every Newborn Study Group. Every newborn: progress, priorities, and potential beyond survival. Lancet. 2014;384:189–205.
11. Bang AT, Paul VK, Reddy HM, Baitule SB. Why do neonates die in rural Gadchiroli, India? (Part I): primary causes of death assigned by neonatologist based on prospectively observed records. J Perinatol. 2005;25 (suppl 1):S29–S34.
12. Lovold A, Stanton C, Armbruster D. How to avoid iatrogenic morbidity and mortality while increasing availability of oxytocin and misoprostol for PPH prevention? Int J Gynaecol Obstet. 2008;103:276–282.
13. Dujardin B, Boutsen M, De Schampheleire I, et al. Oxytocics in developing countries. Int J Gynaecol Obstet. 1995;50:243–251.
14. Ellis M, Manandhar N, Manandhar DS, Costello AM. Risk factors for neonatal encephalopathy in Kathmandu, Nepal, a developing country: unmatched case-control study. BMJ. 2000;320:1229–1236.
15. Chandra S, Ramji S, Thirupuram S. Perinatal asphyxia: multivariate analysis of risk factors in hospital births. Indian Pediatr. 1997;34:206–212.
16. Kaye D. Antenatal and intrapartum risk factors for birth asphyxia among emergency obstetric referrals in Mulago Hospital, Kampala, Uganda. East Afr Med J. 2003;80:140–143.
17. WHO. WHO recommendations for induction of labour. 2011. Geneva: WorldHealth Organization. Available at: https://www.who.int/reproductivehealth/publications/maternal_perinatal_health/9789241501156/en/
. Accessed 6 May 2019.
18. WHO. WHO recommendations for augmentation of labour 2014. Geneva: WorldHealth Organization. Available at: https://www.who.int/reproductivehealth/publications/maternal_perinatal_health/augmentation-labour/en/
. Accessed 6 May 2019.
19. Fronczak N, Arifeen SE, Moran AC, Caulfield LE, Baqui AH. Delivery practices of traditional birth attendants in Dhaka slums, Bangladesh. J Health Popul Nutr. 2007;25:479–487.
20. Bang AT, Bang RA, Baitule SB, Reddy HM, Deshmukh MD. Management of birth asphyxia in home deliveries in rural Gadchiroli: the effect of two types of birth attendants and of resuscitating with mouth-to-mouth, tube-mask or bag-mask. J Perinatol. 2005;25 (suppl 1):S82–S91.
21. Jeffery P, Das A, Dasgupta J, Jeffery R. Unmonitored intrapartum oxytocin use in home deliveries: evidence from Uttar Pradesh, India. Reprod Health Matters. 2007;15:172–178.
22. Matthews Z, Ramakrishna J, Mahendra S, Kilaru A, Ganapathy S. Birth rights and rituals in rural south India: care seeking in the intrapartum period. J Biosoc Sci. 2005;37:385–411.
23. Chalumeau M, Bouvier-Colle MH, Breart G; MOMA Group. Can clinical risk factors for late stillbirth in West Africa be detected during antenatal care or only during labour? Int J Epidemiol. 2002;31:661–668.
24. Bhandari N, Mazumder S, Taneja S, Sommerfelt H, Strand TA; IMNCI Evaluation Study Group. Effect of implementation of Integrated Management of Neonatal and Childhood Illness (IMNCI) programme on neonatal and infant mortality: cluster randomised controlled trial. BMJ. 2012;344:e1634.
25. Mazumder S, Taneja S, Bahl R, et al.; Integrated Management of Neonatal and Childhood Illness Evaluation Study Group. Effect of implementation of integrated management of neonatal and childhood illness programme on treatment seeking practices for morbidities in infants: cluster randomised trial. BMJ. 2014;349:g4988.
26. Aggarwal AK, Jain V, Kumar R. Validity of verbal autopsy for ascertaining the causes of stillbirth. Bull World Health Organ. 2011;89:31–40.
27. Aggarwal AK, Kumar P, Pandit S, Kumar R. Accuracy of WHO verbal autopsy tool in determining major causes of neonatal deaths in India. PLoS One. 2013;8:e54865.
28. WHO. A Standard Verbal Autopsy Method for Investigating Causes of Death in Infants and Children. WHO/CDS/CSR/ISR/99.4. 1999. Geneva: World Health Organization.
29. Rothman KJ, Lash TL, Greenland S. Modern Epidemiology. 2008.3rd ed. Philadelphia: Lippincott Williams & Wilkins.
30. Flandermeyer D, Stanton C, Armbruster D. Uterotonic use at home births in low-income countries: a literature review. Int J Gynaecol Obstet. 2010;108:269–275.
31. Stanton CK, Deepak NN, Mallapur AA, et al. Direct observation of uterotonic drug use at public health facility-based deliveries in four districts in India. Int J Gynaecol Obstet. 2014;127:25–30.
32. IIPS. National Family Health Survey (NFHS-3), 2005-06. 2007.Mumbai, India: IIPS.
33. Frøen JF, Pinar H, Flenady V, et al. Causes of death and associated conditions (Codac): a utilitarian approach to the classification of perinatal deaths. BMC Pregnancy Childbirth. 2009;9:22.
34. Rockhill B, Newman B, Weinberg C. Use and misuse of population attributable fractions. Am J Public Health. 1998;88:15–19.
35. Lee AC, Mullany LC, Tielsch JM, et al. Risk factors for neonatal mortality due to birth asphyxia in southern Nepal: a prospective, community-based cohort study. Pediatrics. 2008;121:e1381–e1390.
36. Igboanugo S, Chen A, Mielke JG. Maternal risk factors for birth asphyxia in low-resource communities. A systematic review of the literature. J Obstet Gynaecol. 2019;11:1–17.
37. Andersson T, Alfredsson L, Källberg H, Zdravkovic S, Ahlbom A. Calculating measures of biological interaction. Eur J Epidemiol. 2005;20:575–579.
38. Sommerfelt H, Steinsland H, van der Merwe L, et al. Case/control studies with follow-up: constructing the source population to estimate effects of risk factors on development, disease, and survival. Clin Infect Dis. 2012;55 (suppl 4):S262–S270.
39. Greenland S. Modeling and variable selection in epidemiologic analysis. Am J Public Health. 1989;79:340–349.
40. Chalumeau M, Salanave B, Bouvier-Colle MH, de Bernis L, Prual A, Bréart G. Risk factors for perinatal mortality in West Africa: a population-based study of 20326 pregnancies. MOMA group. Acta Paediatr. 2000;89:1115–1121.
41. Ronsmans C, Chowdhury ME, Koblinsky M, Ahmed A. Care seeking at time of childbirth, and maternal and perinatal mortality in Matlab, Bangladesh. Bull World Health Organ. 2010;88:289–296.