The PURPOSe cause of death study in stillbirths and neonatal deaths in India and Pakistan: A review

The PURPOSe study was a prospective, observational study conducted in India and Pakistan to determine the cause of death for stillbirths and preterm neonatal deaths, using clinical data together with minimally invasive tissue sampling (MITS) and the histologic and polymerase chain reaction (PCR) evaluation of fetal/neonatal tissues and the placenta. After evaluating all available data, an independent panel chose a maternal, a placental and a fetal/neonatal cause of death. Here, we summarise the major results. Among the most important findings were that most stillbirths were caused by fetal asphyxia, often preceded by placental malperfusion, and clinically associated with pre‐eclampsia, placental abruption and a small‐for‐gestational‐age fetus. The preterm neonatal deaths were primarily caused by birth asphyxia, followed by various infections. An important finding was that many of the preterm neonatal deaths were caused by a nosocomial infection acquired after neonatal intensive care (NICU) admission; the most common organisms were Acinetobacter baumannii, followed by Klebsiella pneumoniae, Escherichia coli/Shigella and Haemophilus influenzae. Group B streptococcus was less commonly present in the placentas or internal organs of the neonatal deaths.

as a baby born alive, who then dies within 28 days of age.A subset of neonatal deaths, those occurring in preterm infants, are defined as deaths occurring in the first 28 days of life to an infant born preterm, or before 37 weeks of gestation.The common feature of these deaths is that they are defined by the time in the fetus' or neonate's life, and not by the cause of death.
For many years, in the published records for causes of infant deaths (i.e.deaths at <1 year of age), the cause of death for infants who died at <28 days of life was often labelled 'neonatal death', whereas for older infants the clinical causes of death, such as pneumonia or an accident, were explicitly listed. 1 Similarly, in the published records of neonatal deaths, the cause of death for infants who were born at <37 weeks of gestation was often ascribed to prematurity; the underlying causes of those deaths were not usually considered. 2 Also, when the overall outcomes of pregnancy were presented, deaths in utero were often listed as 'stillbirths' and their actual causes not described.This was especially true in lowand middle-income countries (LMICs), where investigations into the causes of death for neonates (and especially preterm neonates) and stillbirths were often not performed.
In many LMICs, even with the desire to understand the cause of death for stillbirths and neonates, until relatively recently the tools to accurately determine a cause of death were often not available.Instead, most of the reported study outcomes were based on a single physician's opinion, with little hard evidence to inform the opinion on cause of death.Autopsy has, in some cases, provided some evidence for cause of death, but in LMICs, for a number of reasons, including religious beliefs, costs, lack of trained expertise, and the desire for a quick burial, autopsies were rarely performed on stillbirths or neonatal deaths.Other potentially useful tests were also not often performed.
To overcome some of these obstacles, The Bill & Melinda Gates Foundation (BMGF), along with others, has advocated for more comprehensive investigations into the cause of death for fetuses and infants.][5] It has also been clear for many years that understanding placental pathology, and especially certain types of placental lesions, was key to understanding the causes of stillbirth and neonatal death.However, in most LMICs, evaluating placental histology was rarely performed, in part because there was not a universally accepted classification system for placental lesions, in part because of cost and in part because of an absence of trained placental pathologists.Publication of the Amsterdam criteria in 2016 for the classification of placental lesions partially filled this void. 6And recently, the development of multiplex polymerase chain reaction (PCR) platforms with the ability to identify many pathogens in a small sample of fluid or tissue has made it possible to be more confident about which organisms contributed to the disease or death. 7[10][11] The BMGF has funded several studies to better understand the cause of death for stillbirths, neonatal and child deaths in LMICs, with a goal of using that information to plan interventions to reduce those deaths. 3,8,9One of these projects is the PURPOSe study (Project to Understand and Research Pregnancy Outcomes and Stillbirths in South Asia). 10As can be discerned from its name, this was a study to determine the cause of death in stillbirths and preterm neonatal deaths in India and Pakistan.8][9] The PURPOSe study was funded in 2018 with a primary goal to identify the causes of death in a combined total of 700 stillbirths and 700 preterm neonatal deaths in India and Pakistan. 10Within the available resources, the PURPOSe study used nearly all of the tests available to understand the causes of death and to shed light on a number of other issues related to pregnancy outcomes in LMICs.
Based on their participation in the Study of Illness in Preterms (SIP) study, another BMGF-funded study on the causes of death in preterm infants carried out in Ethiopia from 2017 to 2019, 12 Drs McClure and Goldenberg organised a similar study in south Asia, but expanded the outcomes to include causes of stillbirth, with support from the BMGF.Based on the research capability of the sites at Aga Khan University (AKU) in Karachi, Pakistan, the combined sites of KLE Academy of Higher Research and JN Medical College in Belagavi, India, and at JJM Medical College in Davangere, India, the PURPOSe study was organised and carried out in those locations.The primary goal was to determine the cause of death for stillbirths and preterm neonates in sites in south Asia.

| M ET HODS
Prior to the onset of the PURPOSe study, the investigators reviewed existing studies on cause of death for stillbirths and preterm neonatal deaths for their strengths and weaknesses.Within the available resources, they planned a study using evidence-based investigations to inform the cause of death of stillbirths and preterm neonates.Stillbirths were chosen as one end point because of the paucity of information on the causes of this outcome, especially in LMICs, and preterm neonatal death was chosen as a primary outcome because many studies attributed the cause of death in these infants to 'prematurity', often without considering specific clinical causes such as asphyxia, respiratory distress syndrome (RDS) and infection. 1s noted, in recent years, a number of diagnostic aides to determine the cause of death in stillbirths and neonatal deaths have been developed and promulgated.4][5] This test was a very important addition to the cause of death armamentarium because in many areas, and especially for Muslim populations, there are strong religious prohibitions against mutilation of the deceased.][15][16][17] A further innovation in the PURPOSe study was to ask panel members to determine a maternal cause of death, a placental cause of death and a fetal or neonatal cause of death. 11e chose this approach because in reviewing prior cause of death studies, many approaches required the reviewer to choose one primary cause of death.In doing so, important information about cause of death was often lost.As an example, a frequent cause of death scenario for stillbirths and neonatal deaths involves maternal pre-eclampsia, placental lesions of malperfusion and fetal or neonatal asphyxia.Cause of death attributors are often forced to choose one of those 'causes', whereas a true understanding of the cause of death involves all three.Therefore, in the PURPOSe study, the panel designated a maternal cause, a placental cause and a fetal or neonatal cause of death.For each death, we also asked the panels to make a judgement about whether the death was preventable with the interventions available in the country.
Although not an innovation, we felt it important to know the frequency of certain demographic findings and various histologic lesions and organisms in the available tissues of babies who were not preterm and did not die.Therefore, in both India and Pakistan, we asked that demographic data, a sample of placental tissue for histology and PCR, and a maternal blood sample be collected from 100 term controls at each site. 10,11n reviewing existing cause of neonatal death and stillbirth studies, most came from high-income countries, and many had at least a partial focus on genetic and chromosomal abnormalities. 18Although we could evaluate both external and internal structural abnormalities in the PURPOSe study, funding constraints limited our ability to perform chromosomal or more sophisticated genetic studies.With the high frequency of other causes of mortality, it is likely that genetic and chromosomal abnormalities are not a frequent cause of neonatal death or stillbirth in LMICs.On the other hand, studies from high-income countries strongly suggest that placental conditions were a common cause of stillbirths and neonatal deaths. 19,20Therefore, we were interested in determining whether the same placental conditions could also be causal for an important proportion of the stillbirths and preterm neonatal deaths in LMIC settings.We therefore made every attempt to collect and evaluate all placentas from deceased fetuses as well as from all preterm births, knowing that we would prioritise the evaluation of the placentas from preterm infants who died in the neonatal period.
For the cause of death study, based on preliminary calculations, we set a study sample size goal of 350 stillbirths and 350 neonatal deaths in Karachi, Pakistan, and the same number in Davangere, India.With this preliminary work and with these decisions, we were ready to start the PURPOSe study.The objective of this review article is to summarise the main findings, first for stillbirths and then for preterm neonatal deaths, and to place the PURPOSe findings in the context of the current body of literature on cause of death among stillbirths and newborn deaths.

| R E SU LTS
The final population recruited for this study included 984 stillbirths at ≥20 weeks of gestation, with 607 from Pakistan and 377 from India, and a total of 804 preterm live-born neonatal deaths, with 475 from Pakistan and 329 from India.There were 201 term control pregnancies.

| Stillbirths
The stillbirths that were reviewed by the cause of death panel included stillbirths at ≥28 weeks of gestation, and/or ≥1000 g, with a completed placental evaluation (Table 1).Using all of the information available on cause of death, the first conclusion that the panel made in relation to stillbirth was that the fetal cause of death for the vast majority (81.6%) was fetal asphyxia, with a smaller proportion caused by infection (30.8%) and with an even smaller proportion caused by an observable anomaly (4.9%). 21The clinical associations with the stillbirths caused by asphyxia included pre-eclampsia, fetal growth restriction and maternal haemorrhage, probably as a result of placental abruption.Placental histology indicated that maternal and fetal vascular malperfusion (MVM and FVM, respectively) were common placental findings in stillbirth.
In separate analyses of the placentas, we confirmed that malperfusion was significantly more common in stillbirth compared with control placentas. 22Specifically, placental MVM was present in 58.4% of stillbirth placentas but was found in only 15.4% of term control placentas (Table 2).FVM was also associated with a four-fold increase in stillbirths: 19.0% in stillbirth placentas compared with 5.0% in term control placentas.Various inflammatory lesions, and especially chorioamnionitis, were also more common in stillbirth placentas compared with control placentas.Chorioamnionitis was also significantly increased in the placentas of stillbirths, with 30% in stillbirth placentas versus 3% in controls.
The pathway to asphyxia involved several clinical conditions.Most commonly, these included pre-eclampsia and placental abruption and were often associated with fetal growth restriction.Better obstetric care, especially around delivery, with a focus on pre-eclampsia, placental abruption and fetal growth restriction, was thought to be the most effective way to reduce stillbirths. 22,23he PCR evaluation of organisms in the internal organs of stillbirths produced some interesting results. 24Most important was the observation that only a small proportion of the stillbirths (12.9%) had any pathogens in the internal organs (Tables 3 and 4).The organisms were mainly detected in the fetal brain (12.3%), followed by the cerebrospinal fluid (CSF) (9.5%) and blood (8.4%).Ureaplasma urealyticum/parvum was the organism most frequently detected, found in at least one internal organ sample in 6.4% of the stillbirths.Escherichia coli/Shigella was the next most common organism, present in 4.1% of at least one internal organ sample from stillbirths.None of the other organisms were detected in more than 2% of the internal organs from stillbirths.Thus, infection appeared to play a smaller role in the cause of stillbirth than vascular lesions.
Of the PCR findings in stillbirths, perhaps the most interesting was the more common finding of U. urealyticum/parvum in the fetal brain but not in the fetal lungs (5.7% vs 0%, respectively).Ureaplasma urealyticum/parvum was also by  far the most common organism found in the stillbirth placental tissues. 24Although the high proportion of U. urealyticum/parvum in the brain and CSF of stillbirths was initially surprising, U. urealyticum/parvum was by far the most common organism seen in the placentas and was also common in cord blood.As the largest organ recipient of fetal blood is the fetal brain, we speculated that U. urealyticum/parvum, already frequently present in the placenta, is often transmitted transplacentally in fetal blood to the fetal brain. 24he cause of death panels also determined whether the individual stillbirths were preventable. 23In the opinion of the panels, 55.5% of all stillbirths were preventable with the care currently available in the countries of study.When the preventability analysis was limited to stillbirths at ≥28 weeks of gestation, 73.5% were thought to be preventable.The clinical conditions that contributed the most to preventable stillbirths included fetal growth restriction (52.8%), maternal hypertension (50.2%), placental abruption (31.4%) and deaths in labour, probably through asphyxia (15.7%).
As there were many different types of information potentially available to determine the cause of stillbirth, and because it is expensive and inefficient to collect all possible information, we thought it helpful to evaluate which information was most useful in determining the cause of stillbirth.This is important, as in order for stillbirth cause of death investigations to become more routine, a low-cost and efficient approach is needed.
To determine the most useful tests, we separately evaluated 200 stillbirths in which the key information potentially related to the cause of death was available (Table 5). 25,26The most useful tests included: relevant maternal clinical and obstetric history; a careful placental evaluation, including gross examination and measurement of placental weight (including comparison with an accepted standard to define small and large placentas); histology of the chorioamniotic membranes, umbilical cord and placental body, with a focus on inflammation, haemorrhage and malperfusion; and an external examination of the fetus, including the measurement of weight (compared with some standard to determine fetal growth restriction) and the recording of visible congenital anomalies.Finally, we considered potential data from MITS examinations of internal organ histology and PCR for pathogen evaluation of these same tissues and the placenta.Our first observation was that for organ histology, the lungs are the most informative organs, whereas liver and CNS histology provided the least information.Findings of amniotic fluid debris or meconium in the lung, commonly associated with fetal gasping, were present in somewhat less than half of stillbirths, and often helped the panellists determine a diagnosis of fetal asphyxia.The PCR evaluation of blood, CSF and brain tissue provided the most information.Microbiological analysis of the placenta and membranes were also informative, as was the finding of meconium on any examination.We recognised that this approach will not identify some causes of stillbirth, including some genetic abnormalities and internal organ anomalies, but believe it will identify the most common causes of stillbirth in LMICs, most of the preventable causes of stillbirth in LMICs, and will use the least resources while providing the most accurate information on cause of death. 26

| Preterm neonatal death
Although there were some differences in the cause of preterm neonatal death between Pakistan and India, the most obvious difference was in the preterm neonatal mortality rate itself. 27hereas the populations of preterm neonates were generally similar in birthweight and gestational age, there were very large differences in the mortality rates as well as the settings in which the study was conducted.In India, 17% of the preterm babies admitted to three small neonatal intensive care units (NICUs) associated with the hospital and who were enrolled in the PURPOSe study died, whereas in Pakistan 32% of the preterm infants enrolled in the PURPOSe study and admitted to the NICU affiliated with a large referral hospital died.We explored some potential reasons for the difference in this outcome.Discussions with the Pakistani hospital staff suggested that the number of admissions was beyond the capacity of the staff to provide care, and that medical equipment and drugs were also limited in the NICU.These issues were rarely observed in the Indian NICUs.The data from the PURPOSe study supported these observations.The mean length of stay in the Pakistan NICU was 2 days,   compared with 10 days in the Indian NICUs.Study diagnostic tests to confirm select clinical conditions were rarely performed in the Pakistani NICU but were performed much more frequently in the Indian NICUs.For example, in babies diagnosed with respiratory distress syndrome (RDS), 90% of the infants with this diagnosis in India had a chest X-ray, compared with 9% of similar infants in Pakistan.Thus, the tests expected to be performed under NICU care were rarely performed in the Pakistani NICU but were performed much more frequently in the Indian NICUs.Although we cannot prove that the very short length of stay and the very few tests performed in the Pakistani NICU (likely related to shortages of staff and equipment) were associated with the increased mortality rate, these may be indicators for the overall quality of care that the participants received, which may be associated with the risk of mortality.
The neonatal deaths evaluated by the panellists were restricted to neonates of ≥1000 g.The leading causes of neonatal death designated by the panels were similar in India and Pakistan.Among the maternal causes, the most common was hypertensive disease (36%).The most common placental cause was malperfusion. 27Overall, the most common neonatal cause of preterm neonatal death was birth asphyxia.The next most frequent causes of neonatal death were infections acquired prior to delivery (20.5%) and then infections acquired in the nursery (19.8%).RDS was far less often considered the cause of preterm neonatal death (10.2%).
In separate analyses, we compared the cause of death determined by the NICU physicians compared with the cause of death determined by the panels. 28The panel had access to the entire obstetric history, the placental histological data and placental weight, the internal organ histology and the PCR results, in addition to the clinical course of the neonate.The major difference was the frequent diagnosis of RDS by the physicians, whereas the panels, with much more data, far more often cited the cause of death to be birth asphyxia.In fact, in Pakistan, the discharging NICU physician rarely diagnosed the cause of death as birth asphyxia.We suspect that as nearly all preterm infants who die have respiratory distress, the discharging physician -with far less information than is available to the panel -often chose RDS as the fallback cause of death, whereas the panel, with more information, more often chose birth asphyxia as the cause of death.This observation calls into question the results presented from many prior preterm neonatal cause of death studies performed with limited information, including those with no placental histology and no internal organ histology in which RDS was found to be the most common cause of death among preterm infants. 12The rates of these two diagnoses have important implications for the reduction of mortality.If RDS is the major cause of preterm neonatal death, an approach focused on maternal corticosteroids would seem appropriate, whereas if birth asphyxia was the major cause of preterm neonatal death, a strategy focused on improved obstetrical care and neonatal resuscitation would seem more appropriate.
The PCR results from the placentas and the internal organ biopsies of preterm neonatal deaths obtained by MITS were also interesting (Tables 6 and 7). 29Tissue for pathogen PCR was obtained from liver, lung, brain, blood, CSF and placentas from deceased infants.Altogether, 49.7% of preterm neonatal deaths had at least one organism found in an internal organ, and between 17.6% and 34.1% of each type of organ tissue had at least one organism identified, which are rates much higher than those found in stillbirths.The frequency of organism detection was highest in blood (34.1%), followed by lung (31.1%), liver (23.3%),CSF (22.3%) and brain (17.6%).Acinetobacter baumannii was found twice as often in the internal organs as any other organism, in 28.4% of the neonates compared with: 14.6% for Klebsiella pneumoniae, 11.9% for E. coli/Shigella and 11.1% for Haemophilus influenzae.Of note, A. baumannii was rarely found in the placenta or membranes and was found more commonly in the internal organs of deceased neonates who died later in the neonatal period.That A. baumannii was rarely found in the placenta and was more often found with increasing length of NICU stay strongly suggest that this infection (as well as others) was nosocomial in origin, and could be prevented by better attention to infection control in the nurseries.
An interesting finding from this study applying to mothers, fetuses and newborns in both India and Pakistan was the consistently low rates of group B streptococcus (GBS). 30cross both countries, all fetal and neonatal characteristics, tissues and methodology used to assess GBS, the prevalence of GBS was low.Specifically, GBS diagnosed by PCR was identified in only 0%-6% of internal organs from stillbirths and neonatal deaths and their placentas and control groups, consistent with other reports from South Asia, and lower than the rates reported from Africa, Europe and the USA.Focusing on GBS, and not on the more commonly found A. baumannii, E. coli/Shigella and K. pneumoniae, may miss an important opportunity to reduce neonatal infections in South Asia.

| CONCLUSION
In summary, the PURPOSe study has elucidated a number of issues related to the causes of stillbirth and preterm neonatal death that are likely to be applicable for many LMICs in general, and in South Asia, specifically.The first observation is that a thorough review of the previous studies and their limitations as well as the incorporation of some of the newer technologies is extremely helpful in maximizing the knowledge gained from the next study.A second, and equally important, observation is the importance of appropriate community involvement.Without the effort put into understanding the community issues related to autopsy, and especially MITS, this study could not have been completed.We emphasise that in the PURPOSe study, MITS was accepted as an appropriate diagnostic tool by the parents in nearly half of the cases of fetal or preterm neonatal death.
Birth asphyxia was the most common cause of preterm neonatal deaths -not RDS.In prior neonatal cause of death assessments performed without extensive evaluations, RDS was the condition often named (probably incorrectly) as one of the leading causes of preterm neonatal death.In this study, birth asphyxia was a far more common cause of preterm neonatal death than RDS.
Placental evaluation by PCR and histology is important to obtain accurate causes of death in fetuses and preterm neonates.Acinetobacter baumannii, although rarely found in the placenta, is the most common organism found in internal organ tissues of deceased neonates, but not stillbirths.There is evidence suggesting that this organism is transmitted nosocomially to infants in the NICU.GBS is found much less commonly in the placenta, stillbirths and neonatal tissue in the South Asian sites, compared with reports from Africa, Europe and the USA.
Most stillbirths were caused by fetal asphyxia, often in conjunction with maternal hypertension, fetal growth restriction (FGR) and placental abruption.Placental malperfusion is the most important precursor to stillbirth and is an important precursor to preterm neonatal deaths.Placental malperfusion is also a precursor to FGR, antepartum haemorrhage and hypertension, and together with those conditions plays an important role in the pathways to both stillbirth and neonatal death.Most stillbirths in these locations are preventable with available interventions.Research that focuses on placental malperfusion has the potential not only to determine those at highest risk of stillbirth but also has great potential to develop interventions that result in a decrease in the malperfusion lesions themselves.

AU T HOR C ON T R I BU T ION S
RLG conceived and developed the first manuscript draft; RLG, SS, SSG, JM, GG, VK, SMD, SST, SN, SM, MGK, NKG, SUK, IA, KH, SY, MSS, HY, JK, CMB, RMS, and EMM, coinvestigators on the PURPOSE study, contributed to the study design and data collection; all authors reviewed the final manuscript.

AC K NO W L E D GE M E N T S
The PURPOSe Study Group includes the following contributors: Shivaprasad Goudar, Sangappa M. Dhaded Abbreviation: CSF, cerebrospinal fluid.
Main causes of death among stillbirths investigated in the PURPOSe study.
T A B L E 2 Pathogens identified in placental tissues among stillbirths in the PURPOSe study.
T A B L E 4 Pathogens identified among placentas from preterm neonatal deaths.