Data usefulness in determining cause of stillbirth in South Asia

To evaluate the usefulness of data to determine cause of stillbirth in India and Pakistan.

One reason that so little is known about the causes of stillbirth in LMIC is that until relatively recently, the tools to investigate stillbirths have not been available in many LMIC, thus limiting the accuracy of many cause of stillbirth studies. 4Verbal autopsy, for example, is unlikely to provide useful information on cause of stillbirth.)7][8] The ability to study placental histology has been available for some time, but the publication of the Amsterdam Criteria in 2016 has provided a useful framework to categorise placental lesions. 9nderstanding which data are the most useful for cause of stillbirth determination is difficult, in part because, in recent studies, the cause of stillbirth determination is often made by panel members evaluating all data available for that study.The panellists then choose the most likely cause of the stillbirth, generally using defined guidelines. 10,11Virtually no information is available regarding which data the panel member(s) use to guide their decisions.Untangling the thought process is difficult at best.Available literature suggests that the reasons given for making a specific decision may be subjective and post-hoc to justify the decision. 12However, several studies have attempted to address usefulness of various data to determine cause of stillbirth in HIC settings using different approaches.In particular, the importance of a placental evaluation is illustrated by a US study, The Stillbirth Collaborative Research Network (SCRN) on causes of stillbirth. 13In that study, placental pathology was found to be useful for determining cause of stillbirth in 64.6% of the cases and fetal autopsy in 42.4%, with other tests found to be far less useful. 14tudies from the Netherlands confirm the usefulness of placental pathology in determining cause of death in stillbirths. 15n the US SCRN study, usefulness was defined as data that could either help establish a cause or rule out a cause. 14In another study, a stepwise approach was taken to assess the additive benefit of placental evaluation over clinical data alone. 16iven the limited resources available for studies in many LMICs, we sought to determine usefulness of various types of data to define stillbirth cause of death.For PURPOSe, a study conducted in India and Pakistan, a group of experts prospectively defined the data thought to be useful to inform cause of stillbirth. 10Using the data collected, we conducted analyses to better understand which data were ultimately the most useful.

| M ET HODS
The data used for this endeavour was from the PURPOSe study, conducted from 2018 to 2020 in India and Pakistan. 10rior to that study, we evaluated what was known about the most common causes of stillbirth in LMIC and made sure we had reasonable data to classify a cause of death in those categories.Based on existing literature, we hypothesised that the most common fetal causes of stillbirth in LMIC included asphyxia, infection and congenital anomalies. 4But, because each of those causes could be preceded by a maternal, placental or fetal condition or event, we classified data thought to be useful, if present, under maternal causes, placental causes and fetal causes (Table 1).Thus, the tests that were used to determine stillbirth COD included maternal evaluation, stillbirth evaluation, MITS, placental pathology and histology, and PCR investigation of placenta and tissues. 10ecause this study used data from LMICs, some of the tests recommended in HICs were judged a priori to be either too costly or likely to be less informative regarding the common causes of stillbirth in those locations.For example, studies such as SCRN in the USA 13,17 and TULIP in the Netherlands 15,18 evaluated karyotyping, fetal-maternal haemorrhage and a number of other tests; for the reasons mentioned above, we elected not to perform those tests.
Although there were more than 600 stillbirths enrolled in the overall PURPOSe study, we limited this sub-study to those cases that had MITS, internal organ and placental histology, and PCR evaluation of internal organ tissue and the placenta, as well as maternal clinical history.With a sample size of 200, we estimated that we would be able to detect factors that were useful for at least 10% of the cases with a 95% confidence interval (CI) of 0.06-0.15and in conditions useful for 50% or more, a 95% CI of 0.45-0.57.
We then randomly selected 200 cases for the exercise to define the most useful data.The individual case data included 50 variables which were summarised in a computerised case report format as previously described. 19A study data collection checklist was created to match the data on the case report form.The usefulness reviewer (RLG) reviewed the case report form masked to the panel's choice as to cause of death, and selected the data thought to be useful for determining the cause of death for each case.The criteria to determine the data that were deemed useful, modified based on the SCRN and TULIP studies included (1) data that are feasible to obtain accurately; (2) data that could help determine a cause of death or (3) data that could help eliminate a cause of death. 14,15escriptive statistics were calculated using SAS v.9.4 (SAS).

| R E SU LTS
The maternal characteristics of the mothers of the 200 stillbirths are shown in Table 2. Of these, 119 came from India and 81 from Pakistan.Overall, most of the mothers were 20-30 years of age.Nearly 37% had no formal schooling, with the educational attainment less in Pakistan than in India.Very few of the women in either site worked outside the home.About one-third of the women were primigravidas, more so in India than Pakistan.The table also shows the proportion of women with various clinical conditions.Anaemia (55%), hypertensive disorders (35.5%) and antepartum haemorrhage (19.5%) were especially common.
Table 3 shows the fetal characteristics of the stillbirths evaluated for this study; 51% of the stillbirths were between 1500 and 2499 g, 84% were ≥32 weeks, and 58% of the stillbirths were female.
The data are organised by those variables thought useful to determine a primary fetal cause of death.Table 4 shows the characteristics evaluated for fetal cause of stillbirth usefulness.Of the maternal characteristics, the most common characteristic considered to be useful was maternal hypertension in 72/200 (36.0%), followed by antepartum haemorrhage in 36/200 (18.0%), proteinuria in 20/200 (10.0%), severe anaemia in 17/200 (8.5%) and diabetes in 10/200 (5.0%).Maternal fever, Rh incompatibility, a history of HIV, syphilis or tuberculosis, and detection of fetal heart tones were all thought to provide useful cause of stillbirth data in <5% of the cases.
Table 4 also includes usefulness by the placental findings.From the histology, finding features of maternal vascular malperfusion was deemed useful in 132/200 (66.0%) of the cases, and having features of fetal vascular malperfusion was deemed useful in 41/200 (20.5%).A comparison of placental weight to a standard was judged to be useful in 45% of the cases.Chorioamnionitis (50/200, 25%) and funisitis (15/200, 7.5%) were judged useful in fewer cases than the malperfusion lesions.Evidence of placental abruption was found useful in 20/200 (10%) of the cases.
The data related to external evaluation of the fetus, including birthweight, are shown next.Birthweight for gestational age using the Intergrowth Criteria 20 was deemed useful in 48.5% of the cases.Presence of meconium detected on the external exam was useful in 15.5% of the cases.Various anomalies present on external examination of the fetus including hydrocephaly, hydrops and findings suggestive of a trisomy or a neural tube defect were judged useful in 9/200 (4.5%) of the cases.The presence of a cord complication was helpful in 3/200 (1.5%) of the cases.
Next, the usefulness determinations for organ histology in samples obtained by MITS are summarised.Of these, the finding of aspiration of amniotic debris or meconium in the lung was the finding thought useful most commonly in 39.5% of cases and altogether lung histology was useful in 42.5%.No other internal organ histology finding was considered useful in more than 2% of the cases.Finally, we evaluated the usefulness determinations of the PCR tests of various tissues.Placental PCR findings were thought useful in 26%, brain PCR findings in 15%, cord blood PCR findings in 9%, lung PCR in 4.5%, and liver PCR in 3.5% of the cases.
Distinct from the data in this paper, which focuses on usefulness to determine cause of stillbirth, in a separate analysis of data from the PURPOSe, we present data on the organisms present in the placenta and in the internal organs obtained by MITS.Ureaplasma was by far the most common organism identified in the placenta (27.4%) and was also the most common organism identified in the fetal brain (5.7%),CSF (1.5%) and blood (1.8%).Escherichia coli/Shigella was the next most common in brain (1.5%), CSF (2.9%) and blood (1.5%).The only other organisms found in >1% of any tissue were Group B Streptococcus (1.3%) and Staphylococcus aureus (1.1%).

| DISCUS SION
For the PURPOSe study, we asked the panel to designate a maternal cause of death, a placental cause of death and

Fetal blood X
Cord blood X Placental/cord/membrane X a fetal cause of death, as applicable. 10[23][24][25][26] While there is no universal standard to define 'usefulness', for our purposes, useful data were defined based on prior HIC studies, i.e., useful to determine or to help eliminate a cause of stillbirth, but adapted to focus on the most common causes of stillbirth in LMIC. 14,15For this exercise selecting the data 'useful' to determine cause of stillbirth, we took the approach that an experienced observer, after evaluating all the data available for a given case of stillbirth, but without access to the panel's choice, would note the potentially useful information.We are aware that many other approaches could have been used, but do not believe one would be more valid than the other approaches.From this analysis, certain findings regarding usefulness to determine cause of stillbirth stand out.Of all the findings, placental pathology findings were found to be the most useful, with maternal vascular malperfusion the most useful of all the characteristics evaluated. 27Comparing the weight of the placenta and the weight of the fetus to recognised standards were each found to be useful in nearly half the cases.Another test commonly found useful (in nearly 40% of the stillbirths) was lung histology: aspiration, usually of amniotic fluid debris or meconium on lung histology, was especially useful.Most of the other findings based on internal organ histology were only occasionally deemed useful.Meconium could be discovered in the placental examination, in the external evaluation of the fetus and in the lung histology, and wherever found was considered useful in helping to determine cause of stillbirth. 28Signs of abruption, whether by maternal history or placental histology, if present, were always deemed useful.PCR findings of specific organisms in the placenta, brain, CSF and cord blood were often deemed useful.
Many of the findings that were deemed useful helped to provide evidence for asphyxia as the cause of stillbirth.Maternal hypertension and urine protein, haemorrhage and signs of abruption on placental exam, severe anaemia, a small fetus, a small placenta, meconium, placental malperfusions and fetal lung aspiration all were useful in pushing the cause of death needle toward asphyxia. 29Findings useful to move the needle toward infection were rarer and less often thought to be conclusive.Finding various placental pathogens, some pathogens in internal organs, chorioamnionitis and even funisitis were often not by themselves convincing, and positive lung and brain histology with signs of inflammation together with positive PCR findings, although convincing, were rarely found.
Thus, based on this analysis of data thought to be useful in determining stillbirth cause of death and to recommend an efficient and cost-effective approach for use in LMIC, the authors recommend focusing on the clinical history, the placental evaluation, the external examination of the fetus and, when available, fetal tissue evaluation (obtained by MITS) for histology and organism identification.With MITS and histology availability, we would especially recommend histological evaluation of the fetal lungs.With potential for PCR evaluation using a multiplex PCR platform for determination of organisms, we would recommend evaluation of brain/CSF and cord blood.Because multiplex PCR evaluation for multiple organisms might be beyond the capabilities of a LMIC site, or might be considered too costly, targeting identification of the most common organisms found in this study (Ureaplasma urealyticum, E. coli/Shigella, S. aureus and GBS) either by culture or individual PCR tests, might be a reasonable alternative.

| STR E NGTHS A N D W E A K N E SSE S
This paper has a number of strengths and weaknesses.Among the strengths are the large number of stillbirths with a standard set of data including clinical information, MITS collection of internal organ tissue for histology and pathogen PCR, and histological and PCR evaluation of the placenta.
Weaknesses include the absence of a number of tests used in high-income countries, especially related to genetic abnormalities and full autopsy.We recognise that this approach will not identify some causes of stillbirth, including some genetic abnormalities and internal organ anomalies, but we believe it will identify the most common causes of stillbirth and most of the preventable causes of stillbirth in LMIC 30,31 and may be feasible in many locations.In a paper on usefulness of diagnostic tests for stillbirth in HIC, Page et al. note: 'In a theoretical sense, all tests have some utility with regard to the exclusion of potential causes of stillbirth.However, potential benefits must be weighed against expense, patient and health system burden and potential of a false-positive result'. 14The recommended data to discern stillbirth cause of death in LMIC derived from this report represent an attempt to develop a balanced approach for LMIC cause of stillbirth determination based on these principles.

| CONCLUSIONS
In summary, focusing on the maternal clinical history, the placental histology, the external examination of the fetus and a comparison of placental weight and birthweight to recognised standards were the most informative to determine cause of stillbirth.When available, fetal tissue evaluation (obtained by MITS) for histology and PCR were also informative.Of all the potential tests using MITS specimens, we would first recommend histological evaluation of the lungs, and using a multiplex PCR platform would determine pathogens in blood and brain/CSF.We recognise 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 and most of the preventable causes of stillbirth in LMICs.

AU T HOR C ON T R I BU T ION S
RLG conceived the analyses with support from KH, SS, SSG, RMS, NG and EMM.KH conducted the statistical analyses with EMM.SS, SST, SY, VK, NG, SH, IA, ZU, SSG, GG and SD oversaw study implementation and monitored data quality.All authors reviewed and approved the final article.

AC K NOW L E D G M E N T S
We thank the leadership of the clinical sites and the subjects at each site who participated in this study.

C ON F L IC T OF I N T E R E S T S TAT E M E N T
None declared.

DATA AVA I L A BI L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.

T A B L E 1
Prior to the study, data believed to be useful to help determine asphyxia, infection and a congenital anomaly as the fetal cause of stillbirth.

E
T H IC S A PPROVA L The study protocol was approved by respective ethical review committees of participating institutions from India and Pakistan and the United States.All procedures were conducted per hospital protocol following receipt of informed written consent.ORC I D Elizabeth M. McClure https://orcid.org/0000-0001-8659-5444 R E F E R E NC E S Maternal characteristics of the 200 stillbirths evaluated.
T A B L E 2

Data deemed useful to determine fetal cause of stillbirth (%) Total evaluated (n) 200
Findings deemed useful to determine cause of stillbirth.
T A B L E 4