Criteria for assigning cause of death for stillbirths and neonatal deaths in research studies in low-middle income countries

Abstract

Accurate knowledge regarding cause of death (COD) for stillbirths and neonatal deaths is crucial, especially in low-income countries, in order for public health and medical officials to choose appropriate interventions likely to reduce these deaths. To date, many of the COD studies in these areas have relied only on obstetric or neonatal clinical information and the determination of COD is likely to be inaccurate. Information related to infectious COD is especially lacking. Thus, without more sophisticated testing, data as currently collected only provide a very weak approximation of the COD and may well lead to adoption of interventions of limited usefulness. In this commentary, we propose recommendations regarding the type of data needed to determine with reasonable accuracy the COD for stillbirths and neonatal deaths in low-resource settings. Using these data, and a method to determine the degree of certainty, we then propose definitions for the most common COD. Our goal is to reduce subjectivity and provide more specificity for the tests used in existing classification systems so that the methodology of COD determination is transparent and able to be replicated over time and from location to location.

Introduction

Worldwide, 2.6 million neonatal deaths occur annually [1]. In most countries, the stillbirth rates are equivalent to or greater than the neonatal mortality rates, with about 3 million annual third trimester stillbirths [2,3]. The addition of second trimester stillbirths (20–27 weeks) increases the estimated number of stillbirths worldwide to more than 5 million annually [4]. The vast majority of neonatal deaths occur in low- and some middle-income countries (LMIC) principally in sub-Saharan Africa and South Asia, with rates often exceeding 30/1000 births compared to rates of about 3/1000 in some high-income countries (HIC) [4]. Stillbirth rates are also the highest in South Asia and sub-Saharan Africa, with rates as high as 40–50/1000 births compared to 2–3/1000 in many HIC [3].

Historically, in determining the causes of neonatal death, preterm birth (PTB) has been attributed as one among the causes [1,5]. However, preterm neonates die from a variety of prematurity-related complications such as respiratory distress syndrome (RDS), necrotizing enterocolitis (NEC), and intraventricular hemorrhage (IVH), and also from conditions not specifically caused by a PTB such as asphyxia, infection, and congenital anomalies [6,7]. Stillbirths are caused by infection and congenital anomalies as well as a variety of maternal and fetal conditions leading to asphyxia including placental abruption, obstructed labor, preeclampsia, placental malfunction, and cord complications [3]. The conditions that cause fetal asphyxia also contribute to neonatal mortality through birth asphyxia [8].

Despite this general understanding about causes of death, lack of knowledge about the medical conditions that cause neonatal mortality and stillbirth have impeded efforts to reduce mortality, especially in LMIC. In order to prioritize interventions to improve these outcomes in low-resource areas, it is crucial not only to know the contribution of the major medical or pathological causes of death, but also the sequence of events which led to the death. In HIC, when a stillbirth or neonatal death occurs, clinical as well as postmortem data are generally available to determine the cause of death (COD). Even then, clinicians assigning COD often reach different conclusions about COD [9,10]. In LMIC, clinical data are often sparse, and since laboratory tests, imaging, and autopsies are rarely performed, COD is even more difficult to determine [11].

We have previously proposed algorithms for assigning COD for stillbirths and neonatal deaths with information obtained from lay personnel when limited medical data is generally available [12]. While producing estimates consistent with other LMIC reports, all are hindered by limited data and important causes of death are likely to be missed. The data generally available in LMIC to inform COD often includes only a limited obstetric and/or neonatal history. For hospital admissions, laboratory investigations including white blood cell count, hemoglobin or hematocrit, and C-reactive protein (CRP) may be done, but bacterial cultures to suggest infection and imaging such as chest x-ray or ultrasound are not usually performed. Autopsy is rare in most low-resource settings [13]. Since much of the data required to accurately determine the COD is not routinely collected, the available clinical data may fail to determine the true COD in many LMIC.

To accurately determine COD with reasonable accuracy, testing such as imaging, cultures, polymerase chain reaction (PCR), or other techniques to document infection, placental histology and access to postmortem tissue must be available [14]. Since autopsies are uncommon, minimally invasive tissue sampling (MITS) – a needle biopsy of a specific organ postmortem to obtain tissue for histological examination or organism identification by culture, PCR or other techniques – has been developed as an alternative [15,16]. While still being validated, MITS could provide information similar to the traditional autopsy.

The objective of this commentary is to describe the data needed to determine with reasonable accuracy the COD for stillbirths and neonatal deaths in low-resource settings and using these data to propose definitions for the most common COD. Our goal is to reduce subjectivity and provide more specificity for the tests used in existing classification systems, not to create a new one. Determination of an accurate cause of neonatal death and stillbirth will help prioritization of essential interventions for the most common conditions, and should improve planning and policy decision-making in LMIC.

Materials and methods

Guidelines

As a first step, guidelines for assigning COD for both stillbirths and neonatal deaths are important to improve standardization. Since the World Health Organization (WHO) recommends assigning one primary COD for perinatal mortality, for consistency, this recommendation is used [17]. The primary cause is defined as the condition without which the death would not have occurred [18]. When two or more potential causes are present, we will suggest how one cause can be assigned as primary, with other potential COD designated as contributing causes.

The degree of certainty of the COD assignment is important and is integrated into some classification systems [14,19]. For example, some classification systems assign probable or possible causes to address this issue [14]. To designate a condition as a probable COD requires that all the evidence needed to classify the condition as the COD is present, while to classify a condition as a possible COD requires reasonable evidence that the condition caused the death, without meeting all the criteria. We describe the data we believe are necessary to accurately determine the COD for stillbirths and neonatal deaths in the following sections (Table 1).

Table 1.

Common causes of stillbirth and neonatal death and type of data that may inform designation.

	OB history	Clinical data	Laboratory testing	Imaging	MITS	Autopsy
Stillbirth
Asphyxia	X	X
Infection	X	X	X		X	X
Congenital anomaly		X	X	X		X
Injury	X	X		X		X
Maternal medical condition	X	X	X
FGR	X	X
Neonatal death
RDS	X	X		X	X	X
IVH		X	X	X		X
NEC		X		X		X
Apnea		X
Asphyxia	X	X
Infection	X	X	X		X	X
Congenital anomaly		X	X	X		X
Injury	X	X		X
Anemia		X	X
Hypothermia		X
Pulmonary hemorrhage		X			X	X

Stillbirth

Stillbirths are defined as fetuses delivered at ≥20 weeks gestations who die prior to delivery [1]. Consistent with other LMIC studies, in prior research we found that the most common cause of stillbirth is fetal asphyxia following maternal or fetal conditions including obstructed labor, preeclampsia, abruption and previa, umbilical cord complications, and placental malfunction due to thrombosis, necrosis, and fibrosis [20–23]. Infection is likely the next most common cause of stillbirth. In LMIC, more than 100 different organisms may be responsible for an infectious etiology of stillbirth, but with the exception of syphilis and malaria, the proportion of stillbirths caused by any specific infection including viruses, fungi, and parasites is unknown [24]. Congenital anomalies play an important role in the relatively low stillbirth rate in HIC but represent an unknown proportion in LMIC [25]. Fetal trauma due to maternal accidents or during delivery may cause stillbirth, but its frequency in LMIC is also unknown. The presence of maceration indicates that the death likely occurred prior to labor and is helpful in determining when death occurred and potentially the condition leading to the death [20].

Asphyxia

Since fetal asphyxia is often due to maternal medical or obstetric conditions, the obstetric history, which should include prolonged or obstructed labor, preeclampsia/eclampsia, antepartum or intrapartum hemorrhage, and cord complications, is crucial. Second, histological evaluation of the placenta may provide considerable information related to an asphyxic cause of stillbirth, including thrombosis, necrosis and fibrosis, presence of meconium, abruption and previa and evidence of some cord complications such as a true knot or vasa previa. Nonmacerated stillbirths are also more likely to have had an intrapartum death due to asphyxia than macerated stillbirths. Since autopsy findings related to asphyxia are often nonspecific [26,27], the obstetric history and placental examination are sufficient to assign asphyxia as the probable primary COD. In the absence of a placental examination, the obstetric history may be sufficient to assign asphyxia as the possible COD.

Infection

Among the causes of stillbirth, infection is the most difficult to confirm. Since infection of virtually any fetal organ or the placenta may cause a stillbirth by at least five different mechanisms, and with over 100 potential pathogens, determining that infection is the cause of stillbirth is often problematic [28]. The evidence required to designate an infectious cause of stillbirth varies by type of infection. Multiple sources of data including pathological and microbiological data are generally required. Histological analyses of the placenta or other fetal tissue may suggest infection of a specific organ, with samples obtained from fetal blood, cerebral spinal fluid (CSF), or needle aspiration through MITS or autopsy. Culture, polymerase chain reaction (PCR), or microbiome analyses can then confirm the presence of organisms in specific tissue.

To assign an infection as the probable cause of stillbirth, evidence of the organism being in fetal blood or in a specific organ and some tissue evidence of organ involvement such as white blood cell (WBC) infiltration into an organ or a high WBC in blood is generally needed. Various types of maternal or fetal antibodies might suggest a possible cause of stillbirth, but in general, evidence that the organisms affected the placenta or a fetal organ is required to designate infection as a probable primary cause of stillbirth. As an example, maternal antibodies to parvovirus and signs of a severe fetal anemia should enable parvovirus to be designated as the probable primary COD. Malaria and syphilis are two of the most common infections associated with stillbirth. If the mother has confirmed active malaria during pregnancy with evidence of placental malaria on placental histology, malaria can be considered as the probable primary cause of stillbirth. For syphilis, if the mother has treponemal antibodies and the fetus has phenotypic evidence of syphilis, syphilis can be considered as the probable primary cause of stillbirth.

Congenital anomalies

Assigning a congenital anomaly as a cause of stillbirth requires an ultrasound examination of the fetus in utero, direct examination after birth, imaging the still-birth after birth or an autopsy. Finding a major anomaly such as anencephaly or other lethal anomalies by direct examination of the fetus is sufficient to designate a congenital anomaly as the probable primary cause of stillbirth. If a phenotype is consistent with one of the trisomies, congenital anomaly should be considered as the probable primary COD. Findings obtained at autopsy or postmortem imaging including anomalies of a major organ system should allow the assignment of congenital anomaly as the probable primary cause of stillbirth. Evidence from DNA sequencing or karyotyping, if available, also may confirm a genetic abnormality as a COD.

Fetal injury

While likely rare, the extent of injuries causing stillbirth is unknown. Evidence necessary to assign fetal injury as the probable primary cause of stillbirth includes an obstetric history documenting trauma such as a maternal accident, violence, forceps or vacuum use or breech extraction during delivery and evidence of trauma on visual or imaging inspection of the fetus.

Maternal medical conditions

Many maternal medical conditions are associated with stillbirth including diabetes, renal disease, sickle-cell disease, and other red cell abnormalities [29]. The obstetric history should be evaluated for maternal medical conditions associated with stillbirth. While many women are not routinely screened for these medical conditions in LMIC, if one is present without other likely the causes of stillbirth, the maternal medical condition could be assigned as the probable primary COD.

Fetal growth restriction

Poor fetal growth is often considered an important cause of stillbirth, but whether fetal growth restriction (FGR) causes stillbirth per se or is an underlying condition causing both the decreased growth and the fetal death is often difficult to determine [30]. In order to assess FGR, the best available gestational age and birth weight data must be available. If FGR is diagnosed, it should be assigned as a probable primary cause of stillbirth only if no other potential etiology is present. If other factors are present, FGR should be considered as a contributing COD.

Other causes of stillbirth

Other causes of stillbirth include Rh and other isoimmunizations, antiphospholipid antibodies, fetal/maternal hemorrhage and cholestasis, which are rare in HIC and likely rare in LMIC and often not diagnosed. Nevertheless, when these conditions are documented, they may be considered as the primary cause of stillbirth.

Neonatal death

Neonatal deaths are defined as deaths among live born infants which occur prior to 28 days of life. Available data suggest that the most common causes of neonatal death in LMIC include asphyxia, infection, preterm birth, and congenital anomalies [1]. Being preterm and especially very preterm, places the infant at risk for conditions which are unique to prematurity [5]. As prematurity is often considered itself to be the primary COD, the proportion of preterm neonatal deaths occurring secondary to specific conditions such as RDS, NEC, IVH, infection due to various organisms, as well as asphyxia and congenital anomalies has rarely been determined [5]. Here, we attempt to determine the actual COD among preterm infants as well as among those term infants that die.

Causes of neonatal death related directly to PTB

A preterm infant is defined as a live born infant delivered at <37 weeks gestation. Gestational age may be determined by last menstrual period (LMP), ultrasound or neonatal exam (Ballard) [31]. Algorithms using existing data to choose a “best gestational age” have been described [31–33].

Respiratory distress syndrome (RDS)

To designate RDS as a probable COD requires the infant to be preterm with a clinical history including worsening respiratory distress plus an x-ray compatible with RDS. Histologic findings of hyaline membrane disease in tissue samples obtained by autopsy or MITS in a preterm infant with a history of respiratory distress will also allow RDS to be called the probable primary COD. With a compatible clinical picture, but no autopsy, MITS or x-ray and no laboratory or microbiological evidence of infection, RDS may be considered as a possible COD.

Intraventricular hemorrhage (IVH)

IVH can occur spontaneously in preterm infants or may follow RDS or asphyxia. Autopsy findings of IVH in a preterm infant with a clinical picture such as bulging fontanels, decreasing hemoglobin and seizures will allow IVH to be called a probable primary COD. In the absence of an autopsy, for IVH to be called a probable COD, a clinical picture of IVH plus a cranial ultrasound compatible with IVH is required. With a compatible clinical picture, but no autopsy or cranial ultrasound, and no evidence of infection, IVH should be considered as a possible COD. If IVH occurs following another condition such as RDS or asphyxia, IVH should be considered as a contributing condition, unless the RDS or asphyxia appears to be resolving and the IVH appears to have occurred spontaneously.

Necrotizing enterocolitis (NEC)

NEC can occur spontaneously in preterm infants or may follow RDS or asphyxia. Autopsy or MITS findings of NEC in a preterm infant with a clinical picture of NEC (i.e. a distended abdomen and poor feeding) allows NEC to be called a probable primary COD even without an abdominal x-ray. In the absence of an autopsy, for NEC to be called a probable primary COD, clinical symptoms of NEC plus an x-ray compatible with NEC are required. With a compatible clinical picture only, but no autopsy and no abdominal x-ray, NEC should be considered as a possible COD.

Apnea of prematurity

There are no autopsy, imaging or laboratory tests to designate apnea as a primary COD. Instead, clinical observation or respiratory monitoring showing cessation of breathing may suggest the diagnosis. To designate apnea as the probable primary COD, in addition to observation of breathing cessation, often three or more instances of apnea in 1 hour, all other potential COD must be ruled out. If apnea is present but there is another probable COD such as RDS or asphyxia, apnea should be considered as a contributing COD.

Causes of death among term or preterm infants

In this section we review the predominant conditions that may be a COD among either term or preterm infants, with emphasis on neonatal mortality at <28 days of life.

Asphyxia

Asphyxia may occur in term or preterm infants. To designate asphyxia as a probable primary COD in term or preterm infants, it requires both a clinical picture of asphyxia (including resuscitation and breathing status), and a source of the asphyxia such as an obstetric or neonatal complication. Deaths caused by asphyxia generally occur within the first 2 days of life. If these conditions are met, asphyxia can be designated as the probable primary COD. For infants with multiple organ system failure compatible with severe asphyxia and with obstetric and/or neonatal precursors of asphyxia, asphyxia may also be designated as the probable primary COD. With a clinical picture of asphyxia but no history of an obstetric or neonatal precursor, asphyxia can be designated as a possible COD.

Infections

Infections related to neonatal death include sepsis, tetanus, malaria, and syphilis, but many different infections have been implicated including deaths caused by bacteria, viruses, fungi, and many protozoa [34,35]. The infections can be general (sepsis) or involve specific organs such as the lung, liver, and brain.

Sepsis, pneumonia, meningitis

The clinical picture of infection in infants varies. Tests on tissue samples to diagnose a specific infection include culture, PCR, microbiome, and antibodies to various organisms in maternal serum or cord/neonatal blood. Specific organ cultures can be obtained clinically, at autopsy or by MITS, and cord blood at delivery or neonatal blood [16]. Maternal serum may be obtained for serum antibody analysis. Lung, liver and other organ samples obtained by MITS or autopsy used for histology can be used to diagnose pneumonia as well as other infections.

Autopsy or MITS findings of infection using histology (generally WBC infiltration into the lungs, brain, or other organ) in an infant should allow designation of infection as the probable primary COD. Whenever possible, the organ system most affected by the infection should be named (pneumonia, meningitis, etc.). In the absence of an autopsy or MITS, for infection to be called a probable primary COD, culture or PCR evidence of blood, spinal fluid, lung, urine, or other organ infection is required along with clinical evidence of infection described earlier. With only a compatible clinical picture or culture evidence, infection may be considered as a possible COD.

Tetanus

Designating tetanus as the probable primary COD is usually based on the clinical picture, including umbilical infection, progressive stiffness, inability to feed and seizures, rather than autopsy or other investigations.

Syphilis

If the mother has treponemal antibodies and the newborn has phenotypic evidence of syphilis, syphilis can be considered as the probable primary COD.

Malaria

If the mother has confirmed malaria and there is evidence of placental malaria, and the infant dies an apparent asphyxic death, and there are no other obvious causes of asphyxia, malaria can be considered as the probable primary COD.

HIV

HIV may cause neonatal death following in utero transmission or transmission during breastfeeding. To consider HIV as the probable primary COD, evidence of active neonatal HIV infection must be present and other COD should be ruled out.

Pulmonary hemorrhage

Pulmonary hemorrhage, which may follow many other insults, including RDS, asphyxia, infection and hypothermia, is a common finding on autopsies. For pulmonary hemorrhage to be considered as a probable primary COD, bleeding from the lungs must be present and potential precipitating conditions should be ruled out. For pulmonary hemorrhage to be considered as a possible COD, bleeding from the lungs must be present and any potential precipitating conditions should be unrelated to the hemorrhage. All other cases of pulmonary hemorrhage should be considered as a contributing COD.

Congenital anomalies

A congenital anomaly can be diagnosed by ultrasound examination of the fetus in utero, by direct examination of the neonate after birth, by imaging after birth or by information gained at the time of autopsy [25]. To designate a congenital anomaly as probable COD, one of the following must occur: identifying a major anomaly such as anencephaly or other lethal anomalies by direct examination; a phenotype consistent with one of the trisomies; autopsy findings including anomalies of one of the major organ systems; or, evidence of an anomaly from DNA sequencing or karyotyping.

Injury

Evidence to assign injury as a cause of neonatal death should include an obstetric history documenting potential trauma such as forceps, vacuum, or breech extraction and evidence of trauma on visual inspection or imaging of the newborn.

Anemia

Anemia should be defined based on the gestational age criteria [36]. Anemia generally occurs secondary to another condition such as umbilical cord hemorrhage, fetal maternal transfusion, Rh disease, or a red cell abnormality. Anemia is therefore generally a contributing COD secondary to the condition causing the anemia which will be the probable primary COD. Anemia should be considered as a probable primary COD only in cases with a low hemoglobin or hematocrit and no evidence of a competing etiology. If another likely COD is present, anemia should be considered as a contributing COD.

Hypothermia (neonatal cold injury)

Hypothermia is common; however, infants rarely die of hypothermia alone [37]. Hypothermia would generally be considered as a contributing cause but may be the probable primary COD when severe and prolonged. To designate hypothermia as a probable primary COD, a documented body temperature <35.5 C is necessary over at least 2 days, or a temperature of <32 C on at least one occasion, and depressed vital organs such as anuria, shock, disseminated intravascular coagulation (DIC), or multiorgan damage is required. To designate hypothermia as a possible COD, requires a documented body temperature <35.5 C for at least 2 days, or a temperature of <32 C on at least one occasion, and that no other COD such as RDS, or infection be present. If the infant had sustained body temperatures <35.5 C and a specific competing COD is present, hypothermia should be considered as a contributing COD.

Jaundice/kernicterus

Jaundice is a symptom of hyperbilirubinemia and by itself rarely is a COD. However, extended high levels of bilirubin may lead to kernicterus and neonatal death. With sustained high bilirubin levels and neurological signs compatible with kernicterus and no other competing COD, kernicterus may be considered as the probable primary COD. With clinical jaundice but no bilirubin levels and the presence of neurological signs, in the absence of other COD, kernicterus may be considered as a possible primary COD. Otherwise, jaundice should be considered as a contributing COD. More commonly, jaundice may be a symptom of another condition such as Rh disease and in those cases the condition causing the jaundice should be considered as the probable primary COD while jaundice may be a contributing COD.

Competing causes of death

There are often two or more potential primary COD of death to choose among at the time of designation of the primary COD. While some judgment is required to choose among the potential causes, we found that having some guidelines to choose the primary COD among potential causes of stillbirth and neonatal death is helpful. Our guidelines for choosing among potential causes of stillbirth and neonatal death are shown in Tables 2 and 3.

Table 2.

Recommendations for assigning primary stillbirth COD when two competing stillbirth COD are present.

1. If a major congenital anomaly is present, regardless if any other potential COD is present such as infection or asphyxia, congenital anomaly will be designated as the primary COD.

2. If major trauma is present without a major congenital anomaly, regardless if any other potential COD is present such as infection or asphyxia, trauma will be considered as the primary COD.

3. In the absence of trauma or congenital anomaly, if malaria or syphilis is diagnosed, even if asphyxia is present, malaria or syphilis will be considered the primary COD.

4. In the absence of an anomaly, trauma, malaria or syphilis, if asphyxia is present, asphyxia will be considered the primary COD even if another infection is present.

5. In the absence of trauma, congenital anomaly, malaria, syphilis and asphyxia, if sepsis or other serious infections are present, infection will be considered the COD.

6. Maternal medical conditions will be considered a primary COD only if conditions such as anomalies, trauma, asphyxia, and infection are not present.

7. If any other potential COD is present, FGR will be considered a contributing COD.

Table 3.

Recommendations for assigning primary neonatal COD when two competing neonatal COD are present.

1. If RDS is present and the infant subsequently develops pulmonary hemorrhage and dies, RDS will be considered the primary COD.

2. If RDS is present and the infant subsequently develops apnea and dies, RDS will be considered the primary COD.

3. If RDS is present and the infant subsequently develops IVH and dies, if the RDS had been stable and IVH developed acutely followed by death, IVH will be considered the primary COD.

4. If RDS is present clinically and the infant dies an apparent death from respiratory failure, if RDS and IVH are both found at autopsy, RDS will be considered the primary COD.

5. If RDS is present and the infant subsequently develops IVH and dies, if the RDS had not been stable and IVH developed acutely followed by death, RDS will be considered the primary COD.

6. If RDS is present but stable or improving, and the infant subsequently develops apparent sepsis and dies, sepsis will be considered the primary COD.

7. If sepsis is present and the infant subsequently develops DIC and dies, sepsis will be considered the primary COD.

8. If a major congenital or chromosomal anomaly (such as anencephaly or trisomy 18) is present and the infant subsequently develops RDS or sepsis and dies, the major anomaly will be considered the primary COD.

9. If there is strong evidence for neonatal asphyxia, even if the infant subsequently develops DIC, pulmonary hemorrhage or other complications of asphyxia, asphyxia should be considered the primary COD.

10. Conditions such transient tachypnea of the newborn (TTN), hypothermia, hypoglycemia, anemia and apnea will not be considered the primary COD unless there are no other competing COD identified. In the presence of conditions such as RDS, asphyxia, sepsis and congenital anomaly, these conditions should be considered contributing COD.

11. If neonatal asphyxia is present, and followed by pulmonary hemorrhage, meconium aspiration and/or primary pulmonary hypertension (PPHN), asphyxia will be considered the primary COD.

12. If neonatal asphyxia is present and the infant develops IVH, asphyxia will be considered as the primary COD and IVH a contributing condition, unless the asphyxia appears to be resolved. If so, IVH will be designated as the primary COD and asphyxia a contributing condition.

13. If meconium aspiration and PPHN are both present, PPHN will be considered the primary COD.

14. If meconium aspiration or aspiration of other material is confirmed on autopsy and there is no evidence of PPHN, pneumonia should be considered the primary COD.

15. Unless there is an absence of all other competing COD, fetal growth restriction will only be considered a contributing COD.

16. The COD in infants with a renal or other urinary tract abnormality that results in renal failure will have renal failure designated as the COD with congenital anomaly designated as a contributory COD.

17. If conditions such as Rh disease or any other likely COD is present, jaundice should be considered a contributing condition. With no other likely COD, jaundice with kernicterus can be considered a primary COD.

18. Other than the condition designated as the primary COD by the above guidelines, all other competing conditions that may have contributed to the death should be designated as contributing conditions.

Discussion

In summary, accurately designating the COD for both stillbirths and neonatal deaths, especially in LMIC, is important for planning, prioritizing, and instituting interventions and policies to improve those outcomes. Using a transparent and consistent methodology for COD determination is important so that results from one location to another, and over time in the same location, can be compared. We believe that using a carefully created set of guidelines for assigning COD will eliminate many of the inconsistencies in COD reporting found in previous reports.

While it would be useful to have all appropriate clinical information and testing as well as postmortem investigations including autopsy for all deaths, any COD system has to deal with the issue of missing information. For that reason, we believe that the guidelines used to assign COD need to focus on the data required to designate a specific COD and to describe the degree of confidence in that assignment. We believe that the guidelines described earlier are steps in that direction. We are also aware that many of the tests addressed in these guidelines may not be routinely available in most LMIC settings. Nevertheless, if progress is to be made in reducing stillbirth and neonatal death in LMIC, accurate and complete data to inform the COD determination will be necessary. Sentinel studies, often funded by international organizations, will likely be necessary to acquire accurate COD data such as those described earlier. These studies are essential to ensure that important conditions contributing to the high stillbirth and neonatal mortality rates in LMIC are not missed.