Timing and cause of perinatal mortality for small-for-gestational-age babies in South Africa: critical periods and challenges with detection

Background Little information exists on timing and cause of death for small-for-gestational-age (SGA) babies in low-and-middle-income-countries (LMICs), despite evidence from high-income countries suggesting critical periods for SGA babies. This study explored the timing and cause of stillbirth and early neonatal mortality (END, <7 days) by small-for-gestational age in three provinces in South Africa. In South Africa, the largest category of perinatal deaths is unexplained stillbirth, of which up to one-quarter have intra-uterine growth restriction. Methods Secondary analysis of the South African Perinatal Problems Identification Program (PPIP) database allowed for the analysis of gestational age at death and clinically confirmed diagnosis of stillbirth and early neonatal death (END) (>1000 g and >28 weeks) across gestation. Comparisons by province, size-for-gestational-age, gestational age groups, and maternal condition at death were performed. The provinces investigated were: Western Cape (fortnightly antenatal care visits from 32 to 38 weeks), Limpopo and Mpumalanga (no antenatal care visits between 32 to 38 weeks). Results There were 528,727 births in the study period and 8111 stillbirths and 5792 early neonatal deaths. Similar timing of deaths across gestation was seen for the three provinces with the greatest proportion of deaths for SGA babies at 33–37 weeks (stillbirths 52.9 %; END 43.3 %; p < 0.05). SGA babies had a greater proportion of deaths due to hypertension (SGA22.9 %; AGA 18.6 %; LGA 18.6 %; p < 0.05) and intrauterine growth restriction (SGA 6.8 %; AGA 1.7 %; LGA 1.4 %; p < 0.05). No increase was seen in poor maternal condition for SGA babies and 54.9 % of deaths had a healthy mother. Of mothers that were healthy the greatest proportion of SGA stillbirths were due to unexplained intrauterine death (53.9 %). Conclusion There was a peak in stillbirths for SGA babies 33–37 weeks in all provinces. Detecting SGA is further complicated as in most cases the mother is healthy. Further research into Umbiflow Doppler velocimetry use in low-risk populations is warranted and may be a viable strategy to increase current detection of SGA babies at risk of mortality in LMICs.


Background
Approximately 25 % of children born in low-and-middleincome countries (LMICs) are small-for-gestational-age (SGA) [1]. Babies who are SGA are at increased risk of mortality and neonatal morbidity [2,3], making the detection and clinical management of such infants crucial. The condition may constitute a small but healthy fetus or be due to pathological growth failure (intrauterine growth restriction (IUGR)) [4]. The purpose of identifying SGA fetuses, defined as a birth weight in the lowest decile on standard growth curves [5], is to recognise those most at risk of poor outcomes [3].
Studies to date have indicated that there may be a critical period for increased mortality for SGA babies [3,5,6]. A UK study found that stillbirths between 28 and 36 weeks were increased for SGA babies [5]. Other studies have found that SGA babies are at increased risk of stillbirth compared to non-SGA babies at all gestational ages [3], and that the risk of stillbirth for SGA babies increases with advancing gestational age [6].
In South Africa, the largest category of perinatal deaths is unexplained stillbirth, of which up to one-quarter have IUGR [7]. Early recognition can prevent some of these deaths. Early antenatal detection of SGA babies remains a challenge in LMICs, but is important as most deaths occur in the late preterm or term period, where survival of live born infants in well-resourced units is high [8]. As detection of SGA at the population level is challenging due to resource constraints, identifying critical periods and causes of mortality across gestation may elucidate the best approach. This study explores the gestational age at death and cause of stillbirth and early neonatal mortality (up to 7 days neonatal life) by size-for-gestational age in three South African provinces.

Methods
Secondary analysis of the South African Perinatal Problems Identification Program (PPIP) database allowed for the analysis of gestational age at death and clinically confirmed diagnosis of stillbirth and early neonatal death (END) across gestation. The program also allowed for comparisons between SGA, appropriate-for-gestationalage (AGA) and large-for-gestational-age (LGA) babies from 1 October and 2013 and 31 August 2015 and between three provinces: Western Cape, Limpopo and Mpumalanga. Western Cape has fortnightly antenatal care visits between 32 and 38 weeks, while Limpopo and Mpumalanga do not have visits between 32 and 38 weeks, preventing the opportunity for detection of SGA at these gestations. These provinces were chosen (from nine available) as they have the greatest PPIP coverage, auditing >90 % of perinatal deaths. PPIP is a perinatal quality audit system that has been described in detail elsewhere [7,9]. Briefly, at each clinical site (n = 292) across the three provinces the clinical team perform a review shortly after a death has occurred. The primary obstetric cause of death was defined by the PPIP technical team as the main obstetric event or pregnancy occurrence which was integral in the pathway to perinatal death, as described in other published studies [7]. IUGR was also identified at the time of death through clinical evaluation. Maternal condition at the time of death was also recorded. Categories for the primary obstetric cause of stillbirth and early neonatal death were as follows: antepartum haemorrhage, spontaneous preterm labour (intrapartum stillbirth), unexplained intrauterine death, fetal abnormality, hypertensive disorders, infections, intrapartum asphyxia, intrauterine growth restriction, maternal disease, miscellaneous (rhesus isoimmunisation, twin-to-twin transfusion, extra-uterine pregnancy and other cause of death not described), no obstetric cause and trauma. The maternal condition is defined as either healthy (where the clinician examining her could not find any clinical problems) or the occurrence of a recognised medical or obstetric complication (eg. cardiac, endocrine, respiratory disease or other disease that are an indirect cause of morbidity), categorised as coincidental conditions, medical and surgical disorders, non-pregnancy related infections, extra-uterine pregnancy, pregnancy-related sepsis, obstetric haemorrhage, hypertension, anaesthetic complications, embolism, and acute collapse (cause unknown).
Gestational age was calculated based on date of last menstrual period, ultrasound or clinical examination and cases were excluded if the gestation age was unknown or if the estimated age was considered 'uncertain'. There was no hierarchy employed based on method of gestational age estimation. Detailed data were extracted on all stillbirths >1000 g and 28-42 weeks gestation and early neonatal deaths less than 7 days of neonatal life. Data were only included up to 42 weeks gestation as Theron weight distribution curves are not considered reliable for growth measurements after 42 weeks gestation [10,11]. Only women who had reported receiving antenatal care were included as SGA can be detected antenatally. Data extracted from the three provinces represent 58.6 % of all deaths in South Africa that met the study criteria. Birth weight for gestation was obtained from Theron charts. SGA was defined as neonates with <10th centile for gestational age based on South African specific growth charts [10,11].

Statistical analysis
Stillbirth and early neonatal death cumulative incidence were calculated for each province using the number of reported births (all births for stillbirth, live births for neonatal death rate) as the denominator. The frequencies of deaths occurring across gestation were compared between SGA, AGA and LGA babies as well as between provinces. Comparisons between the proportion of primary cause of deaths were made by size-for-gestational-age, gestational age at death and in relation to maternal conditions. Gestational age was grouped into three categories for analysis: 28-32 weeks, 33-37 weeks and 38-42 weeks. Frequency distributions were performed and Pearson's chi-squared test or Fisher's exact test (where n < 5) were used to determine crude differences between proportions for the key comparisons made (i.e. size-for-gestational-age, gestational age at death, maternal condition and between provinces). Independent t-tests were used to compare means between key groups. A p-value <0.05 was accepted as statically significant.
The PPIP program has ethical approval from the Faculty of Heath Sciences Ethics Committee at the University of Pretoria. The data are collected with permission from the South African Department of Health. This secondary analysis was approved by the PPIP technical task team and UWA Human Ethics Committee.

Gestational age at death
There were no differences between provinces in terms of representation and timing of stillbirths between SGA, AGA, LGA, therefore data from the three provinces were combined and presented as means and standard errors. Stillbirths who were AGA or LGA occurred across gestation without any significant increases or decreases. A larger proportion of SBs occurred for SGA babies in the 33-37 week period in all provinces (Fig. 1). When considering macerated and fresh stillbirths a peak at 33-37 weeks was also seen for SGA babies (Fig. 1). A similar pattern was seen for early neonatal deaths (END) with a peak at 33-37 weeks for SGA babies. The timing of END for AGA and LGA babies increased during the 38-42 week period but remained low at 28-32w and 33-37w.

Primary causes of death
Stillbirths by size-for-gestational-age The primary cause of stillbirth across the sample was unexplained intrauterine death (33.1 %), followed by intrapartum asphyxia (17.2 %) and hypertensive disorders (19.7 %) ( Table 1). Some primary causes of death were more frequent in the SGA group, such as fetal abnormality (37.8 %) and intrauterine growth restriction (59.8 %) which represented up to half of all deaths despite only representing 26.2 % of the total sample (p < 0.05). Intrapartum asphyxia for SGA babies was indicated in 13.3 % of deaths due to intrapartum asphyxia across gestation (p < 0.05).
Causes of stillbirth and condition of mother for SGA babies across gestational groups are presented in Table 5. Causes of early neonatal death and condition of mother for small-for-gestational age babies across gestational groups are presented in Table 6.

Discussion
Our findings indicate that there are specific characteristics unique to SGA babies in terms of gestational age at death and causes of perinatal morality.

Timing of perinatal deaths across gestation
The same pattern of timing of stillbirth across gestation was seen for the three provinces with the greatest proportion of deaths for SGA babies during the 33-37 week period. This was in contrast to AGA and LGA babies where no peaks in the proportion of stillbirths across gestation were seen. This is consistent with observations in high-income countries where peaks have been observed between 34-37 weeks and 32-36 weeks [3,5]. In the current study a greater proportion of ENDs for SGA babies also occurred during the 33-37 week period, suggesting that this is a critical time for both antepartum and postpartum mortality. It is also well-established that there is increased neonatal mortality for SGA preterm babies compared to AGA preterm babies [1], stressing the vulnerability of SGA babies in particular.   Current antenatal care timing differs between the provinces, with Western Cape, the most well-resourced province, continuing fortnightly antenatal care visits between 32 and 38 weeks, while Limpopo and Mpumalanga cease antenatal care visits between 32 and 38 weeks. Previous work has shown that stillbirth risk is increased during periods without antenatal care [12]. The peak in stillbirths for SGA babies seen in all provinces between 33 and 37 weeks gestation suggests that the current detection and management of SGA is not adequate even in Western Cape where frequent antenatal care visits occur.

Causes of death/mother's condition
Growth restriction may not necessarily play a principle role in the cause of death for all SGA babies, as SGA may also be due to slow growth of an otherwise healthy baby [4,13]. In our study a greater proportion of SGA babies had deaths due to hypertension which is known to have placental pathology [13], indicating that growth restriction played a role in these deaths. It is important to consider that all the babies in the current cohort were considered viable (i.e. >1000 g and >28 weeks) and that very severe cases of placental insufficiency and congenital abnormalities could have died as late miscarriages.   Table 6 Cause of Early neonatal death and condition of mother for small-for-gestational age babies, n = 388 Healthy mother Medical surgical Hypertension Other maternal condition 28-32w 33-37w 38-42w 28-32w 33-37w 38-42w 28-32w 33-37w 38-42w 28-32w 33-37w 38-42w Therefore the deaths in the current study could have been potentially avoided if at risk fetuses were detected early.
Detecting SGA is complicated as in most cases the mother is healthy. In the current study there was no increase in poor maternal condition for SGA babies and more than half of all deaths had a healthy mother. Of mothers that were healthy the greatest proportion of SGA stillbirths were due to unexplained intrauterine death. Earlier detection of fetuses at risk antenatally may reduce the number of unexplained intrauterine deaths.

Challenges with detection
Better detection of SGA babies is needed in South Africa and LMICs. Palpation and symphysis fundal height are commonly used in LMICs due to limited alternative resources, despite the limited evidence to support this as an effective method to predict growth restriction [14]. Clinical trials in high-income countries estimate that up to 76 % of SGA cases can be detected antenatally [15]. The use of Doppler velocimetry to measure altered umbilical artery blood flow in high risk women has enhanced the ability to detect fetuses with pathological growth restriction [16], reducing perinatal mortality [17]. However, challenges with Doppler measurement in South Africa are present as there are a large number of perinatal deaths to low-risk healthy mothers [7] and who would therefore not be referred for screening. Studies in South Africa have explored the use of a continuous wave Doppler analyser using a PC (Umbiflow), a simpler alternative to umbilical artery Doppler, which can be operated by nurses and midwives at the primary health care centre level [18,19]. The Umbiflow is able to detect fetuses at risk of stillbirth based on abnormal umbilical artery blood flow. While meta-analyses in high-income countries found Doppler use in low-risk pregnancies to be ineffective to reduce perinatal mortality and morbidity [20], there may be potential benefit of Doppler or Umbiflow use in low risk populations in LMICs where stillbirth rates are higher. However, once a fetus at risk of stillbirth is detected, quality clinical care must also be provided to increase the risk of survival.

Challenges with clinical management
Once SGA has been detected there are challenges in the clinical management of such babies. Currently there are no effective approaches for the reversal or improvement of the growth pattern of a fetus [21], therefore prenatal clinical management is focused on identifying the optimal timing of delivery. The gestational age of the fetus is a critical component of the decision-making process, as fetal mortality is lower than neonatal mortality prior to 31 weeks [22], and delivery after 39 weeks results in increased perinatal mortality [23]. While it is known that the neonatal mortality risk is higher for preterm SGA babies compared to term SGA babies [1], it unclear how much size-for-gestational age impacts the risk of neonatal mortality in the preterm compared to late preterm infant [24,25]. It has recently been suggested that the delivery of SGA babies around 37 weeks is optimal to avoid increased risk of stillbirth occurring after week 37 [26]. Causes of death for SGA babies in other studies have shown different patterns across gestation with a greater proportion of deaths due to congenital abnormality (32-42 weeks) and intrapartum asphyxia (37-42 weeks) at older gestations [5]. We observed similar findings in our data with SGA deaths from intrapartum asphyxia and fetal abnormality occurring more frequently at older gestations. Standard guidelines for managing labour in South African include partogram and monitoring of fetal heart rate with a fetal stethoscope or doptone in Community Health Centres and electronically at district hospitals and above. However it is estimated that in 44 % of deaths these guidelines are not adhered to due to limited resources and staff [27]. Caesarean delivery is not readily accessible in some districts, and long transport times exist between Community Health Centres to facilities that are able to perform caesarean section deliveries. It is likely these factors play a role in the high proportion of deaths due to fetal asphyxia observed in the current study. It is also important to note that spontaneous preterm labour is an underlying cause of early neonatal death where the maternal condition contributes to increased risk of death from disorders associated with prematurity. Early neonatal deaths for preterm neonates reflect access and quality of care after birth for preterm infants. The final cause of death for a preterm neonate could be due to a causes such as hyaline membrane disease, hypoxic ischaemic encephalopathy and meconium aspiration. Doppler measurements in combination with growth charts can be used to inform the management of SGA pregnancies [17]. This has been shown to decrease the induction of labour and hospital admissions in high-income countries [17].

Limitations
There are some limitations to the current study. First, there is likely some ambiguity in the assignment of gestational age from the perinatal audit data. Last menstrual period or ultrasound estimates for gestational age may not be accurate and the SGA category may erroneously capture appropriately sized infants who measure small for their misassigned gestational age. We sought to reduce this issue by using only data where gestational age estimates were considered 'certain'. Second, macerated SGA babies are likely to be overestimated as death may have occurred up to 2 weeks prior [28] thus the fetus may have been appropriate-for-gestational age at the time of death. Third, as we had access to aggregated data at a centre level, and data for the number of live births for SGA, AGA and LGA were unavailable we were unable to calculate stillbirth risk, relative risks between live and stillborn babies or perform individual unit analysis using multivariable regression. Fourth, both multiple and single gestations were included in the current study. These were not separated due to limited statistical power and therefore any differences in patterns of mortality between multiple and single gestations are not presented in the current study. Deaths directly related to multiple gestation such as twin-to-twin transfusion only represented 0.3 % of all deaths in the current study, therefore it is unlikely that the inclusion of multiple gestations introduced significant bias.

Conclusion
Mortality from SGA in South Africa accounts for a considerable number of deaths, yet there is little research on the timing, causes and detection of SGA in low-resource settings. The detection and management of SGA is important as most deaths occur in late preterm or term infants, where the chance of survival is high when adequate care can be delivered. If SGA infants can be detected antenatally and managed effectively there is an opportunity to significantly reduce the burden of perinatal mortality in LMICs. Further studies on the relationship between antenatal detection and outcome of SGA are needed, especially in LMICs.