Trends in the causes of stillbirths over 20 years in Southern Japan

Abstract

Background

This study aimed to determine the temporal trend in the causes of stillbirths over a period of two decades using a regional perinatal database in order to facilitate the development of perinatal strategies to prevent stillbirths in Japan.

Methods

This was a population-based retrospective study. Cases of perinatal death and neurological damage were reported by perinatal centers and primary birth clinics, followed by their peer review and audit, and final registration in the database. Data for stillbirths, defined as fetal death at ≥ 22 weeks of gestation between January 1, 2001, and December 31, 2020, were extracted from the database. Causes of stillbirths were reclassified according to the ReCoDe system. Temporal trends in the causes of stillbirths per 1,000 births and proportion of the causes were evaluated using the Cochran-Armitage test.

Results

Over the 20 years, in the study region, a total of 205,025 were delivered at ≥ 22 weeks of gestation, and 569 were stillborn (2.8 per 1,000 births). The most common cause of stillbirth was “no relevant condition identified” in 39.5% cases, followed by “abruption” in 12.3%, “lethal congenital anomaly” in 9.5%, and “umbilical cord, other” in 5.8%. The trends in stillbirths caused by “fetal growth restriction,” “abruption,” “asphyxia,” and “no relevant condition identified” significantly decreased. However, no change in trend due to “lethal congenital anomaly” was seen. The stillbirth trend caused by “cord, other” significantly increased. The proportion of stillbirths related to unidentified causes remained unchanged.

Conclusions

Over the 20-year period, the rate of stillbirths caused by abruption, fetal growth restriction, and asphyxia, which can be reduced by early detection and intervention, decreased. The incidence of stillbirths caused by cord constriction increased. Investigations to prevent cord-accident stillbirths would be required to further reduce stillbirths in the study region. Establishment of algorithms that allow the identification of the causes of stillbirths would be crucial to reduce instances of stillbirths due to unidentified causes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12884-025-07794-8.

Background

Stillbirth remains one of the most tragic adverse pregnancy outcomes, despite the improvements in perinatal care systems and medical devices over the decades. Identification of the causes of stillbirth is important to facilitate the counseling of families and management of subsequent pregnancies. Furthermore, understanding the temporal changes underlying the causes of stillbirth is necessary to develop regional strategies for the prevention of stillbirth, neonatal death, and neurological damage as a sequential chain of poor perinatal outcomes [1–3].

In Japan, perinatal mortality, defined as fetal death after 22 weeks of gestation and neonatal death before 7 days of life, was 3.2 per 1,000 births, and stillbirth, defined as fetal death after 22 weeks of gestation, was 2.5 per 1,000 births in 2020 [4]. A retrospective study from Japan using the perinatal registry database of the Japan Society of Obstetrics and Gynecology from 2013 to 2014 showed that the most common cause of stillbirth is still unknown (32.8%), followed by placental disorders (21.7%), umbilical cord abnormalities (15.6%), congenital malformation (15.1%), fetal disorders (7.9%), maternal disorders (3.8%), infection (2.1%), and intrapartum events (1.1%) [5]. Although approximately 80% of deliveries occurred in primary birth clinics during the study period in Japan, 70% of the cases (stillbirths) were registered from perinatal centers in the registry database. Therefore, the stillbirth rate in the study, of 6.2 per 1,000 births, was higher than the national rate of 3.0 per 1,000 births. While the exact percentage is unknown, most stillbirths are managed in perinatal centers, and some are conducted in primary birth clinics in Japan. This suggests that the results do not represent the general population as mentioned by the authors. Furthermore, the trends of the causes of stillbirths in Japan have not yet been evaluated.

The current study aimed to determine the temporal trend of the causes of stillbirths over two decades using a regional perinatal database to facilitate the development of perinatal strategies to prevent stillbirths in Japan.

Methods

This was a population-based retrospective study. In Japan, a pregnant woman who delivers at > 12 weeks of gestation has a legal obligation to submit birth certificate or stillbirth certificate to the regional government. Accordingly, the annual total number of live births and stillbirths are published by the government. However, there is no publicly available data on the backgrounds and causes of stillbirths. Therefore, we, the Miyazaki Perinatal Data Group, have been supporting all perinatal centers and primary birth clinics (7 perinatal centers and 18 primary birth clinics in 2022) in the region to report cases of perinatal death and neurological damage since 1997. Towards that, an audit is done twice a year to determine the causes and contributing factors of death or neurological damage [2, 6, 7]. The audit consists of at least 10 maternal-fetal medicine specialists and 5 neonatologists from the perinatal centers. The cases are registered in our original perinatal database. Since the database is based on the reporting system from the perinatal centers and primary birth clinics, cases of missed report, home birth without medical assistance, and those of residents who had given birth outside the region are not registered in the database. We extracted data for stillbirths (defined as the fetal death at ≥ 22 weeks of gestation) from the database between January 1, 2001 and December 31, 2020. The exact total stillbirths in the study region were extracted from publicly available reports by the government [4]. Since the causes of deaths in the database are not classified by any system, the causes of stillbirth were reclassified by the authors according to the ReCoDe system [8]. We used the ReCoDe classification system since it is easy to apply, has good consistency, and has low interobserver variability [9]. ReCoDe system allows only one identified cause of stillbirth. The primary cause of stillbirth is entered at two levels. One anatomical group is chosen from nine categories–fetus, umbilical cord, placenta, amniotic fluid, uterus, mother, intrapartum, trauma, and unclassified–and then subdivided into 37 pathophysiological conditions. For the unclassified group, the registered cases with undetermined cause of stillbirth were classified as “no relevant condition identified.” Unregistered cases with unreported detailed clinical data were classified as “no information available. In case of a discrepancy between the number of stillbirths published by the government and that in the database, we marked them as “no information available.” The causes of stillbirth are often uncertain and whether a condition is direct cause or contributing factor is hard to determine [10–12]. Therefore, we only determined the severe and uncontrolled conditions that led to stillbirth. For example, uncontrolled diabetes with hyperglycemia ≥ 200 mg/dL or ketoacidosis at stillbirth was considered a cause. Only cases of fetal growth restriction (FGR) with deteriorated fetal well-being according to Doppler studies, biophysical profile score, non-stress test at antenatal testing, or severe small-for-gestational age < 3 percentile at stillbirth were considered as causes. FGR was determined by antenatal ultrasound examination [13] or the Japanese optimal birth weight standard by gestational age [14], in accordance with the obstetrical practice guidelines in Japan [15]. The cause of stillbirth with < 10 cases over the 20 years was excluded from the trend analysis. The temporal trends of the causes of stillbirth per 1,000 births and the proportion of the causes were evaluated using the Cochrane-Armitage test. Statistical significance was set at p < 0.05. Statistical analyses were performed using JMP Pro version 16.0.0 (SAS Institute Inc., Cary, NC, USA).

Results

A total of 205,025 neonates were delivered at ≥ 22 weeks of gestation, and 569 were stillbirths (2.8 per 1000 births) over the 20-year period in the study region. Out of the 569 stillbirths, 487 (85.6%) underwent peer review, auditing, and database registration.

Backgrounds of the 487 stillbirth cases from the database are shown in Table 1. The 82 unregistered cases were not included in the backgrounds. Stillbirths in preterm infants accounted for 68.2% of all stillbirths. Ten twins (2.1%) were included in the study. The rates of postmortem and placental pathological examinations were 25.9% and 54.8%, respectively.

Table 1.

Backgrounds for the cases obtained from the database (n = 487)†

Characteristics	Patients (n = 487)
Maternal age, mean ± SD (years)	31 ± 5.6
< 20 (%)	7/346 (2)††
≥ 35 (%)	95/346 (27.4)††
≥ 40 (%)	20/346 (5.8)†
Nullipara (%)	160/309 (51.8)†
Multiple pregnancies	10/487 (2.1)
Gestational week
Preterm (22–36 wk) (%)	332/487 (68.2)
22–27 wk (%)	123/487 (25.3)
28–33 wk (%)	130/487 (26.7)
34–36 wk (%)	79/487 (16.2)
Term (37–41 wk) (%)	116/487 (23.8)
Post-term (≥ 42 wk) (%)	11/487 (2.3)
Unknown	28/487 (5.7)
Fetal weight, median (range) (g)	1590 (176–4250)
< 500 g (%)	57/487 (11.7)
500–999 g (%)	97/487 (19.9)
1000–1499 g (%)	67/487 (13.8)
1500–2499 g (%)	129/487 (26.5)
2500–3999 g (%)	107/487 (22)
≥ 4000 g (%)	1/487 (0.2)
Unknown	29/487 (6.0)
Intrapartum (%)	17/487 (3)
Autopsy (%)	126/487 (25.9)
Placental pathological examination (%)	267/487 (54.8)

SD, standard deviation

†The unregistered 82 cases were not included

††Denominators are different due to missing data

The most common cause of stillbirth was “no relevant condition identified” in 39.5% cases, followed by “abruption” in 12.3%, “lethal congenital anomaly” in 9.5%, and “umbilical cord, other” in 5.8% (Table 2). Among the 33 cases of “umbilical cord, other”, umbilical cord constriction was the most common (32/33), followed by amniotic band syndrome.

Table 2.

Proportion of the causes of stillbirth according to recode system in the population (n = 569)

Causes of stillbirth	Patients (n = 569) (%)
Fetus	91 (16)
Lethal congenital anomaly	54 (9.5)
Infection, chronic/acute	4 (0.7)
Non-immune hydrops	0 (0)
Isoimmunization	1 (0.2)
Fetomaternal hemorrhage	1 (0.2)
Twin-twin transfusion	2 (0.4)
Fetal growth restriction	29 (5.1)
Umbilical cord	44 (7.7)
Prolapse	2 (0.4)
Constricting loop or knot	9 (1.6)
Velamentous insertion	0 (0)
Other	33 (5.8)
Placenta	72 (12.7)
Abruption	70 (12.3)
Previa	0 (0)
Other “placental insufficiency”	2 (0.4)
Other	1 (0.2)
Amniotic fluid	10 (1.8)
Chorioamnionitis	10 (1.8)
Oligohydramnios	0 (0)
Polyhydramnios	0 (0)
Other	0 (0)
Uterus	4 (0.7)
Rupture	3 (0.5)
Uterine anomalies	0 (0)
Other	1 (0.2)
Mother	22 (3.9)
Diabetes	4 (0.7)
Thyroid diseases	1 (0.2)
Essential hypertension	3 (0.5)
Hypertensive diseases in pregnancy	5 (0.9)
Lupus or antiphospholipid syndrome	0 (0)
Cholestasis	0 (0)
Drug misuse	0 (0)
Other	9 (1.6)
Intrapartum	16 (2.8)
Asphyxia	16 (2.8)
Birth trauma	0 (0)
Trauma	2 (0.4)
External	2 (0.4)
Iatrogenic	0 (0)
Unclassified	307 (54)
No relevant condition identified	225 (39.5)
No information available	82 (14.4)

The total stillbirth rate declined significantly from 3.98 per 1,000 births in 2001 to 2.20 per 1,000 births in 2020 (p < 0.01). The causes of “lethal congenital anomaly,” “FGR,” “other umbilical cord,” “abruption,” “asphyxia,” “chorioamnionitis,” “no relevant condition identified,” and “no information available” were included in the trend analysis. The incidence of each cause of stillbirth per 1,000 births according to the ReCoDe classification by year is shown in Fig. 1. The trends of stillbirths caused by “FGR,” “abruption,” “asphyxia,” “no relevant condition identified,”and ”no information available” were significantly decreased. Furthermore, there has been no case of stillbirth due to asphyxia since 2013. In contrast, there was no change in the trend of stillbirths caused by “lethal congenital anomalies”. The stillbirth caused by “cord, other” significantly increased (detailed data in Additional file 1).

Fig. 1.

Trends in each cause of stillbirth (/1,000 births)

The trends of stillbirths caused by fetal growth restriction, abruption, asphyxia, no relevant condition identified, and no information available were significantly decreased while the stillbirth caused by cord, other significantly increased. There was no change in the trend of stillbirths caused by lethal congenital anomalies.

Trends in the proportions are listed in Fig. 2. The proportion of “asphyxia”-related stillbirth significantly decreased while those of “lethal congenital anomaly” and “cord, other” significantly increased (detailed data in Additional file 2).

Fig. 2.

Trends in the proportions of each cause (%)

The proportion of asphyxia-related stillbirth significantly decreased while those of lethal congenital anomaly and cord, other significantly increased.

Discussion

From 2001 to 2020, 569 stillbirths occurred in the study region located in southern Japan, and the total stillbirth rate declined significantly. The incidences of stillbirths due to FGR, abruption, and asphyxia declined significantly while those due to cord increased significantly. Stillbirths from FGR, abruption, and asphyxia are potentially avoidable by educating the pregnant women, identifying the pregnant women who are at risk during pregnancy, improving the management by ultrasound imaging and fetal heart rate monitoring, early intervention, including cesarean section, and transportation to a higher-level perinatal center. The most concerning findings from the study was that the proportion of stillbirths with no relevant condition identified was the most common and remained unchanged over the study period.

Ideally, the cause of stillbirth should be identified. Stillbirths related to unidentified causes have been the most frequent, and remained unchanged for more than 20 years. Its proportion of 39.5% is consistent with that from high-resource settings reported in a systematic review (32.1%) [16]. A review paper reported that full assessment, including autopsy and placental investigations according to local guidelines, contributed to a decrease in the proportion of unidentified causes from 30% to only 5% [17]. Although we did not analyze the trend of the uptake of autopsy and placental examination, we presumed that the low uptake affected the high frequency of unidentified causes of stillbirths over time. Although the guidelines from Japan recommend postmortem examination, histopathological placental and umbilical cord examination, skeletal radiography, chromosomal examination, and maternal blood tests, there are not required. except for external fetal examination [15]. Therefore, assessment decisions depend on the attending physician, which can lead to a high incidence of stillbirths with unidentified causes. Placental pathological examination and fetal autopsy are considered the gold standards for assessing the cause of stillbirth. Two large studies investigating the utility of tests for identifying the causes of stillbirth have shown that placental pathological examination is the most helpful in either confirming or excluding the cause of stillbirth in 64.6% and 95.7% of cases, followed by fetal autopsy in 42.4% and 72.6%, respectively [18, 19]. A new strategy to evaluate the cause of stillbirth based on placental pathological examination, fetal autopsy, and clinical clues, such as those proposed by the Stillbirth Collaborative Research Network in the US [18], should be established.

The incidence of stillbirths caused by abruption, the second most common cause in the study region, was remarkably reduced, although its proportion did not change significantly. The trends in the incidence and proportion of stillbirths related to abruption vary across different studies. In a single-center study in Ireland, the incidence significantly decreased from 1.13 per 1,000 in 1989 to 0 per 1,000 in 2018 [20]. Although placental abruption is not available in the classifications used in New Zealand and the UK, the proportion of stillbirths related to antepartum hemorrhage reduced significantly from 10.2% in 2000 to 8.15% in 2007 in the UK while remaining unchanged in New Zealand [21]. The prevalence of abruption itself has been declining in European countries for unknown reasons [22]. Although recognition of high-risk patients, early diagnosis of placental abruption, and prompt transfer to a higher-level perinatal center may contribute to the reduction of stillbirths caused by abruption, some stillbirths with placental abruption are not potentially preventable [23]. In future studies, we plan to address the factors that prevent abruption and discover predictive markers to reduce abruption itself.

Asphyxia was the most remarkable change in the study region, since both the incidence and proportion significantly decreased and ceased after 2013. The proportion of stillbirths caused by asphyxia decreased from 4.6% in 2007 to 2.2% in 2014 in New Zealand, and from 10.9% in 1993 to 6.4% in 2007 in the UK [21]. A trend study regarding stillbirth etiology from 1989 to 2018 in a single center in Ireland had shown that stillbirths caused by hypoxia did not occur after 2003 [20]. The findings suggested that stillbirths caused by asphyxia can be decreased by improved intrapartum management. In future, we need to investigate the factors that impact the reduction in stillbirths caused by asphyxia in Japan.

The reasons underlying the increment of stillbirth related to “cord, other,” in which most of them were constriction in this study, are uncertain. Although numerous studies have assessed the association between umbilical cord abnormalities and adverse pregnancy outcomes, including stillbirth, most have focused on knots or the nuchal cord [24]. A large study investigating the association between umbilical cord abnormalities and stillbirths had shown that 19% of the stillbirths (94/496) were caused by umbilical cord abnormalities; the incidence of composite knots or constrictions was only 27% (26/94) [25]. In the study region, most umbilical cord abnormalities were constricted rather than knotted (32 vs. 9). Generally, confirmation of umbilical cord abnormalities as the cause of stillbirth should be performed with caution since many cord abnormalities are commonly observed in live births [10, 24]. Furthermore, some researchers have claimed that constriction represents an artifact change after stillbirth [26]; the incidence of stillbirths caused by constrictions may have been underestimated at the beginning of the study period.

Prevention of stillbirth caused by lethal congenital anomaly is complicated since it occurs with a certain frequency and inevitable. Some proportion of stillbirth caused by congenital anomalies may be reduced with preconceptual care and counseling, and adequate prenatal counseling and diagnostic testing [27]. Although the incidence of stillbirth caused by lethal congenital anomaly had tended to decline since 2011, it was increased in 2020. We have no idea of the reason for it.

A strength of our study was that the registry data were subjected to a multidisciplinary regional audit. A study from Italy has shown that a regional audit system could allow defining the causes of death [28]. Since all perinatal centers and birth clinics in our region contributed to the database, the presented data could be more representative of the Japanese population than the previous studies from Japan [5].

However, this study had some limitations. International comparisons of stillbirth rates and specific causes of death are difficult due to the lack of a common definition of stillbirth and the use of different classification systems [9, 21]. In New Zealand, stillbirth is defined as 20 weeks of gestation or older and/or 400 g birth weight greater. The Perinatal Society of Australia and New Zealand Perinatal Death Classification System is used to determine the cause of stillbirth. In the UK, stillbirth is defined as 24 weeks of gestation or older [21]. Since the classification system changed several times from 1993 to 2014, interpretation and comparison are more complicated. The absolute values of the incidence were not provided in either report. A single-center study from Ireland including 840 stillbirths (defined as singleton fetus weighing ≥ 500 g) over 30 years is more comparable to our study [20], although the former used a different classification system.

Conclusions

In conclusion, the rate of stillbirths caused by abruption, FGR, or asphyxia, which can be avoided by early detection and intervention, decreased in our region over the 20-year period. The incidence and proportion of stillbirths caused by cord constriction increased, and the proportion of stillbirths related to unidentified causes remained unchanged. To reduce stillbirths further, investigations to prevent cord-accident stillbirths would be required in future. Establishment of algorithms that allow the identification of the cause of stillbirths could potentially reduce stillbirths due to unidentified causes in the study region.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

FGR

Fetal growth restriction

ReCoDe

Relevant condition at death

Author contributions

EK: Data collection, data analysis, and manuscript drafting; YM: Study design, data interpretation, and critical manuscript revision; NY: Data collection; YK: Data collection, interpretation, and supervision; SK: Data interpretation and supervision; HS: Data interpretation and scientific comment; TI: data interpretation and scientific comments.

Funding

This study was supported by a Clinical Research Support Grant from University of Miyazaki Hospital.

Data availability

Data used in this study are available from the corresponding author upon request.

Declarations

Ethics approval and consent to participate

The ethics committee of the Miyazaki Medical Association Hospital approved this study, and the requirement of written informed consent was waived due to the use of a deidentified database (approval no. 2022-45). The study was conducted according to Helsinki declaration.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.