Comparative trends in cause-specific fetal and neonatal mortality in twin and singleton births in the North of England, 1982–1994
Correspondence: Dr S. V. Glinianaia, Department of Epidemiology and Public Health, School of Health Sciences, The Medical School, University of Newcastle, Framlington Place, Newcastle upon Tyne NE2 4HH, UK.
Objective To examine trends in cause- and birthweight-specific fetal and neonatal mortality rates in twins and singletons.
Design Descriptive analysis based on a regional register.
Setting The Northern Health Region of England, 1982–1994.
Sample Two hundred and thirty-six fetal and 356 neonatal twin deaths; 2687 fetal and 2301 neonatal singleton deaths from a population of 10,734 twins and 505,477 singletons.
Main outcome measures Fetal and neonatal autopsy rates, cause- and birthweight-specific fetal and neonatal mortality rates in twins and singletons.
Results The extended perinatal mortality (including stillbirths and neonatal deaths) rate (EPMR) was 55.2 per 1000 in 1982–1994 in twins compared with 9.9 per 1000 in singletons. The relative risk for twin compared with singleton deaths was 5.6 (95% CI 5.1-6.1) being highest for immaturity (12.9, 95% CI 11.1–15.0). A significant decrease in the EPMR occurred in both twins and singletons in 1988–1994 compared with 1982–1987. The EPMR decreased mainly due to a reduction of deaths from antepartum asphyxia in twins and intrapartum asphyxia and trauma in singletons, as well as a reduction in congenital malformations in both groups. In both twins and singletons, birthweight-specific mortality rates improved between 1982–1987 and 1988–1994.
Conclusion The higher relative risk for twin deaths remained stable due to a similar decrease in the EPMR for both twins and singletons. The cause-specific relative risk in twins declined for antepartum asphyxia. The mortality rate resulting from lethal congenital malformations decreased in twins and singletons mainly due to earlier detection and subsequent termination of pregnancy.
The risk of perinatal death is much higher among twins than singletons1–6, and this difference for twins has not changed over time. The commonest cause of poor outcome for twin pregnancies is immaturity7–10. Preterm births (< 37 weeks of gestation) occur approximately 5-5.5 times more frequently in twins compared with singletons3,5,11,12. However, evidence suggests that pulmonary maturity develops earlier in twins that in singletons which results in a lower incidence of bronchopulmonary dysplasia in preterm twins than in singletons of comparable gestational age13,14. Furthermore, recent population-based studies15,16 have demonstrated that optimal perinatal survival occurs two weeks earlier in twins compared with singletons. The difference in ‘ideal’ gestational age between twins and singletons is even more pronounced; the lowest fetal mortality for twins is at 36–37 weeks and 2500–2800 g compared with 40–41 weeks and 3700–4000 g for singletons17. Intrauterine growth retardation is a higher risk factor for fetal death than immaturity17. The second leading cause of perinatal death in twins is intrauterine asphyxia which, proportionally, occurs less frequently in twins compared with singletons, but accounts for 8%-10% of all antepartum deaths17,18.
A significant improvement in perinatal outcome for twin pregnancies during the late 1970's to early 1980's was attributed mainly to an increase in survival of very preterm and very low birthweight newborns1,3. In contrast, the stillbirth rate in twins did not change significantly3. In fact, there was an increase in stillbirth rate due to a rise in antepartum deaths1. The results of our previous study on twin mortality rates in the north of England between 1982–1994 showed a significant decreasing trend for stillbirth but not for early neonatal mortality rate, a finding which contrasts with the above studies19. The achievements in prenatal management of twin and singleton gestation during the past 10–15 years may have affected the distribution of causes of death. An analysis of immediate causes of twin and singleton death and their contribution to an improvement in perinatal mortality of both categories, is warranted.
It has been shown also that perinatal autopsy affects the distribution of immediate causes of death, although its rate in most of the UK fails to reach the nationally recommended minimum level of 75%20,21. In Wales, the contribution of perinatal and infant postmortem examination to the final diagnosis altered clinicopathological classification in 13% of cases and revealed the cause of death in 18%21. In Northern Ireland, the autopsy findings resulted in alteration to the pre-autopsy Wigglesworth classification in 20.7% of cases20.
This study was designed to examine cause- and birth-weight-specific mortality rates in twins and singletons in the north of England between 1982–1987 and 1988–1994. Autopsy rates were compared in twins and singletons, and the distribution of causes of death, depending on whether an autopsy was performed, was explored.
The former Northern Region is a well-defined geographical area with a population of over three million and approximately 40,000 births per year. Data from a population-based register, the Northern Region Perinatal Mortality Survey, which has been in operation continuously since 1981, were used. The Perinatal Mortality Survey is a collaborative exercise involving individual maternity units in 16 health districts in the region based on the mechanisms already in place for the routine review of perinatal deaths. Each maternity unit has a review team with an obstetrician, a paediatrician, and a midwife on the panel (together with a pathologist, general practitioner, and community physician as appropriate) under an elected convenor. The team completes a survey form for every stillbirth and neonatal death to mothers resident in the Northern Region. Survey forms are normally completed by the team in whose unit the baby dies. If an infant is transferred to another unit before death the reporting team liaises with the clinicians who previously cared for the mother and baby to produce a single, joint return.
Completed forms are sent to the regional medical officer, who is responsible for maintaining medical confidentiality. One of the Perinatal Mortality Survey Steering Group members reviews each form to verify its accuracy and consistency. Receipts of the survey forms are always acknowledged, and a duplicate copy of the acknowledgement is sent to the district medical officer in the district where the mother is normally resident. This helps to ensure that all relevant deaths are included in the survey and avoids duplication. The data collection system in the Perinatal Mortality Survey has been described previously in greater detail22.
All stillbirths (as defined below) and all neonatal deaths of liveborn infants delivered after gestational week 20, both in twin and singleton pregnancies, occurring in the Region between 1982–1994 were identified from the database. To compare trends in twin and singleton mortality rates, the time period was divided into two approximately equal spans: 1982–1987 and 1988–1994. The selection of the time spans was arbitrary, simply aiming to ensure adequate numbers in each group for meaningful statistical analysis.
The total number of twin and singleton births in the Northern Region was obtained from the Office for National Statistics. All data are based on the date of birth. Registrations of some multiple births may be incomplete because of the legal definition of what is registerable. However, the number of missing multiple births is likely to be small and the effect on our results marginal.
Data on terminations of pregnancy due to antenatally detected congenital malformations were obtained from the population-based Northern Congenital Abnormality Survey (NorCAS), which is also an ongoing register, collecting data on major congenital abnormalities occurring after 1 January 1984 (irrespective of whether they occur as spontaneous abortions, terminations of pregnancy or registered births, and regardless of whether they are diagnosed antenatally or not). This applies to all infants born to mothers resident within the boundaries of the former Northern Region, even if they were delivered outside the region. The cases are notified from multiple sources, thus enabling high case ascertainment. Subsequently, the cases are checked for duplication and then entered onto the register. Further details of data collection have been published previously23,24. As data on terminations of pregnancy are available since 1984, the comparison of losses due to congenital anomalies with an allowance for differences in the number of terminations of pregnancy in two periods was made for 1984–1987 and 1988–1994 only.
Before October 1992 all fetal deaths were registered as stillbirths if they were delivered after gestational week 28. After October 1992, the gestational age for stillbirth registration was lowered to 24 weeks as a result of a change in legislation in England and Wales25. Neonatal death is a death, following live birth, of a baby before completing 28 days of life (death within 0–27 days). The extended perinatal mortality rate (EPMR) is the total number of stillbirths and neonatal deaths per 1000 (live plus still) registered births. The stillbirth rate is the number of stillbirths per 1000 total births. The neonatal mortality rate is the number of neonatal deaths per 1000 live births.
Immediate causes of stillbirths and neonatal deaths were classified using the ‘Extended Wigglesworth Classification’26. Immaturity combines all the pathological conditions associated with prematurity: severe pulmonary immaturity, hyaline membrane disease and periventricular haemorrhage. According to the ‘Extended Wigglesworth Classification’, congenital anomaly is defined as any genetic or structural defect arising at conception or during embryogenesis incompatible with life or potentially treatable but causing death26. This category included congenital anomalies from chapter XIV of the ninth edition of the International Classification of Diseases27 (codes 740.0–759.9), inborn errors of metabolism, genetically determined diseases or congenital tumours. Expert members of the Perinatal Mortality Survey Steering Group reviewed all stillbirths and neonatal deaths. Death was ascribed to a congenital malformation if the anomaly corresponded to the above definition.
Descriptive analysis was carried out using the Statistical Package for the Social Sciences (SPSS for Windows)28. The χ2 test was used to test differences in proportions. Thus, the autopsy rates for twin and singleton stillbirths and neonatal deaths were compared between 1982–1987 and 1988–1994, and the total autopsy rates were compared between twins and singletons by the time period (Table 1). The stillbirth rate, neonatal mortality rate and EPMR, as well as cause- and birthweight-specific EPMR were tested for differences between 1982–1987 and 1988–1994, separately for twins and singletons (Tables 3–6). Statistical significance was accepted as P < 0.05. The relative risk (RR) for twin deaths compared with singleton deaths was calculated using the statistics software package Epi Info29.
Table 1. Autopsy rates in twin and singleton deaths in 1982–1987 and 1988–1994. Values are given as n or n (%).
|1982–1987|| || || || |
| Stillbirths||96 (76.2)||126||1012 (78.3)||1293|
| Neonatal deaths||94 (56.0)||168||766 (63.8)||1200|
| TOTAL||190 (64.6)||294||1778 (71.3)||2493|
|1988–1994|| || || || |
| Stillbirths||90 (81.8)||110||1036 (74.3)||1394|
|1988–1994Neonatal deaths||96 (51.1)||188||567 (51.5)||1101|
|1988–1994TOTAL||186 (62.4)||298||1603 (64.2)||2495|
Table 3. Cause-specific rates of fetal and neonatal mortality in twins and singletons and the relative risk (RR) of twin deaths compared with singletons. Values are given as n (rate per lo00 births) or RR [95% CI]. NA = not applicable; NNDs = neonatal deaths.
|Congenital anomaly|| || || || || || || |
| Twins||13 (2.7)||8 (1.4)||26 (5.5)||16 (2.8)||39 (8.0)||24 (4.1)||0.01|
| Singletons||139 (0.6)||100 (0.4)||440 (1.9)||331 (1.2)||579 (2.4)||431 (1.6)||< 0.001|
| Combined 1982–94 twin/singleton RR||4–1 [2.6–6.5]||2.6 [1.9–3.6]||2.7 [2.1–3.4]|| |
|Antepartum anoxia|| || || || || || || |
| Twins||84 (17.3)||69 (11.7)||−||−||84 (17.3)||69 (11.7)||0.02|
| Singletons||940 (4.0)||1076 (4.0)||2 (0.0)||2 (0.0)||942 (4.0)||1078 (4.0)||0.99|
| Combined 1982–94 twin/singleton RR||3.6 [3.0–4.3]||NA||3.6 [3.0–4.2]|| |
|Intraparturn asphyxia, anoxia, trauma|| || || || || || || |
| Twins||17 (3.5)||7 (1.2)||19 (4.0)||22 (3.8)||36 (7.4)||29 (4.9)||0.10|
| Singletons||177 (0.8)||156 (0.6)||167 (0.7)||161 (0.6)||344 (1.4)||317 (1.2)||0.01|
|Combined 1982–94 twin/singleton RR||3.4 [2.2–5.1]||6.0 [4.3–8.3]||4.6 [3.6–6.0]|| |
|Immaturity†|| || || || || || || |
| Twins||−||−||96 (20.3)||117 (20.3)||96 (19.8)||117 (19.9)||0.96|
|Singletons||−||−||378 (1.6)||401 (1.5)||378 (1.6)||401 (1.5)||0.29|
|Combined 1982–94 twin/singleton RR||NA||13.1 [11.3–15.2]||12.9 [11.1–15.0]|| |
|All causes†|| || || || || || || |
| Twins||126 (26.0)||110 (18.7)||168 (35.5)||188 (32.6)||294 (60.6)||298 (50.7)||0.03|
| Singletons||1293 (5.5)||1394 (5.2)||1200 (5.1)||1101 (4.1)||2493 (10.6)||2495 (9.2)||< 0.001|
Table 4. Number of cases and stillbirth rates in 1982–87 and 1988–94 in twins and singletons due to the change in stillbirth definition in October 1992. Values are given as n (rate per 1000 births).
|Twins||126 (26.0)||104 (17.7)||110 (18.7)||84 (17.3)||66 (11.2)||69 (11.7)|
|Singletons||1293 (5.5)||1293 (4.8)||1394 (5.2)||940 (4.0)||1024 (3.8)||1076 (4.0)|
Table 5. Deaths (numbers and extended perinatal mortality rates (EPMR)) and terminations of pregnancy due to congenital anomalies in twins and singletons in 1982–87 and 1988–94. Values are given as n (rate per l000 births).
|Twins||22 (6.7)||5||27 (8.2)||24 (4.1)||12||36 (6.1)|
|Singletons||358 (2.3)||280||638 (4.0)||431 (1.6)||860||1291 (4.8)|
Table 6. Table 6. Birthweight (BW) distribution and birthweight-specific extended perinatal mortality rates (EPMR) in twins and singletons in 1982–87 and 1988–94. Values are given as n (rate per 1000 births), unless otherwise indicated.
|Twins|| || || || || || |
| ≤1000||139||132 (949.6)||215||162 (753.5)||−196.1||−264.0 to-128.0|
| 1000–1499||286||65 (227.3)||321||55 (171.3)||−55.9||−120.0 to 7.8|
| 1500–1999||665||39 (58.6)||802||34 (42.4)||−16.3||−39.0 to 6.4|
| 2000–2499||1431||29 (20.3)||1709||21 (12.3)||−8.0||−17.0 to 1.0|
| 2500–2999||1613||22 (13.6)||1925||19 (9.9)||−3.8||−10.9 to 3.4|
| >3000||707||7 (9.9)||832||7 (8.4)||−1.5||−11.1 to 8.1|
| Missing BW||13||−||76||−|| || |
| TOTAL||4854||294 (60.6)||5880||298 (50.7)||−9.9||−18.6 to-1.1|
|Singletons|| || || || || || |
| ≤1000||685||468 (683.2)||989||624 (630.9)||−52.3||−9.8 to −6.2|
| 1000–1499||1319||398 (301.7)||1481||355 (239.7)||−62.0||−9.5 to −29.2|
| 1500–1999||2579||351 (136.1)||3033||286 (94.3)||−41.8||−58.6 to −25.0|
| 2000–2499||9586||342 (35.7)||10,495||305 (29.1)||−6.6||−11.5 to-1.7|
| 2500–2999||41,187||354 (8.6)||43,838||351 (8.0)||−0.6||−1.8 to 0.6|
| 3000–3499||90,915||345 (3.8)||99,720||324 (3.2)||−0.5||−1.1 to4.01|
| >3500||88,748||235 (2.6)||107,709||250 (2.3)||−0.3||−0.8 to 0.1|
| Missing BW||632||−||2561||−|| || |
| TOTAL||235,651||2493 (10.6)||269,826||2495 (9.2)||−1.3||−1.9 to −0.8|
Mortality rates in twins
There was a total of 10,734 twin births (5367 twin maternities) in the Northern Region between 1982–1994 (mean twinning rate 10.5 per 1000 maternities). A total of 236 stillbirths (198 antepartum and 38 intrapartum), 297 early neonatal and 59 late neonatal deaths, from 459 twin pregnancies, were notified to the Perinatal Mortality Survey. There were 2687 stillbirths (2227 antepartum and 460 intrapartum), 1807 early neonatal and 494 late neonatal deaths among 505,477 singleton births during the same period. The overall EPMR was 55.2 per 1000 in twins and 9.9 per 1000 in singletons. The ratio of stillbirths to neonatal deaths in twins was 1:1.51 compared with 1:0.86 in singletons. The EPMR decreased significantly from 60.6 in twins and 10.6 in singletons in 1982–1987 to 50.7 (P= 0.03) and 9.2 (P < 0.001) in 1988–1994 for twins and singletons respectively.
Perinatal autopsy in twins and singletons over time
Table 1 shows the number of stillbirths and neonatal deaths and proportion of postmortem examinations performed in twin and singleton deaths over time. In twins, there was no significant difference in autopsy rates between 1982–1987 and 1988–1994 for both stillbirths and neonatal deaths. The autopsy rates were higher for stillbirths than neonatal deaths in both periods. In contrast, in singletons, there was a significant decrease in autopsy rates from 78.3% in 1982–1987 to 74.3% in 1988–1994 for stillbirths (P= 0.018) and from 63.8% to 51.5% for neonatal deaths (P < 0.001). A comparison of autopsy rates between twins and singletons revealed a significantly higher rate for singletons in 1982–1987 (P= 0.02) but this difference disappeared in 1988–1994 due to the decrease in proportion of autopsies performed in singletons.
Table 2 shows the distribution of immediate causes of death in twins and singletons based on autopsy results and on clinical details only. Both in twins and singletons, autopsies were more likely to have been performed in certain categories. Thus, for both groups, the highest autopsy rate was in fetal deaths due to antepartum anoxia (81% in twins and 77.1% in singletons). In deaths due to immaturity, an autopsy was performed only in half of the cases. The bias in the autopsy rates was reflected in the difference in the distribution of the leading causes of death between the groups with and without autopsy, which is more noticeable in twins. Therefore, due to a higher autopsy rate in stillbirths compared with neonatal deaths (Table 1), antepartum anoxia displaced immaturity as the leading cause of twin death in those that were examined post mortem. However, the difference in the distribution of the leading causes of twin death between the groups with and without autopsy should not invalidate the comparison of cause-specific mortality rates in twins over time because the autopsy rates in twins did not differ significantly during the 13 years.
Table 2. Cause of death in twins and singletons based on postmortem examination (PM) and without PM. Values are given as n (%).
|Congenital malformation||32 (8.5)||31 (14.4)||63 (10.6)||627 (18.5)||383 (23.8)||1010 (20.2)|
|Antepartum anoxia||124 (33.0)||29 (13.4)||153 (25.8)||1557 (46.1)||463 (28.8)||2020 (40.5)|
|Intraparturn asphyxia, anoxia, trauma||46 (12.2)||19 (8.8)||65 (11.0)||437 (12.9)||224 (13.9)||661 (13.2)|
|Immaturity†||105 (27.9)||108 (50.0)||213 (36.0)||396 (11.7)||383 (23.8)||779 (15.6)|
|Infection||12 (3.2)||11 (5.1)||23 (3.9)||146 (4.3)||86 (5.4)||232 (4.6)|
|Twin-to-twin transfusion||43 (11.4)||11 (5.1)||54 (9.1)||—||—||—|
|Other specific causes||12 (3.2)||4 (1.8)||16 (2.7)||152 (4.5)||48 (3.0)||200 (4.0)|
|Sudden infant death||2 (0.5)||—||2 (0.3)||34 (1.0)||16 (1.0)||50 (1.0)|
|Unexplained||—||3 (1.4)||3 (0.5)||32 (0.9)||4 (0.2)||36 (0.7)|
|All causes||376(100.0)||216 (100.0)||592 (100.0)||3381 (100.0)||1607 (100.0)||4988 (100.0)|
Cause-specific mortality in twins and singletons
Table 3 shows cause-specific stillbirth and neonatal mortality rates for twins and singletons over the periods 1982–1987 and 1988–1994, as well as the RR of twin deaths compared with singletons in 1982–1994. A significant reduction in the twin stillbirth rate in the later period (P= 0.013) affected the time trend in the EPMR. Cause-specific EPMR declined significantly for congenital malformations (P= 0.011) and antepartum asphyxia (P = 0.019). There was also a decrease in twin mortality due to intrapartum asphyxia and trauma but it was not statistically significant. As seen in Table 3, the twin neonatal mortality rate did not improve significantly between 1982–1987 and 1988–1994, primarily due to a constant rate of deaths ascribed to immaturity, the leading cause of twin neonatal deaths. In contrast to twins, the EPMR in singletons decreased significantly over time due to a decrease in neonatal mortality rate (5.1 to 4.1 per 1000 live births, P < 0.001) but not in stillbirth rate (5.5 to 5.2 per 1000 births, P= 0.12). The EPMR in singletons decreased mainly due to a highly significant reduction in deaths due to congenital malformations for both stillbirths and neonatal deaths (P < 0.001) and intrapartum asphyxia and trauma (P < 0.01). There was no improvement in singleton stillbirth rate due to antepartum anoxia, the major cause of deaths in singletons.
The RR for twin deaths compared with singletons was 5.6 (95% CI 5.1–6.1), ranging from 2.7 (95% CI 2.1–3.4) for congenital abnormalities to 12.9 (95% CI 11.1–15.0) for immaturity (Table 3). The decline in the twin EPMR paralleled that of singletons showing similar RR in the two time periods (1982–1987: RR 5.7, 95% CI 5.1–6.4; 1988–1994: RR 5.5, 95% CI 4.9–6.2). The RR for antepartum anoxia in twins compared with singletons decreased from 4.3 (95% CI 3.5–5.4) in 1982–1987 to 2.9 (95% CI 2.3–3.7) in 1988–1994.
Table 4 presents the effect of the change in stillbirth definition in 1992 on the stillbirth rate in twins and singletons in 1988–1994. In twins, if the old definition of stillbirth (28 completed weeks of gestation) is applied to both time periods, there will be a more significant decrease in both the overall stillbirth rate and stillbirth rate due to antepartum anoxia (26.0 vs 17.7 per 1000, P= 0.01; 17.3 vs 11.2 per 1000, P= 0.01, respectively). In singletons, no improvement in stillbirth rate can be attributed to the change in stillbirth definition in 1992. Thus, using the 28th week of gestation, as a cut off for stillbirth definition in 1988–1994, resulted in a highly significant decrease in the stillbirth rate from 5.5 per 1000 in 1982–1987 to 4.8 per 1000 in 1988–1994 (P < 0.001). However, for singleton deaths due to antepartum anoxia, the decrease in the cause-specific stillbirth rate even after applying the old stillbirth definition was small and not statistically significant (P= 0.28).
To examine factors affecting mortality rates due to congenital malformations, data on terminations of pregnancy for congenital abnormalities were analysed (Table 5). As data on terminations of pregnancy were available only from 1984, the comparison between 1984–1987 and 1988–1994 was made. For twins, there was a decrease in mortality due to congenital abnormalities from 6.7 per 1000 births in 1984–1987 to 4.1 per 1000 in 1988–1994, although the difference was not statistically significant (P= 0.12). After the number of terminations was added to the number of deaths, the mortality rate changed to 8.2 per 1000 births and terminations in 1984–1987 and 6.1 per 1000 in 1988–1994, showing a smaller decrease over time (P= 0.29). For singletons, there was a significant decrease in mortality rates due to malformations from 2.3 per 1000 in 1984–1987 to 1.6 per 1000 in 1988–1994 (P < 0.001). However, when an allowance was made for the greater proportion of terminations of pregnancy in singletons in the second period, the mortality rate due to congenital malformations rose significantly (P < 0.001) over time (Table 5).
Birthweight-specific EPMR in twins and singletons
Table 6 shows birthweight distribution and birthweight-specific EPMR between 1982–1987 and 1988–1994 for twin and singleton births. The EPMR in both twins and singletons decreased over time despite a higher proportion of infants with extremely low birthweight (< 1000 g) in the second period. This increase is partially attributed to the use of the lower gestational age for stillbirth definition since 1992, which led to a higher proportion of extremely small fetuses in both twin and singleton populations. When stillborn fetuses delivered after October 1992 at a gestational age 24–27 weeks with a birthweight < 1000 g were excluded from the analysis, the proportion of extremely low birthweight babies still showed a rise of 24.5% for twins and of 13.8% for singletons from the first to the second period. Table 5 also shows that the EPMR was lower in both twins and singletons for each birthweight category in the later period. Although the crude EPMR was 5.6 times higher for twin births than for singleton, the EPMR for birthweight categories 1000–1499 g, 1500–1999 g and 2000–2499 g was lower for twins than for singletons. As data on gestational age for twin and singleton populations are not available, it is not possible to adjust for gestational age.
There was a significant decrease in the birthweight-specific stillbirth rate due to antepartum asphyxia in twins from 127.1 per 1000 in 1982–1987 to 74.6 per 1000 in 1988–1994 (P= 0.009) for fetuses with a birth-weight < 1500 g, although a nonsignificant decrease was also found in higher birthweight categories. In singletons, there was a nonsignificant decline in stillbirth rate due to antepartum asphyxia in 1988–1994 for low birth-weight infants, whereas the rates increased significantly from 1.4 per 1000 in 1982–1987 to 1.7 in 1988–1994 (P= 0.03) for babies with a birthweight ≥ 2500 g. The similar trend was observed for the overall stillbirth rate in singletons. Unlike the stillbirth rate, the singleton neonatal mortality rate significantly declined over time in all birthweight categories.
In the Northern Region the EPMR in twins during the 13 years was 55.2 per 1000 twin births, 5.6 times higher than in singletons. This is consistent with that reported by other authors5,15. Neonatal death was more common than stillbirth in twins but this ratio was reversed for singletons which may in part be explained by the predominance of antepartum asphyxia as a cause in singletons.
The autopsy rate for singleton deaths at 71% in 1982–1987 was lower than the 75% minimum recommended level by a joint working party of the Royal College of Obstetricians and Gynaecologists and Royal College of Pathologists in 198830 and declined to 64% in 1988–1994. As autopsy findings affect the distribution of causes, caution must be exercised when analysing changes in cause-specific mortality rates if autopsy rates are not considered.
There was a significant decrease in the overall EPMR for both twins and singletons in 1988–1994 compared with 1982–1987 with the decrease in the rates in each birthweight category. There was no decline in the proportion of very low birthweight babies in both twin and singleton births, which eliminates its potential positive effect on the EPMR. On the contrary, there was a higher proportion of extremely low birthweight babies in 1988–1994 than in 1982–1987 in twins and singletons. Even after allowing for the change in stillbirth definition, the increase in the proportion of babies less than 1000 g remained. Therefore, the decline in the birth-weight-specific EPMR contributed to the observed improvement in the crude EPMR in the Region.
A significant decrease in the stillbirth rate but not in the neonatal mortality in twins is the opposite of that reported in other studies13,31. Other groups, however, have analysed outcomes from births which occurred prior to those in the present study. To our knowledge, this report is the first to identify a recent alteration in the trends for twin mortality. The decrease in twin stillbirths was mainly attributed to a significant reduction in deaths from antepartum asphyxia and congenital malformations. In singletons, there was no reduction in deaths from antepartum asphyxia. Thus, for twins compared with singletons, there was a substantial decline in the relative risk for antepartum asphyxia. This decrease in twin stillbirth rate might have been due to enhanced ultrasound surveillance and consequent preterm delivery. No data are available, but it is likely that biometric ultrasonographic monitoring was performed more frequently for twin pregnancies during the later time period. In addition, Doppler ultrasonography was introduced into practice from about 1990 onwards. The use of Doppler ultrasonography has been shown to significantly decrease the antepartum stillbirth rate in high risk pregnancies32, as well as increasing the incidence of premature (< 34 weeks) delivery. It would also have led to the increase in the proportion of extremely low birth-weight babies as a consequence of ultrasound based iatrogenic delivery.
The rise in singleton deaths due to antepartum asphyxia in the normal birthweight categories (≥ 2500 g) during the later time interval is concerning. At the same time that ultrasonographic surveillance for twins has increased, hospital-based assessments for singleton pregnancies have been reduced. From about 1991 onwards, many units in the region devolved a substantial component of antenatal care into the community. However, extreme caution is needed when extrapolating from one time-based trend to another. Continued monitoring is necessary to determine whether this trend is maintained.
In singletons, there was a significant decline in deaths due to congenital malformations and intrapartum asphyxia and trauma. The decrease in deaths due to congenital malformations in singletons was due to earlier antenatal diagnosis and subsequent termination of pregnancy. Thus, early antenatal detection of neural tube defects (i.e. spina bifida), which are the major component of severe congenital abnormalities that are terminated, considerably improved in the Region during 1985–199433.
In contrast to the increase in the stillbirth rate of singletons with normal weight at birth, the neonatal mortality rate decreased significantly for all birthweight categories in 1988–1994 compared with 1982–1987. This observation confirms nationally34 and internationally35,36 reported improvement in birthweight-specific neonatal mortality rate.
Comparison of the distribution of immediate causes of fetal and neonatal death in twins and singletons showed a different distribution. This can be explained by a higher proportion of preterm and low birthweight twins and due to the specific conditions associated with twinning (e.g. twin-to-twin transfusion). The contribution of antepartum asphyxia and congenital malformations to singleton deaths was 1.6 and two times, respectively, greater than in twins. In contrast, immaturity accounted for 15.6% of deaths in singletons compared with 36.0% in twins.
The enhanced efficacy in prenatal management of twin gestation and in neonatal intensive care due to technological advances during the past 20 years has led to a considerable improvement in both twin fetal mortality and survival of extremely preterm infants37,38. Early diagnosis of twin pregnancy, antenatal determination of chorionicity, serial ultrasound scanning, evaluation of fetal wellbeing with the biophysical profile score and umbilical arterial Doppler velocimetry might all have contributed to a decrease in twin perinatal mortality9,32,39–42. Furthermore, with accurate prenatal diagnosis, intensive fetal surveillance, and appropriately timed delivery, perinatal survival of the most vulnerable category of twins—monoamniotic—has improved43. However, despite the decrease in the twin EPMR in the region, the mortality rate due to immaturity, the major cause of twin neonatal death, has not improved. Further research is required to determine whether recent promising investigations and interventions such as transvaginal cervical sonography (± cerclage) and uterine activity monitoring have the potential to decrease twin neonatal mortality from immaturity.
This study has shown that the extended perinatal mortality rates in twins and singletons have decreased over the last 13 years. Examination of the cause-specific mortality rates for twins and singletons has revealed interesting differences in the reasons for the reductions. Less twins are dying as a result of antepartum asphyxia, whereas fewer singletons are succumbing due to congenital malformations and intrapartum asphyxia and trauma. Immaturity remains the greatest threat to twin survival. Future studies will determine whether or not newer strategies as transvaginal cervical monitoring or ambulatory uterine monitoring will reduce twin deaths due to complications of preterm delivery.
We are grateful to Dr C. Wright, Clinical Director of the Regional Maternity Survey Office, for comments on the paper and would like to acknowledge the dedication of Mrs M. Renwick, Operational Manager, and other staff at the Regional Maternity Surveys Office. We thank the Perinatal Mortality. Survey Steering Group for access to the data and are grateful to all the district convenors and coordinators in the Northern Region for their continued collaboration and support of the Perinatal Mortality Survey. Lastly we wish to acknowledge the help of staff at ONS for providing national data for twins and singletons. S.G. is funded by the NHS Executive (Northern and Yorkshire).