Changing prognosis for babies of less than 28 weeks' gestation in the north of England between 1983 and 1994BMJ 1997; 314 doi: https://doi.org/10.1136/bmj.314.7074.107 (Published 11 January 1997) Cite this as: BMJ 1997;314:107
- a South Cleveland Hospital, Middlesbrough, Cleveland TS4 3BW
- b Royal Victoria Infirmary, Newcastle upon Tyne NEW 4LP
- Correspondence to: Dr Tin
- Accepted 28 October 1996
Objective: To investigate the changing prognosis for babies of less than 28 weeks' gestation.
Design: A prospective, collaborative, population based survey.
Setting: The former Northern Regional Health Authority.
Subjects: All the births between 1983 and 1994 at 22 to 27 completed weeks' gestation to women normally resident in the region.
Main outcome measures: Miscarriage, stillbirth, death in the first year of life, and disability in survivors.
Results: There were 479 070 registered births in the study period. No baby of 22 weeks' gestation survived; only eight (4%) of the 197 babies of 23 weeks who were alive at the onset of labour survived for a year–a proportion that did not change during the study period. Survival among other babies of less than 28 weeks improved progressively between 1983-6 and 1991-4, but administration of artificial surfactant to babies requiring ventilation from mid-1990 was associated with further improvement in survival only in those over 25 weeks' gestation. Babies of 24 weeks required three times as much high dependency care per survivor as babies of 27 weeks (76 v 26 days). The rate of severe disability in the one year survivors of less than 26 weeks' gestation (30/123; 24%) was similar to that seen in the sampled survivors of 26 and 27 weeks (29/108; 27%); the proportion disabled did not change significantly during the study period. All the children born in 1983, 1987, and 1991 were later reassessed in greater detail: 10% (13/136) seemed destined for a continuing life of total dependency.
Conclusions: Gestation, if accurately assessed, can give a woman facing very preterm delivery a clear indication of the prognosis for her baby and help her judge the appropriateness of accepting obstetric intervention and sustained perinatal support.
Survival in babies of 24-27 completed weeks of gestation has improved in the past 12 years
The proportion of survivors with severe disability (25%) has not changed
The survival (viability) of babies born before 24 weeks' gestation (about 4%) has not changed, and most survivors are severely disabled
One in 10 of all survivors has a disability so profound that he or she is never likely to become independently mobile or to communicate effectively with others
Gestation, if accurately assessed, can help women facing very preterm delivery (and their attendants) to assess the likely prognosis for the baby at the onset of labour
The prognosis for very preterm babies was substantially improved 25 years ago by effective artificial respiratory support. In the past few years there has been much debate as to whether more recent developments in perinatal care have had any further impact on the limits of viability.1 The issue has important implications for obstetric care, for the wise use of limited neonatal resources, and for the provision of sensitive and realistic nursing support for babies born 15 or more weeks early.2 3 We therefore looked at the immediate and long term outcome of all babies delivered either dead or alive at 22 to 27 weeks' gestation in the north of England over a 12 year period.
Subjects and methods
A voluntary, collaborative, region-wide survey of perinatal mortality has been in operation in the area served by the former Northern Regional Health Authority since July 1980.4 This started with the collection of confidential information on the outcome of all pregnancies resulting in a registrable birth, but information was also obtained on the outcome of every pregnancy lasting at least 22 weeks in 1983,5 even if it ended in a miscarriage or termination, and similar information was collected routinely on all mothers normally resident in the region every year after 1985. Pathology and gynaecology records were reviewed to ensure that no late miscarriages were missed. Information on gestation was obtained from the obstetric notes and calculated from the mother's menstrual history unless this was uncertain or differed by at least 14 days from the estimate obtained from at least one reliable ultrasound scan undertaken before 20 weeks' gestation. Information on every stillbirth and infant death was cross correlated with the independent registration returns made to the Office of Population Censuses and Surveys, which also provided information on the total number of registered births to mothers normally resident in the region each year.
In addition, a record was kept of the progress of all surviving babies of less than 26 weeks' gestation after 1982, and this was supplemented with a more detailed assessment of the developmental attainment of every child born before 28 weeks in 1983, 1987, and 1991, undertaken in a standardised way at 2 years by one of four clinicians. Severe disability was defined by using the criteria listed in the 1994 consensus report published by the Oxford Regional Health Authority jointly with the National Perinatal Epidemiology Unit.6 The Griffiths's developmental quotient was assessed in each child seen at 2 years. No child was lost to follow up.
Attitudes to the registration of very immature infants who showed only transient signs of life after birth are not always consistent.7 Neonatal mortality can be influenced by the vigour with which these babies are resuscitated after delivery, while intensive care may merely delay death.8 Studies that include only babies admitted to the neonatal unit after delivery often quote misleadingly high rates of survival. A more informative and potentially more consistent statistic relates survival to the number of babies alive at the onset of labour. Babies in this study were so classed if a normal fetal heart rate had been documented by using Doppler or real time ultrasound, or both, after the induction or spontaneous onset of labour but the baby was not registered as live born (although there were two singleton and three twin births in which advanced maceration after delivery suggested that this observation had been in error). When it was not possible to time the onset of labour precisely death was classed as having occurred during labour if the fetus was alive when the mother first came under observation in the delivery area with threatened labour or a suspected abruption.
Comparative data on neonatal mortality for the period 1966-9 were abstracted from case records of the Princess Mary Maternity Hospital, where carefully validated information on gestational age for every live birth was collected for research purposes by staff from the University of Newcastle upon Tyne from mid-1965.
There were 479 070 registered births to mothers normally resident in the region between 1983 and 1994 inclusive, including 39 registered live births at 19 to 21 weeks' gestation (10 with a birth weight of ≥500 g) all of whom died on the day of birth. The outcomes for all deliveries at 22 to 27 weeks (other than those due to a notified termination of pregnancy) are shown in table 1. Reliable ultrasound information with which to validate the menstrual information on gestation was available for 96% of these pregnancies, and in all but a small minority of cases this was a measurement of biparietal diameter at between 14 and 18 weeks. The number delivered at each week of gestation rose as pregnancy progressed (fig 1).
Figure 2 shows how neonatal survival has improved for babies of short gestation over the past 25 years. The statistic of greatest relevance to the obstetrician and to the family, however, relates survival to the number of babies alive at the onset of labour: information on non-registrable late miscarriage was not available for the years before 1983, but figure 3 shows how the prognosis for these babies rose between 1983-6 and 1991-4. No baby of 22 weeks survived to 1 year (table 1), and only 7.5% (8/121) of the live born babies of 23 weeks survived (a proportion that did not increase during the 12 year study period). Of the babies of 22-23 weeks who did not survive, 92% died within two days of delivery. The widespread availability and free use of artificial surfactant in the context of the OSIRIS trial (open study of infants at high risk or with respiratory insufficiency–the role of surfactant)9 from March 1990 was associated with a further improvement in survival in babies of 26-27 weeks in the third period of four years, but there was no comparable improvement among babies less mature than this (table 1). Birth at 27 weeks became progressively more common between 1983 and 1994, and an increase in the number of induced labours (excluding inductions after fetal death) and caesarean deliveries before the onset of labour possibly explains at least half this rise (21% of deliveries at 27 weeks were elective at the end of the 12 year period). No such trend with time was seen in babies born earlier than this. Deaths after discharge accounted for a third of all the postneonatal deaths in the first year. Table 2 shows how the amount of high dependency care10 received varied with gestation at birth.
Severe disability among the one year survivors of less than 26 weeks' gestation (24.4%; 95% confidence interval 17.1% to 33.0%) was similar to that seen in the sampled survivors of 26 and 27 weeks (26.9%; 18.8% to 36.2%). The proportion with severe disability did not change during the 12 year period (table 3). The children born in 1983, 1987, and 1991 were all reassessed in greater detail when they were 2 years old. Several then had a developmental quotient of less than 50 and disabilities likely to prevent their ever developing any independent mobility or any ability to communicate intelligibly, or both. Such profound disability was not seen in any of the nine survivors of less than 25 weeks' gestation but was seen in 13 of the 136 survivors of 25-27 weeks (9.6%; 5.2% to 15.8%).
Changes in maternity care, probably associated with the introduction of prolonged respiratory support, seem to have lowered the threshold of “viability” by two weeks during the past 28 years (see fig 2). The survival rate for 1966-9 shown here is very similar to that reported by Lubchenco et al from Denver, Colorado, for much the same time period.11 There is, however, no evidence that the improving prognosis for babies of 24 or more weeks' gestation over the past 12 years has had any further impact on the lower limit of viability (see fig 3). No baby of 22 weeks' gestation survived during that time and only eight of the 197 babies of 23 weeks alive at the onset of labour. Four of these eight survivors had severe disability on subsequent follow up, but none was so disabled as to seem destined for a continuing life of total dependency–that is, unlikely to achieve independent mobility or an ability to communicated freely with others in later childhood, or both.
The present gestation specific survival rates for live births are very similar to the only other available area based rates for the United Kingdom.121314 Survival has been a little higher in some of the hospital based studies published from centres in Australia,15161718 Canada,192021222324 and the United States,25 26 but it is widely recognised that the results achieved by a specialist institution do not always reflect those found in the population at large. Only a minority of studies report survival to 1 year rather than survival to discharge, and many exclude lethal malformation. Nevertheless, all these reports conclude that survival is never seen in babies of less than 23 weeks' gestation and is only to be expected in about 5% of all live births at 23 weeks. Other reports from Europe have found no evidence of long term survival in babies born before 24 weeks' gestation.2728293031 Only one previous study has documented death during delivery.23 Nine of these reports included information on disability in long term survivors, but only one was a population based study,12 and in only four was there no loss to follow up among babies of less than 28 weeks' gestation.16 20 27 30 The failure of surfactant to have any detectable impact on survival in babies of less than 26 weeks conflicts with the interpretation placed on several other studies that have used historic controls,323334 but is in keeping with the only randomised, multicentre, placebo controlled trials of surfactant specifically designed to study its efficacy in babies of less than 750 g at birth.35 36
The findings on the limits of viability reported here stand in stark contrast with those published from the research network supported by the National Institute of Child Health and Human Development (NICHHD) in America,37 38 which have formed the basis for the advice on viability recently issued in Canada2 and in the United States.3 The hospitals in this network have recently reported survival to discharge in 19% of all in-born live births of 22 and 23 weeks' gestation and a survival rate of 47% at 24 weeks' gestation.39 These discrepancies are hard to explain, but more than 18% of the mothers in the initial NICHHD studies had received no antenatal care; there must, therefore, be some doubt about the accuracy of some of the gestational assessments. Significant errors may also creep in if such information is not abstracted directly from the obstetric record.40 It should also be noted that nearly two thirds of the mothers in the early NICHHD studies were black and that nearly a quarter of the babies were classified as “light for dates.” It is known (from other NICHHD data) that both these factors significantly enhance rates of survival specific for birth weight.33 These factors may well explain why NICHHD birthweight survival rates are also higher than in most other studies.
Birth weight provides an alternative yardstick for the assessment of viability and one that can be used reliably even when gestation is not known for certain. It is not, however, one that can be used to influence management before delivery. There is also evidence to suggest that gestation is a better predictor of death or disability41 42 (and also of length of stay before discharge5 13) than birth weight.
The risks of severe disability in survivors born at 24-25 and at 26-27 weeks' gestation in this study were not dissimilar. Others have found a clear diminution in risk as gestation advances,12 23 but this is far from universal.16 19 43 At least half the very few survivors of 23 weeks' gestation, on the other hand, were severely disabled, as in every other published study.44 While it is accepted that cognitive disability is not always recognisable in children as young as 2, the likelihood of survival and of survival without severe disability is certainly higher than many clinicians currently assume.45 46 Boys were more likely to die, both during and after delivery, and more likely to survive with disability. Babies who were “light for dates”47 were more likely to die (as in most other studies33 43 48) but no more likely to be disabled if they did survive. The proportion delivered by section rose from 12% of all live births in 1983 to 23% in 1994, but there was no significant correlation between mode of delivery and outcome after allowance was made for gestation at birth.15 49 The disability rate in babies of less than 28 weeks' gestation was double that seen in babies of 28-29 weeks in this region (25.6 v 11.1%), as in another recent large community based study.50
These findings point to the importance of establishing gestation reliably to assess the viability of the very preterm baby and the risk of disability in survivors. They suggest that current recommendations,2 3 based on NICHHD data from a predominantly black population,37 38 cannot be used without modification to guide practice in a largely white community. They also suggest that improvements in mortality in the United Kingdom in the past 12 years associated with the arrival of surfactant therapy have not been associated with any change in the threshold of viability or in the chance of severe disability in survivors. These data should not be used to establish an arbitrary “threshold” below which resuscitation should never be offered at birth. They can be used, however, to influence the vigour with which further respiratory support is pursued in consultation with the family and in the knowledge that in the United Kingdom most babies of less than 25 weeks' gestation who do not survive currently die within two days of delivery (fig 4).
We are grateful to the Office of Population Censuses and Surveys (now the Office for National Statistics) and the coordinating group responsible for the region's maternity survey for access to mortality data, and to our medical, midwifery, and nursing colleagues in the Neonatal Network for the other information on which this report is based. We are particularly grateful to Drs Donald Bell and Monica Placzek for providing information on children born in South Cumbria, many of whom received their early neonatal care in Lancaster, Leeds, or Manchester. Neonatal survival data for Newcastle for the years 1966-9, analysed by gestation, was provided by the late Professor JM Parkin. Drs John Beesley, Vona Ellis, and Sue Thomas provided information from their unpublished follow up study of all the children born in 1987. An expanded version of table 1, showing the data for each calendar year separately, is available on request.
Conflict of interest: None.