Survival and neurodevelopmental morbidity at 1 year of age following extremely preterm delivery over a 20-year period: a single centre cohort study

Outcome measures

At a postnatal age equivalent to 1 year corrected for postmenstrual age, the infants were recalled for further assessment. Verbal consent to the assessment was obtained from the parent who attended. The assessment consisted of a structured neurological examination using a modification of the scheme described by Amiel-Tison (1), a developmental assessment performed by a developmental psychologist using the Griffiths scale, and the recording of clinical data on visual and auditory function. Classification of neurodevelopmental outcome was undertaken blind to the perinatal and neonatal data. The original Griffiths scales (2) were used for infants born between 1981 and 1995. For infants born between 1996 and 2000 the revised Griffiths scales (3) were employed.

In addition to developmental assessment using the Griffiths scale, two further outcome measures were employed:

Neurodisability was classified as either: (a) impairment without disability or (b) impairment with disability. Definitions for these categories are provided in the online supplementary material.

Statistics

Statistical analysis was performed using Statistical Package for Social Sciences (SPSS for Windows version 10.1; SPSS Inc., Chicago, IL, USA). Data were analysed using four sequential 5-year periods (birth dates 1981–1985, 1986–1990, 1991–1995 and 1996–2000). Chi-squared tests were used for categorical variables and ANOVA and Students t-test used for continuous variables.

This study was part of a long-term follow-up programme for very preterm infants for which approval was given by the Joint University College London/UCLH Committee on the Ethics of Human Research.

Maternal and neonatal basic data

During the period January 1981–December 2000, a total of 357 infants between 22- and 25⁺⁶-week gestation were delivered and admitted to the neonatal unit. In addition, in the period from January 1991–December 2000, a further 16 infants were identified as showing signs of life on labour ward at UCLH, but were not deemed viable and hence did not undergo active resuscitation. None of these 16 infants was admitted to the neonatal unit and all died shortly after birth. In a small number of other cases where infants had died on labour ward, no information on gestational age was available and these were excluded from our analysis.

The data were divided into four consecutive 5-year periods to aid analysis. Table 1 shows the total numbers of admissions to the neonatal unit for each time period. There were similar numbers of infants recruited during each time period. There was no significant difference in mean maternal age or in the multiple birth rates. From 1991 and onwards, there was a significant increase in the incidence of caesarean section. Prior to 1992, the use of antenatal corticosteroids was not routinely recorded. There appeared to be a significant increase in the use of antenatal corticosteroids in the last 5-year period 1996–2000.

Of the 357 infants admitted to the neonatal unit, 248 were inborn at UCLH and 109 were born at other hospitals and transferred postnatally. There was a trend towards a decreasing rate of postnatal transfer over the 20-year period. There was a significant reduction in mean birthweight over the 20-year period (p = 0.006) with the most marked reduction occurring between the latter two time periods.

Survival

Table 2 shows survival to discharge from hospital in each time period according to gestational age. The denominator represents all admissions to the neonatal unit within the gestational age range. Whilst overall survival remained fairly constant between 1981 and 1995, a significant increase was seen in the last period 1996–2000 (p ≤ 0.001). There was a trend for increased survival in inborn infants compared with outborn (see online supplementary material). Table 3 shows the survival of UCLH born infants between 1991 and 2000 using a denominator which includes those infants born alive on labour ward but not resuscitated and transferred for intensive care. Although the numbers are small, there was a clear trend towards increased survival with time. Figure 1 illustrates the changes in survival for different gestational ages, as a percentage of all admissions.

Supplementary information on maternal and neonatal variables and on survival in inborn and outborn infants is presented in the online version of this paper.

Outcome at 1 year

Of the 357 infants admitted to the neonatal unit, 163 (45%) survived to discharge from the unit. A further two infants died following discharge from the unit, but before the age of 1 year and 161 were available for assessment at 1 year of age. Of these 159 (99% of those available) underwent a Griffiths developmental assessment, 149 (93%) had a structured neurological examination and the overall neurodevelopmental status was determined for 148 (92%). The mean age at the time of the Griffiths assessment was 53.7 weeks corrected for prematurity (range: 45–72, standard deviation [SD]: 3.9).

The mean totals and sub-quotient scores for the Griffiths assessment are shown in Table 4. A major restandardization of the Griffiths developmental scales was carried out prior to 1996 and the final 5-year period of our study employed the new Griffiths scales. Statistical analysis of the variation in mean score between the time periods was therefore not possible. The data show an apparent fall in the mean scores for both the general quotient and the sub-scales in the final period. This is discussed below. Table 5 shows the results for overall neurodevelopmental assessment expressed as a percentage of all those assessed. No significant temporal trend was apparent with regard to the percentage of children with evidence of neurodisability at 1 year of age.

The study population

The study population included all infants who were admitted to the neonatal unit at UCLH within a gestational age range of 22–25⁺⁶ weeks over a 20-year time period. Throughout this time, there have been major advances in the clinical management of these infants. However despite this, the overall philosophy of care has remained constant. The general policy of the unit over this period was to offer intensive care to all extremely preterm infants who, in the opinion of the senior clinical staff, had a realistic possibility of benefiting from this treatment. When infants were born alive around the limits of viability, the usual policy was to admit the infant to the neonatal unit where, following a period of assessment, a decision would be taken on whether to continue intensive care.

As the study infants included a substantial number who were transferred after birth from a range of hospitals in London and South England, it is not possible to define the denominator population from whom the infants were derived. However, there was no evidence of any systematic selection bias and there is no reason to think that the study infants are unrepresentative of those admitted to tertiary regional neonatal centres in countries with advanced healthcare systems. It is possible that the infants transferred from other hospitals might represent an atypically ‘healthy’ population and hence the overall survival figures were inflated. However, in our cohort, the outborn infants had lower survival rates in general compared with the inborn infants. As a consequence, the inclusion of outborn infants tended to depress the overall survival rates. The improved outcome of inborn compared with outborn infants has been reported previously (4), and is reflected in the accepted policy of transferring mothers at risk of extremely preterm delivery to perinatal centres wherever possible. It is likely that increasing implementation of this policy was the cause of the progressive fall in the percentage of outborn admissions, from 46% in 1981–1985 to 22% in 1996–2000.

A further possible explanation of the increased survival following admission to the neonatal unit is that an unusually large number of extremely preterm infants who were born alive at UCLH died without being admitted to the neonatal intensive care unit. However, when the survival of infants born at UCLH was calculated as a proportion of all live births at the hospital within the period 1991–2000, the overall survival rate was not greatly altered (Table 3). It is possible that we failed to identify a significant number of infants who were born alive at UCLH but who died in labour ward. However, this is unlikely because the policy adopted by the senior clinicians in this period was generally to admit babies who showed signs of life to the neonatal unit for further assessment.

A limitation of our study is that the outcome was assessed at a corrected age of 1 year and hence it is likely that the true incidence of neurodisability in the cohort was underestimated. Longer-term outcome studies of extremely preterm infants have demonstrated an increased incidence of educational and behavioural problems at school age (5,6). A previous study from our group found that the developmental assessment at 1 year was generally predictive of outcome at the age of 8 years if the children were either classified as normal or as having impairment with disability at 1 year (7). However, about half of children who showed impairment without disability subsequently required extra educational provision at the age of 8 years.

The Griffiths scales were restandardized in 1996 to reflect the rise in mean scores observed in control populations because the scales were originally standardized in the 1940s (3). As a result of this restandardization, there was a tendency for scores to fall. An infant who obtained a mean developmental quotient of 111 on the previous Griffiths scales would score approximately 100 on testing with the 1996 edition. It is therefore not possible to make a direct comparison between the Griffith scores obtained in the years 1996–2000 with those obtained between 1981 and 1995. However, it is notable that the assessment of neurological impairment and overall neurodevelopmental status did not show any significant increase in the incidence of disabling impairment in the period 1996–2000.

Comparison with other published data on neonatal survival

Improvements in survival with time have been documented by other groups (8–15). Single centre studies have tended to show higher survival figures, as do those which include infants of greater gestational ages. Hoekstra et al. (16) reported the outcome of infants between 23 and 26 weeks gestation over a 15-year period from 1986 to 2000 in single tertiary level neonatal unit in the United States. Their data showed a progressive improvement in overall survival as a percentage of admissions from 53% in 1986 to 89% in 2000. Survival at 23 weeks gestation rose from 40% to 66% and at 24 weeks gestation from 49% to 81%. The Victorian Infant Collaborative Study Group (11,12) also showed similar results for a population of infants with birthweights 500–999 g between 1979 and 1997. Survival at 2 years rose from 25% in 1979 to 73% in 1997. In the same geographical region, survival as a percentage of live births at 23 weeks rose from 10% to 41% between 1991 and 1997, and at 24 weeks from 33% to 41% (17).

Survival data from large population-based studies have shown lower overall survival rates compared with single centre studies. The EPICure study undertaken in the whole of United Kingdom and Ireland in 1995, gave overall survival rates of 11% of total livebirths at 23 weeks, 26% at 24 weeks and 44% at 25 weeks of gestation (18,19). As a proportion of admissions to a neonatal unit, the relevant survival figures were 20%, 33% and 52%. Figure 2 illustrates survival and outcome as a proportion of all admissions to a neonatal unit in the EPICure study, compared with similar data from UCLH for the period 1996–2000.

Large population-based studies have the advantage of avoiding problems of selection bias and enable the accumulation of large numbers to reduce statistical errors. However, they inevitably represent the outcome from a very heterogeneous range of perinatal facilities. The EPICure study incorporated data from all 276 maternity units throughout United Kingdom and Ireland. It is therefore inevitable that these units encompassed a wide range of staffing, resources and expertise in the care of extremely preterm infants. In addition, it seems likely that the maternity units encompassed a range of management policies (both obstetric and paediatric) towards infants born at the limits of viability, ranging from ‘proactive’ to ‘noninterventionist’. Hence, it is not surprising that average survival figures obtained from large population based studies will be different from those observed from single centre cohorts.

Effect of management policies on survival rates

A large population-based study from Sweden showed that regions that carried out ‘proactive’ management of extremely preterm infants had significantly higher survival rates than those associated with a ‘selective’ policy (20). In the period 1995–1999, survival as a percentage of liveborns at 23 weeks ranged from 16% to 41% with different management policies. At 24 weeks, the comparable figures were 49% and 70%. A study of outcome in 12 tertiary neonatal centres in United States found significant differences between centres (21). Variations in ‘active’ resuscitation of extremely preterm infants ranged from 5% to 28% and overall survival ranged from 51% to 72%. Hence, differences in management policies between units are likely to result in marked variations in the rates of survival around the limits of viability.

Neurodisability

Our data do not show a significant rise in the percentage of surviving infants with neurodevelopmental impairments leading to disability at 1 year of age (Table 5). However, over the 20-year period, there was a rise in the absolute numbers of survivors, both of those who were classified as ‘normal’ and of those who were classified as impaired with disability. The total number of infants classified as ‘normal’ rose from 12 in the first period to 29 in the last period of the study. At the same time, the total of survivors with disability rose from 5 to 13. Hence, advances in neonatal care led to an increase in both the total number of ‘normal’ infants at 1 year of age as well as an increase in the total number of infants classified as disabled. Longer-term outcome studies are required to assess the number of children with significant impairments at school age and beyond.

Wilson-Costello et al. (22) assessed two cohorts of infants born with birthweights of <1000 g during 1982–1989 and 1990–1998. They found that the survival increased from 49% in 1982–1989 to 67% in 1990–1998, but that the rate of overall neurodevelopmental impairment at 20 months rose from 26% in 1982–1989 to 36% in 1990–1998. However, in a subsequent study (23) comparing the outcomes of infants born in 2000–2002, the same authors found improved neurodevelopmental outcome without a further change in survival. The Victorian Infant Collaborative Group (12) reported a marked increase in survival over two decades without an associated increase in overall disability, which remained approximately constant. O'Shea et al. (13) studied infants with birthweights of <800 g born in North Carolina between 1979 and 1994. Their results showed that survival increased from 20% to 59% while the incidence of major neurodevelopmental impairment remained relatively constant at 21–28%. It is clear that no consistent relationship has been demonstrated between changes in survival and the overall incidence of neurodevelopmental disability, and our data support the conclusion that increased survival may occur without any overall increase in the percentage of disabled survivors.