Article Text
PDF
Statistics from Altmetric.com
Introduction
Survival after very preterm birth has increased across the developed world, but a consistent theme to emerge in published studies is that the increase in survival may come at the cost of increased morbidity.1,–,3 Hack and Fanaroff4 summarised the predominantly North American experience from the 1990s and concluded that costs of salvaging ever more immature babies were beginning to outweigh the benefits. Their data are predominantly based on birth weight rather than gestation, are derived from heterogeneous populations (unit based, regional) and reflect the situation in a healthcare system (in the USA) in which the provision of tertiary neonatal care has become steadily more regionalised since the 1970s.5 Regionalisation of perinatal care in Europe is less developed and the demographics and models of care delivery vary between the different European countries.6 7 This article is an attempt to draw together the European literature on population-based outcomes of very preterm babies from the past 25 years.
Methods
A literature search was carried out on MEDLINE and EMBASE databases for all studies citing outcomes of preterm populations over the past 30 years. Studies that referred to cohorts that were not population based (for this purpose one confined to a defined geographical region or country) were excluded. Articles cited in the retrieved literature and not identified by other means were located separately. In addition, reference was made to the author's own collected literature and through direct correspondence with two of the authors.
Findings
A total of 29 population-based studies published since 1986 relating to cohorts born between 1983 and 2006 were identified; 13 studies reported survival outcomes only8,–,19 (table 1); 12 reported developmental outcomes on an immature subset1 and at least 217 19 will be reporting long-term outcomes in the future. Information on neurodevelopmental outcome was available for 17 cohorts20,–,59 (also W Tin, personal communication) with latest outcomes reported varying from 2 to 19 years of age (table 2). A total of 13 reports32,–,34 36 37 39 45,–,49 51 57 34 36,–,42 from 6 cohorts compared outcomes with classmates or age matched term born children. The studies related to countrywide populations in 14 instances and regional populations in the remainder from 13 European countries. One study (MOSAIC, for ‘Models of OrganiSing Access to Intensive Care for very preterm births’) included populations from more than one country. All studies apart from five15 30 34 36 54 (two of which reported some outcomes by discrete gestational age37 55) used gestational age as the primary criterion for inclusion and some also used birth weight. The most commonly used gestation range was from 20 to 31 weeks. Nine studies focused on very immature babies (<27 weeks) and one study included babies of 32 weeks.
Studies reporting mortality and short-term outcomes only to date
Studies reporting long-term outcomes
Ascertainment
There is compelling evidence that those children who are traceable but whose families do not engage with follow-up studies when they are approached have a significantly higher rate of morbidity than those whose families cooperate.60,–,62 Ascertainment in the majority of studies was good (median 89%, range 52% to 100%). For most studies, there is not enough information about the missed children to tell whether they were representative of the reported group. There was no consistent relationship between ascertainment and the size or date of the studies.
Mortality
All studies found that gestational age was the most powerful predictor of mortality and several reported an additional significant contribution from birth weight.7 30 55 56 It was possible to ascertain numbers of live births in all but 1 study28; denominators for deaths during labour and admission to neonatal intensive care unit were reported in 16 studies. All but one study reported mortality before discharge (1–36 weeks corrected age (CA)) and all follow-up studies reported the number of post discharge deaths and of children not assessed. Studies comparing mortality over time within populations,8 10 12 15 showed a progressive decrease in mortality over time in each gestational age band except at <24 weeks where mortality was unchanged. There was also evidence of a gradual increase in age at death over time although more than 95% of deaths still occurred before discharge from hospital.2 24 25 Comparisons of similar cohorts between Trent, UK with Denmark11 and France13 revealed differences in mortality which were largely accounted for by differences in the preterm birth rate, attitudes to recording and resuscitating very immature babies and case mix. The MOSAIC study reported wide variations in mortality between 10 European regions17: as much as twofold for babies alive at the onset of labour and fourfold for those babies admitted to neonatal units and these differences were not confined to the most immature babies. Obstetric attitudes to certain key interventions were a contributing factor63 and mortality was highest in the region with the most centralised neonatal intensive care service and lowest in the one with the highest ratio of maternity units to total births.
Cerebral palsy
Differences in definition of disability between studies and over time make direct comparisons between them difficult25 64 but all studies report an excess of cerebral palsy (CP) in the very preterm group compared to a reference group or normal term population. Rates of CP increase with decreasing gestation (6% to 9% for all <32 weeks but 16% to 28% for those <26 weeks). Spastic diplegia (44% to 54%) and quadriplegia (24% to 28%) were the commonest types seen and remained so over time. Hemiplegia and monoplegia occurred in about 10% and other movement disorders in about 12%. There did not appear to be a relationship between gestation and the type of CP seen. In one study57 half of the CP children were able to walk unaided, 14% were able to walk with a stick and the remainder were non-ambulant. In all, 3% to 4% of children with moderate or severe CP had visual or hearing impairment. The incidence of CP was increased in children who were growth restricted at birth.36 55
Cognitive and intellectual function
A consistent finding across all studies is that cognitive deficits became more apparent with increasing age. There is disagreement between investigators on the ability to predict later cognitive outcome from a Bayley assessment at around 2 years but there may be better prediction for children with neurosensory impairment.25 65 Cognitive ability appears to have more relationship with gestation than CP but children with growth restriction achieved lower scores at 5 years than their appropriately grown counterparts.55 In the POPS (for ‘collaborative Project On Preterm and Small for gestational age’) cohort, 12.6% of children had cognitive or neurosensory impairment at 19 years but their mean IQ was 97.8. Some of the variance in IQ was explained by gestational age (approximately 1.3 IQ points for each gestational week) but most was explained by parental intelligence.26 These results may reflect the very small proportion of extremely immature survivors in the POPS study (33 of <27 weeks and only 5 of <26 weeks). EPICure children at 11 years of age had significantly lower scores than classmates for cognitive ability, mathematics and reading and were twice as likely to need special educational input.46
Neuromotor and executive function
Subtle motor problems and deficits in executive function (broadly integrative functions necessary to achieve precise goal directed behaviour) were shown to have an important contribution to overall impairment in EPICure children at 6 years of age.45 Study children without CP did significantly less well than term classmates on posting coins, heel walking and standing on one leg and had more associated motor overflow and mirror movements during tasks. Non-right handedness was almost three times as common. A total of 54% of the variance in teacher ratings was explained by cognitive, visuospatial and sensorimotor deficits.
Language
In all the cohorts from which detailed language assessment was reported31 32 37 47 extremely preterm children scored significantly lower than controls or tests norms. Scores on speech articulation, prereading skills and understanding simple instructions were principally affected in two studies.32 37 EPICure children were nearly 10 times as likely to have language problems across all domains at 6 years compared with controls and two to five times as many of the abnormal group were boys.47
Vision and hearing
Significant disability from visual or hearing loss is gestation dependent and rates are moderately consistent across cohorts. In all, 10% of EPIPAGE survivors born at ≤26 weeks had severe visual disability compared to 2% of those born at ≥28 weeks. Severe hearing loss is less common (1% to 2% of ≤26 weeks and <1% in those ≥28 weeks). Poor visual perception (>2 SD above the mean) in extremely preterm Swedish children born in the early 1990s was five times as common as in controls (21% vs 4%).37 The importance of the contribution to severe visual loss from retinopathy has become clearer in the cohorts born since systematic screening was introduced (1988 in UK). In the EPICure cohort five of the six babies with severe visual disability had been treated for retinopathy. However, although the increased survival of the more susceptible infants has meant that rates and absolute numbers of retinopathy of prematurity (ROP) needing treatment are on the increase, gestation-specific rates of visual impairment over a 15-year period to 1998 had decreased,66 which may reflect the efficacy of ROP screening programmes.
| Study | Date | GA (weeks)/BW (g) | Age at assessment | Disability diagnosis | Disability rate |
|---|---|---|---|---|---|
| The Netherlands POPS | 1983 | <32 | 2 years | ‘Major handicap’: no definition (57% of these had central motor deficit, mainly CP). | 42/656 (6%) |
| <32 | 5 years | ‘Major handicap’: World Health Organization (WHO) definition. | 47/648 (7%) | ||
| <32 | 10–11 years | Unable to function without help due to visual or hearing impairment, unable to walk, eat or dress secondary to CP or estimated IQ <55. | 45/433 (10%) | ||
| <32 | 14 years | >1 moderate or severe problem in cognition (4%), hearing (2%), vision (2%), neuromotor (8%), education (32%). | 36/475 (8%) | ||
| <32/<1500 g | 19 years | 227/705 (32%) | |||
| Northern UK | 1983 | <32<1500 g | 2 years | CP limiting physical activity, hearing loss needing aid, visual acuity <6/60, DQ <70 | 20/230 (7%) |
| Scotland UK LBW study | 1984 | <1500 g only | 4.5 years | Severe neuromotor impairment, blindness, deaf with aids | – |
| S Bavaria, Germany, Bavarian longitudinal study | 1985–6 (13 months) | <32 subset | 6.25 years | Serious impairment on Kaufman-ABC (controls 0.8% to 1.9%) or language (HSET) (controls 0.8%) | 1.6% to 2.9%, 1.9% |
| Sweden | 1985–86 | <29 | 10.5 years | WISC-III-R full scale (vs control): 89.8 vs 106.5, VMI 93.3 vs 109.6, CBCL 26.1 versus 15.5 | – |
| Sweden (south-east) | 1987–8 (15 months) | ≤1500 g | 4 years | Scheffzek category 3 or 4 | 3/82 (4%) |
| Northern UK NNN | 1990–1 | 23–32 | 2 years | CP limiting physical activity, hearing loss needing aid, visual acuity <6/60, DQ <70 | 64/566 (11%) |
| Sweden | 1990–2 (2 years) | ≤1000 g | 3 years | CP grade 3–4 (7%), severe visual impairment (4%), severe hearing defect (0%) | 40/362 (11%) |
| Sweden | 1990–2 (2 years) | 23–25 | 11 years | Mod or disabling CP, visual impairment <20/200, hearing deficit plus aids, special education | 18/86 (21%) |
| Liverpool UK | 1991–2 | <32 | 7–8 years | At least one of: MABC <fifth percentile, VMI score <85, COMPS score (only 227 tested) <0 | 128/227 (56%) |
| Trent UK | 1991–3 | <26 | 2 years | Severe disability (WHO) | 19/55 (34%) |
| Denmark ETFOL | 1994–5 | <28 | 2 years | Scheffzek score ≥2, visual impairment, deaf with aids, psychomotor retardation | 31/183 (17%) |
| UK and Ireland EPICure | 1995 (10 months) | <26 | 2.5 years | Unable to walk without assistance, blind, impaired hearing even with aids, no clear speech. | 64/283 (23%) |
| 6 years | Non-ambulant CP, IQ>3SD below mean, profound sensorineural hearing loss, blindness. | 53/241 (22%) | |||
| 6 years | Serious impairment (<2 SD) on Pre-School Language Scale 3 (PLS-3 (UK)). | 31/199 (16%) | |||
| 6 years | MABC (+1.7 s), NEPSY (−0.76 SD), teacher assessment compared with classmates. | 37/191 (19%)/195–219 | |||
| 6 years | Rutter Behaviour Scales, Child Behaviour Checklist. | ||||
| 11 years | ‘Serious impairment’: maths 44%, reading 30%; cognitive 40%; SEN statements 34%. | ||||
| 11 years | Risk of psychiatric diagnosis 27% versus 9% in controls. ASD 8% vs 0%. ADHD 7% vs 1%. | ||||
| Switzerland | 1996 | <32 | 2 years | Neurological impairment including visual and hearing abnormalities | 10/92 (11%) |
| Finland | 1996–7 | <1000 g | 5 years | Major disability | 41/172 (24%) |
| The Netherlands LFUPP | 1996–7 | <27 | 2 years | Bayley I MDI/PDI <68, Hempel neurology score ‘definitely abnormal’ | 10/28 (36%) |
| France EPIPAGE | 1997 | <33 | 5 years | Non-ambulant CP, MPC score <55, severe visual or hearing difficulty | 83/1600 (5%) |
| Belgium EPIBEL | 1999–2000 | <27 | 3 years | Bayley II MDI <55or PDI<55, ‘severe’ CP, deafness even with aids, severe visual loss | 27/89 (30%) |
Summary of disability rates in individual studies
Behaviour and psychosocial adaptation
Teenage children from the POPS study demonstrated less risk taking behaviours than peers and were less socially adept.26 Swedish 10-year-old children had an increased incidence of mental or emotional delay, were less able to play and socialise and had more trouble understanding simple instructions compared with controls.37 Five studies describe an excess of behaviour problems, distractibility and attention deficit hyperactivity disorder.26 31 33 48 51 Hyperactivity and conduct problems in the EPICure children at 6 years could be explained by cognitive deficits but this was not the case for attention deficits, peer and emotional functioning (three to five times the rate in control children).40 At 11–12 years of age EPICure children had a twofold to threefold higher risk of a psychiatric diagnosis than normal gestation classmates50 and were at higher risk for autistic spectrum disorder (ASD).51 There was an association between ASD and neurocognitive outcomes.
Growth
All studies reporting growth describe increasingly negative standard deviation scores (SDS) over the early years for height, weight and head circumference.30 33 36 37 52 55 There is conflicting evidence thereafter. Children in the POPS cohort had caught up to their predicted percentile range at 9–10 years so long as they had birth weights between the 10th and 90th percentiles. Growth-restricted children were more disadvantaged55 67 with SDS of −0.2967 in contrast to the findings in the Scottish cohort in which only head circumference was different from children who were appropriately grown at birth.30 The EPICure cohort (much more immature on average) had significantly impaired growth at 30 months with SDS for height, weight and head circumference ranging from −1.39 to −1.76. At 6 years there had been some catch up in height and weight (by approximately 0.4 SDS), but very little in head circumference (0.13 SDS) even when children with CP were excluded68. These findings were replicated in the Finnish cohort both at 655 and 11 years.69 Body mass index (BMI) remained unchanged between 30 months and 6 years (SDS −0.88).68 Elevated BMI in the POPS cohort at 19 years was associated with an increased rate of weight gain in the postnatal period and, to a lesser extent, in early infancy.70
Blood pressure
A constellation of factors (umbilical arterial catheters, renal impairment from hypoperfusion, early programming) raises the possibility that systemic blood pressure (BP) may be raised in survivors of extremely preterm birth. Neither POPS nor EPICure bears this out. EPICure children assessed at 6 years had BP values within the normal range67 and, although POPS survivors at 19 years had higher blood pressure than the normal population this was not explained by gestation, birth weight or weight for gestational age but was associated with an increased BMI and weight at the time of assessment.71
Bronchopulmonary dysplasia
Findings from the European studies replicate those from other literature: as mortality decreases (an effect seen mainly at the lower end of the gestational spectrum) the incidence of bronchopulmonary dysplasia (BPD) (however it is defined) increases.2 15 72 Survivors with BPD continue to have an excess of respiratory symptoms at school age and more abnormalities on pulmonary function testing than preterm survivors without BPD.73 74 In the MOSAIC study the incidence of BPD varied twofold between European regions and the differences remained when adjustment was made for gestational age and gender.72 There was a relationship in the most immature babies between survival rates and the incidence of BPD (supporting the notion that the BPD ‘cost’ increases with decreasing gestation) but there was no consistent relationship with respiratory support practice (L Gortner, personal communication).
Discussion
The ability to directly compare outcomes between studies is limited by the different gestational age ranges in the cohorts, the lack of gestation-specific follow-up information in many of the reports and the differing ages at which the populations were studied. In general, studies of the most immature babies (<27 weeks) report higher disability rates, which is not a surprise. The four studies that have reported findings from their cohorts when they have reached middle school age and beyond underline the importance of continuing follow-up long enough to uncover the high rate of more subtle disabilities that impact on learning and social integration. Even comparisons between identical gestation groups over the same timeframe cannot avoid biases, which lead to difficulties in interpretation. In the Trent/Denmark and Trent/France studies coding differences, ascertainment and attitudes to treatment were all cited but the most fundamental difference, which was hypothesised to account for the differences in outcome seen, was more qualitative and described as ‘reproductive health’. The MOSAIC study further highlights the difficulties of comparing outcomes across populations that differ with respect to service infrastructure and healthcare policy and underlines the importance of taking these into account. Absolute differences remain important of course; in the MOSAIC study it was estimated that there would be a twofold variation in CP rate in the preterm populations studied, something that we will only know for certain when later outcome data become available. And if some of the differences are at least partly attributable to models of care they may be relatively easily amenable to change. The continually improving survival of all but the most immature babies and the relatively constant rate of major disability72 mean that absolute numbers of children with significant disability attributable to very premature birth are increasing. The ‘softer’ morbidities that become increasingly apparent as the children grow older, particularly behavioural disorders and the need for special educational input at school, are a concern and are no less important in terms of their social and health economic impact. However, the absolute number of children who survive without major disability is also increasing76 and children with the commoner (and less severe) forms of CP rate their quality of life highly77 so this side of the balance needs to be carefully weighed when policy making decisions are being made.
References
Footnotes
-
Provenance and peer review Commissioned; externally peer reviewed.