Since bronchopulmonary dysplasia (BPD) was first described in the late 1960s following the introduction of mechanical ventilation, there has been a steady increase in the prevalence of the condition with an apparent reduction in severity.1 2 Initially, BPD was associated with barotrauma and oxygen toxicity, but its prevalence appears to have risen with the increasing survival of extremely preterm infants, in whom the incidence of BPD is highest. These infants may develop BPD by different routes, with significant contributions from inflammatory and developmental lung changes leading to alveolar simplification.3 Despite increasing survival at extremely low gestational ages, however, there is little gestational age-based information regarding outcome in this population. Many studies report data derived before the now widespread use of antenatal steroids and surfactant replacement therapy, which may well have changed the nature of the disease. There is debate as to whether BPD is associated with worse developmental outcomes over and above the disability associated with preterm birth.4 5

We conducted a prospective study of all babies born between 20 weeks and 25⁺⁶ weeks gestational age in the United Kingdom and Ireland over a 10-month period beginning in March 1995 (the EPICure Study).6^–8 The duration of postnatal steroid therapy (which we believe may be a surrogate marker for the most severe respiratory disease in this population) was independently associated with poor growth9 and motor disability,7 and the use of supplemental oxygen at 36 weeks postmenstrual age with Bayley psychomotor and mental development indexes7 at 30 months. BPD and its treatment are therefore key variables in relation to neonatal illness and outcome in this population.

In the early years of life, continuing respiratory morbidity from BPD can considerably increase the need for continuous medication and hospital admission.10 In this paper we describe longitudinal respiratory health in our population of extremely preterm children from birth to 6 years. We hypothesised that, despite improving health, respiratory outcomes would still be related to neonatal lung disease and show significant differences from term-born comparison children at 6 years, even after allowing for known environmental risk factors.

METHODS

Population

The method of data collection has been described previously.6 8 11 Information for all births between 20 weeks and 25 completed weeks of gestation in the United Kingdom and Ireland was collected between March and December 2005. Figure 1 shows how the survivors seen for this report were derived. Similar proportions of children discharged from hospital and seen at 30 months or 6 years of age were receiving supplemental oxygen at 36 weeks postmenstrual age, the pragmatic definition of BPD adopted for this paper (fig 1). Medical and respiratory histories and socio-demographic details were obtained through a semi-structured interview and questionnaire.

Assessment at 30 months

Each child had a paediatric assessment in a hospital outpatient clinic (n = 235) or at home (n = 48).8

Assessment at 6 years

Each child had a physical and neuropsychological assessment either at school (n = 215), in a hospital outpatient clinic (n = 2) or in the family home (n = 24) at 62–87 months age.11 12 Age and sex matched classmates were recruited as a comparator group as previously described.11 Peak expiratory flow (PEF) was measured using a Wright flow meter (Clement Clarke, Harlow, Essex, UK) using a standardised technique. The best of three attempts was accepted following a period of instruction and practice. At this age, history was based on core questions from the International Study of Asthma and Allergies in Childhood (ISAAC).13 Questionnaires were returned for 92% of children evaluated at 6 years.

The protocol for the 6-year assessment was approved by the Trent Multicentre Research Ethics Committee and signed parental consent was obtained for each follow-up evaluation.

Statistics

Data were extracted from the perinatal, 30-month and 6-year databases. Information on oxygen use over the first 2 years was available for 283 children assessed. Longitudinal comparisons are reported using only those 236 children (77% of the cohort) whom we had evaluated at all three time points. The proportions at each age with clinical problems do not differ from the whole populations seen for follow-up at either age. Because the outcomes are not rare, the relative risks (RR) and odds ratios (OR) are not similar. Relative risks have been used for comparisons between extremely preterm and comparison children, and odds ratios for the analysis of risk factors in the extremely preterm children. Adjustments for gestation were made by including actual gestational age in the regression or logistic regression analyses. Data were analysed using SPSS for Windows (v 14.0) and Stata 9. The associations with gestational age were tested for non-linearity using fractional polynomials in Stata 9.

RESULTS

Supplemental oxygen

Of the 283 extremely preterm survivors evaluated at 30 months, 210 (74%) and 146 (52%) were receiving supplemental oxygen at 36 weeks postmenstrual age (PMA) and at their estimated date of delivery (40 weeks PMA), respectively; 101 children (36%) went home with supplemental oxygen. At 36 weeks PMA, the proportion of children receiving supplemental oxygen, that is with BPD, was not significantly different across all gestational age groupings. However, at 40 weeks PMA, less mature children were more likely to be receiving oxygen supplementation (⩽23 weeks: 78% (23/31); 24 weeks: 61% (61/98); 25 weeks: 44% (82/186); χ²_trend: p<0.001). This trend was again reflected in those children discharged home with additional oxygen (χ²_trend: p<0.001). After adjustment for gestational age, boys were marginally more likely to need oxygen supplementation at 36 weeks PMA (OR 1.62; 95% CI 0.96 to 2.7; p = 0.068) or at 40 weeks PMA (OR 1.57; 95% CI 0.99 to 2.50; p = 0.057), while singletons were less likely to require oxygen at 40 weeks PMA compared to multiple births (OR 0.48; 95% CI 0.28 to 0.83).

Figure 2A presents the proportions of the 283 assessed children receiving oxygen after 40 weeks PMA, with a median further duration of oxygen therapy of 2.5 months (75th percentile: 8.5 months). The duration of supplemental oxygen therapy was similar for the least mature (log-rank test for trend: p = 0.27; fig 2B) despite differences in the proportions receiving oxygen at 40 weeks PMA.

Severe morbidity

At 30 months and again at 6 years, two children were receiving supplemental oxygen for at least part of the day; one of these children also had severe developmental impairment. No child required artificial respiratory support. Three children had required tracheostomy but only one tracheostomy remained in situ (at 30 months and at 6 years).

Longitudinal respiratory morbidity

For these analyses, the denominator population is the 236 children seen at 30 months and 6 years. There were no significant differences in the rates of supplemental oxygen use at 36 or 40 weeks PMA among those seen at both follow-up evaluations, compared with the 72 children who dropped out (RR at 36 weeks: 1.07; RR at 40 weeks: 1.03) or compared with the 47 who dropped out between 30 months and 6 years (RR 1.03 and 1.01, respectively). Slightly fewer children seen at both ages received home oxygen (RR 0.78; 95% CI 0.54 to 1.13). Children who dropped out were more likely to be males, have more neurological impairment and have younger mothers.11 These biases may underestimate morbidity in the extremely preterm population.

Symptoms

By 30 months, 197 extremely preterm children (84%) had received some treatment for “chest problems” and for 10 children immunisations were delayed because of illness. Children meeting our criteria for BPD received more treatments than those who did not, but had similar atopy histories, environmental triggers and reported symptoms (table 1).

At 6 years (table 2), extremely preterm children with BPD had wheeze and associated symptoms more frequently than those without BPD and were 2.5 times more likely to be under the surveillance of a paediatrician. We compared the outcomes of extremely preterm children without BPD with comparison children (table 2). Although the extremely preterm children are more likely to have symptoms or to have received treatments, fewer extremely preterm children were currently taking inhaled medicines. They were, however, twice as likely to have been diagnosed with “asthma”.

Although wheezing was more frequent with shorter gestation (χ²_trend: p = 0.004), there were no significant effects of gestational age on bronchodilator and steroid use, although after the first year, less medication use was reported for children born at 25 weeks than at <25 weeks. There were no significant differences between the frequency of wheeze and medication use for boys and girls. Most children receiving medication at 6 years had received it at 30 months, but 12 children had been commenced on bronchodilators and 10 children on steroids.

Hospital admissions

The majority of hospital admissions over the first 2 years were for respiratory symptoms. A total of 150 (64%) extremely preterm children were admitted for any cause up to the 30-month evaluation; of those, 33 (42%) were admitted once or twice and 52 (22%) three or more times with a respiratory problem. Lower respiratory tract infection, bronchiolitis and dyspnoea/wheeze were the most common reasons for admission. Children with BPD were no more likely to require multiple admissions (⩾3 episodes) in the first 30 months compared to those without, in contrast with children discharged with home oxygen who were at increased risk (OR 2.70; 95% CI 1.39 to 5.27).

At the 6-year assessment, of 45 extremely preterm admissions to hospital over the preceding 12 months, only five were for respiratory indications (the majority were for surgical reasons). Among comparison children there were eight admissions, with none for respiratory illness.

Clinical examination

At 6 years 106 extremely preterm children (44%) were considered to have abnormal examination findings: 65 had a deformity, including 45 with Harrison’s sulcus, 13 had signs of hyperinflation, two had tracheal deviation and the remainder acute signs. Among the comparison children, five (3%) had chest deformity (two with Harrison’s sulcus). The mean resting respiratory rate was 22.9/min (SD 5.0) in extremely preterm children and 20.9/min (SD 4.8) in comparison children (p<0.001). These findings were similar to those observed at 30 months (not shown).

Peak flow measurements were successful in 200/236 of the extremely preterm children and 154/160 comparison children; the lower frequency of success in extremely preterm children was primarily because of the inability of children with serious disability to perform the test. PEFs for these extremely preterm children, after adjusting for height differences using linear regression, were on average 39 l/min lower (95% CI 30 to 47; p<0.001) than those for comparison children. extremely preterm children without BPD have significantly reduced adjusted PEF (−26 l/min; 95% CI −39 to −13; p<0.001); adjusted PEF was lower still in those with BPD but subsequently discharged home without oxygen (BPD vs no BPD: −13 l/min; 95% CI −26 to 0; p = 0.05) and lower still for children discharged home with oxygen (vs no BPD: −25 l/min; 95% CI −39 to −11; p = 0.001). Adjusted PEF was higher in comparison boys compared to girls (18 l/min; 95% CI 6 to 30; p = 0.005) but not significantly so in extremely preterm boys (5 l/min; 95% CI −6 to 15; p = 0.41). Considering the whole 6-year cohort (n = 241) did not substantially alter these findings (not shown).

Environmental and perinatal factors

A family history of atopy (wheeze, diagnosis of asthma and/or hay fever) was not significantly associated with BPD (OR 1.2; 95% CI 0.67 to 2.2), duration of supplemental oxygen therapy (mean difference: 0.91 months; 95% CI −1.70 to 3.52), or with bronchodilator or inhaled steroid use in the second or sixth years, respectively. Reports of mould in the house at 30 months were uncommon (<4%) but still marginally associated with wheezing, night cough and wheeze on exercise, with odds ratios of 4.15, 2.99 and 5.14, respectively. There was no evidence to suggest the influence of any environmental factors on the differences between the preterm children and their peers at 6 years (tables 1 and 2).

Night cough and exercise induced wheeze were evaluated alongside “wheeze” as parents reported these symptoms independent of wheezing. Univariate associations between risk factors, symptoms and peak flows at 6 years are shown in table 3. In the multivariate analysis (table 4), there was a significant negative interaction between the effect of maternal smoking in pregnancy and the presence of BPD, such that each of the three groups (maternal smoking/no oxygen at 40 weeks, maternal smoking/oxygen at 40 weeks and non-smoking mother/oxygen at 40 weeks) all have very similar increased odds ratios compared with those with non-smoking mothers/no oxygen at 40 weeks and thus similar long term effects in this extremely preterm population.

In spite of the association of postnatal steroid use and duration of postnatal oxygen, there was no relationship between wheeze and postnatal steroid use. Birthweight for gestation, singleton status and chorioamnionitis became significant after adjusting for gestation; these were not included in the final models because of concern of over-fitting, as all the coefficients became more extreme.

Univariate associations with PEF are shown in table 3. Independently, only supplemental oxygen at 36 weeks reached conventional significance (−14 l/min; 95% CI −27 to 1; p = 0.036). Home oxygen (p = 0.074), antenatal steroids (p = 0.075) and male sex (p = 0.070) all had similar sized independent effects after adjustment for oxygen at 36 weeks and remained close to significance. Substantially similar findings were obtained using PEF after adjustment for height (not shown).

DISCUSSION

In this geographically based cohort of children born at 25 weeks of gestation or less, a continuum of respiratory morbidity is demonstrated from the perinatal period through to 6 years where children have increased risk of poor respiratory health and reduced PEF. One of the most important determinants of later respiratory problems remains the severity of neonatal lung disease. Within the extremely preterm cohort there are strong effects from smoking in pregnancy, but in multivariate analysis these effects are very similar to the effect on respiratory outcome accruing from BPD. Indeed, the interaction could suggest that there may be a ceiling to symptom frequency reached by any of the three combinations. Smoking in pregnancy may itself be associated with very preterm birth, but in a study such as this it is difficult to tease out any modulating effect that smoking may have on the severity of neonatal lung disease. Having a smoker in the home in childhood is strongly associated with respiratory health in the child.

Compared to other preterm populations, this is undoubtedly one at very high risk: three quarters have moderate to severe BPD and almost all of the remainder have mild BPD by the most recent definition.14 The least mature are at the greatest risk of neonatal lung disease and the highest risk of wheeze and poor respiratory health as they grow up. Although extremely preterm children without BPD, as defined, have reported outcomes which are not statistically different from the comparison group, for most symptom measures they have higher odds ratios, while their adjusted PEF is still significantly lower, indicating clinically significant deficits. Several children were unable to carry out PEF testing because of other serious impairments; technique may have been less good in the extremely preterm children, who have an excess of subtle motor and executive function deficits12 which could impair their technique.

The most dramatic change in the population between 30 months and 6 years is in the reduction in hospitalisation and medication use. The need for re-hospitalisation for respiratory illness is well described,15^–17 and may be exacerbated by RSV infection18 which is itself associated with wheezing into middle childhood. Nonetheless, the prevalence of inhaler use at 6 years (23%–26% in the past 12 months) remains over twice that of the comparison group. Other longitudinal studies have shown a similar increased prevalence of symptoms in middle childhood19 but with a reduction in symptoms as adolescents.20

Other studies that have considered outcome following BPD or respiratory outcomes for cohorts of very low birthweight infants have relatively few babies born at such low gestational ages and very few represent cohorts born after the common use of antenatal steroids and surfactant. Previous studies with births in the 1970s and 1980s have really studied children with more classical BPD resulting from primarily ventilator induced lung injury,20^–24 whereas BPD among current extremely preterm survivors is characterised by an arrest in alveolar development.2 3 As survival has increased, it has become difficult to study the pathology of this “new BPD”, but studies in animals suggest that these changes lead to long term problems.25 Thus, this cohort is distinguished by the severity of the neonatal illness as judged by the ongoing use of supplemental oxygen, which itself may worsen early outcomes over the first year.26

Although the label “asthma” had been applied to twice as many extremely preterm children as to their classmates, it is unclear to what extent this has a bearing on the symptoms reported. Children growing up after neonatal lung disease have a continuum of symptoms and the extent to which reversible airway components contribute is unclear. In terms of environmental triggers, there were few differences in exposure to smoke, mould, damp, pets and heating type at home (latter two not shown) or history of atopy, which does suggest continuing disease and seems to reject the hypothesis that atopy predisposes to BPD.27 We also have noted an association between clinical chorioamnionitis and exercise induced wheeze at 6 years which may represent continuing morbidity from the prenatal environment.14

In common with previous studies, the prevalence of symptoms and need for treatment/hospitalisation reduced significantly between 30 months and 6 years. Despite their improving respiratory health, chest deformity was observed in 44% of the index children, indicating chronic respiratory problems, compared to only 3% of the comparison children. These results parallel those of a regional study of children born at <1000 g or <28 weeks at 8–9 years of age with similar rates of “asthma” (25% vs 14% in controls).28 In this study the children were older and formal respiratory function tests could be undertaken. A 14% reduction in forced expiratory volume and similar changes in flow measures were observed with larger changes in those who had had BPD, again similar to our observations with PEF.

This is a unique study population as it is based on an entire cohort from the UK and Ireland, but this strength has restricted our choice of outcomes. We were unable to undertake more detailed respiratory function testing because of the difficulties in reliably performing spirometry in a variety of settings at such a young age.

We have described persisting high respiratory morbidity over the first 6 years in a geographical cohort of babies born before 26 weeks gestation in 1995. The poor performance on peak flow, high prevalence of chest deformity and continuing symptoms of wheeze make it imperative to continue to study the longer term effects of neonatal lung disease through adolescence into adult life to understand the implications that this has for adult respiratory health and for the wider participation of these children in sport and social activities.