Poor longitudinal growth in childhood is common in children born at lower gestational ages. We recently reported growth to 30 months of age for a national cohort of extremely preterm (EP) children, identifying poor longitudinal growth, poor weight gain and smaller head size compared with population norms.1 Poor postnatal growth was associated with a range of clinical factors, including weight for gestation at birth, long courses of postnatal corticosteroids and later feeding difficulties. Head growth in particular was associated with later serious disability, but overall, it was not related to developmental test scores at 30 months of age corrected for prematurity in those able to do them.

Growth after birth2 and over the first year3 4 may affect later cardiovascular risk, but this risk needs to be balanced for all babies against the need for reasonable postnatal growth, particularly of head size to achieve adequate neurological5 or cognitive outcomes.6 7 Although developmental outcome has been shown to be related to postnatal growth, most evidence is for children with low birth weight for gestation.8 Many children who survive birth at borderline viability have significant periods of postnatal growth failure, probably owing to inadequate early nutritional intake.9 Thus the long-term consequences of early growth failure in EP children remain to be identified.

Our earlier study1 is part of a continuing longitudinal evaluation of children born in the United Kingdom and Republic of Ireland at 25 completed weeks of gestation or less (the EPICure study). Over 10 months in 1995 we identified 1289 live births between 20 and 25 weeks of gestation; 308 survived beyond the first year after birth. We have reported a range of outcomes for this group at discharge from hospital10 and at 30 months age corrected for prematurity.1 11^–13 Recently we re-evaluated this cohort at 6 years of age and demonstrated persisting high levels of disability and emerging cognitive impairments.14 We also measured growth and blood pressure (BP) with the objectives of comparing their outcomes with a comparison group of contemporary term-born children, of investigating their relationships with earlier predictors, and reporting catch-up growth.

METHODS

The derivation and characteristics of this EP cohort have been described,1 10 12 as have the details of the 6-year evaluation.14 The population represents all surviving children born at 25 weeks 6 days gestational age or less from March through December 1995. Of the 308 children known to be alive at 30 months, the parents of 241 children consented to the study.

Two hundred and four children were in mainstream education. For each of these, we sought an age- and sex-matched classmate for comparison.14 We were able to assess 160 term-born children. All children were assessed by a paediatrician and a psychologist after training in the techniques used for the study.

We carried out physical and neuropsychological evaluations, supplemented by a clinical history obtained by parental questionnaire, as previously described.14 Weight was measured on identical weighing scales (Salter Housewares Ltd, UK) or on clinic equipment. Height was measured using a standard stadiometer (Child Growth Foundation). Measurements of maximum occipito-frontal head circumference (OFC) and mid-arm circumference were made with a LASSO-O tape respectively. BP was measured at the end of each assessment in the right arm of a sitting child, using a mercury sphygmomanometer taking the first and fifth Korotkoff sounds as systolic and diastolic pressures, respectively.

Standard deviation scores (“z” scores) for weight, height, body mass index (BMI) and head circumference at 6 years were calculated from UK population norms using age since birth.15 Comparisons were made with data already collected at birth and expected date of delivery (weight and OFC), and at 30 months (weight, height, OFC and BMI).

This study was approved by the Trent Multicentre Research Ethics Committee. Informed signed consent for this assessment was obtained from all the families.

Statistical analysis

Data were double entered and outliers were verified before combination with the main study data set for analysis with SPSS for Windows software (release 14.0, SPSS, Chicago, USA). Univariate and multivariate regression analyses were performed using STATA, version 9.0 (Stata Corp, College Station, Texas, USA). A forward stepwise procedure was used to establish independent factors associated with growth, as previously.11 For regression analyses, to avoid extreme outliers, SD scores at 6 years were truncated at the following values: weight +2.5 (n = 1), head circumference −4.5 (n = 5), BMI −4 (n = 4), and for change in SD scores from 30 months, head circumference −2.5 (n = 3), +2.5 (n = 2) and height −2.5 (n = 2), +2 (n = 1).

In analysing associations with BP, BMI for height (the residual of linear regression) rather than BMI was used. For investigation of earlier predictors of later BP in the EP cohort, the residuals of BP on height and BMI rather than actual BP values have been used to avoid confounding by the effects of perinatal variables on growth. No adjustments have been made for multiple testing.

RESULTS

Two hundred and forty-one children (78% of 308 survivors) participated at a median age of 76 months (range 62–87 months; 51% boys). They were representative of the whole population of survivors over a range of perinatal variables and of 47 children not assessed at 6 years but evaluated at 2½ years.14 The comparison group had a median age of 74 months (range 61–86 months; 44% boys).

Of the EP children assessed at 6 years there was information from birth for weight (n = 241), OFC (n = 166); from expected date of delivery for weight (n = 225), height (n = 66), OFC (n = 214), and from the 30-month assessment for weight (n = 226), height (n = 216), OFC (n = 231) and BMI (n = 212). Of the 160 comparison children, measurements were obtained for weight (n = 159), height (n = 159) and OFC (n = 160).

The mean birth weight (relative to UK standard data) of the EP children who were assessed at 6 years was for the gestations ⩽23, 24 and 25 weeks, +0.70 SD, +0.37 SD and +0.07 SD, respectively.

Figure 1 shows raw data for weight and OFC of the EP and comparison children, relative to the 0.4th, 50th and 99.6th centiles for girls and boys. At 6 years extremely preterm children were still smaller than their peers in each of the four main growth measures: mean difference in SD scores for weight was 1.25 SD (95% CI 1.0 to 1.50), for height 0.97 SD (0.75 to 1.19), for head circumference 1.30 SD (1.07 to 1.53) and for BMI 0.95 SD (0.70 to 1.20) (all p<0.001; table 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Measured growth parameters at 6 years for EPICure index children and for the comparison group: Weight (upper graphs) and head circumference (lower graphs) for boys (left graphs) and girls (right graphs). Reference lines are taken from growth standards.15

These differences remained after elimination of children with cerebral palsy (weight: 1.29 SD, height: 0.87 SD, head circumference 1.16 SD; all p<0.001). Further excluding children with other severe disabilities at 6 years and adjusting for height, the comparison children’s head circumference remained 0.69 SD (95% CI 0.46 to 0.92) greater than that of the remaining EP children, whose head circumference was itself 0.33 SD (95% CI −0.03 to 7.0; p = 0.07) greater than those with cerebral palsy or severe disability.

SD scores from the 30-month assessment were recalculated using chronological age so that intervening growth could be evaluated. The change in weight standard deviation score from 30 months to 6 years was +0.37 SD (95% CI 0.23 to 0.50) (p<0.001), for height +0.42 SD (0.31 to 0.52) (p<0.001) and for OFC +0.13 SD (0.00 to 0.25) (p = 0.05; table 2). There was no change in BMI over this period (95% CI −0.18 to 0.18). Thus, some catch-up was observed in weight and height but less in head growth. Figure 2 shows the mean changes in SD scores from birth for weight and OFC for those children for whom we had measures at each age relative to chronological age; correcting for prematurity would raise these values slightly closer to the mean.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Change in weight and head circumference standard deviation scores between birth, expected date of delivery (EDD), 30 months and 6 years of age (uncorrected for prematurity). Data are shown only for children who had measures at each age.

The relationship between perinatal variables and growth parameters was examined in sequential regression analysis, as described previously11 The relationships between outcomes and sequentially occurring variables were similar to the results for the 30-month analysis (table 3).10 There were some differences in growth measures between ethnic groups: at 6 years Afro-Caribbean children were taller, the white children lighter and the children with mother’s from the Indian subcontinent no longer had a significantly smaller head circumferences.

Birth weight for gestational age remained strongly associated with all growth outcomes at 6 years except BMI, where the independent effect was 0.11 (−0.06 to 0.28; p = 0.2). The effects per SD of birth weight after adjustment for factors arising before leaving hospital are similar for head circumference 0.32 SD at 30 months and 0.31 SD at 6 years, while for weight (0.33 SD, 0.27 SD) and height (0.32 SD, 0.20 SD) the respective associations are smaller at 6 years. Gestational age was related only to head circumference (0.24 SD per gestational week at 6 years; 0.29 SD at 30 months). Problems while in hospital, particularly necrotising enterocolitis, severe changes on first chest radiograph and longer courses of steroids were related to the same growth outcomes as at 30 months, except that steroids were no longer related to weight at 6 years. Head growth is related to height (r = 0.32 EP, r = 0.42 comparison) so the independent effects for head circumference adjusted for height are also shown in table 3, to indicate which factors specifically affect relative head growth rather than those mediated through association with linear growth. Fewer factors are associated with relatively poor growth velocity between 30 months and 6 years (table 3).

As head growth is often compromised in those with severe brain injury, the analysis was repeated for children without cerebral palsy or severe disability at 6 years. While the effects of birth weight for gestation, feeding difficulties at 2 years, necrotising enterocolitis and ethnicity remained very similar, the effects of gestational age and postnatal steroids were weaker and no longer statistically significant.

From EP children we recorded BP for 208 children and systolic pressures only in a further six children. The blood pressures from 158 comparison children became our reference data. Mean (SD) systolic and diastolic pressures for EP children were 89.3 (11.8) mm Hg and 57.6 (9.5) mm Hg, respectively, compared with 91.7 (9.9) mm Hg and 60.1 (9.4) mm Hg for the comparison group. These differences were significant (systolic: −2.3 mm Hg (−4.62 to −0.07) p = 0.043; diastolic: −2.4 mm Hg (−4.4 to −0.47), p = 0.015). Mean differences diminished and became non-significant when adjusted for height and BMI for height (table 4). Among the EP children diastolic BP adjusted for height and weight was slightly higher in children with cerebral palsy but not significantly so (systolic difference: 0 mm Hg (−4.0 to 4.0), p = 1.0; diastolic difference: 2.7 mm Hg (−0.7 to 6.0), p = 0.11). EP children had greater variability in systolic BP than the comparison children, both unadjusted (p = 0.02) and after adjustment for height and BMI (p = 0.004, Bartlett’s test for equal variances). For diastolic BP the relationship with common variables (BMI, height, mid upper arm circumference, age, sex and ethnicity) was similar in the two groups, but there was a suggestion that the relationship of height (shorter in EP children) (adjusted for BMI) and systolic BP differed between EP and comparison children (coefficients 0.78 vs 0.45 mm Hg per cm).

Within EP children, multiple regression was undertaken to assess whether there were significant independent relationships of adjusted blood pressure (for height and BMI) with perinatal variables. The four children who had had a stoma or intestinal resection for necrotising enterocolitis had markedly higher BP than the others (+10 mm Hg for diastolic and +16 mm Hg for systolic pressure); they are excluded from further analyses of perinatal associates with BP. Adjusted systolic BP was independently associated with maternal smoking in pregnancy (−4.1 mm Hg (−7.1 to −1.1)), birth <24 weeks (10.8 mm Hg (5.6 to 16)) and marginally with an antenatal cervical suture (4.6 mm Hg (−0.4 to 9.6)). Adjusted diastolic BP was independently associated with maternal smoking in pregnancy (−3.0 mm Hg (−5.5 to −0.5)) and use of steroids postnatally (3.2 mm Hg (0.5 to 5.8)) and cervical suture (6.0 (1.8 to 10.0)). Growth velocity during the three time epochs (birth to full term; to 30 months; to 6 years) was not associated with adjusted blood pressure values at 6 years.

DISCUSSION

These data demonstrate that being born extremely preterm is associated with poor longitudinal somatic growth, with little catch-up over the preschool years. By 6 years the EP children remain significantly shorter, lighter and have a smaller head circumferences than reference data or, a group of classmates. Between 2½ and 6 years there has been some catch-up in height (0.42S D) and weight (0.37 SD) but less in head circumference (0.13 SD). These findings are mirrored in the subset of children for whom we have data at each age (fig 2).

By 30 months head size catch-up was significantly worse for babies born at 23 weeks of gestation or less than at 24 or 25 weeks1 which persisted at 6 years—mean head circumference at 6 years rising by 0.21 SD for each gestational week. It might be expected that head growth would be proportionate to height and indeed many factors are associated with both, however, gestational age and birth weight for gestational age had significant independent additional effects on head circumference. The effect of gestation and postnatal steroid use was weaker in children without cerebral palsy or severe disability, reflecting the increased risk of these outcomes with lower gestational age or the need for longer periods of postnatal steroid use, or both. We identified few associates of poor growth from 30 months to 6 years; the group of children who needed food supplementation at 30 months, presumably because of earlier growth concerns, are falling further behind at 6 years.

Our data appear at variance with a smaller study of 52 children born in 1988–91 at <29 weeks of gestation, where catch-up to the mean was demonstrated in sequential measurements of weight and height over the first 7 years.16 Most of these children were more mature at birth than our EP children, but those children <27 weeks of gestation were not significantly different from those born at 27–28 weeks. In contrast other population studies of children <32 weeks of gestation have demonstrated similar poor growth over the first 5 years of life.17

Most studies of growth in middle childhood have described children born at very or extremely low birth weight (<1500 g (VLBW) and <1000 g (ELBW), respectively). It is difficult to make direct comparison with such studies as they include an excess of more mature but growth-restricted infants for whom later growth performance may be different.17 18 Indeed, in our study birth weight for gestational age was highly associated with growth parameters. Almost all studies have identified that VLBW or preterm children are relatively slim and this was true of our population. Furthermore, mean BMI SD score has remained constant between 2½ and 6 years. A low fat mass may place preterm children at decreased risk for adult obesity and for the associated increased risk of cardiovascular morbidity.19

Studies of ELBW children to adolescence19 20 have not demonstrated complete catch-up in height, although one has shown catch-up for head growth for 12–16 year olds,10 so it is plausible that further follow-up may show improvements in head growth.

No studies have evaluated BP in middle childhood in EP populations, in contrast to VLBW children in adolescence or adult life.20^–25 Generally, studies at older ages show a significant increment in systolic and diastolic pressures but, in contrast, Saigal and colleagues reported no differences in BP in their adolescent ELBW cohort with mean gestational age of 27 weeks.26 It should be noted that differences in BP in middle childhood are often small and even whole-population studies such as ours, where BP is measured as part of a global assessment may be too small to detect such differences.

Some studies suggest that it is only preterm, small for gestational age, infants who are at risk of increased BP.27 28 Although our EP children were heavier than average at birth, they demonstrate serious extrauterine growth failure. Many preterm growth-restricted infants in previous studies would have had poor growth at the same maturational age

The extent to which BP tracks from middle childhood to adult life is poorly understood. Studies have reported conflicting results, although none have studied such an immature population as ours. The effects of low weight at birth and high weight velocity after birth may be additive and weight gain from birth may be the stronger predictor.29 30 Indeed lower birth weight and greater weight gain between 1 and 5 years of age have been associated with higher systolic BP in young adult life,31 and endothelial function in preterm adolescents may be enhanced following slower growth in the first 2 weeks after delivery.32 This suggests that even short periods of early catch-up growth are bad for later cardiovascular health. Rapid weight gain in childhood has been associated with a further increase in risk, but only in boys who were light at birth.33 Further analysis suggested that this “adiposity rebound” occurs at around 2 years of age.34 Given the poor immediate postnatal growth of EP children,10 the lack of significant catch-up over the first 2½ years,1 the lower BPs seen in this study, and the finding of similar BPs in adult life for ELBW children,26 these EP children may not be at higher risk of later cardiovascular disease. Current EP children may have better growth outcomes because of enhanced attention to nutritional care.

We also demonstrated a negative association between smoking during pregnancy and high BP, consistent with a study that showed smoking reduced BP at age 7.5–8 years if gestation was <33 weeks, but increased BP if gestation was >33 weeks.35 Women who stop smoking in pregnancy have heavier babies than those who do not36 and term babies who have growth retardation in the first trimester are heavier than those who do not.37 It might be speculated that the babies whose mother’s smoked were at a potential growth advantage once born, an effect negated by EP birth.

This cohort of extremely preterm children remain much shorter, lighter and slimmer than their classroom peers, and head growth has not shown significant catch-up over the past 3–4 years. Despite these growth changes, we did not observe any significant differences in BP after correction for height and BMI. BP must be monitored in further follow-up studies, but currently our findings seem to indicate little risk for increased BP in adult life provided that the adults stay relatively slim.