Of 6,761 eligible children, 26 lacked data on learning difficulties and 44 children lacked information regarding deprivation (Appendix 1). This left 6,691 in the cohort for the primary analysis with a median age of 5 years at time of survey (IQR 5, see appendix 2). 19 lacked data on behavioural problems and 3,335 missing data on SEN and consequently the number of children in each analysis varied with outcome assessed. Infants in the primary cohort for analysis were more likely to be male than those not analysed due to missing data (p = 0.02) but were otherwise similar in terms of demographics (appendix 3).
Children born preterm were more likely to be male (p = 0.05), more likely to be born to a mother who smoked (p = 0.01), from multiple pregnancies (p < 0.001) and be born by unplanned LSCS (p < 0.001); they had lower birthweights (p < 0.001) and were less likely to have started breastfeeding at birth (p < 0.001). Children in the more deprived deciles had younger mothers (p < 0.001), were of a younger age at time of survey and were more likely to be of an ethnic minority group (p < 0.001). They were more likely to be born to mothers who smoked (p < 0.001), from a singleton pregnancy (p = 0.01) and by normal vaginal delivery (p < 0.001). More deprived children had lower birthweights (p < 0.0001) and were less likely to have started breastfeeding at birth (p < 0.001) (Table 1).
Table 1
Characteristics of study group split by gestational age and deprivation measure
Measure | Number with data | Preterm (n = 4023) | Term (n = 2668) | P value | 1st and 2nd WIMD decile (Most deprived) (n = 1158) | 3rd and 4th (n = 1322) | 5th and 6th (n = 1397) | 7th and 8th (n = 1348) | 9th and 10th (Least Deprived) (n = 1466) | P value |
Demographic | | | | | | | | | | |
Maternal age (years) | 5956 | 30.2 (6.0) | 30.4 (5.6) | 0.3* | 28.105 (6.1) | 29.040(6.1) | 30.503(5.7) | 31.223(5.3) | 31.979(5.2) | < 0.001† |
Sex (Male) | 6691 | 2223 (55.3%) | 1408 (52.8%) | 0.05 | 626 (54.1%) | 719 (54.4%) | 766 (54.8%) | 749 (55.6%) | 771 (52.6%) | 0.6 |
Ethnic minority groups | 5974 | 214 (6.0%) | 148 (6.1%) | 0.8 | 94 (8.9%) | 81 (6.7%) | 47 (3.8%) | 63 (5.3%) | 77 (5.9%) | < 0.001 |
Age at survey (y)- median (IQR) | 6691 (100%) | 5 (5) | 5 (5) | 0.4** | 3 (5) | 5 (5) | 4(5) | 4 (5) | 4 (5) | 0.008†† |
Pregnancy | | | | | | | | | | |
Smoking in pregnancy | 6409 | 541 (14.0%) | 299 (11.8%) | 0.03 | 270 (24.4%) | 223 (17.6%) | 159 (11.8%) | 104 (8.1%) | 84 (6.0%) | < 0.001 |
Multiple pregnancy | 6691 | 894 (22.2%) | 41 (1.5%) | < 0.001 | 133 (11.5%) | 194 (14.7%) | 179 (12.8%) | 197 (14.6%) | 232 (15.8%) | 0.01 |
Mode of Delivery | 5956 | | | < 0.001 | | | | | | < 0.001 |
Normal Vaginal Delivery | | 1494 (41.7%) | 1446 (61.0%) | | 564 (53.7%) | 589 (49.7%) | 605 (49.2%) | 559 (47.2%) | 623 (47.6%) | |
Breech | | 83 (2.3%) | 10 (0.42%) | | 21 (2.0%) | 19 (1.6%) | 19 (1.6%) | 17 (1.4%) | 17 (1.3%) | |
Instrumental | | 270 (7.5%) | 279 (11.8%) | | 64 (6.1%) | 107 (9.0%) | 115 (9.4%) | 114 (9.6%) | 149 (11.4%) | |
Elective CS | | 553 (15.4%) | 307 (13.0%) | | 133 (12.7%) | 193 (16.3%) | 172 (14.0%) | 154 (13.0%) | 208 (15.9%) | |
Unplanned CS | | 1185 (33.0%) | 329 (13.9%) | | 268 (25.5%) | 277 (23.4%) | 318 (25.9%) | 338 (28.6%) | 313 (23.9%) | |
Infant and postpartum | | | | | | | | | | |
Birthweight (kg) | 6691 | 2.216 (0.63) | 3.439 (0.48) | < 0.0001* | 2.616 (0.8238) | 2.668 (0.8234) | 2.713 (0.8049) | 2.762 (0.8439) | 2.744 (0.8428) | < 0.0001† |
Breastfeeding initiated at birth | 5693 | 2050 (61.8%) | 1630 (68.7%) | < 0.001 | 468 (46.2%) | 644 (57.6%) | 806 (67.9%) | 819 (73.1%) | 943 (75.3%) | < 0.001 |
Values are numbers (%) or mean (SD)unless otherwise stated. Comparisons are made by Chi2 ,*t-test, †ANOVA, **Mann-Whitney U and ††Kruskal-Wallis, as appropriate |
There was strong evidence (p < 0.001) that term infants were more likely to live in the less deprived areas and some less strong evidence of a relationship with deprivation for preterm infants (p = 0.05) (Fig. 1, data in appendix 5).
[Fig. 1 Percentage of children living in each WIMD Decile by gestational group with p for trend]
In univariable analysis, preterm infants were more likely than term infants to have a learning difficulty (10.5% vs 4.5%, p < 0.001), a SEN (7.3% vs 2.5%, p < 0.001) and a behaviour problem (11.3% vs 5.8%, p < 0.001) (Table 2). There was evidence for a high risk of global developmental delay and general learning difficulties (both p < 0.001) in preterm infants, but insufficient evidence for an association with specific speech or communication problems, autism or dyslexia (Table 2).
Table 2
Association between gestational age, and deprivation decile, and the risk of learning difficulties or developmental problems.
| Gestation | Deprivation by WIMD Decile | |
| N total | n per group | Preterm | Term | P value | 1st and 2nd | 3rd and 4th | 5th and 6th | 7th and 8th | 9th and 10th | ptrend |
Learning Difficulties | | | | | | Least | | | | Most | |
All Learning Difficulties | 6691 | 541 | 421 (10.5%) | 120 (4.5%) | < 0.001 | 128 (23.7%) | 112(20.7%) | 110(20.3%) | 91 (16.8%) | 100 (18.5%) | < 0.001 |
GDD* | 149 | 128 (3.2%) | 21(0.8%) | < 0.001 | 36 (24.2%) | 32 (21.5%) | 29 (19.4%) | 26 (17.5%) | 26 (17.5%) | 0.02 |
Speech/Communication | 89 | 56 (1.4%) | 24(0.9%) | 0.07 | 20 (25%) | 10 (12.5%) | 27 (33.8%) | 12 (15%) | 11 (13.8%) | 0.06 |
Autism | 47 | 30 (0.8%) | 17(0.6%) | 0.6 | 11 (23.4%) | 14 (29.8%) | 12 (25.5%) | 3 (6.4%) | 7 (14.9%) | 0.01 |
Learning difficulty (general) | 201 | 162 (4.0%) | 39(1.5%) | < 0.001 | 53 (26.4%) | 45 (23.4%) | 32 (15.9%) | 33 (16.4%) | 38 (18.9%) | 0.002 |
Dyslexia | 65 | 46 (1.1%) | 19 (0.7%) | 0.08 | 8 (12.3%) | 11 (16.9%) | 10 (15.4%) | 17 (26.1%) | 19 (29.2%) | 0.05 |
Other Outcomes | | | | | | | | | | | |
Education Statement | 3356 | 179 | 145 (7.3%) | 34 (2.5%) | < 0.001 | 42(23.5%) | 38(21.2%) | 40(22.3%) | 25(14.0%) | 34(19.0%) | 0.001 |
Behavioural Problem | 6672 | 611 | 456(11.3%) | 155(5.8%) | < 0.001 | 181(29.6%) | 158(25.9%) | 117(19.1%) | 84(13.8%) | 71(11.6%) | < 0.001 |
Values are number (%) |
*GDD = Global developmental delay |
Comparisons are by Chi2 or extended Wilcoxon rank-sum test as appropriate. |
Children living in more deprived areas also had a higher prevalence of learning difficulties, SEN and behaviour problems (each p < 0.001). There was strong evidence that global developmental delay (p = 0.02), general learning difficulties (p = 0.002) and autism (p = 0.01) were associated with increasing deprivation. There was weak evidence of increasing speech or communication needs (p = 0.06) but less dyslexia (p = 0.05) in more deprived deciles. The combined results are shown in Fig. 2.
[Fig. 2 Proportion of children with a learning disability by gestational age and WIMD decile]
In the logistic regression, compatible with the univariable model, preterm infants (OR 2.57 (2.08–3.17)) and children in increasingly deprived areas (OR 1.08 (1.05–1.12), p < 0.001) had increased odds of learning difficulties (Table 3). There was no evidence of interaction between the two exposures (pinteraction=0.4). There was little change in the point estimates with the addition of potential confounders to the model and little evidence of interaction in the final adjusted model (pinteraction=0.3).
Table 3
Associations between both gestational age, and deprivation measures, and developmental outcomes; along with measures of interaction/modification between the two exposures.
Neurodevelopmental Measure | Unadjusted model | Adjusted for demographics factors* | Adjusted for demographics* and clinical factors** |
| OR (95% CI) | Pinteraction | OR (95% CI) | Pinteraction | OR (95% CI) | Pinteraction |
Learning Difficulties | n = 6691 | | n = 5443 | | n = 4563 | |
All Learning Difficulties | | | | | | |
Preterm birth WIMD Decile | 2.52 (2.04–3.12) 1.08 (1.04–1.11) | 0.4 | 2.64 (2.07–3.37) 1.07 (1.03–1.11) | 0.5 | 2.67 (2.02–3.53) 1.08 (1.03–1.12) | 0.3 |
GDD |
Prematurity WIMD Decile | 4.06 (2.55–6.46) 1.06 (1.00-1.13) | 0.5 | 4.80 (2.79–8.28) 1.08 (1.01–1.15) | 0.4 | 4.88 (2.73–8.74) 1.10 (1.02–1.18) | 0.3 |
Speech |
Preterm birth WIMD Decile | 1.53 (0.95–2.49) 1.09 (1.01–1.18) | 0.2 | 1.58 (0.91–2.75) 1.09 (1.00-1.20) | 0.2 | 1.53 (0.79–2.95) 1.04 (0.94–1.16) | 0.2 |
LD (general) |
Preterm birth WIMD Decile | 2.91 (2.04–4.17) 1.10 (1.04–1.16) | 0.8 | 2.78 (1.83–4.23) 1.07 (1.01–1.14) | 0.4 | 2.72 (1.68–4.40) 1.07 (1.00-1.15) | 0.5 |
Autism |
Preterm birth WIMD Decile | 1.14 (0.63–2.08) 1.14 (1.03–1.27) | 0.2 | 1.00 (0.52–1.91) 1.18 (1.04–1.34) | 0.04 | 0.94 (0.44–2.12) 1.23 (1.06–1.40) | 0.02 |
Dyslexia |
Preterm birth WIMD Decile | 1.72 (1.001–2.96) 0.94 (0.86–1.02) | 0.02 | 2.14 (1.12–4.08) 0.87 (0.78–0.97) | 0.04 | 2.38 (1.08–5.23) 0.90 (0.80–1.02) | 0.05 |
Educational Statement | n = 3356 | | n = 2594 | | n = 2062 | |
Preterm birth WIMD Decile | 2.99 (2.04–4.38) 1.09 (1.03–1.15) | 0.2 | 2.68 (1.75–4.11) 1.11 (1.05–1.19) | 0.6 | 2.44 (1.50–3.98) 1.14 (1.06–1.23) | 0.5 |
Educational statement in Children with Learning Difficulties | n = 389 | | n = 291 | | n = 231 | |
Preterm birth WIMD Decile | 1.45 (0.87–2.43) 1.04 (0.97–1.12) | 0.8 | 1.26 (0.70–2.29) 1.13 (1.04–1.25) | 0.4 | 1.06 (0.54–2.11) 1.14 (1.02–1.28) | 0.6 |
Behavioural problems | n = 6672 | | n = 5429 | | n = 4550 | |
Preterm birth WIMD Decile | 2.01 (1.67–2.44) 1.19 (1.15–1.22) | 0.1 | 2.14 (1.73–2.67) 1.14 (1.11–1.19) | 0.1 | 2.07 (1.62–2.65) 1.11 (1.07–1.16) | 0.3 |
Values are OR (95% CI) from the multi-level logistic regression model (random effects variable was the age at the time of the survey).* Adjusted for maternal age at birth, sex and ethnicity ** Adjusted for smoking in pregnancy, multiple births, mode of delivery, birthweight and whether breastfeeding was initiate |
In the fully adjusted model, there was strong evidence that preterm babies were more likely to have GDD (p < 0.001), general LD (p < 0.001) and dyslexia (p = 0.03) but no clear association with speech delay (p = 0.2) or autism (p = 0.9). Deprivation appeared to be associated with GDD (p = 0.01) and Autism (p = 0.004) but not speech disorders (p = 0.5) or dyslexia (p = 0.09).
There was some evidence that the relationship between preterm birth and autism (p = 0.02) and dyslexia (p = 0.05) was modified by deprivation. The relationship between autism and deprivation was seen in term (OR 1.54 (1.18-2.00)) but not preterm infants (OR 1.09 (0.93–1.29)). In contrast, the relationship between lower rates of dyslexia with increasing deprivation was seen in preterm (OR 0.84 (0.73–0.97)) infantsbut not in term infants (OR 1.11 (0.87–1.40)).
Comparable patterns were seen with the analysis of the secondary outcomes. In the logistic regression, preterm infants had an increased odds of SEN (OR 4.08 (1.68–9.88)) as did children in increasingly deprived areas (OR 1.14 (1.01–1.28)). Preterm birth increased the odds of behavioural problems (OR 2.36 (1.45–3.85)) as did living in increasingly deprived areas (OR 1.21 (1.14–1.29)). There was no evidence of interaction between the exposures for SEN (p = 0.2) and behavioural problems (p = 0.1).
There was little evidence for any interaction between preterm birth and deprivation and the risk of SEN(pinteraction=0.6) when restricting the cohort to those with learning difficulties. Sensitivity analyses for primary outcome gave compatible results to the main analyses; using Townsend score (pinteraction=0.3) or the WIMD rank at birth (pinteractiont=0.5) as the measure of deprivation, or using gestaional age split into five levels (pinteraction=0.7).