Early childhood parent-reported speech problems in small and large for gestational age term-born and preterm-born infants: a cohort study

Objective (1) To assess if preterm and term small for gestational age (SGA) or large for gestational age (LGA) infants have more parent-reported speech problems in early childhood compared with infants with birth weights appropriate for gestational age (AGA). (2) To assess if preterm and term SGA and LGA infants have more parent-reported learning, behavioural, hearing, movement and hand problems in early childhood compared with AGA infants. Design Cohort study. Setting Wales, UK. Participants 7004 children with neurodevelopmental outcomes from the Respiratory and Neurological Outcomes of Children Born Preterm Study which enrolled 7129 children, born from 23 weeks of gestation onwards, to mothers aged 18–50 years of age were included in the analysis. Outcome measures Parent-reported single-answer questionnaires were completed in 2013 to assess early childhood neurodevelopmental outcomes. The primary outcome was parent-reported speech problems in early childhood adjusted for clinical and demographic confounders in SGA and LGA infants compared with AGA infants. Secondary outcomes measured were parent-reported early childhood learning, behavioural, hearing, movement and hand problems. Results Median age at the time of study was 5 years, range 2–10 years. Although the adjusted OR was 1.19 (0.92 to 1.55) for SGA infants and OR 1.11 (0.88 to 1.41) for LGA infants, this failed to reach statistical significance that these subgroups were more likely to have parent-reported speech problems in early childhood compared with AGA infants. This study also found parent-reported evidence suggestive of potential learning difficulties in early childhood (OR 1.51 (1.13 to 2.02)) and behavioural problems (OR 1.35 (1.01 to 1.79)) in SGA infants. Conclusion This study of 7004 infants in Wales suggests that infants born SGA or LGA likely do not have higher risks of parent-reported speech problems in early childhood compared with infants born AGA. To further ascertain this finding, studies with wider population coverage and longer-term follow-up would be needed.


Pl e a s e n o t e:
C h a n g e s m a d e a s a r e s ul t of p u blis hi n g p r o c e s s e s s u c h a s c o py-e di ti n g, fo r m a t ti n g a n d p a g e n u m b e r s m a y n o t b e r efl e c t e d in t his ve r sio n.Fo r t h e d efi nitiv e ve r sio n of t hi s p u blic a tio n, pl e a s e r ef e r t o t h e p u blis h e d s o u r c e.You a r e a d vis e d t o c o n s ul t t h e p u blis h e r's v e r sio n if yo u wi s h t o cit e t hi s p a p er.
Thi s v e r sio n is b ei n g m a d e a v ail a bl e in a c c o r d a n c e wit h p u blis h e r p olici e s. S e e h t t p://o r c a .cf. a c. u k/ p olici e s. h t ml fo r u s a g e p olici e s.Co py ri g h t a n d m o r al ri g h t s fo r p u blic a tio n s m a d e a v ail a bl e in ORCA a r e r e t ai n e d by t h e c o py ri g h t h ol d e r s .

INTRODUCTION Background
2][3] Population studies have shown trends that infants are being born heavier.In England and Wales, there was an average increase of 40 g over 26 years for all live births recorded between 1986 and 2012 with a 8%-10% increased risk of being born with a high birth weight during this period. 4Trends for heavier birth weights were also observed in other countries including Canada, USA and Sweden, and the cause for this shift is not understood. 4This drives a need to better understand predictors of birth weight and its association with longerterm outcomes.
Outcomes for certain subgroups such as prematurity and small for gestational age (SGA) have been well explored but remains unclear for large for gestational age (LGA) infants particularly mid-term to long-term outcomes.
LGA is associated with birth complications and indirect effects persisting into adulthood. 5Around birth, larger term infants have increased risk of shoulder dystocia, meconium aspiration, lower 5 min

STRENGTHS AND LIMITATIONS OF THIS STUDY
⇒ This study has a relatively large sample size, although this represents around 26% of all eligible invitees to Respiratory and Neurological Outcomes of Children Born Preterm Study (n=26 722).⇒ Major domains of early childhood developmental outcomes were explored.⇒ Potentially using parent-reported questionnaires may introduce some degree of recall bias.⇒ Like most cohort studies, missing data in those enrolled, particularly in confounders, were present.⇒ Data were collected between 2003 and 2011.

Open access
Apgar scores and death. 6 77][8] In preterm births, the prognosis for being LGA is unclear, with some literature suggesting an advantage; reporting lower perinatal mortality in preterm LGA infants, but with higher risks of early-onset sepsis and intraventricular haemorrhage. 9verall, data on the mid-term to long-term effects of being LGA across all gestational ages are lacking, and little is known about neurodevelopmental outcomes.A retrospective cohort study by Moore et al found an increased risk of autism in term infants born SGA between 23 and 31 weeks, whereas being born large may have conferred some protective effect. 10In view of speech being a dynamic product of higher function cognitive and sensorimotor feedback processes involving multiple cortical and subcortical areas for planning, selecting, sequencing and motor programming, it was selected as the primary outcome of interest.Due to the complex interaction between neurodevelopmental domains incuding speech, this study also evaluated learning, behavioural, hearing, movement and hand problems. 11jectives 1.To assess if preterm and term SGA or LGA infants have more parent-reported speech problems in early childhood compared with infants with birth weights appropriate for gestational age (AGA).2. To assess if preterm and term SGA and LGA infants have more parent-reported learning, behavioural, hearing, movement and hand problems in early childhood compared with AGA infants.

Study design
This study was conducted using data collected from the Term infants were selected to be comparable with the preterm infants, for date of birth, sex and locality.A total of 7129 responses were received including consent for data usage and access to health databases.Characteristics between those who enrolled and those that did not are shown in online supplemental file 3. Parents with preterm infants were more likely to respond to the questionnaire than those with term births (p<0.001) and responders were less likely (39.5% vs 53.8%) to live in the most deprived half of Wales.However, the proportions of males (p=0.52) and those with low, normal or high birth weights for their gestation were similar (p=0.61).
Parent-reported answers on neurodevelopmental outcomes were collected in 2013 for all ages.For example, 'Does your child have any problems with their speech?';followed by a yes, or no, option.In the event neither option was selected, the response was recorded as unsure.][13] The primary neurodevelopmental outcome was parentreported early childhood speech problems. 11Secondary neurodevelopmental outcomes were parent-reported early childhood learning, behavioural, hearing, movement and hand problems.For this study, all 'unsure' responses were re-coded as 'no' and included in the primary analysis.

Study population
The eligible population were all children enrolled in RANOPS, born from 23 weeks of gestation to mothers from 18 to 50 years of age with available speech outcomes (n=7004).Exposure measures were gestational age at birth and birthweight centile by category.Birthweight centiles were calculated for each sex and gestation (in weeks) using the LMS Growth programme (Medical Research Council). 14SGA was defined as <10th centile on the UK-WHO growth charts and >90th centile for LGA.These are values generally accepted across England and Wales, with Scotland using the 5th and 95th centiles. 15 16estational age was categorised (as per WHO definitions) as extremely preterm (<28 weeks), very preterm (28-31 weeks) and moderate to late preterm (32-<37 weeks). 17ost-maturity was defined as gestational age at birth greater than or equal to 42 weeks. 18SGA and LGA infants were compared with AGA infants across gestations.

Covariates
Covariates included neonatal and maternal influences known a priori to influence birth weight in-utero.Neonatal factors are singleton or multiple births, gestational age at birth and sex.Maternal factors accounted for, included smoking during pregnancy, the Welsh Index of Multiple Deprivation (WIMD) score (a measurement of relative deprivation for small areas with scores of 1-1909, 1 being the most deprived) and their age. 19hese overlap with potential influences on the primary outcome, parent-reported speech problems in early childhood.While complex and multifactorial in potentially influencing the primary outcome, maternal socioeconomic status was accounted for using WIMD scores. 20redictors of low birth weight include low birth weight, 21 multiple pregnancies 22 and foetal sex. 23Mode of delivery was not assumed to impact birth weight and considered likely multifactorial due to perinatal clinical practice surrounding estimated birth weight or centile, birth complications and maternal preference. 15 24 25The determinants of high birth weight are, however, less clear.

Open access
Evidence suggests that active smoking is associated with increased risk of low birth weight and preterm birth, and prevalence of smoking is associated with socioeconomic status. 26 27There is also an increased risk of low birth weight with younger mothers, perhaps influenced by socioeconomic status.However, evidence surrounding birth weight and advanced maternal age is inconsistent, with some studies showing increased prevalence of infants born LGA, while others showing a U-shaped trend of low birth weight with increasing age. 26 28-31atistical analysis Duplicates and infants with missing exposure or primary outcome data were removed.Initially, neonatal and maternal characteristics at birth for all infants were compared across their birthweight centile category.Comparisons were performed using the Kruskal-Wallis equality of populations ranks test for birthweight centile category, gestational age and WIMD score at birth.Sex, number of births, mode of delivery and maternal smoking during pregnancy were compared using the χ 2 test and maternal age at delivery using the analysis of variance test.
Next, the proportion of children with speech problems between birthweight centile categories was compared.Using a logistic regression model, the unadjusted and adjusted for potential confounders, ORs for speech problems, comparing SGA and LGA infants to those born AGA were derived.Stratifying by preterm and term infants and using a logistic regression model, the ORs for speech problems were also derived for SGA and LGA infants compared with AGA infants.In view of a single primary outcome, p values corresponding to each statistical test were not corrected.
Six sensitivity analyses were performed.First, we repeated the logistic regression using a random-effects model to cluster by week of gestational age, and second by the age of the child at time of the survey.We then repeated using the 5th and 95th centile cut-offs, and then again removing all responders where 'unsure' was coded for the primary outcome. 16The main analysis was also repeated with an ordinal logistic regression analysis looking at the odds of an increasing number of reported neurodevelopmental disorders.Finally, the analysis was repeated using a missing data technique (Multiple Imputation with Chain Equations (details in online supplemental file 4)) to assess the impact of missing outcome and covariate data on the association seen. 17Likelihood ratio tests were used to compare models.Analysis was performed using Stata SE V.17 (Statacorp LLC).

Patient and public involvement
Patients or the public were not involved in the design, conduct, reporting, or dissemination plans of this research.Open access

RESULTS
Four thousand two hundred and eighty-four preterm and 2865 term infants were enrolled in RANOPS (n=7149). 12wenty duplicates were removed, leaving 7129 children.Those born to mothers younger than 18 or above 50 years of age were excluded (n=83), leaving 7046 eligible children.Eight infants had missing birthweight centiles, and a further 34 had missing data on the primary outcome, early childhood speech problems, leaving a study population of 7004 participants for the primary analysis (99.40% of the eligible responders).
Stratifying by birthweight centile categories of SGA, AGA and LGA infants, baseline neonatal and maternal characteristics of the study cohort are shown in table 1.The median age at time of the survey was 4.08-5.17(p=0.69).Besides the distribution of sex across birthweight centile categories, there were significant differences in other baseline neonatal and maternal characteristics analysed.
Distribution of early childhood speech problems in preterm and term infants by birthweight centile is as shown in figure 1. Figure 2 depicts the number of neurodevelopmental disorders stratified by gestation age at birth, birthweight centile category and deprivation measures or WIMD category (quintile).Analysis of the distribution of the primary outcome, parent-reported speech problems showed a significant difference between different birthweight centile categories, p=0.05.There were significant differences between birthweight centile categories and the secondary outcomes, except for parent-reported hearing problems (p=0.59)(table 2).
Logistic regression adjusted for neonatal and maternal features of the primary outcome (parent-reported speech problems in early childhood) showed that SGA and LGA infants were not more likely than their AGA counterparts to have problems, OR 1.19 (0.92 to 1.55) and OR 1.11 (0.88 to 1.41), respectively.Analysis of parent-reported secondary outcomes found evidence that SGA infants have an increased risk of learning difficulties in early childhood compared with those born AGA, OR 1.51 (1.13 to 2.02).There was also some evidence that SGA infants were more likely to have early childhood behavioural problems, OR 1.35 (1.01 to 1.79).This study did not find that infants born SGA or LGA were more likely to have hearing, movement or hand problems in comparison to AGA infants (table 3).
Analysis was repeated using a random-effects regression model, stratified by preterm-born or term-born, and the adjusted ORs did not demonstrate that preterm infants born SGA or LGA were more likely to have parentreported speech-problems than infants born AGA.There was also no evidence of interaction between the exposures measured (P interaction =0.999) (table 4).
Multilevel regression model of SGA and LGA babies compared with those born AGA across each gestational age category (using child age as the clustering variable) also showed compatible results (SGA, OR 1.19 (0.92 to 1.55); LGA, OR 1.11 (0.87 to 1.41)) with the main analysis.When 'unsure' responses excluded from analysis, logistic regression repeated showed no increased likelihood in primary neurodevelopmental outcome, parent-reported

Open access
early childhood speech problems in babies born SGA, OR 1.20 (0.92 to 1.56) or LGA, OR 1.13 (0.90 to 1.44) compared with those born AGA.Repeat of the analysis using the 5th and 95th centiles as cut-offs for SGA showed that babies born SGA were not more likely than babies born AGA to have parent-reported early childhood speech problems, OR 1.35 (0.98 to 1.86). 16There was also no evidence to suggest more parent-reported early childhood speech problems in babies born LGA, OR 1.11 (0.84 to 1.48).Repeat of the main analysis examining the odds of an increasing number of reported developmental disorders found results compatible with the primary analysis (SGA, OR 1.17 (0.96 to 1.43); LGA, OR 1.11 (0.93 to 1.32)).Finally, repeating the main analysis

DISCUSSION
][34][35][36][37][38][39][40] Example of such is the validated PARCA-R (Parent Report of Children's Abilities) for very-preterm infants at 2 years. 413][44] While the absolute numbers reported here are large, they do only represent a relatively low proportion of those invited to enrol in the study; although they had similar low or high birthweigths compared with those who did not enrol.However, they also appeared to come from less deprived areas than the wider population, and interpretation of our findings should consider this.The production of speech is complex, relying on not only intact cognitive, motor and sensory functions but also complicated by hearing loss, particularly the child's age at time of hearing loss. 32 45Reassuringly, in this work,

Open access
we were unable to identify clear associations between SGA, or LGA, infants and adverse parent-reported speech outcomes in early childhood; in either the unadjusted or adjusted models, as well as in the analysis of hearing problems.Also reassuringly, this study also did not find an increased risk of adverse parent-reported behavioural and hand problems in SGA and LGA infants, movement problems in SGA infants and learning difficulties in LGA infants.However, CIs are relatively wide and important increases in morbidity cannot be excluded and more work with precise estimate may be warranted.
Evidence from this study suggest that SGA infants may have higher risks of parent-reported learning difficulties and behavioural problems in early childhood but not hearing, movement or hand problems.A single-centre cohort study of term infants in Australia born to women presenting antenatally between 1981 and 1984 showed that children born SGA had significantly more learning difficulties when followed up at 14 years of age using a parent-reported survey and academic achievement test.This study reported a comparable 19.8% and 20.5% incidence for those who completed psychometric testing and behavioural questionnaires, respectively.This study also demonstrated long-term attention difficulties in extremely SGA (3rd centile and below) term-born female adolescents. 46Similar findings of poorer attention, executive function and memory were reported in a small cohort study of SGA preterm-born and term-born young adults who underwent neuropsychological assessment. 47imilarly, a cross-sectional study of 5181 childrens' behaviour, between ages 4 and 15 years in England, using a validated parent-reported questionnaire and stratifying for sociodemographic factors, suggested an association between birth weight and behavioural problems in children. 6 48e also saw that the proportion of all parent-reported early childhood disorders showed a strong relationship with gestational age at birth (seen in figure 2).This appears consistent with the wider literature, [49][50][51] and this work was designed to adjusted, in part, for this by using birthweight centiles.Alternatively, this may, in part, reflect the complexities and co-dependency of many of these outcomes as gestational age lowers, and further work is currently underway to look at the phenotypes and interactions of challenges these infants demonstrate.
Analysis of baseline characteristics noted a significant difference in maternal age, and demographics at birth between birthweight centile categories.9][30][31] In addition, the association between maternal smoking, deprivation and birthweight centile categories was sobering, with 20% of SGA infants exposed to in-utero smoking.In this work, more than 1 in 5 mothers smoked during pregnancy, identifying an important public health focus on a modifiable risk factor for low birth weight to address, 26 with even first trimester cessation of smoking having substantial benefits. 26 52 53here was also a significant difference in the mode of delivery across birthweight centile categories likely due to clinical practice managing high-risk pregnancies. 15 24 25his may include singleton or multiple pregnancies, gestational age and estimated birth weight or centile at time of birth.Sex was not associated with SGA or LGA, compatible with findings of a large cohort study in Netherlands. 23

CONCLUSIONS
Findings from this work, on a subset of 7004 infants in Wales suggest that infants born SGA or LGA may not have higher risks of speech problems in early childhood when compared with AGA infants.While enrolment was achieved on only a subset of less deprived infants, important differences may still exist, and we found that some infants being born SGA may have increased parentreported learning difficulties and behavioural problems compared with AGA infants.Further longer-term studies on infants born SGA and LGA would be of value to better understand the association of birth weight on neurodevelopment.

Study Design: The mor bidity and health utilisation of late pr eter m infants
In Wales, the aver age number of live-bor n deliver ies over the last 5 year s is 34,464 per annum. 2 Appr oximately 2,000 each year ar e bor n pr ematur ely w ith 500 bor n extr emely pr eterm and 1,500 late pr eter m.For late pr eter m infants, w e w ill addr ess: (a) r espir ator y mor bidity, (b) neur odevelopmental problems and (c) health car e utilisation, especially due to r espir ator y r easons, via a questionnair e study and also the w ell-linked health databases in Wales w ith the collabor ation of local neonatologists.
The data w ill be compar ed to ter m bor n and extr emely pr eter m-bor n groups.Our team, including Pr ofessor John Hender son as w ell as Pr ofessor Fr ank Dunstan, has gr eat exper tise in questionnair e studies assessing health of children and adults. 4,5  modified ISAAC questionnaire focussing on r espir ator y symptoms (r ather than aller gy) w ill be used to assess r espir ator y health including w heezing, dr ug usage and physical activity as w ell as visits to hospitals and gener al pr actitioner s in school-aged childr en 6 and a validated questionnair e by Pow ell et al w ill be used for pr eschool childr en .7,8 The questionnair es w ill be used to assess r espir ator y health and for any neur odevelopmental difficulties in a cr oss-sectional sur vey of childr en aged 1, 2, 3, 5, 7 and 9 year s.The questionnair es w ill allow us to assess the impact of late pr eter m bir th on r espir ator y outcomes, developmental delay and health ser vice utilisation.The questionnair es w ill be age-appr opr iate (one for <5 year s of age and another for 5-9 year s of age).The tw o age-appr opr iate questionnair es ar e based on pr evious extensively validated questionnair es for both r espir ator y symptoms and neur odevelopmental outcomes in both pr eschool and school-aged childr en.[6][7][8][9][10][11] We shall send 2,000 questionnair es w ith an invitation letter and par ent information sheet for each age-gr oup for the pr eter m groups (including 1,500 late pr eter m-and 500 extr emely pr eter mbor n childr en) together w ith a similar number for age-matched ter m controls.The matching of the ter m controls w ill be based on sex, place of bir th and chronological age (DOB), although in the analysis of the data, the cor r ected postnatal age (this is the age a pr ematur e baby w ould be if bor n at term) of the pr eter m infants w ill be taken into account as a potential confounding factor .The total of 24,000 questionnair e packs w ill be sent via the NHS Wales Infor matics Ser vice (NWIS), w hich have exper tise for identifying subjects fr om the NHS administrative r egister and then pr oviding DST, the mailing house w ith a database to send out the questionnair es.Fur thermor e, childr en who have moved fr om the ar ea or have died ar e identified and excluded, by using data fr om the All Wales Per inatal Sur vey (AWPS) w hich is dir ected by Pr ofessor Sailesh Kotecha w ith Pr ofessor Fr ank Dunstan a member of the steer ing committee.2 AWPS maintains data on childr en w ho die befor e their first bir thday, and this data w ill be used along w ith the Welsh Demographics Ser vice, w hich also holds information on childr en w ho have passed aw ay.
BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) Pr ior to any r eminder packs being sent out to the families, NWIS w ill be up-dated of all families w ho have r etur ned a questionnair e pack.NWIS w ill then up-date the contact data base and also r epeat the safe guar d of w ho has died, and then send the new data base of contact infor mation to DST, w ho will send out the r eminder packs (1 st after 1 month fr om the initial mailing date and 2 nd aft er 3-4 months).
Secondly, using linked databases (including the Patient Episode Database for Wales, PEDW and the National Community Child Health Database, NCCHD and the Welsh Demogr aphics Ser vice, WDS) for hospital episodes, w e w ill identify the general health ser vice usage of these childr en.For those childr en w ith a r eturned, signed (consented) questionnair e w e w ill be able to link the hospital data to their completed questionnair e to allow for more accur ate analysis.The infor mation gather ed fr om PEDW, NCCHD and WDS w ill also contain data on all childr en less than 10 year s of age in Wales (anonymised) and this w ill allow us to deter mine the r epr esentativeness of the sample r etur ning questionnair es.The social status of a family w ill be deter mined by location using the w elsh index for multiple depr ivation child index scor e 2011 (taken fr om WDS).
The linked cr oss sectional data at 5 age points w ill be summar ised in ter ms of r ates of w heezing, r ates of hospital admissions, etc.The explor ation of the data w ill be hypothesis led to investigate if being bor n late pr eter m has an effect on the health outcome of childr en w hen compar ed to ter m bor n controls.The health outcomes w ill include hospital admissions, visits to GP, w heezing episodes, use of pr escr ibed medication, doctor diagnosed conditions, long ter m disability and activity levels.The formal analysis w ill use gener al linear models to compar e r ates betw een differ ent gestational age gr oups adjusting for confounders such as social class, mater nal smoking during pr egnancy, bir th w eight, gender etc.
Sample size: For a question w ith binar y r esponse (e.g.symptom pr esent or absent), assuming a 50-60% r esponse r ate to the questionnair e, w e w ill have 95% pow er for identifying a differ ence betw een the pr eter m-and ter m-bor n childr en if the tr ue symptom r ate is 15% in pr eter m and 10% in ter m childr en.We have estimated the r esponse r ates based on pr evious similar studies in Wales. 4,12

NWIS collaboration
We have alr eady liaised w ith NWIS and shall ask their team to identify childr en bor n betw een 1/ 01/ 2003 to 31/ 12/ 2011.NWIS w ill identify all pr eter m infants (<37 w eeks of gestation at bir th) for each age gr oup (1,2,3,5,7 & 9 year s of age) using the national database (NCCHD and WDS).NWIS w ill then select the ter m controls by identifying gender matched childr en bor n on the same day and in the same ar ea (hospital or midw ifer y led unit) as each pr eter m infant.So ther e w ill be appr oximately 2,000 ter m controls for each age group.For each age (1,2,3,5,7 and 9 year s of age) ther e w ill be thr ee gr oups; childr en bor n at full ter m (37-43 w eeks of gestation), childr en born late pr eter m (33-36 w eeks of gestation) and childr en born extr emely pr eter m (≤32 w eeks of gestation).Each child w ill be assigned a study number based on their DOB/ gestational age/ a unique r efer ence number .
The information obtained from the health databases (WDS, NCCHD and PEDW) for each child w ill include the child's name, cur r ent addr ess, gestational age, DOB, bir th w eight, br east feeding at bir th and at 8 w eeks of age, w elsh index depr ivation scor e, hospital admissions of the last year and dischar ge diagnosis.The postcode for the curr ent addr ess w ill be used to identify the local health boar d (so that the appr opr iate local PI is identified on the invitation letter being sent to the family).This information w ill need to be checked to r emove any childr en that have died via AWPS and the WDS database.This w ill involve AWPS sending data to NWIS w ith the name, DOB and addr ess of each child that has passed aw ay over the last 10 years (1/ 1/ 03 to 31/ 12/ 12).None of this data will be made available to the r esear ch team.NWIS w ill supply DST, the mailing house w ith a study BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) DST w ill divide each data set into 7 sections accor ding to the local health boar d in which the child r esides.DST w ill then pr int the individual questionnair es (including study ID number s), invitation letter s including the family addr ess and signed w ith the appr opr iate LHB w ith details of the local PI, and infor mation sheets and place these in envelopes w ith w indow s that allow the letter to show the family addr ess, and these packs w ill then be sent by r oyal mail to the families.Stamped addr essed envelopes (to our r esear ch depar tment) w ill be included w ithin the study pack.DST also pr ovide a fur ther data health check, infor ming us of any duplicates and of families w ho may have moved, thus avoiding unnecessar y/ inappr opr iate mailing.No per sonal data will be r eleased by NWIS to the r esear ch team at this stage.
NWIS w ill send study ID number s w ith the child's gestational age to the r esear ch team, so that w hen questionnair es ar e r etur ned the cor r ect gestational age for the child can be confir med.Families w ill be given 4-8 w eeks to r etur n the questionnair es.A list of families who have r etur ned questionnair es (by study ID number s) will be sent to NWIS and they w ill send fur ther infor mation on each child that has a signed (consented) questionnair es, to include bir th w eight, antenatal details, hospital of bir th, hospital admissions and dischar ge diagnosis over the last 12 months.NWIS, follow ing a r e-check of the health data base for any deaths, w ill cr eate a list of any families that have not r etur ned a questionnair e.This up-dated study database w ill be given to DST and the fir st r eminder pack w ill be sent to these families at 1 month after the initial pack w as sent out.If the r esponse r ate r emains low then this process w ill be r epeated w ith a second r emainder pack being sent after 3-4 months, w hen the r esear ch team w ill give NWIS a list of all families w ho have r etur ned questionnair es and NWIS w ill collate an up-dated list, having r emoved any families w her e a child has died in the inter im per iod since last checking.This up-dated study database w ill be given to DST and the second r eminder pack w ill be sent to those families w ho have not yet r esponded.
Questionnair es w ill be r etur ned to the Depar tment of Child Health at Car diff Univer sity and pr ocessed by automated scanner s and specialised softw ar e (Remar k Office OMR 8).The questionnair es will be coded by the softw ar e, which also identifies any problems such as missing infor mation.Any pr oblems detected by the softw ar e w ill be r eview ed by the r esear ch team.This w ill for m par t of the data quality control checks.Anonymised data w ill be available for the w hole cohor t to allow us to deter mine the r epr esentativeness of the r esponder s.The r etur ned questionnaires w ill be pr ocessed by Dr Mar tin Edw ar ds and all data stor ed secur ely in the r esear ch office at the Depar tment of Child Health based at the Univer sity Hospital of Wales, Heath Par k, Car diff.By r etur ning the completed questionnair es it w ill be assumed that the family is consenting to take par t in the study.If the questionnair e for m is signed the family w ill be consenting to be contacted w ith r egar ds to clar ifying any issues with the r etur ned questionnair e; contacted in the futur e to take par t in fur ther r esear ch in this ar ea; and for access to patient identifiable infor mation in the health databases to link hospital admission and GP r ecor ds information w ith that gather ed from the questionnair es.It is clear ly stated in the infor mation leaflet w hat w ill happen w ith the data collected.

Ethical approval
The pr oject has ethical and R&D approval fr om w ithin Wales.The global gover nance check r efer ence is I RAS 91349 and the Ethics r efer ence is 12/WA/0155.The sponsor for the project is Car diff Univer sity (SPON1038-11).
BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s)

Timescale
We aim to mail the questionnair es betw een Januar y and Febr uar y 2013.We shall allow the families 6 -8 w eeks to r espond initially and send a second mailing of questionnair es betw een Apr il and May 2013 and if necessar y a thir d mailing of questionnair es in betw een July and August 2013, dependent upon the r esponse r ate.14]

Resources
The study is cur r ently funded by Depar tmental funds and r esear ch gr ant funding is being sought fr om sever al char ities that suppor t medical r esear ch -decision expected Jan/ Feb 2013.

Documents
i.
Letter of invitation to par ents -1 side of A4 paper ii.
Questionnair e for pr eschool childr en (<5 year s old) -4 sides of A4 paper iii.
Questionnair e for school aged childr en (5-9 year s old) -4 sides of A4 paper iv.
Infor mation sheet for families -2 sides of A4 paper v.
Flow diagr am of r esear ch protocol

Figure 1
Figure 1 Proportion of preterm and term infants with parent-reported speech problems in early childhood by birthweight centile (Gaussian smoothed).

Figure 2
Figure 2 Proportion of neurodevelopmental disorders stratified by gestation age at birth, birthweight centile category and WIMD category (quintile).AGA, appropriate for gestational age; LGA, large for gestational age; SGA, small for gestational age; WIMD, Welsh Index of Multiple Deprivation.

Table 1
Neonatal and maternal birth characteristics for all including preterm-born and term-born infants Values are n (%), mean (SD) or median(IQR)as appropriate.Denominator between measures differs due to missing data.*Maternal age at the time of delivery.†Welsh Index of Multiple Deprivation/Deciles of WIMD (lower values reflecting more deprivation).‡Maternal smoking during pregnancy.AGA, appropriate for gestational age; LGA, large for gestational age; SGA, small for gestational age; WIMD, Welsh Index of Multiple Deprivation.on May 9, 2023 by guest.Protected by copyright.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065587 on 27 April 2023.Downloaded from

Table 3
Unadjusted and adjusted ORs for early childhood neurodevelopmental outcomes in SGA and LGA infants compared with AGA controls *Adjusted for foetal sex, singleton or multiple birth and gestational age at birth.†Adjusted for maternal age and WIMD score at the time of birth and maternal smoking status during pregnancy.AGA, appropriate for gestational age; LGA, large for gestational age; SGA, small for gestational age; WIMD, Welsh Index of Multiple Deprivation.

Table 4
Unadjusted and adjusted ORs for primary neurodevelopmental outcome of early childhood speech problems stratified by preterm or term in SGA and LGA infants compared with AGA controls *Adjusted for foetal sex, singleton or multiple birth.†Adjusted for maternal age and WIMD score at time of birth and maternal smoking status during pregnancy.AGA, appropriate for gestational age; LGA, large for gestational age; SGA, small for gestational age; WIMD, Welsh Index of Multiple Deprivation.on May 9, 2023 by guest.Protected by copyright.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2022-065587 on 27 April 2023.Downloaded from database split into two sections (> 5 year s of age and > 5 year s of age).The database w ill contain the study ID number , name, addr ess and local health boar d of each study par ticipant.Data w ill be tr ansfer r ed betw een NWIS and DST via File Tr ansfer Pr otocol Secure (FTPS).