Children with CHD are at increased risk for neurodevelopmental delay after paediatric cardiac surgery. Reference Gunn, Beca and Hunt1–Reference Newburger, Wypij and Bellinger3 A statement issued by the American Heart Association in 2024 highlighted the need for early detection of developmental delay and the provision of supportive interventions for children with CHD. Reference Marino, Lipkin and Newburger4 Follow-up is important, given that young children with developmental delays are more likely than their peers to have emotional, behavioural, and health problems as adults. Reference Ilardi, Ono, McCartney, Book and Stringer5 In the United Kingdom, there are no guidelines for the screening and monitoring of children with CHD for neurodevelopmental delays beyond what healthy children receive. In a prospective study of preschoolers with CHD by Hoskote et al, Reference Hoskote, Ridout and Banks6 parents of 60.4% of children with developmental delays and no known predisposing syndrome, did not report using child developmental, neurology, or special senses services, highlighting the need for improved assessment of neurodevelopmental progress in these children, to enable their appropriate access to services. Children with CHD are treated by cardiac specialists with little exposure to child development training and practice. At young ages, motor delays are most commonly apparent; however, more subtle impairments may not be identified until children are older. Reference Schiller and Tibboel7,Reference Hövels-Gürich, Seghaye and Schnitker8 The fact that the gold standard assessment of neurodevelopment involves face-to-face testing with a trained examiner using validated assessment tools such as Bayley Scales of Infant Development Reference Albers, Grieve and Bayley9 represents a further challenge given that these specialist resources are limited in scope. Therefore, the use of a parent-completed screening method such as the Ages and Stages Questionnaires-3, Reference Squires, Bricker and Potter10 which are widely used in healthy children and have been used for children with CHD to a limited extent, Reference Billotte, Deken and Joriot11–Reference Lépine, Gagnon and Prud'homme13 is of interest since this might be deliverable at a lower cost to a health service. In a previous study, we found that the Ages and Stages Questionnaires were sensitive against a gold standard neurodevelopmental assessment over a range of testing thresholds but lacked specificity at some thresholds. Reference Brown, Ridout and Pagel14 We therefore wished to evaluate this measure further since it might prove a useful adjunct that could help the right children to access early intervention for neurodevelopmental delays.
This study aimed to explore the use of the Ages and Stages Questionnaires prospectively amongst a multi-centre cohort of children with CHD who were followed up after cardiac surgery as part of a longitudinal study, Reference Pagel, Brown and McLeod15–Reference Read, Ridout and Johnson17 including describing outcomes based on this measure and risk factors for an abnormal outcome.
Materials and methods
Study design
The ‘cardiac impact study‘ Reference Wray, Ridout and Jones16–Reference Brown, Ridout and Pagel18 is a prospective cohort study that included children who underwent an index heart surgery at five United Kingdom children‘s heart centres between October 2015 and June 2017. The current study represents a secondary analysis involving the Ages and Stages Questionnaires data collected in the cardiac impact study.
Patient population
Children were recruited as morbidity cases to the ‘cardiac impact study‘ following a cardiac surgery if they experienced at least one important postoperative morbidity, as defined previously, Reference Pagel, Brown and McLeod15 including acute neurological events, unplanned re-intervention, feeding problems, renal replacement therapy, major adverse events, extracorporeal life support, and necrotising-enterocolitis. The control children with no morbidities were matched by measures of CHD complexity and chronological age and identified at discharge after heart surgery as having experienced no morbidities. Reference Wray, Ridout and Jones16,Reference Brown, Pagel and Ridout19
Study variables collected
The design of the cardiac impact study was such that all children who were admitted for cardiac surgery were eligible for enrollment, whether or not they had undergone prior cardiac surgery. CHD is highly variable and heterogenous, and children undergo surgery at a range of ages, and may also undergo serial surgeries and reoperations. All these interventions may have an impact on subsequent outcomes, hence, by necessity we used patient-level characteristics of CHD complexity and any additional non-cardiac conditions to account for case mix.
We recorded pre-selected and defined baseline clinical and demographic variables prospectively at baseline and reviewed these at follow-up. Reference O'Brien, Clarke and Jacobs20–Reference Cassedy, Drotar and Ittenbach22 The clinical variables are known to be related to the risk of mortality Reference Pagel, Brown and McLeod15,Reference Brown, Ridout and Pagel18 and included CHD complexity category (graded in complexity from A to E), bypass time, presence of functionally univentricular heart, history of non-cardiac congenital comorbidities, history of acquired comorbidities (including acquired brain injuries), prematurity (birth <37 completed weeks of gestation), additional cardiac risk factors, and high severity of illness prior to surgery (a requirement for ventilation or mechanical circulatory support prior to surgery). Other variables known to be related to measures of functional outcome were also collected: race, family income, and number of carers and highest parental educational level. Reference Knowles, Ridout and Crowe23–Reference Harkness, Gregg and Fernández-Salgado25
We considered age at assessment in our analyses and given that all children had their first postoperative assessment 6 months after cardiac surgery, age at assessment was in line with the age at cardiac surgery. Age was grouped into bands of neonatal age at surgery/ 6 months old at postoperative assessment, infant age at surgery/ >6 months to 18 months old at postoperative assessment and child age at surgery/ >18 months old at postoperative assessment.
Follow-up procedures
Ages and Stages Questionnaires data were collected at two timepoints:
Postoperative assessment, 6 months after surgery
We undertook detailed follow-up at 6 months (± 2 weeks) after the index cardiac surgery, Reference Pagel, Brown and McLeod15,Reference Brown, Pagel and Ridout19 between May 2015 and December 2017, irrespective of patient age (we refer to this as the postoperative assessment). We offered this assessment to all children in the cardiac impact study (both cases and controls) alongside quality of life questionnaires which were part of the wider study objectives. Reference Wray, Ridout and Jones16 For children who were aged under 60 months at the time of surgery, the Ages and Stages Questionnaires were undertaken by a cardiac nurse with the parent. The assessment was undertaken at the patient's cardiac centre alongside a clinic visit, or over the phone based on preference.
Toddler assessment – at aged two to three years
Within the scope of a subsequent follow-up study, Reference Read, Ridout and Johnson17 we further assessed children who were under 12 months of age at their first cardiac surgery, when they were aged between two and three years (between September 2018 and March 2020). At this toddler assessment, we excluded children with genetic conditions known to cause neurodevelopmental delay. Reference Read, Ridout and Johnson17 The rationale for this was that children who have a genetic condition known to cause neurodevelopmental delay are routinely referred to community paediatrics, which is the main route to being offered early intervention for neurodevelopmental delays in the United Kingdom. In contrast, children with CHD are not offered any such assessment in the United Kingdom, and hence we focused this second evaluation on children with much more complex CHD who were least likely to have been offered early intervention. We aimed to explore and describe their neurodevelopment with a view to improving access to services in future. Two trained psychology researchers approached families who had previously agreed to be contacted and asked for consent to assess children at home. Parents were sent an information sheet about the study and the Ages and Stages Questionnaires were undertaken either during a home assessment or by parents online or on paper, and the completed responses were posted to the research team if a home assessment was declined.
The ages and stages tool
The outcome measure for this study was the Ages and Stages Questionnaires-3 result. Reference Squires, Bricker and Potter10 The Ages and Stages Questionnaires consist of 21 different age versions applicable to children between the ages of birth and 66 months. Six questions per domain, across 5 domains are answered as yes (10), sometimes (5), not yet (0), and a total score is calculated for each domain. The domains assessed are the following: gross motor, fine motor, communication, personal social, and problem solving. Additional yes/no and free text questions are included. According to the manual, if a child scores below 1 standard deviation of the normative mean in any single domain, developmental activities should be provided, and if they score below 1 standard deviation in two or more domains or below 2 standard deviations in any domain, they should be referred for further assessment. For all children assessed with Ages and Stages Questionnaires in our study, we provided a report of the results to the parents and sought their permission to share the report with appropriate health professionals.
Defining patient scores in the current study
Any Ages and Stages Questionnaires domain score falling between 1 standard deviation and 2 standard deviations below the normative mean (per the manual) was coded Amber; any domain score falling >2 standard deviations below the normative mean were coded Red. All other domain scores were coded Green. As well as individual domain scores, patients were given an overall color score (Red, Amber, or Green) corresponding with their worst performing domain. If there were any Red domains, overall score was Red. If there were Amber domains but no Red domains, overall score was Amber. If all domains were Green, overall score was Green.
Data completeness and quality
We assessed the quality of the Ages and Stages Questionnaires data for completeness, with particular interest in the post-operative (first) time point which was a larger sample and involved data collection using an approach closer to that which could be used in routine practice. Study data were considered ineligible if the incorrect version of the Ages and Stages Questionnaires was completed for the patient's age band, or if the questionnaires were not completed (lost to follow-up). Patients who died during follow-up were noted.
Data analysis
Study outcomes for the postoperative assessment
Primary outcome – patients with an Ages and Stages Questionnaires total score of Red versus a result of Amber + Green
Secondary Outcome – patients scoring Red versus Amber + Green in each of the five individual Ages and Stages Questionnaires domains.
Assessing systematic bias at the postoperative assessment
A Chi square test was used to compare the participating patients with eligible non-participating patients.
Risk factors for abnormal outcome at the postoperative assessment
We explored links between patient risk variables and both primary and secondary outcomes at the postoperative time point using univariable and multivariable logistic regression. Complexity of CHD was ranked from most severe (A), to least severe (E) using categories developed for paediatric cardiac surgery early outcome monitoring. Reference Wray, Ridout and Jones16 We present these outcomes as odds ratios with 95% confidence intervals. Missing data were not included in the logistic regression, and we have reported the number of observations for each category. Primary and secondary outcomes results where p < 0.05 were included in the multivariable analysis.
Evaluation of outcome data for children who had two assessments
Only a small subset of patients participated in an assessment at two sequential time points; therefore, our analysis was limited and descriptive. This group of children had cardiac surgery in infancy and excluded those with congenital non-cardiac malformations. We compared the Ages and Stages Questionnaires colour scores at the postoperative, and toddler age time points in terms of the number of patients who had improved, stayed the same, or worsened. We described the characteristics of patients who had improved, stayed the same, or got worse based on the presence of patient demographic and clinical variables most strongly related to the study outcomes at the postoperative time point.
Results
Study sample
From 666 patients who were recruited to the cardiac impact study, 554 children who were less than 60 months of age and survived to the postoperative time point at 6 months were eligible for inclusion in the postoperative assessment. Of these 554 children, 306 (60.3%), were included, and the reasons for non-inclusion were completion of the wrong version (by age) of the Ages and Stages Questionnaires for 37 children and parents elected not to participate for 211 children (supplementary materials figure 1).
Systematic bias in participating versus non-participating patients at the postoperative assessment
The only significant differences between the groups were in ethnicity and mother's highest education level. There were more white patients in the participating cohort, and the mother's highest education level was higher in the participating cohort. Details provided in Supplementary Table S1.
Post-operative assessment ages and stages questionnaires outcomes
In the sample of 306 children, 171 (55.9%) were boys,124 (40.5%) had their index cardiac surgery as a neonate, 138 (45.1%) had their index surgery as an infant, and 44 (14.4%) had their index surgery as a child; 45 (14.7%) had a univentricular heart, 70 (22.9%) had very complex CHD, and 37 (12.1%) had non-cardiac congenital malformations (further details in Supplementary Table S2).
Overall Ages and Stages Questionnaires outcomes at the postoperative assessment (6 months) were Total = 306, Green = 117 (38.3%), Amber = 57 (18.6%), and Red = 132 (43.1%). The youngest children, who were aged 6 months at the time of assessment (neonatal surgery), were most likely to score Red (n = 113/124, 85.6%), whereas children who were in the two older age groups were less likely to score Red (n = 32/182, 17.6%) (supplementary material figure 2.)
Risk factors for postoperative outcome
The univariable logistic regression comparing patients scoring Red versus Green + Amber found that patients with univentricular heart were more likely to score Red compared to patients without univentricular heart (odds ratio 2.1 (95% confidence interval 1.1, 4.1) p = 0.02). Children who were assessed at age 6 months after a neonatal cardiac surgery were more likely to score Red than children who were assessed at ages over 6 months after they had surgery as infants or children (odds ratio 54.2 (95% confidence interval 26.0, 113.1) p < 0.001). Surprisingly, patients with congenital non-cardiac malformations were less likely to score Red, compared to patients without (odds ratio 0.5 (95% confidence interval 0.2, 0.9) p = 0.03). Other risk variables, including complexity level of CHD, were not significant at p = 0.05 significance level.
The multivariable logistic regression comparing patients scoring Red versus Green + Amber (Table 1) on the Ages and Stages Questionnaires incorporated the risk variables of univentricular heart, congenital non-cardiac malformations, and age group. In this multivariable analysis, only the age group was significant at p = 0.05 significance level and as we show in Table 1 univentricular heart, and congenital non-cardiac comorbidities were non-significant. Those of neonatal age at surgery (aged 6 months at follow-up) were more likely to score Red than those of infant or child age at surgery (odds ratio 52.3 (95% confidence interval 25.0, 109.6) p < 0.001).
Results of the univariate logistic regression for primary outcomes (red versus amber and green total scores) and the multivariable logistic regression for the primary outcome, including risk variables below p = 0.05 in the univariate analysis: univentricular heart, congenital non-cardiac malformations, and age
Data were missing in the following variables: Ethnicity (13 missing), Mother's educations (34 missing), Household income (36 missing), Number of carers (33 missing).
Postoperative assessment secondary outcomes
We show patient numbers for each colour score of the Ages and Stages Questionnaires by the individual domains shown in Figure 1. The patient numbers for each domain are divided into neonate versus non neonate as the neonatal age band contributes heavily to the Red scores.
Ages and stages questionnaires colour scores by domain, divided by neonates versus infants + children.
In Table 2, we present univariable and multivariable logistic regression comparing patients scoring Red versus Amber + Green in individual domains of the Ages and Stages Questionnaires that were statistically significant in the univariable analysis – full details of all domains and risk variables are found in supplementary materials Table S3. With univariable analysis, we found that across all domains, neonatal age at index surgery increased the risk of scoring Red compared to infant or child age at the index surgery. Univentricular heart increased the risk of obtaining a Red score in communication and gross motor categories (odds ratio 2.5 (95% confidence interval 1.1–5.4) p = 0.02 and odds ratio 2.3 (95% confidence interval 1.2–4.4) p = 0.01, respectively). A more complex CHD category was significant in the communication domain as a predictor for scoring Red (odds ratio 2.2 (95% confidence interval 1.1–4.4) p = 0.03). The multivariable analysis for the secondary outcomes in the communication and motor domains (domains where more than one risk variable had a p < 0.05) found that only age group was significant in both domains, with neonatal age at surgery/assessment age of 6 months significant for both (p < 0.001).
Results of the univariate and adjusted analysis for secondary outcome: red versus amber and green across individual domains in domains and risk variables statistically significant in the univariate analysis
*The coefficient and odds ratio for age: neonate versus infant and child are omitted due to perfect/quasi-complete separation issues in the data.
Children who completed a second ages and stages questionnaire
Of the 258 children eligible for the toddler assessment, 127 completed the Ages and Stages Questionnaires (the other 131 did not consent to participate). We had data from two time points from only 87 children, unfortunately, the remaining 40 had not completed the Ages and Stages Questionnaires at the first post-operative time point. Of the 87 children who had two assessments, 53 (60.9%) were male, 59 (67.8%) had surgery as a neonate, 28 (32.1%) had surgery as an infant (none of them were from the child age group), and 7 (11.9%) had a univentricular heart. The overall Ages and Stages Questionnaires outcomes at the second assessment at toddler age were Green = 46 (52.9%), Amber = 18 (20.7%), and Red = 23 (26.4%).
Serial assessment of Ages and Stages Questionnaires outcomes
The group that completed both the postoperative assessment and the toddler assessment were compared to the postoperative assessment group, and we present the results in Supplementary Table S4. There was a significant difference between the groups in the following categories: presence of congenital non-cardiac malformations, type of operation, and household income. Only 1 (1.1%) patient had congenital non-cardiac malformations in the toddler group, compared to 37 (12.1%) in the postoperative group. The toddler group had 9 (10.3%) children who had a reparative procedure and 45 (51.7%) who had a palliative procedure and 33 (37.9%) who had an ambiguous procedure compared to the postoperative assessment group who had 173 (54.5%) who had a reparative procedure compared to 54 (17.7%) who had a palliative procedure and 79 (25.8%) who had an ambiguous procedure. The toddler group had 60 (69%) patients with a household income over £25k/pa compared to 177 (61%) in the postoperative assessment group. When the Ages and Stages Questionnaires colour scores were compared between the two time points for the 87 children, 43 (49.2%) patients had improved, 30 (34.5%) stayed the same, and 13 (16.1%) got worse. We present the breakdown of changes between the two assessments in Supplementary Table S5. The most notable changes were by age: of 59 children who had surgery as a neonate 39 (66.1%) improved, 17 (28.8%) stayed the same, and only 3 (5.1%) got worse. Of 28 children who had surgery as an infant 5 (17.9%) improved, 12 (42.9%) stayed the same and 11 (39.3%) got worse. Figure 2 describes the changes between the postoperative assessment and the toddler assessment for each individual domain. The largest contributor to improved colour scores was the gross motor domain where 45 (51.7%) of the patients improved between the first and second assessment. In all domains except the gross motor domain, the majority of patients remained the same between the two time points.
Figure showing ages and stages questionnaires colour score changes between the postoperative assessment and the toddler assessment by domain.
Discussion
Summary of our findings
At a postoperative assessment 6 months following index cardiac surgery in 306 children, overall, 43% of children had a Red Ages and Stages Questionnaires result indicating that they should be referred for specialist assessment of child development. A notable finding was that most children who were assessed at 6 months old after undergoing neonatal cardiac surgery (85.6%) had a Red result and the gross motor domain was the greatest contributor to these Red scores. A subset of 87 children was reassessed as toddlers, and at this second assessment, 66.1% of children who had their first assessment after neonatal surgery had an improved Ages and Stages Questionnaires outcome, whereas amongst children who had a first assessment later in infancy very few improved (17.9%), and 39.3% got worse. The multiple model indicated that young age group versus the older age groups at both index surgery and assessment appeared more influential than CHD complexity, univentricular status, or non-cardiac morbidities in the occurrence of a Red score. A potential contributor to the Red scores amongst children who were assessed 6 months after neonatal surgery is that parental responses may capture issues beyond neurodevelopment, including medical / health issues related to heart disease at this young age. This explanation might be supported by the prominence of gross motor issues amongst the Red Ages and Stages Questionnaires results in these children. We were somewhat surprised by the improvement in scores at the second assessment that was conducted in a subset of children, given that children with CHD have been reported to grow into their deficits. Reference Gerstle, Beebe, Drotar, Cassedy and Marino26–Reference Davidson, Amso, Anderson and Diamond28 One interpretation of these findings is that complex CHD and serial staged palliation and/or other CHD treatments in the period following neonatal cardiac surgery, led to children's physical abilities at the age of 6 months being affected by their health, which in turn influences the Ages and Stages Questionnaire assessment at this age. An alternative interpretation of these findings is that the Ages and Stages Questionnaire findings at the age of 6 months after neonatal cardiac surgery, including motor delays, were an early feature of neurodevelopmental deficits. With both explanations, there is the potential for improvement with physical therapy input, if resources are available to provide this for young babies with CHD.
Our results in context
The Ages and Stages Questionnaires are a screening tool completed by parents often with the support of a trained nurse to assess whether further specialist assessment and intervention are needed to support neurodevelopment. Hence, this can be a useful screening test to identify possible neurodevelopmental delays in a timely way. Our work builds on previous studies detailing the use of Ages and Stages Questionnaires in this population. Reference Squires, Bricker and Potter10–Reference Lépine, Gagnon and Prud'homme13 Billotte et al used the Ages and Stages Questionnaires to assess 210 children with complex CHD, although with a median age of 34 months, their cohort was older, they found 86 (41%) had a score >2 standard deviations below the normative mean (i.e., Red). Reference Billotte, Deken and Joriot11 This was more likely with greater cardiac complexity and non-cardiac syndromes. Lepine et al. Reference Lépine, Gagnon and Prud'homme13 (146 children) and Noeder et al. Reference Noeder, Logan and Struemph12 (163 children) assessed the performance of the Ages and Stages Questionnaires in detecting developmental delay in children with CHD, based on a gold standard assessment. Both studies found that the Ages and Stages score of ‘Red' performed well as a threshold for referral based on the American Association of Pediatrics Guidelines for a developmental screening tool (sensitivity and specificity above 70–80%). Reference Noeder, Logan and Struemph12,Reference Lépine, Gagnon and Prud'homme13
Strengths and limitations
The prospective multi-centre nature of our data collection is a strength of our study, since our experience might be indicative of what could happen with the potential future use of the Ages and Stages Questionnaires in clinical practice if they were to be used after cardiac surgery, noting that children often receive serial surgeries as part of their treatment. A limitation of our study is that our baseline first assessment enrolled all eligible children who were admitted for cardiac surgery during the study period, irrespective of their prior history. Therefore, we assessed children with a range of ages with differing prior clinical histories. In children who were assessed after cardiac surgery in infancy and childhood, we had no prior earlier assessment of their development. The low capture of serial data points makes the interpretation of the data more challenging. Moreover, we did not collect details of children's prior developmental assessments and interventions, and hence these cannot be used to interpret our data.
Considering the representativeness of the sample, there was a greater likelihood of participation in the study from white race patients, high-income households, families with two carers, and university educated mothers. The small numbers of study participants outside this demographic could be a likely explanation for the lack of association found with social variables. In addition, the number of patients with data for both the postoperative and toddler assessments was small in comparison to the eligible patients for the study and limited our ability to draw generalised conclusions.
In our study, the 37 children with congenital non-cardiac malformations were unexpectedly less likely to score Red. The most common malformations were chromosomal abnormalities, renal abnormalities, and non-cardiac abnormalities, which have been linked to greater chances of a Red Ages and Stages Questionnaire outcome in prior studies. Reference Noeder, Logan and Struemph12,Reference Lépine, Gagnon and Prud'homme13 Of children with congenital non-cardiac malformations, at the time of surgery, eight were neonates, twenty-two were infants, and seven were children: the low proportion of neonatal surgery patients in this group might explain why on average the chance of a Red result was lower than average for this cohort of children. Another possible explanation could be response shift where parents make adjustments for their children and score them higher, particularly in subjective domains, e.g. communication and problem solving, which is a limitation with a questionnaire-based measure.
Thirty-seven patients were excluded due to the incorrect Ages and Stages Questionnaire for their age group being completed. There is an online tool to aid choosing the correct version of the Ages and Stages Questionnaires – the administration date is entered along with the child's date of birth and the number of weeks the baby was born premature (if any). This finding emphasises the importance of training on this tool being implemented to the healthcare professionals and parents using it.
Conclusion
Given the importance of accessing early intervention for neurodevelopmental impairments in young children with CHD, within a healthcare system which is resource constrained, we believe that experience gained with the Ages and Stages Questionnaires is vital. Our prospective multi-centre study of outcomes based on the Ages and Stages Questionnaires amongst children undergoing cardiac surgery, flags up some important learning points for its future use in this setting: the Ages and Stages Questionnaires, which have been previously found to perform well as a measure of developmental delay in children with CHD, may add to the timely evaluation of these children at all ages; although the interpretation of Ages and Stages Questionnaires results in the youngest children aged less than 6 months, who are assessed after neonatal surgery, may be more challenging. This may relate in part to the impacts of medical complexity on these children at the time of assessment. This finding supports the benefits of serial assessment, which is recommended by the American Heart Association. Our study emphasises the importance of training for nurses who use the Ages and Stages Questionnaires with parents.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1047951124026477.
Acknowledgements
We wish to thank all those who contributed to the Cardiac Impact Study from Great Ormond Street Hospital, Bristol Children‘s Hospital, Evelina London Hospital, Bristol Children's Hospital, and University College London.
Financial support
This study was funded by Great Ormond Street Children's Charity (GOSHCC V0139). This project benefited from funding by the National Institute for Health Research Health Services and Delivery Research programme (Project No: 12/5005/06). The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the NIHR HS&DR programme or the Department of Health. Research at Great Ormond Street Hospital is supported by the GOSH NIHR BRC.
Ethical standard
This study was approved by London City Road Research Ethics Committee (14-LO-1442).
Open access
