APIC indicates Adults Born Preterm International Collaboration; RECAP, Research on European Children and Adults Born Preterm; and VPT/VLBW, very preterm or very low birth weight.
Diamonds represent pooled estimates from either the IPD or non-IPD subgroup analysis or from all cohorts; diamond size indicates the 95% CI for the pooled estimate. The arrow indicates that the lower 95% CI (1.54) for the Linsell et al EPICure study7 is further than the axis limit of 1.5. Horizontal lines represent the 95% CIs of the estimates for each cohort. Box size represents the weighting given to the study. SMD indicates standardized mean difference.
eTable 1. Childhood Neurosensory Impairment in VPT/VLBW Participants From Each IPD Cohort
eTable 2. Newcastle Ottawa Criteria and Ratings for Each IPD Cohort
eTable 3. Linear Mixed Model Demonstrating Reducing Gestational Age by Birth Year Among VPT/VLBW Participants
eTable 4. Linear Mixed Model Demonstrating Reducing Birth Weight by Birth Year Among VPT/VLBW Participants
eTable 5. IQ and Demographic Information of All Participants From Each IPD Cohort
eTable 6. Neonatal and Demographic Data for VPT/VLBW Participants From Each IPD Cohort
eTable 7. Study Characteristics of VPT/VLBW Cohorts Not Included in the IPD Meta-analysis
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Eves R, Mendonça M, Baumann N, et al. Association of Very Preterm Birth or Very Low Birth Weight With Intelligence in Adulthood: An Individual Participant Data Meta-analysis. JAMA Pediatr. 2021;175(8):e211058. doi:10.1001/jamapediatrics.2021.1058
Is very preterm birth or very low birth weight vs term birth associated with intelligence in adulthood?
In this meta-analysis of individual participant data from 8 cohorts comprising 2135 adults with and without very preterm birth or very low birth weight in 7 countries, IQ was significantly lower among adults who were born very preterm or with very low birth weight compared with adults who were born at term, with a mean between-group difference of approximately 12 IQ points. Lower gestational age, lower birth weight z scores, the presence of neonatal bronchopulmonary dysplasia or intraventricular hemorrhage, and lower maternal educational levels were significantly associated with lower IQ among adults born very preterm or with very low birth weight.
This individual patient data meta-analysis suggests that very preterm birth or very low birth weight may be associated with a clinically relevant difference in IQ, relative to term birth, in adulthood.
Birth before 32 weeks’ gestation (very preterm [VPT]) and birth weight below 1500 g (very low birth weight [VLBW]) have been associated with lower cognitive performance in childhood. However, there are few investigations of the association of neonatal morbidities and maternal educational levels with the adult cognitive performance of individuals born VPT or VLBW (VPT/VLBW).
To assess differences in adult IQ between VPT/VLBW and term-born individuals and to examine the association of adult IQ with cohort factors, neonatal morbidities, and maternal educational level among VPT/VLBW participants.
Systematic review of published data from PubMed and meta-analysis of individual participant data (IPD) of cohorts from 2 consortia (Research on European Children and Adults Born Preterm [RECAP] and Adults Born Preterm International Collaboration [APIC]).
The meta-analysis included prospective longitudinal cohort studies that assessed the full-scale IQ of adults born VPT or VLBW and respective control groups comprising term-born adults.
Data Extraction and Synthesis
The study followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline for analyses of individual participant data and identified 8 studies that provided data from 2135 adults (1068 VPT/VLBW and 1067 term-born participants) born between 1978 and 1995. Meta-analyses of IPD were performed using a 1-stage approach, treating VPT birth or VLBW and cohort as random effects.
Main Outcomes and Measures
Full-scale IQ scores were converted to z scores within each cohort using the combined SD of VPT/VLBW participants and a control group of term-born participants, with scores centered on the mean of the control group.
A total of 426 records were identified and screened. After exclusions, 13 studies were included in the aggregate meta-analysis. The IPD meta-analysis included 8 of the 9 RECAP and APIC cohorts with adult IQ data. The mean (SD) age among the 8 IPD cohorts was 24.6 (4.3) years, and 1163 participants (54.5%) were women. In unadjusted analyses, VPT/VLBW participants had mean adult IQ scores that were 0.78 SD (95% CI, −0.90 to −0.66 SD) lower than term-born participants, equivalent to a difference of 12 IQ points. Among VPT/VLBW participants, lower gestational age (score difference per week of gestation, 0.11; 95% CI, 0.07-0.14), lower birth weight z scores (score difference per 1.0 SD, 0.21; 95% CI, 0.14-0.28), the presence of neonatal bronchopulmonary dysplasia (score difference, −0.16; 95% CI, −0.30 to −0.02) or any grade of intraventricular hemorrhage (score difference, −0.19; 95% CI, −0.33 to −0.05), and lower maternal educational level (score difference, 0.26; 95% CI, 0.17-0.35) were all significantly associated with lower IQ scores in adulthood.
Conclusions and Relevance
In this IPD meta-analysis, lower gestational age, lower weight for gestational age, neonatal morbidities, and lower maternal educational levels were all important risk factors associated with lower IQ among young adults born VPT or VLBW.
An important life outcome after very preterm (VPT) birth (<32 weeks’ gestation) or very low birth weight (VLBW; <1500 g) is intelligence, defined as “the capacity to learn from experience, using metacognitive processes to enhance learning and adapt to the surrounding environment.”1(p751) Standardized intelligence tests in the general population provide an IQ score with a normative mean (SD) of 100 (15) points. An individual’s IQ is associated with a range of life course outcomes, including physical health, premature death, educational attainment, and socioeconomic success.2-5 Thus, adult IQ is a global factor associated with long-term outcomes among individuals born VPT or VLBW (VPT/VLBW).
While individual studies typically indicate that VPT/VLBW individuals have lower IQ in adulthood than term-born individuals,6-8 to our knowledge, no specific meta-analysis of adult IQ among VPT/VLBW individuals has been published. Meta-analyses of childhood IQ have reported that the scores of VPT/VLBW children are approximately 11 to 13 IQ points lower on tests than term-born children.9-12 Sustained differences into adulthood cannot be assumed, as indicated by the smaller IQ differences in adulthood that have been reported among individuals with normal vs low birth weight.13 Furthermore, considerable variation across VPT/VLBW cohorts has been found, which is potentially explained by factors such as gestational age or birth weight inclusion criteria and later selective attrition.11 Variation in IQ may be associated with individual-level neonatal or demographic factors, such as sex, low birth weight for gestational age, neonatal morbidities, or maternal educational level.14-17
To investigate cohort- and individual-level factors, meta-analyses of individual participant data (IPD) have been proposed as superior to traditional meta-analyses of aggregated data.18 Meta-analysis of IPD allows for accurate harmonization of data across cohorts and increased statistical power for detecting individual-level risk factors.19 Furthermore, the associations of cohort-specific factors, such as rates of attrition, can be investigated.
We performed an IPD meta-analysis of IQ in adulthood with 3 objectives. The first was to compare the difference in adult IQ among VPT/VLBW participants vs a control group of term-born participants, the second was to examine cohort- and individual-level factors associated with adult IQ among VPT/VLBW participants, and the third was to conduct a sensitivity analysis assessing whether adult IQ scores from the IPD cohorts were representative of all cohorts of VPT/VLBW individuals.
This study was conducted as part of the Research on European Children and Adults Born Preterm (RECAP) Consortium.20 Along with 7 adult RECAP cohorts, 6 non-European cohorts from the Adults Born Preterm International Collaboration (APIC) Consortium21 were invited to participate. All studies had received country-specific ethical reviews, with participants providing written informed consent, and all adhered to the Declaration of Helsinki.22 This IPD meta-analysis was registered with the International Prospective Register of Systematic Reviews (PROSPERO; registration number CRD42020162043) and followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline for analyses of individual participant data.
Prospective longitudinal cohort studies of VPT/VLBW adult participants who had completed a standardized IQ test at a mean age of 17 years or older were eligible for inclusion. All cohort studies were required to include a control group of term-born participants to allow for computation of harmonized and comparable IQ z scores, which minimized bias owing to different versions of tests or secular trends (ie, the Flynn effect).23
To assess whether the cohorts from RECAP and APIC were representative of all VPT/VLBW participants reported in the literature, we performed a PubMed search using the search string (intelligence OR IQ OR cognition OR cognitive) AND (adult OR adulthood OR late adolescence) AND (preterm OR gestation OR birthweight OR birth weight). The last search was performed on July 9, 2020.
Eligibility for inclusion was assessed by 2 authors (R.E. and Y.N.). Any disagreements regarding eligibility were resolved by discussion.
After initial data scoping, encrypted data from each cohort were transferred to the University of Warwick. Data including IQ scores, neonatal variables, maternal educational levels, the presence of neurosensory impairment (NSI) in childhood, and attrition rates were collected for all cohorts. All data were only accessible to authorized personnel from the RECAP Consortium.
To harmonize results, IQ scores were converted to z scores within each cohort using the combined SD of both groups (VPT/VLBW and control), with scores centered on the mean of the control group. Neonatal data included gestational age at birth, sex, birth weight, presence of bronchopulmonary dysplasia (BPD), presence of intraventricular hemorrhage (IVH), and multiple birth. The definitions of BPD varied, with some studies defining BPD as oxygen dependency at 36 weeks’ postmenstrual age and some defining it as oxygen dependency more than 28 days after birth. For each definition, separate subanalyses were performed to assess each criterion’s independent association with IQ. The definition of IVH was classified according to criteria provided by Papile et al24 (ie, IVH is categorized from grades 1-4, with 1 indicating hemorrhage limited to germinal matrix, 2 indicating blood noted within the ventricular system but not distending it, 3 indicating blood in the ventricles with distension of the ventricles, and 4 indicating intraventricular hemorrhage with parenchymal extension); however, some cohorts provided either IVH grades 3 and 4 or IVH grades 2 and 3 combined. Thus, IVH was harmonized into no IVH vs any IVH (grades 1-4), and a subanalysis was performed to compare no IVH or IVH grades 1 to 2 with IVH grades 3 to 4 among cohorts for which analysis was possible. Multiple birth was classified as a binary variable, with 0 indicating singleton birth and 1 indicating multiple birth. Birth weight z scores were determined using the Fenton international growth chart for preterm infants.25
Maternal educational level was harmonized according to the International Standard Classification of Education (ISCED) into low (ISCED levels 0-2), medium (ISCED levels 3-5), and high (ISCED levels 6-8).26 Evidence of childhood NSI was collated from data indicating severe visual impairment (blind in both eyes), hearing impairment (uncorrected by assistive devices), nonambulatory cerebral palsy, or childhood cognitive impairment (IQ <70 points). If data regarding a certain NSI variable were missing for a participant, the individual was categorized as having no evidence of the presence of that impairment. Data indicating the presence of NSI were combined into a binary childhood NSI variable (any evidence of impairment vs no evidence of impairment) (eTable 1 in the Supplement). In addition, 4 cohort-level factors were assessed. For each cohort, the percentage of eligible VPT/VLBW participants who did not have adult IQ scores was calculated (ie, the percentage of attrition among VPT/VLBW participants). To focus on selective attrition, the percentage of VPT/VLBW participants with diagnoses of childhood NSI who did not have adult IQ scores was also calculated (ie, the percentage of attrition among VPT/VLBW participants with childhood NSI). Data on year of birth and mean age at assessment among VPT/VLBW participants in each cohort were also recorded.
Data were assessed for consistency with previous studies of the included cohorts, with any discrepancies resolved by communication with the respective study investigators. Two authors (R.E. and Y.N.) assessed cohort quality and comparability using the Newcastle-Ottawa Scale27 (score range of 0-9, with higher scores indicating higher quality) (eTable 2 in the Supplement). The primary outcome of interest was the full-scale IQ z score of VPT/VLBW participants compared with term-born participants.
All participants with adult IQ scores were included. Missing neonatal data were imputed solely for VPT/VLBW participants, and missing data on maternal educational levels were imputed for both VPT/VLBW and term-born participants using multiple imputation by chained equations (mice),28 which resulted in less than 5% of the data being imputed (Table 1).7,8,29-34 In the first analysis, a simple comparison of IQ scores between VPT/VLBW and term-born participants was conducted using a 1-stage linear mixed model. We analyzed the association of VPT birth or VLBW with IQ using a random-intercept model for each cohort and a random slope for the association of VPT birth or VLBW with IQ by cohort, which was estimated using maximum likelihood via the lme4 package, version 1.1-21, in R software, version 3.6.1 (R Foundation for Statistical Computing).35 The association between VPT birth or VLBW and IQ was then examined after adjusting for sex and maternal educational level and after removing VPT/VLBW participants with childhood NSI or differentiating between VPT/VLBW participants with and without neonatal morbidities (ie, IVH and BPD). All analyses used a 1-stage approach with random intercepts and slopes.
To explore antecedents of IQ scores among VPT/VLBW participants, a 1-stage IPD analysis was performed. Cohort factors (age at IQ assessment, birth year, percentage of attrition among VPT/VLBW participants, and percentage of attrition among VPT/VLBW participants with childhood NSI) were added as fixed effects. Individual-level neonatal factors and maternal educational levels were then also added as fixed effects. Beta estimates from all factors were reported for both the univariable and multivariable analyses to assess their independent and combined associations. Statistical significance was set at P < .05.
A sensitivity meta-analysis using aggregate data was performed by combining all cohorts used in the IPD analyses, summary data from the Hack study,6 and additional cohorts identified through the PubMed search for whom IPD were not requested. The standardized mean difference (SMD) in IQ scores between VPT/VLBW and term-born participants in each cohort were pooled using a random-effects meta-analysis via the meta package, version 4.9-7, in R software.36 Heterogeneity across cohorts was assessed using Cochran Q and I2 statistics, and a subgroup analysis was performed to differentiate between IPD and non-IPD cohorts to test for selection bias.
A total of 426 records were identified and screened; of those, 413 records were identified through a PubMed search, and 13 records were known studies conducted by the RECAP and APIC consortia. Overall, 342 records were excluded based on titles and abstracts, and 84 full-text articles were assessed for eligibility (Figure 1). Of 7 potential RECAP cohort studies of adults,7,8,29-31,37,38 2 were excluded: the Suikkanen et al study,37 which did not perform a full-scale IQ test, and the Weisglas-Kuperus et al study,38 which did not include a control group. Among 6 potential APIC cohort studies,6,32-34,39,40 3 were excluded: the Saigal et al39 and Doyle et al40 studies, which did not assess adult IQ, and the Hack study,6 for which only summary data were available. After exclusions, 13 studies (9 RECAP or APIC cohorts and 4 cohorts identified through PubMed) were included in the aggregate meta-analysis. The IPD meta-analysis included 87,8,29-34 of the 9 RECAP and APIC cohorts with adult IQ data, comprising a total of 2135 adults (1068 VPT/VLBW participants and 1067 term-born participants). The mean (SD) age among the 8 IPD cohorts was 24.6 (4.3) years, and 1163 participants (54.5%) were women. Summary data for each cohort included in the IPD analyses are shown in Table 1.
The 8 IPD cohorts were from 7 high-income countries (Australia,34 Finland,29,30 Germany,8 Ireland,7 New Zealand,32 Norway,31 and the United Kingdom7,33); 6 of the studies were regional,8,29-31,33,34 and 2 were national.7,32 The mean (SD) gestational age at birth was 28.3 (2.8) weeks among VPT/VLBW participants and (mean [SD] gestational age was not available for term-born participants because some participants were recruited in childhood or adulthood), and birth years ranged from 1978 to 1995. The mean (SD) age at IQ assessment was 24.4 (4.6) years among VPT/VLBW participants and 24.8 (4.3) years among term-born participants. Among VPT/VLBW participants, more recent birth year (post-1978) was associated with lower gestational age (difference per year, −0.32 weeks; 95% CI, −0.60 to −0.04 weeks) and lower birth weight (difference, −29.85 g; 95% CI, −58.78 to −0.91 g) (eTable 3 and eTable 4 in the Supplement). In total, 557 VPT/VLBW participants (52.2%) and 606 term-born participants (56.8%) were women (eTable 5 and eTable 6 in the Supplement).
Based on the Newcastle-Ottawa Scale27 (eTable 2 in the Supplement), the mean (SD) study quality score was 7.9 (0.6), and studies were rated highly with regard to representativeness, ascertainment of exposure, and assessment of outcomes. However, studies differed in inclusion criteria and attrition rates among VPT/VLBW participants, which were higher than 50% in 47,8,29,30 of 8 cohorts (Table 1).
The analysis of all participants from IPD cohorts7,8,29-34 indicated that mean IQ z scores among VPT/VLBW participants were 0.78 SD (95% CI, −0.90 to −0.66 SD) lower than those of term-born participants. When sex and maternal educational level were included, the estimate of the association of VPT birth or VLBW with IQ scores was minimally reduced, from a difference of −0.78 SD to a difference of −0.74 SD (95% CI, −0.85 to −0.63 SD). Excluding participants with childhood NSI reduced the IQ difference from −0.78 SD to −0.65 SD (95% CI, −0.76 to −0.55 SD). With regard to neonatal morbidities, the presence of any grade of IVH among VPT/VLBW participants was associated with a larger difference in IQ scores than no IVH (−0.99 SD [95% CI, −1.19 to −0.79 SD] vs −0.70 SD [95% CI, −0.84 to −0.57 SD], respectively). A similar difference among VPT/VLBW participants with and without neonatal BPD was also found (−0.93 SD [95% CI, −1.10 to −0.76] vs −0.67 SD [95% CI, −0.80 to −0.55], respectively).
Table 2 shows the results of the IPD meta-analysis examining the association of neonatal factors, maternal educational level, and cohort factors with IQ scores among VPT/VLBW participants. Significant associations with IQ z scores in the multivariable analysis were gestational age (score difference per week of gestation, 0.11; 95% CI, 0.07-0.14), birth weight z score (score difference per 1.0 SD, 0.21; 95% CI, 0.14-0.28), the presence of neonatal BPD (score difference, −0.16; 95% CI, −0.30 to −0.02) or any grade of IVH (score difference, −0.19; 95% CI, −0.33 to −0.05), and maternal educational level (score difference, 0.26; 95% CI, 0.17-0.35) (Table 2). These findings indicated, for example, that among VPT/VLBW participants, each extra week of gestation was associated with an increase in IQ z score of 0.11, which was equivalent to 1.65 IQ points. In contrast, neither sex nor singleton or multiple birth significantly altered adult IQ scores among VPT/VLBW participants. In addition, none of the cohort-level factors were significant in the multivariable analysis. The association with birth year was significant in the univariable analysis (score difference, −0.02; 95% CI, −0.03 to 0), which suggested that more recent birth year was associated with lower IQ in adulthood.
The PubMed search of 413 records identified 4 additional non-RECAP and non-APIC cohorts41-45 with extractable adult IQ data. Adding the summary data from the Hack study45 produced 5 non-IPD cohorts and 8 IPD cohorts (Figure 1). Characteristics of the non-IPD cohorts are shown in eTable 7 in the Supplement.
Using aggregate data, the SMD between VPT/VLBW and term-born participants was −0.61 (95% CI, −0.93 to −0.29) in non-IPD cohorts41-45 and −0.84 (95% CI, −0.97 to −0.71) in IPD cohorts7,8,29-34 (Figure 2). According to the results of the Cochran Q test, this finding suggested no significant differences between the IPD and non-IPD cohorts (Q = 1.80; P = .18). However, the heterogeneity was larger among non-IPD cohorts (I2 = 75%) compared with IPD cohorts (I2 = 41%).
This IPD meta-analysis examined the association between VPT birth or VLBW and adult IQ. Among 8 cohorts contributing IPD,7,8,29-34 VPT/VLBW participants scored 0.78 SD lower in IQ score than term-born participants, which is equivalent to a between-group difference of approximately 12 IQ points. This substantial difference was marginally reduced after adjustment for sex and maternal educational level. Even when participants with childhood NSI were excluded (which also removed those with low childhood IQ), the IQ score difference between VPT/VLBW and term-born participants was 0.65 SD (equivalent to 9.8 IQ points). The addition of cohorts for which IPD were not available41-45 did not substantially alter the findings. Among VPT/VLBW participants, individual-level factors associated with lower IQ were earlier gestational age, lower birth weight z score, the presence of neonatal BPD or IVH, and lower maternal educational level.
The IQ score difference of −0.78 SD between VPT/VLBW and term-born participants is a larger standardized difference than that reported for other functional outcomes, such as mental and physical health and social functioning.46 The IQ differences between VPT/VLBW and term-born adults found in this study are also similar to those previously reported in studies of childhood IQ.9-12 Three prospective studies found moderate to high stability in IQ scores from childhood to adulthood among VPT/VLBW individuals.7,32,47 The adult cohorts in the present IPD analysis were followed up for decades, producing a higher risk of selective attrition over a long period. Individuals who are more socially disadvantaged or who have NSI are more frequently lost to follow-up,48 which may be associated with the smaller difference in IQ between VPT/VLBW and term-born participants.49 However, the cohort differences in rates of attrition among all VPT/VLBW participants or among VPT/VLBW participants who specifically had childhood NSI were not associated with IQ, nor was the age at IQ assessment. Furthermore, the association between IQ and birth year was not significant after including individual-level factors in the multivariable analysis. This finding suggests little change in mean adult IQ scores among VPT/VLBW individuals born between 1978 and 1995 after adjustment for the lower birth weights and gestational ages among individuals born more recently. In more recent VPT/VLBW cohorts, no improvements in childhood IQ by birth year have been reported.12 Improvements in IQ have also not been observed in successive cohorts of children born extremely preterm from the same regions.50,51 Thus, given the stability of IQ over time among these cohorts, changes in neonatal care and reduced mortality52 do not appear to have translated into long-term improvements in IQ among VPT/VLBW individuals during this period.
For individual-level factors (Table 2), even when adjusting for other neonatal factors and maternal educational levels, a dose-response association was found between gestational age and IQ, replicating individual study findings.53,54 The association of neonatal IVH and BPD with lower IQ is also consistent with previous meta-analyses of IQ in childhood and individual studies of IQ in adulthood.12,55 After other factors were controlled for, neonatal BPD was associated with a mean reduction in IQ of 2.4 points (95% CI, −4.5 to −0.3 points; difference in z score, −0.16), and the presence of any grade of IVH was associated with a mean reduction in IQ of 2.9 points (95% CI, −5.0 to −0.8 points; difference in z score, −0.19) (Table 2). Severe IVH (grades 3-4) was only investigated in a univariable analysis that did not include all cohorts; however, severe IVH was associated with a mean decrease in IQ of 9.9 points (95% CI, −13.8 to −6.2 points; difference in z score, −0.66). However, this finding warrants cautious interpretation because other neonatal factors and maternal educational levels were not controlled for. Birth weight z scores had a significant association with IQ after other factors were controlled for. Results from the multivariable model suggest that a birth weight of −2 SD for gestational age is associated with a decrease of 6.3 IQ points (95% CI, −8.4 to −4.2 points) compared with a birth weight that is appropriate for gestational age. This finding provides additional support for the association between low birth weight for gestational age and lower IQ in adulthood among VPT/VLBW individuals.56
In contrast to neonatal factors, maternal educational levels and similar factors have been largely overlooked in research on VPT birth or VLBW and outcomes.17 Compared with low maternal educational level, VPT/VLBW participants who had mothers with medium or high educational levels had IQ scores that were generally 0.26 SD and 0.52 SD (3.9 and 7.8 IQ points) higher, respectively. These associations are equivalent in extent to those associated with serious neonatal complications (ie, BPD and IVH). The association of maternal educational level with adult IQ may reflect an amalgam of different factors. These factors may include genetic traits,57 maternal smoking,58 breastfeeding rates,59 and parental behaviors.55 Some factors could be modified postnatally and have been reported to have implications for academic achievement and development among both the general population and VPT/VLBW groups.60,61
This study has several strengths. The harmonization and use of IPD to assess neonatal factors and maternal educational levels, the presence of childhood NSI, and adult IQ scores for 8 cohorts of VPT/VLBW participants allowed for reliable comparison across cohorts. In addition, we tested a range of specific cohort factors, which is a challenging task in an aggregate meta-analysis because these details are rarely available in published studies.
The study also has limitations. These include the differences in cohorts with regard to eligibility criteria, such as the stricter inclusion criterion of less than 26 weeks’ gestational age in the Linsell et al study,7 the use of maternal educational level rather than broader factors (such as socioeconomic status or combined parental educational level), and the different methods used for recruiting participants for the control groups. Individuals in the control groups were typically recruited in infancy; however, in some cohorts, recruitment occurred in childhood or adulthood, and neonatal data were unavailable. Thus, we could not evaluate whether factors such as birth weight z scores are similarly associated with IQ among term-born participants. Given that the mean age at assessment was 24.6 years, the findings of the present study reflect IQ in young adulthood only.
Adults who were born VPT/VLBW had IQ scores that were a mean of 12 points lower than adults who were term born. This finding is similar to those reported in meta-analyses of IQ in childhood7,9-12,47 despite the present study’s greater risk of selective attrition owing to a longer follow-up period. The provision of antenatal and neonatal care that is associated with reductions in BPD and IVH62,63 and parenting or educational interventions that help to decrease the social disparities associated with maternal educational levels61 may improve cognitive outcomes among adults who were VPT or VLBW.
Accepted for Publication: March 16, 2021.
Published Online: May 28, 2021. doi:10.1001/jamapediatrics.2021.1058
Corresponding Author: Dieter Wolke, PhD, Department of Psychology, University of Warwick, University Road, Coventry CV4 7AL, United Kingdom (email@example.com).
Author Contributions: Mr Eves had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Eves, Mendonça, Anderson, Hovi, Räikkönen, Zeitlin, Wolke.
Acquisition, analysis, or interpretation of data: Eves, Mendonça, Baumann, Ni, Darlow, Horwood, Woodward, Doyle, Cheong, Anderson, Bartmann, Marlow, Johnson, Kajantie, Nosarti, Indredavik, Evensen, Räikkönen, Heinonen, Wolke.
Drafting of the manuscript: Eves, Mendonça, Marlow, Wolke.
Critical revision of the manuscript for important intellectual content: Eves, Mendonça, Baumann, Ni, Darlow, Horwood, Woodward, Doyle, Cheong, Anderson, Bartmann, Marlow, Johnson, Kajantie, Hovi, Nosarti, Indredavik, Evensen, Räikkönen, Heinonen, Zeitlin, Wolke.
Statistical analysis: Eves, Ni, Wolke.
Obtained funding: Horwood, Woodward, Johnson, Kajantie, Räikkönen, Zeitlin, Wolke.
Administrative, technical, or material support: Eves, Mendonça, Baumann, Ni, Cheong, Anderson, Bartmann, Evensen, Zeitlin, Wolke.
Supervision: Bartmann, Marlow, Nosarti, Wolke.
Conflict of Interest Disclosures: Mr Horwood reported receiving grants from the Health Research Council of New Zealand and the National Institutes of Health during the conduct of the study and grants from Cure Kids (New Zealand) and the National Health and Medical Research Council (Australia) as well as personal fees from the University of Otago at Christchurch outside the submitted work. Dr Doyle reported receiving grants from the Centres of Research Excellence, National Health and Medical Research Council (Australia), during the conduct of the study. Dr Anderson reported receiving grants from the National Health and Medical Research Council (Australia) during the conduct of the study. Dr Marlow reported receiving consulting fees from Novartis, RSM Consulting, and Takeda Pharmaceutical outside the submitted work. Dr Kajantie reported receiving grants from the Academy of Finland, the Finnish Foundation for Pediatric Research, the Signe and Ane Gyllenberg Foundation, and the Sigrid Juselius Foundation during the conduct of the study and grants from the Finnish Diabetes Research Foundation, the Finnish Foundation for Cardiovascular Research, the Juho Vainio Foundation, and the Novo Nordisk Foundation outside the submitted work. Dr Wolke reported receiving funding from the New Opportunities for Research Funding Agency Co-operation in Europe, Dynamics of Inequality Across the Life-course Program outside the submitted work. No other disclosures were reported.
Funding/Support: This work was supported by grant 733280 from the European Commission as part of the Research on European Children and Adults Born Preterm (RECAP) Consortium Preterm Project, an EU Horizon 2020 study to construct a platform combining data from cohort studies of very preterm birth in Europe.
Role of the Funder/Sponsor: The European Commission had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: Stef Van Buuren, PhD, of the University of Utrecht/ Netherlands Organisation for Applied Scientific Research, Leonhard Bakker, MSc, of the Netherlands Organisation for Applied Scientific Research, and Henrik Singmann, PhD, of University College London provided advice regarding statistical analysis of the individual participant data. None of the contributors received compensation for their assistance.