General cognitive functioning is the most common domain assessed in long-term outcome studies of children born very preterm, and this is typically done by administering a measure of general intelligence, or IQ. A recent meta-analysis found that at a group level, the IQ of children born very preterm was 13 points lower than children born at term, representing a group difference of approximately 0.9 standard deviation (SD).⁴ The authors also performed a meta-regression examining the standardized mean difference according to the birth year of the pooled cohorts (1990–2017), and found no evidence that the IQ discrepancy between children born very preterm and term is reducing in more recent cohorts. While IQ tests are generally sensitive at identifying cognitive problems, these measures are not particularly helpful for determining the nature of the cognitive deficits. Neuropsychological evaluations are required to establish the profile of cognitive strengths and weaknesses in children born very preterm, which is critical for determining appropriate remediate strategies.

Sensory and motor systems are impacted following very preterm birth. Meta-analyses report poorer performance in children born very preterm of 0.6 to 0.9 SD across tasks assessing visual perception and visual-motor integration when compared with term peers,⁵ with the rate of impairment being 2 to 3 times higher for those born extremely preterm when compared with children born at term.⁶ With regards to performance on standardized tests of motor functioning, the performance of children born very preterm is approximately 0.5 to 0.9SD below term peers.⁷ In contrast to a rate of 0.1% to 0.2% in the general population, the prevalence of cerebral palsy (CP) is approximately 15% for infants born before 28 weeks’ gestation and 6% for those born between 28 and 31 weeks’ gestation.⁸ Furthermore, it has been reported that approximately half of 5-year-olds born very preterm have functionally impaired motor coordination and balance which is not related to CP, and meeting criteria for developmental coordination disorder (DCD).⁹

Inattention is often noted to be an area of concern for children born very preterm by parents and teachers,¹⁰ which is supported by numerous neuropsychological studies that have identified poorer performance by children born very preterm on formal tests of attention compared with term controls.^11–14 In a cohort of children born extremely preterm, it was estimated that 75% had at least a mild attention impairment.¹³ Deficits have been reported across all facets of attention functioning including selective, sustained, shifting, and divided attention,^13,14 and while there is inter-individual variability,¹⁵ at a group level, the evidence suggests that very preterm birth is associated with a generalized attention impairment.

Executive function refers to those cognitive processes that are critical for goal-directed behavior and includes inhibitory control, working memory, cognitive flexibility, planning, and reasoning ability.^11–18 Meta-analyses have reported deficits across all elements of executive functioning, with the magnitude of the group differences with term controls ranging from 0.3 to 0.9 SD.^11,12,19 Executive dysfunction has been reported in preschoolers born very preterm,²⁰ but there is no evidence that these difficulties decline with age.^21,22

Less research has studied memory following very preterm birth, but the evidence to date suggests that memory is also an area of concern.^23–25 The rate of verbal and visuospatial episodic deficits was approximately three-fold in a cohort of 7-year-olds born very preterm compared with term born controls.²⁵ Similar findings were reported in an older cohort (13-year-olds), although this study found the recall of visuospatial information to be particularly problematic for children born very preterm with a six-fold higher rate of impairment.²³ Prospective memory is also more likely to be impaired in children born very preterm, especially time-based tasks in which children are required to remember to perform a future action at a specific time.²³

Language delay is common in infants and toddlers born very preterm, however, language challenges are ongoing with a meta-analysis revealing that school-aged children perform 0.6 SD below term controls for expressive language and 0.8 SD below for receptive language.²⁶ Children born extremely preterm seem particularly vulnerable, with reports of a 10-fold higher odds of a moderate to severe language impairment.²⁷

Long-term outcome studies of children born very preterm often assess behavior and emotional status using parent-reported questionnaires. While differences are rarely found between children born very preterm and term on scales of externalizing behavior problems (i.e., aggression, conduct problems, defiance), there is a tendency for parents of children born very preterm to report more internalizing (i.e., anxiety, depression, social withdrawal)¹² and attention problems. Similarly, in adulthood, those born very preterm self-report more internalizing difficulties and fewer externalizing difficulties than term born peers.²⁸ Consistent with these findings, higher rates in specific psychiatric disorders have been reported in those born very preterm when structured clinical interviews have been conducted.^29,30 In a recent individual participant meta-analysis, those born very preterm were 10 times more likely to meet criteria for a diagnosis of autism spectrum disorder (ASD), 5 times more likely to meet criteria for attention-deficit hyperactivity disorder (ADHD), and twice as likely to meet criteria for anxiety.³¹

Given the increased risk for deficits across cognitive, motor, and behavioral domains, it is not surprising that children born very preterm are more likely to have academic difficulties. Multiple meta-analyses have reported poorer performance on standardized tests of reading and maths in children born preterm compared with term controls, especially those born very and extremely preterm, with group differences ranging from 0.5 SD to 0.8 SD.^12,32,33 Related to these findings, studies have reported a marked increase in the rate of children born very preterm who are receiving extra support and resources at school.^34,35

This summary of the long-term outcomes following very preterm birth focuses on group-level findings and paints a pretty grim picture. However, it is important to note that the profile and severity of neurodevelopmental impairments for each child are unique. For example, a child may present with CP and intellectual impairment, another with ADHD and learning impairment, and a third with DCD and language impairment. There is also a significant proportion of children born very preterm who present with no developmental concerns. To demonstrate the considerable variability in outcomes at the individual level, the next two sections will highlight the importance of (1) analyzing the true rate of impairments and not just focusing on mean group differences and odds ratios and (2) utilizing person-centered analyses to examine profiles of functioning.

Caution is needed when interpreting mean group differences as they can mask the breadth of outcomes at an individual level within a group. To illustrate this point, let’s examine the findings from an 8-year follow-up of a cohort of children born extremely preterm or extremely low birth weight in the state of Victoria in 1997.³⁶ This cohort was assessed on measures of IQ, academic achievement and behavior problems, with their functioning compared with a group of children born term and normal birth weight, matched on birth hospital, due date, sex, mother’s country of birth and health insurance status. The difference between the preterm and control groups for full-scale IQ was 12.5 points, or 0.83SD, which is widely considered a large effect.³⁷ Based on this finding alone, it is tempting to assume that a large proportion of this extremely preterm/extremely low birth weight cohort will have an intellectual impairment, which in turn could influence clinical decision-making and the way information is communicated to families. However, when using the control group distribution to classify intellectual impairment, nearly 50% of the preterm cohort had an IQ in the average range or above, and when using the test’s norms to classify intellectual impairment as per normal clinical practice, over 75% of the preterm cohort had an IQ in the normal range or above.³⁶ Based on the control group distribution, only 15% of the preterm group had a major impairment (defined as <–2SD) and an additional 36% had a mild impairment (defined as between –1 and –2 SD); 13% of the control group had an intellectual impairment. This observation is not unique to IQ. In the same study, the extremely preterm/extremely low birth weight group performed 8 and 9 points lower on standardized tests of reading and spelling, respectively than the control group (0.5–0.6 SD).³⁶ Yet more than 70% of the preterm group performed in the average range or above for reading and maths, even when using a strict impairment classification. This data underlines the heterogeneity observed in long-term outcome, with the very preterm population having outcome distributions similar to that of the general population, albeit with a downward shift. We suggest that understanding the true rate of impairment, as described in this example, is equally, if not more, valuable than knowing that the preterm group had mean IQ deficit of 13 points.

Similarly, care is needed when evaluating odds ratios and relative risks, especially for low-prevalence disorders. A landmark study examined the association between preterm birth and psychiatric disorders using hospital psychiatric admissions for all individuals born in Sweden between 1973 and 1985 (n = 1,301,522).³⁸ In comparison to individuals born at term, individuals born very preterm had a 2.5-fold higher risk to be hospitalized for psychosis, 2.9-fold higher risk to be hospitalized for depressive disorder, and 7.4-fold higher risk to be hospitalized for bipolar disorder. On the surface, these findings are alarming, and in isolation can significantly distort their significance. This whole population data-linkage sample comprised 5125 individuals born very preterm, and the actual number hospitalized for psychosis, depressive disorder, and bipolar disorder was only 6 (0.12%), 22 (0.43%), and 4 (0.8%), respectively. Therefore while the relative risk of being hospitalized for these psychiatric disorders is higher for those born very preterm, the true rate is very low (<1%) and this context is crucial for reassuring families regarding the long-term prognosis for their children.

In summary, while the very preterm population perform significantly lower than term peers and are at increased risk for impairment across most neurodevelopmental domains, at an individual level, a significant proportion of the children are functioning within normal expectations and only a small minority experience severe impairment.

Clinicians monitoring the development of children born very preterm appreciate the heterogeneity in long-term neurodevelopmental outcomes in this population; however, this is difficult to illustrate when focusing on group-based data. Over the past 10 years, a number of groups have been applying person-centered multivariable approaches to investigate the existence of distinct subgroups of children born very preterm based on their neurodevelopmental profiles. While the studies vary in terms of the outcome domains of interest, the age of assessment, and sample size, all have reported marked heterogeneity with the identification of between 3 and 8 subgroups.^15,39–43

The French EPIPAGE-2 study applied Latent Profile Analysis (LPA) to 1977 children born very preterm assessed at 5 years of age using a wide range of outcome domains including motor, cognitive, behavioral, and social functioning (15 outcome variables).⁴³ Four distinct profiles were identified for the very preterm cohort, the largest labeled as “favorable outcomes” (45%), with mean scores all above the test mean. The remaining three profiles included a subgroup that exhibited reasonable motor and cognitive functioning but specific behavioral difficulties (14%), a subgroup with mild cognitive and motor difficulties but good behavior and social functioning (31%), and the final profile reflected marked motor and cognitive difficulties. These findings highlight the considerable heterogeneity in the type and severity of impairments for children born very preterm.

Some studies have applied similar clustering approaches to neuropsychological data. For example, the ELGAN study is a large U.S. longitudinal observational study of children born before 28 weeks’ gestational age.³⁹ Using LPA with nine measures of attention and executive function outcomes at 10 years of age, four different profiles or subgroups were identified. One-third of the preterm cohort had a “normal” cognitive profile, represented by group means in the average range for all outcomes. The largest subgroup (41%) was labeled “low-normal”, with means falling between 0.5 and 1.0 SD below the test mean. The other two subgroups were children classified as having “moderately impaired” and “severely impaired” functioning, representing only 17% and 8% of the cohort, respectively. A study from Victoria, Australia, used LPA to examine distinct attention subgroups based on assessments performed at 7 and 13 years.¹⁵ Three subgroups were identified at both timepoints, with the largest subgroup consisting of children exhibiting age-appropriate attentional skills, with the other two groups exhibiting suboptimal attention across different domains. Taking advantage of the longitudinal data, the very preterm children were categorized into four transition groups: (1) stable average attention (35%), (2) stable low attention (25%), (3) improving attention (23%), and (4) declining attention (17%).

At a group level, children born very preterm are reported to exhibit emotional and behavioral difficulties in the areas of anxiety, attention, and peer/social relations, leading to this pattern being labeled the “preterm behavioral phenotype.”⁴⁴ Recently, a profile analysis was performed on a Victorian cohort of 8-year-old children born extremely preterm to determine whether the difficulties attributed to this phenotype co-occur at an individual level based on the four problem scales from the Strengths & Difficulties Questionnaire (SDQ).⁴¹ The analysis revealed four profiles, including one profile that was similar to the “preterm behavioral phenotype”, although this represented only 20% of the children born extremely preterm. The majority of the children (55%) were classified in a profile that reflected minimal emotional or behavioral problems. Therefore while there is strong evidence that children born very preterm are at increased risk for difficulties in attention, anxiety, and peer/social relations, this cluster of difficulties rarely co-occur within individual children.

In summary, there is considerable evidence from studies using techniques such as profile analysis that there is significant heterogeneity in the neurodevelopmental outcome for children born very preterm. These studies demonstrate that a significant proportion of this population perform within or above expected levels across neurodevelopmental domains, and for those who are exhibiting difficulties, the pattern varies both in terms of the severity and the type of deficits.

The neurodevelopment of children born very preterm is typically monitored by structured follow-up programs in early childhood but formal surveillance rarely extends beyond the preschool period. This is inadequate as many important cognitive and behavioral attributes do not emerge until the preschool period, followed by an extended developmental trajectory throughout childhood and into adolescence. As such, many skills and behaviors are difficult to reliably assess in early childhood. For example, the Bayley Scales for Infant and Toddler Development is only moderately predictive of school-aged outcomes.⁴⁵

Understanding whether early neurodevelopmental issues reflect a delay or an ongoing deficit has important ramifications as to how to manage a child with developmental delay. There is considerable evidence that cognitive and behavioral difficulties are present in adults born very preterm,⁴⁶ suggesting that these children are unlikely to grow out of the neurodevelopmental problems that surface in early childhood. Furthermore, longitudinal studies have generally reported that the group differences observed between children born very preterm and term remain relatively stable at the group level, with little evidence of neurodevelopmental challenges resolving with age among survivors of very preterm birth.^47,48 However again, group-level data may mask significant changes occurring at the individual level.

A longitudinal study in Christchurch, New Zealand assessed the IQ for 110 children born very preterm and 113 children born at term at 4, 6, 9, and 12 years of age.⁴⁷ Despite some shift in IQ measures across these timepoints, the mean IQ for both groups remained very stable (mean IQ for very preterm group ranged from 94.7 to 96.6; mean IQ for term group ranged from 104.0 to 106.9), and as such the magnitude of the group difference also remained constant across these four timepoints (Cohen’s d ranged from 0.6 to 0.9). Linear mixed-effects growth curve analysis was performed, which confirmed that the pooled trajectory for the groups was stable. However, the model also found evidence for marked individual variability, demonstrating that profiles at an individual level often do not mirror developmental profiles based on group-level data.

The inconsistency between group-level and individual-level developmental trajectories has been observed in outcome domains other than IQ. A Melbourne study examined language functioning in a cohort of children born very preterm and term at 2, 5, 7, and 13 years of age.⁴⁸ The linear effects model demonstrated (1) that the term group scored higher than the very preterm group at each timepoint, (2) the overall performance of both groups was remarkably similar at all timepoints resulting in very stable trajectories, and (3) there was no evidence that the language developmental trajectories differed between the very preterm and term groups. Further analysis of this dataset was undertaken to investigate the presence of distinct subgroups of children based on different developmental trajectories, with the latent growth mixture modeling revealing five profiles.⁴⁹ Approximately 30% of children born very preterm had consistently low to very poor language functioning between 2 and 13 years. The remaining children had better language functioning, although there was a subgroup who had very stable functioning, a subgroup who displayed strong early language skills that decreased across development, and another subgroup who displayed increasing language development between 2 and 7 years.

Instability across development at an individual level has also been observed in mental health disorders. The rate of any DSM-5 disorder was found to be 24% at age 7 years and 28% at age 13 years in a cohort of children born very preterm.⁵⁰ For specific disorders at 7 and 13 years, the rate of ADHD remained stable at 11%, while anxiety (11%–14%), mood (1%–3%), and autism (3%–6%) disorders increased marginally. Importantly, however, there was a considerable proportion of children who either moved into or out of a mental health diagnosis group. There are numerous explanations for this instability including a large group of subthreshold children who are on the borderline for a diagnosis. The instability may also be explained by normal maturational changes such as reductions in separation anxiety with age but an increase in generalized anxiety disorder and social anxiety during adolescence and early adulthood. Furthermore, the functional implications of subtle impairments, which are relevant when making a mental health diagnosis, may not become obvious until later in development with an increase in social and functional demands.

While the magnitude of group differences and rate of impairments may remain constant across childhood for the very preterm population, at an individual level, there is marked variability with some children displaying significant catch-up while others displaying decline. The nature of problems may also vary with age. For example, early expressive language delay may evolve into difficulties with higher-level pragmatics and discourse in adolescence. This highlights the care needed when counseling families regarding the long-term expectations for their child, as well as the importance of ongoing surveillance throughout childhood and into adolescence.

Long-term outcomes studies of children born very preterm generally report on cohorts born 10 years or more earlier, and during that time, substantial improvements in perinatal and neonatal care may have occurred. While we have no choice other than to translate those findings of past cohorts to contemporary cohorts, an understanding of how medical advances have altered long-term neurodevelopmental outcomes is important. In particular, it is unwise to assume that long-term neurodevelopmental outcomes have improved alongside decreasing mortality because (1) it is possible that with more high-risk infants surviving, an increase in impairment will be observed, and (2) interventions that enhance survival may be associated with adverse long-term effects.^51,52

There are only a few groups that have assessed long-term outcomes in geographic cohorts of children born early across multiple eras. The Victorian Infant Collaborative Study (VICS) group is unique in that they have adopted the same methodology to assess outcomes of children born extremely preterm/extremely low birth weight in four discrete eras, 1991–1992, 1997, 2005, and 2016–2017, alongside contemporaneous term/normal birth weight controls. A recent paper comparing the 2-year outcomes across these eras reported that the rate of CP has declined from 11%–12% in the 1990s to 6% in 2016–2017.¹ While there was no decline in developmental delay and neurodevelopmental disability over this 25-year period, there was a linear improvement in survival without major disability. Similarly, the EPICure group reported an 11% increase in survival without disability between 1995 and 2006 in infants born <26 weeks’ gestation in England,⁵³ and the EPIPAGE group reported a 7% increase in survival without neuromotor or sensory disability and a 3% decrease in rate of CP between 1997 and 2011 in infants born very preterm in nine regions of France.⁵⁴ Thus when taking survival into account, there is considerable evidence to suggest that early neurodevelopmental outcomes have improved with advances in medical care.

However, as already noted, tools that assess neurodevelopment in early childhood cannot evaluate skills that are yet to emerge, and are not strongly predictive of long-term outcomes. While there are only a few reports of change across eras in terms of long-term outcomes, based on research to date, there is little evidence to suggest improved functioning in more recent cohorts. The VICS group examined IQ in the extremely preterm group at age 8 years relative to the matched control group in the 1991–1992, 1997, and 2005 cohorts and found that the IQ deficit remained relatively stable across eras after adjustment (1991–1992: –8.0; 1997: –10.5; 2005: –10.2).⁵⁵ There was also no improvement in academic functioning across eras, and in fact, there was some evidence that relative to term controls, the deficit in reading, spelling, and mathematics was greater in more contemporary cohorts.⁵⁵ The VICS group also reported an increase in executive function impairments at age 8 years for the 2005 cohort in comparison with the earlier cohorts, in particular in the working memory and the planning/organization domains.⁵⁶ Similarly, motor impairment increased from 23% in 1991–1992 to 37% in 2005⁵⁷ which was due to an increase in non-CP motor impairment and not an increase in CP. At 11 years of age, the EPICure group also found no significant improvement between the 1995 and 2006 cohorts in IQ and academic achievement for infants born <26 weeks’ in England.⁵⁸

While survival without early disability seems to be improving, there is currently no evidence that advances in care are associated with better long-term neurodevelopmental outcomes. More research is needed to investigate changing neurodevelopmental outcomes across eras, especially mental health and adaptive outcomes in adolescence and early adulthood.

Moving forward, we need to understand the reasons for the marked variability in long-term outcomes in children born very preterm. Knowing the key factors associated with good and poor outcome across different domains will enhance the capacity to identify high-risk children early in life, enabling surveillance and interventions tailored to the needs of individual children to be initiated in early childhood and before the emergence of problems. At present, the capacity to predict developmental outcome following very preterm birth is poor. Gestational age, birth weight, small for gestational age, bronchopulmonary dysplasia, sepsis, necrotizing enterocolitis, and brain pathology are examples of neonatal clinical factors associated with neurodevelopmental outcomes, yet independently and collectively these clinical factors are unable to accurately predict long-term outcome.⁵⁹ This may be explained by the critical role of genetic predisposition, parenting, and the social environment in shaping child development, such that prediction models will always lack precision unless these additional factors are also included.

At a population level, social factors such as socio-economic status, maternal education, and family structure are known to strongly influence child development.^60,61 This is also the case for children born very preterm such that these factors are likely to explain some of the heterogeneity in long-term outcome.^62–65 Determining the independent role of social factors on child neurodevelopment is challenging as they are interrelated and often co-occur.⁶⁶ Research to date suggests that social risk factors have both independent and additive effects on long-term developmental outcomes in children born very preterm. For example, a study with children born very low birth weight found higher maternal education, higher paternal education, and caregiver employment were all associated with an increase in cognitive functioning over time of 3 to 4 points, but the effect increased to 11 points for children in which all three social factors were present.⁶⁷

The combined effect of clinical and social-environmental factors on cognitive and academic outcomes in children born very preterm has been examined across development.⁶⁸ Neonatal clinical factors associated with poorer neurodevelopment included lower gestational age, small for gestational age, male sex, Grade 3/4 IVH, postnatal corticosteroid treatment, and neonatal surgery, while important social factors included lower social class, maternal education, and language spoken at home. There is some evidence that the contribution of neonatal clinical factors on neurodevelopment diminishes with increasing age, while the impact of social factors is cumulative with the contribution increasing throughout development.⁶⁸ When examining language outcomes of children born very preterm, the contribution of social-environmental factors far outweighs the contribution of neonatal clinical factors, with parenting explaining the most variance.⁶⁹

Very preterm birth affects the whole family, and there is strong evidence of the significant distress experienced by parents that can endure for many years.^70,71 The impact on the broader family needs to be considered when evaluating the long-term consequences of very preterm birth, especially given that parental mental health and parenting are likely to influence child outcome.⁷² In fact, the family environment, including parenting practices and behaviors, is likely to be the strongest predictor of child development even in high-risk children such as those born very preterm. More research is needed investigating family outcomes and how environmental factors can play a protective role in supporting the child’s development.