Relative Age and Psychotropic Drug Use in Preterm and Term-Born Children and Young Adults

The study was based on a linkage between the Medical Birth Registry of Norway (NMBR),²⁰  the Norwegian Prescription Database,²¹  and Statistics Norway,²²  using the unique Norwegian personal identification number. The NMBR includes all Norwegian citizens and provides maternal and perinatal variables. The Norwegian Prescription Database provides information about all prescribed drugs dispensed by pharmacies. Information on education was collected from Statistics Norway. We followed individuals born from 1989 through 1998, with registered gestational age (GA) between 23 + 0 and 42 + 6 weeks, who had no registered congenital birth defects, were alive at 10 years, and had registered maternal variables. Only individuals with birth weights more than 400 g and birth weights considered likely for GA were included.

The study was assessed and approved by the Regional Committees for Medical and Health Research Ethics.

Individuals were followed between 2004 and 2016 with annual registrations, from the year they turned 10 until the year of their 24th birthday, emigration, or death, whichever occurred first.

Gestational age (in week + days) was categorized in 2 groups according to the mother’s last menstrual period; preterm (GA 23 + 0 to 36 + 6) and full term (GA 37 + 0 to 42 + 6). For sensitivity analyses, GA was further subcategorized in 4 groups (GA 23 + 0 to 33 + 6, GA 34 + 0 to 36 + 6, GA 37 + 0 to 38 + 6, and GA 39 + 0 to 42 + 6).

Relative age was measured by month of birth in the year and categorized in 2-month intervals. Individuals born in January/February were defined as relatively older, having a high relative age, whereas those born in November/December were defined as relatively younger, having a low relative age.

Outcomes were defined according to the Anatomical Therapeutic Chemical system: N06B psychostimulants for ADHD. Secondary outcomes were prescription status of 4 other categories of psychotropic drugs: N06A antidepressants, N05CD/N05CF/N05CH hypnotics and sedatives, N05B anxiolytics, and N05A antipsychotics (Supplemental Table 2). We registered prescription status (one [or more] prescription(s) vs no prescription) for each outcome every year from age 10 to 23 years.

We included covariates considered as potential confounders in the relationship between relative age and GA and mental health. Child variables collected from the NMBR included birthyear, birth weight, multiple birth and sex. We created a z-score for birth weight according to Marsal et al’s fetal growth standards,²³  and identified individuals with birthweights more than 6 SDs below or more than 3 SDs above the z-score (mean value), according to GA. Maternal variables, including parity and relationship status, were also collected from NMBR. Details of covariates are presented in Supplemental Table 3.

We used generalized estimating equations logistic regression models to compare the use of psychostimulants (primary outcome) per year from age 10 to 23 between exposure groups with different time of birth in the year (relative age) and different GA (preterm versus term). All analyses were repeated for each of the 4 secondary drug outcomes.

The primary analysis included the full study sample and assessed time of birth in the year in six 2-month intervals, with an interaction term between time of birth category and GA group to explore differences in the impact of relative age between term- and preterm-born individuals. Analyses were performed for males and females separately and adjusted for participants’ age (during follow-up), year of birth, and multiple birth status and mothers’ parity, relationship status, age in years and age in years squared, educational level, and county of birth. Estimates from the regression analyses were used to calculate and graphically present the percentage with the outcome in each exposure group, using average marginal effects (with covariates as observed).

In a subsample, we compared the group with the lowest relative age (born November/December) with the group with the highest relative age (born January/February) and assessed outcomes in 3 periods according to age in follow-up (10–14 years, 15–19 years, and 20–23 years) by adding an interaction term between period and relative age groups in the analyses. Analyses were performed for term and preterm separately and adjusted for participants’ year of birth and multiple birth status and mothers’ parity, relationship status, age in years and age in years squared, educational level, and country of birth.

We performed sensitivity analyses where preterm individuals were stratified into subgroups to explore differences in associations among very preterm and later preterm individuals.

All analyses were done using Stata statistical software version 15.1 (StataCorp).

A total of 488 470 individuals were included in the primary analyses (251 525 [51.5%] male participants and 6.1% were born preterm). Participants’ birth month was evenly distributed over the year. Figure 1 and Table 1 show the study population and population characteristics, respectively.

Figure 2 displays annual psychostimulant use by categories of birth month throughout the year, showing gradually higher proportions with psychostimulant prescription with increasing birth month from January/February to November/December for both preterm- and term-born males and females. Annual prescriptions for boys born in November/December was 1.0% (95% confidence interval [CI], 0.4–1.7) higher for preterm born and 1.3% (95% CI, 1.1–1.4) higher for term born, compared with boys born in January/February (corresponding odds ratios [OR], 1.37 [95% CI, 1.12–1.68] and 1.74 [95% CI, 1.63–1.86], P for interaction between birth month category and preterm status .36). Corresponding figures for girls were 0.5% (95% CI, –0.0 to 1.0) higher for preterm and 0.4% (95% CI, 0.3-0.6) higher for term born (ORs, 1.39 [95% CI, 0.98-1.96] and 1.43 [95% CI, 1.30–1.57], P for interaction between birth month category and preterm status .93); details provided in Supplemental Table 4.

The subsample used for further comparisons consisted of 152 725 individuals, 79 022 born in January/February (4741 [6.0%] were born preterm) and 73 703 individuals born in November/December (5076 [6.9%] were born preterm) (Supplemental Table 5).

Figure 3 shows ORs of psychostimulant prescription at ages 10 to 14, 15 to 19, and 20 to 23 years for the relatively younger (born in November/December) group compared with their relatively older peers (born in January/February) in term and preterm born (P for interaction between relative age, age group and preterm status <.001). Although OR of prescription for the relatively younger group compared with the relatively older group decreased with increasing age in the term population (from 1.80 [95% CI, 1.69–1.91] at ages 10 to 14 years to 1.17 [95% CI, 1.08–1.27] at ages 20 to 23 years), we observed a stable association over age in the preterm population (ORs, 1.39 [95% CI, 1.13–1.69] at ages 10 to14 years and 1.34 [95% CI, 1.00–1.78] at ages 20 to 23 years). The relative age effect over age/time differed between males and females among term born (P for interaction between relative age, age group, and sex <.001), but less so in the preterm population (P for interaction between relative age, age group, and sex .10) (Supplemental Fig 7).

The results indicated that relatively younger born late preterm (GA 34-36) have about 50% increased risk of psychostimulant use from ages 10 through 23 years when compared with their relatively older peers (Fig 4). However, the corresponding comparison for preterm born before week 34 did not indicate strong relative age effects, but estimates were imprecise because of a relatively low number in this group.

There were no relevant changes in prescriptions of the 4 other psychotropic drug groups with increasing birth month from January/February to November/December when studying the entire period from 10 to 23 years (Fig 5), neither for preterm nor term males or females. However, at ages 10 to 14 years, ORs were increased for prescription of several drugs for the relatively younger individuals, compared with relatively older peers, both in the term and to some extent in the preterm-born groups (Fig 6) (eg, ORs for antipsychotics at ages 10 to 14 years 1.39 [95% CI, 1.18-1.64] in term born and 2.43 [95% CI, 1.39–4.27] in preterm born). Such relative age effects were not present among the older age groups. Supplemental Fig 8 shows the corresponding results stratified by sex.

Our findings showed that young relative age was associated with higher psychostimulant prescription across ages 10 to 23 years. Overall, psychostimulant use was higher in preterm than in term born, and the relative age effect for psychostimulants was seen within the preterm- and the term-born group. However, although the relative age effect for psychostimulant prescription decreased over age for the term population, we saw a stable trend over ages 10 to 23 years for the preterm population. Relatively younger term and preterm groups were more often prescribed antidepressants and antipsychotic drugs at 10 to 14 years compared with peers born early in the year, but this did not persist at later ages.

Our findings show an explicit relative age effect for psychostimulant prescription in both preterm and term boys and girls. Earlier literature supports the same tendency, without taking gestation into account, finding that children born late in the academic year are more often diagnosed with, and more often prescribed medication for, ADHD.^12–15,24 

Earlier research on term-born individuals suggests that relatively younger age is also related to other adverse mental health effects.²⁵  A recent study including 10 million people found that low relative age was associated with diagnoses of anxiety, depressive disorders, ADHD, and with prescription of ADHD medication and antidepressants.²⁶  Other studies have found increased risk of depression,¹⁴  lower life satisfaction, more psychosomatic complaints, and increased risk of being overweight among relatively younger children.¹⁰ 

Our findings show relative age effects related to antidepressants and antipsychotics in ages 10 to 14 years for both preterm and term peers.

For the preterm born, we are not aware of earlier studies assessing the impact of relative age effects on mental health, but findings from studies on academic performance are relevant to consider because academic performance is linked to mental health and mental health disorders.²⁷  A British cohort study from 2013 conducted by Odd et al²⁸  looked at academic outcomes and showed a gradual reduction in scores on reading, writing, and mathematics from oldest to youngest students in class in ages 5 to 7 years, including for preterm individuals. The same authors found that special educational needs were maintained until the age of 16 years among preterm pupils enrolled in school based on date of birth, compared with those enrolled based on expected date of delivery.²⁹  Individuals born preterm experienced some catch-up on test results to their term peers during ages 5 to 16 years but did not totally close the gap.³⁰ 

Whether relative age effects persist into adolescence and young adulthood, indicating difficulties of a more chronic nature, is of essential concern. As children grow older and the developmental differences between those born late and early in the year become smaller, one also would expect diminishing relative age effects. Most of the earlier research in this field concludes that relative age effects for ADHD diagnosis and prescription are largest during early years of school.^15,31–33  Also, relative age effects related to internalizing symptoms, poorer peer relationships and mental health impairment decreased over time from ages 11 to 12 years⁹  in one study. However, other studies suggest sustained effects of relative age on mental health (eg, a Japanese study showing increased mortality from suicide at ages 19 to 21 years).³⁴  For the term population in our study, we observed decreasing relative age effects regarding psychostimulant prescription from ages 10 to 14 to ages 20 to 23 years, in concordance with most earlier studies. However, for the preterm population, we found a more stable relative age effect for psychostimulants across ages 10 to 23 years, supporting findings of the relatively long-term educational outcomes Odd et al found for preterm individuals.^28–30 

It is well known that adolescents born preterm are more prone to having an ADHD diagnosis.^3,4  Also, there is increasing evidence of a “preterm behavioral phenotype” associated with symptoms of anxiety, inattention, and social difficulties.^35,36  These vulnerabilities could possibly explain less resilience to suboptimal or inappropriate social and educational environments, and to the experience of coming up short compared with peers in terms of social skills and athletic and academic performance in preterm born. This could further contribute to lasting effects with reduced level of functioning.

For antidepressants, hypnotics, anxiolytics, and antipsychotics, our findings did not affirm sustained relative age effects through adolescence and into young adulthood. Although this could indicate that there is no lasting connection between relative age and mental health disorders other than ADHD, it is also possible that medication use is a less robust indicator of these disorders, especially in older adolescents and young adults.

In line with findings from numerous countries and earlier studies,^{12,13,33,37,38}  the total proportion of boys getting a psychostimulant prescription was higher than for girls in early ages (10–14 and 15–19 years), but the association between younger relative age and psychostimulant prescription was more pronounced among female than male participants. Surprisingly, this trend was not seen among the youngest preterm girls (10-14 years). A possible explanation for a small relative age effect regarding psychostimulant prescription among the youngest preterm girls could be that relative immaturity in this group for some reason to a lesser extent is interpreted as ADHD or rather interpreted as other types of psychopathology. However, relatively broad CIs in the preterm group does not support any firm conclusion from this observation.

A minority of studies, including from Denmark,¹⁸  show no relative age effects concerning ADHD diagnosis/medication. A suggested explanation for this is that a considerable proportion of children born late in the year delay school start in Denmark, thus being more mature at the time. Also, enrolling preterm pupils in school based on expected date of delivery instead of actual delivery date has shown to reduce the need of special education.²⁹  Findings suggest that children with initial learning difficulties predominantly linked to slow maturation and lack of self-regulation could benefit from delayed school enrollment.³⁹  On the other hand, there are studies suggesting that delayed school entry could deprive children with developmental difficulties from one important year of educational support.⁴⁰ 

Strengths of this study is the study design, with a large naturally selected population across all GAs, with complete follow-up using high-quality registry data over several years across youth.

Although we have adjusted for several perinatal and maternal covariates considered possible confounders in this context, GA is a complex phenomenon and residual confounding is still likely to be present. Although we cannot rule out that some people plan birth to a specific month of the year, we assume that birth month is more or less randomly distributed. We therefore consider confounding by GA as less important for the comparison between relatively younger and older groups.

Prescription of medication must be considered as one of several possible ways of measuring mental health in young people, reflecting a certain functional impairment (lack of sleep, anxiety, mood, hyperactivity/concentration), although it is not necessarily correlated with the prevalence of psychiatric disorders. However, psychostimulants in Norway must meet diagnostic criteria and be initiated by a specialist in child and adolescent psychiatry, pediatrics, or neurology, and an earlier Norwegian study has shown a high correlation between ADHD diagnoses and dispensed ADHD medication.¹² 

Finally, for subgroup analyses some ORs in the study are narrow, and although relevant at the population level, individual risk must be interpreted with caution.

Our findings suggest that the preterm population has sustained relative age effects, compared with term peers. Currently, the cause of this remains uncertain, but may be linked to the higher prevalence of developmental and cognitive difficulties in preterm children. More research into mechanisms for and interventions to reduce relative age effects in the preterm population is necessary. Nevertheless, our findings suggest need for approaches at various levels. Universal and system level approaches are needed to reduce relative age effects among school children in countries where such effects exist. Examples could be delayed/flexible school entry or more inclusive school practices. In addition, both health care and educational professionals should give particular attention to preterm children born late in the school year in the transition to school.

ADHD

attention-deficit/hyperactivity disorder

CI

confidence interval

GA

gestational age

NMBR

Medical Birth Registry of Norway

OR

odds ratio