Relative age and ADHD symptoms, diagnosis and medication: a systematic review

Attention-deficit/hyperactivity disorder (ADHD) is a common childhood neuro-developmental disorder, characterised by three core symptoms: inattention, hyperactivity and impulsiveness causing an impairment in functioning [1]. Although epidemiological studies suggest that globally ADHD affects around 5% of school-age children, diagnosis and prescription rates are heterogeneous between countries [2‐7] and estimated prescribing rates vary, for example, from 0.9% in Denmark to 4.6–4.7% in Canada and Iceland [8].

The receipt of a clinical diagnosis of ADHD depends on evidence of symptoms affecting functioning in more than one setting, for example, both at school and at home. The diagnostic process, therefore, usually involves the collection of information from those who encounter the child in different contexts, for example, the child’s parents and school teachers, as well as the observations and interpretation of the health care professionals conducting the assessment [1].

There is an overwhelming literature documenting neurobiological, clinical and pharmacological evidence for the validity of ADHD as a diagnosis [9]. Despite the operationalization of the diagnostic process, since diagnosis involves clinical judgment, without an objective test, there remain a number of areas of debate within the literature [10, 11]. This systematic review focuses on one debate, whether relative age within the academic year affects the likelihood of a child being diagnosed with and/or receiving medication for ADHD. In many countries, there is a set age at which a child starts their first year of school, with a chronological date cut-off, e.g. 1st of September. This means that one child, born early in September may be 5 years old when the academic year starts, however, a child born at the end of August will only recently have turned 4 years of age. It would be expected that the older child will be more developmentally mature than the younger child, however, the academic and developmental expectations for the two children are likely to be similar, especially at school. The relative age effect is well evidenced within sport [12] and academic achievement [13]. It has also been studied with regard to child mental health problems [14]. There is no single accepted definition of younger relative age within the literature. Here we refer to children born at least in the younger half of the school year, however, others have defined this as the youngest one, two, three or 4 months of the year.

In relation to ADHD, the relative age effect has usually been demonstrated in countries with high prescribing rates for ADHD [15, 16], whereas findings from countries with lower prescribing rates have been mixed [2, 17, 18]. This area has important implications for diagnostic and prescribing practice as well as school entry policy. It has been argued that the relative age effect may represent the more immature behaviour of younger children being diagnosed and treated as ADHD and, therefore, more relaxed school entry policies may be able to offset this [15, 17]. This systematic review aims to investigate the strength of evidence for a relative age effect, distinguishing countries known to have higher and lower prescribing rates [19, 20]. It addresses three key questions—Is there an association between younger relative age, defined as being in the second half of the academic year, and: (1) the presence of high levels of ADHD symptoms, (2) receiving a clinical diagnosis of ADHD and (3) receiving medication for ADHD?

A literature search was conducted with the assistance of an information specialist. This covered articles published from the 1st of January 2000 to the search date of the 7th September 2017. Databases searched included: MEDLINE, EMBASE, PsycINFO, Web of Science, ERIC, Psychology and Behavioural Sciences Collection and The Cochrane Library. Search keywords comprised: (1) Various terms for ADHD including: Attention Deficit Disorder with Hyperactivity, ADHD, ADDH, ADHS, hyperkinesis, hyperactive* and inattention* and (2) Relative age, relative maturity, relative immaturity, young for grade, young for year, old for grade or old for year. Additional studies were identified through checking reference lists of obtained articles. A further update search using the same search terms and databases was conducted on the 23rd of November 2017.

A total of 123 references were retrieved through initial database searches and four through reference checking. A further two references were identified through the update search. After duplicates were removed, 63 abstracts were screened. Thirty records were excluded on the basis of: not relating to ADHD (9), response letters/reviews (4), case reports (5), no birth month information (8), adult data only (2) and tests of intervention (2).

The remaining 33 full-text articles were reviewed. A further 13 of these were excluded due to: no birth month information (1), ADHD not being separate from other child mental health disorders (1) and conference abstracts (11).

Twenty studies were assessed for the review and data extraction, the characteristics of these studies are shown in Table 1. Six of these could not be included in the quantitative synthesis initially due to: information not being presented on the total number of children, with and without a diagnosis/medication receipt [8, 19, 22, 24, 25] or comparisons only being made between the starting school age not age within the school year [26]. However, following communications with the authors, data were provided for three studies [8, 20, 25] and were therefore included. The PRISMA flowchart for study selection is presented in Fig. 1.

For the three studies which included measures of ADHD symptoms, two showed evidence of a relative age effect. This was from symptom reports from teachers [27, 35] with weaker evidence of an effect from parents’ reports [35]. In contrast, the third did not show evidence of a relative age effect in parent-reported levels of symptoms [25]. A meta-analysis was not possible since the results were not directly comparable due to the use of different types of rating scales, e.g. Strengths and Difficulties Questionnaire, social rating scales and Autism–Tics, ADHD and other Comorbidities inventory.

For studies investigating the proportion of children receiving a diagnosis and/or medication, data were extracted and meta-analyses were conducted, separating studies into subgroups based on the outcomes studied (diagnosis or prescription) and whether the country was known to have higher or lower rates of ADHD prescribing. However, heterogeneity estimates were too high, for an overall analysis (diagnosis I² = 97%, prescriptions I² = 95%) to be presented.

A meta-analysis of the studies investigating the proportion of children receiving medication in higher prescribing countries showed a significant relative age effect with those younger in the academic year being more likely to receive medication for ADHD (I² = 74%, RR 1.27 (1.19–1.35)), as shown in Fig. 2. However, the meta-analysis of studies reporting the proportion who received a diagnosis showed high heterogeneity (I² = 91%) and therefore is not presented here, Fig. 3 shows the risk ratio from each study.

For the other countries, heterogeneity estimates were too high for reporting of the meta-analysis for either diagnosis or medication, I² = 98% and I² = 97%, respectively, as shown in Figs. 4 and 5.

As shown within the risk-ratio plots, almost all studies have shown, to some extent, higher proportions of ADHD diagnosis and prescriptions amongst the youngest in the academic year. The studies which did not show a significant relative age effect were from Denmark (RR 0.91, 95% CI (0.86–0.96); [17] and RR 1.02, 95% CI (0.97–1.07); [18]), and one from Germany (RR 1.55, 95% CI (0.51–2.95); [35])). However, there is variation in the magnitude of the risk ratio estimates between different studies, e.g. one finding that children born in the youngest month of the academic year were over 1.6 times more likely to be diagnosed with ADHD (RR 1.61, 95% CI (1.46–1.78)) and to receive medication (RR 1.75, 95% CI (1.55–1.98)), compared with the children born in the month after the cut-off [29]. In comparison, a study from Israel showed an increased risk ratio of around 1.2 for the youngest third of the school year, compared with the oldest third (RR 1.17, 95% CI (1.12–1.23) [23].

This review has found that the majority of studies show evidence of a relative age effect influencing both the diagnosis of and receipt of medication for ADHD. This was demonstrated most clearly within studies from higher prescribing countries, with a modest pooled risk estimate of 1.27 for medication amongst the youngest in the school year compared with their older peers in the same school year. Data from the other countries were more mixed, with high levels of heterogeneity.

Differences between study results might reflect methodological differences. As shown in Table 1, studies differed by sample size, years studied, ages studied and methods of reporting and recording ADHD diagnosis and medication. However, a number of other factors may contribute to the differences found across studies.

First, as noted above, there are significant variations in the estimated rates of ADHD diagnosis and medication provision between countries [8]. Schwandt and Wuppermann [20] plotted the relative effect sizes of the relative age effect against diagnosis prevalence rates. They found a positive correlation, with countries with higher overall prevalence rates showing a larger relative age effect, an effect which was also shown between different regions in Germany. This suggests the possibility of misdiagnosis in relation to children with young relative age. However, this explanation does not fit with findings from a study using data from Finland, a country with low diagnosis and prescription rates but with evidence of a relative age effect [19].

Second, linked to the above, there are differences in diagnostic practice across countries, for example, which professionals are able to give a diagnosis [15, 17]. The culture of diagnostic practice within a country’s health system may have an influence on the relative age effects found.

Third, school entry regulations and policy may play a role. Some studies have highlighted the possibility of children being held back an academic year as a possible factor influencing the magnitude of relative age estimations and contributing to heterogeneity [15, 17, 29]. Not only do countries differ in their age of school entry, but also in the extent to which these regulations are adhered to. For example, in Taiwan, children may possibly attend school 1 year early because of an arrangement between parents and teachers, purposefully rendering them the youngest [29]. In the US in the mid-1990s, around 10% of pupils delayed entry into kindergarten, this was more common for boys and for those with developmental delay [15]. In comparison, in Denmark, only 60% of children born in the last quarter before the cut-off date are enrolled in school for their assigned year [17]. In Israel, parents often opt or are recommended by the child’s kindergarten teacher to delay the commencement of school [23]. However, in the study from Iceland [16] just 0.7% of children were estimated to be either a year ahead or behind. If immaturity being mistaken for ADHD is a cause of higher rates of diagnosis and medication amongst those youngest in the year, encouraging greater awareness of this amongst parents, pre-school staff and clinicians may be useful in addressing the relative age effect. However, research is needed on the potential benefits and harms of holding children back a year, e.g. moving children to be the oldest within a year group could increase adults’ expectations of them [36, 37].

Fourth, teacher perceptions, in particular, may play a role. Elder [27] demonstrated that teachers’ ratings of ADHD symptoms showed a significant relative age effect, having a much greater magnitude than parents’ ratings. This suggests that teachers are more likely to compare children with others in the same school year rather than by chronological age and thus may contribute to the possible over-diagnosis of ADHD in younger children. In support of this finding, other studies have found no relative age effect in parental reports of symptoms or self-reported symptoms from adults with ADHD [25].

To our knowledge, this is the first systematic review of the association between relative age and ADHD. This review has identified and brought together existing research in a rigorous and systematic manner, enabling meta-analyses of the data, where possible. However, there are a number of potential limitations. First, since there are a number of different ways in which the relative age effect can be flagged within a study’s title and abstract there is a risk that some studies may have been missed. Second, some studies did not publish their data in a form which could be inputted to the meta-analysis. This meant that some large studies within the field could not be included in the meta-analysis [19, 24]. Others excluded children born in the first or last months of the academic year due to their likelihood of being held back [23, 31], this may have introduced bias within the data. Third, high levels of heterogeneity meant that it was not always possible to conduct a meta-analysis.

This systematic review and meta-analysis has drawn together worldwide studies investigating the relative age effect in the symptoms, diagnosis and medication treatment of ADHD amongst children and adolescents. It has shown that the relative age effect is evident in the majority of countries, however, there is considerable variation in its magnitude. Possible explanations include: overall diagnostic rates, national differences in diagnostic practice and school entry regulations and the influence of different informant sources. In isolation, none of these theories are able to fully explain the differences shown. Further research is needed to better understand the reasons for the relative age effect. At an individual level it is crucial for clinicians to consider a child’s relative age when assessing for ADHD.