Attention-deficit hyperactivity disorder (ADHD) is a common neurodevelopmental condition, usually first apparent in childhood, and with core features of inattention, hyperactivity and impulsive behaviour. Reference Kim, Kim, Lee, Jeong, Lee and Lee1,Reference Posner, Polanczyk and Sonuga-Barke2 Historically, ADHD was thought to abate during late adolescence or early adulthood. Reference Laufer and Denhoff3 However, this assumption has been challenged in recent decades. There is now a large body of evidence from longitudinal studies showing that many children with ADHD have problems that persist into adulthood. Reference Di Lorenzo, Balducci, Poppi, Arcolin, Cutino and Ferri4 The long-term impairment associated with ADHD may be explained by the core symptoms of the disorder, associated deficits in emotional regulation Reference Shaw, Stringaris, Nigg and Leibenluft5 and executive functioning Reference Sadozai, Sun, Demetriou, Lampit, Munro and Perry6 or the emergence of coexisting problems including substance use disorder, Reference Mohr-Jensen, Müller Bisgaard, Boldsen and Steinhausen7 criminal offending Reference Mohr-Jensen, Müller Bisgaard, Boldsen and Steinhausen7,Reference Silva, Colvin, Glauert and Bower8 and psychosis. Reference Shyu, Yuan, Lee, Yang, Yang and Lee9
The 2021 Global Burden of Disease study reported considerable global disability from ADHD in young adulthood, with almost half of the estimated 85 million people affected being aged 15–40 years. Reference Ferrari, Santomauro, Aali, Abate, Abbafati and Abbastabar10 Because the presentation of ADHD may be more subtle in adults than in children, and is often overshadowed by coexisting problems, its true impact on disability may be underestimated. Reference Kooij, Bijlenga, Salerno, Jaeschke, Bitter and Balázs11 Increased awareness of this disability points to a need to measure the spectrum of outcomes experienced by people with ADHD. This would guide targeted support and secondary prevention efforts for young people with ADHD. It would also help identify subgroups of the adult population with an excess of undiagnosed or untreated ADHD – for example, people in contact with the justice system.
It remains unclear whether long-term outcomes in people with ADHD are explained by factors in the early family environment that give rise to ADHD. However, a study of sibling pairs that accounted for the effect of shared familial factors showed that children with ADHD had much worse long-term economic outcomes than their unaffected siblings. Reference Fletcher12 A further consideration is whether the long-term effects of ADHD are accounted for by other childhood externalising disorders (notably conduct disorder and oppositional defiant disorder (ODD)) that commonly coexist with ADHD. Few studies have examined this question. Reference Di Lorenzo, Balducci, Poppi, Arcolin, Cutino and Ferri4
In the present study, we use prospective data from a 40-year New Zealand birth cohort to estimate associations between adolescent ADHD symptoms and a broad range of mental health and psychosocial outcomes in early adulthood up to age 40 years. We aimed to examine associations after accounting for potential confounding by early individual and familial factors. Because childhood ADHD symptomatology commonly coexists with other externalising disorders that are associated with poor long-term outcomes, including conduct disorder and ODD, we also fitted models adjusting for symptoms of those disorders to estimate the independent contribution of ADHD symptoms.
Method
Study design and setting
The Christchurch Health and Development Study (CHDS) is a longitudinal cohort study that enrolled at birth 1265 people born in Christchurch, New Zealand in 1977. The study has assessed study members annually from birth to age 16 years, then at age 18, 21, 25, 30, 35 and 40, via survey interviews by trained lay interviewers, with a response rate between 75 and 80% of surviving cohort members in the last three assessment waves. The study has approval from New Zealand Southern Regional Health and Disability Ethics Committee (reference no. 16/STH/188/AM01). Study members and/or their parents gave written consent at each wave of data collection. A statement from Strengthening the Reporting of Observational Studies in Epidemiology is available in Supplementary Table 1 available at https://doi.org/10.1192/bjp.2025.10475.
Measures
ADHD symptoms (ages 14–16 years)
At ages 15 and 16 years, parental and self-reports about cohort members’ behaviour were obtained using the Revised Behaviour Problems Checklist, Reference Quay13 supplemented by self-report items from the Diagnostic Interview Schedule for Children, Reference Costello, Edelbrock, Kalas, Kessler and Klaric14 to assess whether cohort members had experienced symptoms of ADHD. From these reports, we created continuous measures of the number of ADHD symptoms reported by summing relevant items in the three scales across the two waves of data collection at ages 15 and 16 years. For the purposes of the present analysis (see Statistical analysis, below), we classified participants according to their level of ADHD symptoms by dividing the sample into those in the upper quartile of ADHD symptoms and those in the lower three quartiles. Those participants with six or more symptoms endorsed were classified as being in the top quartile for ADHD symptoms.
Conduct disorder and ODD symptoms (ages 14–16 years)
Also at ages 15 and 16 years, parent and self-report items concerning conduct disorder and ODD, derived from the two scales described above and augmented by parental self-report items on conduct problems from the Self-Report Early Delinquency Scale, Reference Moffitt, Moffitt and Mednick15 were used to create continuous measures of conduct disorder and ODD symptoms, created by summing relevant items across the two waves of data collection at ages 15 and 16 years. For the purposes of the present analysis, we classified participants according to their level of conduct disorder and ODD symptoms by dividing the sample into those in the upper quartile of conduct disorder/ODD symptoms and those in the lower three quartiles. Those participants with two or more conduct disorder symptoms, or three or more ODD symptoms endorsed, were classified as being in the top quartile for conduct disorder or ODD symptoms.
Substance use disorder outcomes (ages 16–40 years)
The Composite International Diagnostic Interview (CIDI) Reference Robins, Wing, Wittchen, Helzer, Babor and Burke16 was used to identify alcohol and cannabis use disorder (abuse or dependence) at each wave, for the period since the previous assessment of the cohort, from ages 16–18, 18–21, 21–25, 25–30, 30–35 and 35–40 years. Nicotine use disorder (past 30 days) was ascertained from custom-written items on cigarette smoking. Reference Fergusson, Goodwin and Horwood17 A history of regular illicit stimulant use (likely to have been predominantly amphetamines based on New Zealand drug market data) was defined as any period of at least weekly since the previous assessment. Any other illicit drug use was also recorded.
Criminal offending and victimisation (ages 18–40 years)
Offending behaviour over a 12-month period prior to each interview was evaluated using the Self-Report Delinquency Inventory, Reference Elliott, Huizinga and Klein18 supplemented by custom survey items. Behaviours were categorised as involving either property offences or violence. Intimate partner violence (IPV) perpetration and victimisation was assessed through the Conflict Tactics Scale, Reference Straus19 which included items pertaining to physical IPV, for the assessment periods 21–25, 25–30, 30–35 and 35–40 years. Data on arrests during each assessment period from ages 18–21 years to 35–40 were obtained via self-report.
Relationship breakdown, unemployment and socioeconomic indicators
At each interview from age 25 to 40 years, participants were asked whether they had experienced a period of unemployment lasting at least 3 months while actively seeking work at any point during a calendar year since their last assessment. A measure of relationship dissolution during each assessment period from ages 18–21 to 35–40 years was developed using annual life events data based on items from the Feeling Bad Scale. Reference Lewis, Siegel and Lewis20 At each interview from age 25 years, participants reported on their income in the past 12 months (equivalised to New Zealand dollars using the Purchasing Power Parity Index), 21 and whether they owned their own home. Living standards were measured with the Economic Living Standard Index (Short Form) Reference Jensen, Spittal and Krishnan22 at ages 30 and 35 years, which was updated by the Material Wellbeing Index Reference Perry23 at age 40 years.
Internalising mental health problems (ages 16–40 years)
Symptoms of DSM-III-R or DSM-IV anxiety disorders and major depressive disorder were assessed for each assessment period from ages 16–18 to 35–40 years using CIDI. Binary variables were created corresponding to anxiety disorder or major depression at each assessment period. The presence of suicidal ideation in each period was generated using the suicidal thinking item from the CIDI depression scale.
Prescribed stimulant medication use
At age 15 and 16 years, study members’ parents reported on cohort members’ regular prescribed medications. Data on stimulant medication dispensing in adulthood (age 25–40 years) were obtained by linkage to New Zealand pharmaceutical dispensing records, using cohort members’ National Hospital Identification Number.
Covariate factors
Covariate factors were selected from the CHDS database, based on a priori knowledge about their probable associations with the exposure and outcomes. The following individual and familiar factors were identified as potential confounders in estimates of the association between ADHD symptoms and adult outcomes: measures of family socioeconomic positioning in childhood (maternal age, maternal education, family living standards to age 10 years, socioeconomic status at birth, average family income ranking to age 10 years); family functioning (parental illicit drug use, parental offending, parental alcohol problems, changes in parental figures to age 15 years, and an omnibus count measure of family problems); exposure to physical and sexual abuse in childhood; and individual factors (gender at birth, childhood cognitive ability, school performance in middle childhood, meeting criteria for major depression/anxiety disorder in middle adolescence) and conduct disorder and ODD at ages 14–16 years (see above). Details about these measures are provided in Supplementary Material 1.
Statistical analyses
All analyses were conducted using Stata version 18.5 for Windows (StataCorp, College Station, TX; www.stata.com). As noted above, the measures of ADHD, conduct disorder and ODD symptoms were quartilised, with participants being classified according to whether they were in the top quartile for symptoms of each disorder over the period 14–16 years. Initial attempts to model the associations between ADHD symptoms and life outcomes (controlling for both conduct disorder and ODD symptoms) resulted in significant model instability when ADHD (and conduct disorder and ODD) were treated as either (a) diagnostic classifications or (b) symptom count measures, due to collinearity between the measures. It proved possible to fit stable and parsimonious models when comparing those in the upper quartile of each measure with those in the lower three quartiles.
In the first step of the modelling, bivariate associations between ADHD symptom quartile (age 14–16 years) and adult outcomes were tabulated, comparing those in the top quartile with those in the bottom three quartiles for ADHD symptoms. Next, we calculated associations between ADHD symptom quartile and covariate factors antecedent to, or concurrent in time with, the measurement of ADHD symptoms.
It is possible that any associations between ADHD symptom quartiles and life course outcomes might be explained by (a) family and individual factors from childhood that confound the associations between ADHD symptoms in adolescence and outcomes; and (b) co-occurring conduct disorder and/or ODD symptoms in adolescence. To examine these possibilities, a series of repeated-measures generalised estimating equation (GEE) models, with an unstructured correlation matrix and robust (sandwich) estimators, were fitted to generate population-averaged odds ratios for associations between ADHD symptom quartile and adult outcomes age 18–40 years: model 1, without covariate adjustment; model 2, with adjustment for antecedent individual and familial covariate factors; and model 3, with adjustment for antecedent individual and familiar covariate factors plus concurrent other externalising symptoms (being in the highest quartile for conduct disorder and ODD). All models were adjusted for age (wave). All childhood covariates were entered into the models in a single block.
Finally, we repeated the analyses following exclusion of the five individuals who had reported being prescribed stimulant medication for ADHD to age 40 years; exclusion of these cases did not materially alter the results of the analyses.
Sample size
Complete case analyses were based on the 995 cohort members (496 male, 499 female) for whom full data were available on ADHD symptoms in adolescence (ages 14–16 years) and outcomes over the period 16–40 years. This sample represented 81.6% of the surviving cohort at age 40 years. The ethnic composition of the sample was 82.2% New Zealand European and 17.8% New Zealand Māori.
Analyses suggested that there were small, but statistically significant (P < 0.05), tendencies for the final sample to underrepresent individuals from socially disadvantaged backgrounds, characterised by low maternal education and low socioeconomic status and family living standards (see Supplementary Table 2 for details). To address the possible bias in estimation process due to sample losses, the inverse-probability data weighting methods described by Carlin et al Reference Carlin, Wolfe, Coffey and Patton24 were used to examine the potential implications of selection effects arising from the pattern of missing data. These analyses produced essentially the same pattern of results as those reported here, suggesting that the conclusions of this study were unlikely to have been influenced by selection bias.
Results
Bivariate associations between ADHD symptoms and life course outcomes, ages 16–40 years
At age 14–16 years, 7.4% of cohort members met the criteria for ADHD. The number of symptoms of ADHD reported during the period 14–16 years ranged from 0 to 15, and those participants with 6 or more symptoms endorsed were classified as being in the top quartile for ADHD symptoms. As noted in the Method section, to facilitate analyses and improve model stability we dichotomised the sample into those with high ADHD symptoms (top quartile) and those with low symptoms (bottom three quartiles). Table 1 shows adult outcomes (percentages or means) for those with high and low ADHD symptoms. For all outcomes, there were statistically significant (P < 0.05) differences between the two groups. To age 40 years, those with higher ADHD symptoms had elevated levels of substance use problems, mental health problems (major depression, anxiety disorder and suicidal ideation), arrests, property and violent offending and IPV victimisation. Those with higher ADHD symptoms also had lower levels of employment, home ownership, relationship stability, income and living standards. Supplementary Table 3 provides data for each assessment period summarised in Table 1.
Table 1 Associations between attention-deficit hyperactivity disorder (ADHD) symptoms in adolescence (ages 14–16 years) and mental health and psychosocial outcomes (ages 16–40 years)
IPV, intimate partner violence.
Associations between ADHD symptoms and potential confounding factors
It could be argued that the associations shown in Table 1 reflect the influence of covariate factors from childhood and adolescence that increase the likelihood of both ADHD symptoms and the outcomes. Table 2 shows the associations between ADHD symptoms and a series of potential confounding factors obtained from the study database (see Method), and the mean scores or percentages for each confounding factor over ADHD symptom level (top quartile versus bottom three quartiles). The table shows that, apart from the participant’s gender recorded at birth, there were statistically significant (P < 0.05) differences between those in the top quartile for ADHD symptoms and those in the lower three quartiles on a range of individual and family measures, suggesting that those in the highest quartile were more likely to have been exposed to socioeconomic disadvantage, family instability and maladaptive parental behaviour, and abuse, than those in the lower three quartiles.
Table 2 Associations between attention-deficit hyperactivity disorder (ADHD) symptoms (ages 14–16 years) and potential confounding factors
SES, socioeconomic status.
a. P-values derived from Mantel–Haenszel test of independence for percentages, and F-tests from generalised linear models for means.
Adjusted associations between ADHD symptoms and life course outcomes (ages 16–40 years)
To examine the role of potential confounding in the relationship between ADHD in adolescence and adult outcomes, a series of three GEE models were fit to the associations between ADHD and each outcome. Table 3 shows the results of the three models for each outcome: model 1, in which outcomes are modelled as a function of ADHD and age; model 2, in which outcomes are modelled as a function of ADHD symptoms, age and the set of confounding factors; and model 3, which in which outcomes are modelled as a function of ADHD symptoms, age, confounding factors and co-occurring conduct disorder and ODD. Figure 1 shows the covariate adjusted means and percentages for each outcome, classified by ADHD symptoms level. Table 3 shows the following results:
-
(a) The results of model 1 parallel the findings shown in Table 1. They show that ADHD symptoms were significantly associated with poorer adult outcomes across all domains.
-
(b) After adjusting for confounding factors (model 2), all substance use and offending outcome measures remained significantly (P < 0.05) associated with ADHD symptoms, with odds ratios ranging from 1.96 to 2.86. For mental health outcomes, the size of effects attenuated, and only major depression remained significantly associated (P < 0.05) with ADHD symptoms. None of the economic outcomes were significantly associated with ADHD symptoms after controlling for confounding.
-
(c) In models adjusted for both confounding factors and co-occurring conduct disorder or ODD, the association between ADHD symptoms and mental health and socioeconomic outcomes further attenuated, with odds ratios close to 1. However, higher ADHD symptoms remained associated with elevated odds of substance use and criminal offending outcomes, as well as physical IPV perpetration. Odds ratios ranged from 1.35 to 1.48 for substance use, 1.49–1.57 for offending and 1.67 (95% CI: 1.03–2.66) for physical IPV perpetration. As shown in Supplementary Table 4, for these outcomes the magnitude of associations for ADHD symptoms was similar to that for ODD symptoms, while effect sizes for conduct disorder symptoms were in general larger than for either ADHD or ODD symptoms across most outcomes for substance use, mental health and offending.
Fig. 1 Predicted percentages and mean values for life outcomes, by attention-deficit hyperactivity disorder (ADHD) quartile grouping. IPV, intimate partner violence.
Table 3 Associations between attention-deficit hyperactivity disorder (ADHD) symptoms (ages 14–16 years) and life course outcomes to age 40 years, before and after control for confounding, and co-occurring conduct disorder and oppositional defiant disorder
IPV, intimate partner violence; GEE, generalised estimating equation.
a. Analysed using linear GEE models rather than logistic GEE.
Figure 1 shows the predicted percentages and mean values for all outcomes, classified by ADHD symptom grouping (top quartile versus bottom three quartiles), based on the fully adjusted model (model 3). The figure highlights large gaps in the predicted percentages of several outcomes, including those with nicotine dependence (34.2% for those in the top quartile as opposed to 14.8% in the lowest three quartiles) and violent offending (13.8% for those in the top quartile as opposed to 4.9% in the lowest three quartiles).
Sensitivity analysis: effects of stimulant medication use
Only one cohort member took prescribed stimulant medication (methylphenidate) at ages 15–16 years, while 4 were dispensed methylphenidate between ages 25 and 40 years; exposure to stimulant medication in the cohort was therefore minimal. However, to account for the possibility that the results of the study might have been influenced by those taking stimulant medication for ADHD, the analyses described above were repeated after leaving out the five participants who reported being prescribed stimulant medication. There was no change to the pattern of results in this analysis.
Discussion
ADHD symptoms in adolescence were associated with a broad range of adverse psychosocial outcomes up to age 40 years. These included mental health and substance use problems, criminal offending, relationship breakdown, unemployment and lower economic attainment. In general, effect sizes were moderate. This was a mostly untreated sample, and outcomes therefore reflect the natural history of ADHD symptoms.
In adjusted models, ADHD symptoms remained associated with substance use problems and criminal offending, with effect sizes in the range 1.4–1.6. The adjusted models included symptoms of ODD and conduct disorder. Given the very high rate of comorbidity between ADHD symptoms and these conditions, Reference Thapar, Harrington and McGuffin25 it is unsurprising that the inclusion of these covariates substantially attenuated the associations. Further elucidation of the contributions of ADHD, conduct disorder and ODD symptoms to later mental health and economic outcomes is not possible for our data because these conditions are so closely linked. Nonetheless, our findings add to a growing body of literature suggesting that young people with ADHD symptomatology often have problems that persist into adulthood. For example, it has been reported that ADHD is associated with worse long-term academic performance, Reference Arnold, Hodgkins, Kahle, Madhoo and Kewley26 social functioning Reference Harpin, Mazzone, Raynaud, Kahle and Hodgkins27 and economic achievement. Reference Fletcher12 Our study is also consistent with a recent meta-analysis showing that substance use problems and antisocial behaviours are among the most common long-term correlates of childhood ADHD symptomatology. Reference Di Lorenzo, Balducci, Poppi, Arcolin, Cutino and Ferri4
Expert opinion suggests that ADHD in adults has been underdiagnosed and undertreated. Reference Kooij, Bijlenga, Salerno, Jaeschke, Bitter and Balázs11,28 For example, in Australia a 2023 Senate Inquiry into ADHD found there was high unmet need for treatment, and the inquiry recommended increased funding. 28 A 2019 European consensus statement similarly noted that ADHD is a ‘lifelong impairing condition’. Reference Kooij, Bijlenga, Salerno, Jaeschke, Bitter and Balázs11 While our findings, in the New Zealand context, suggest that ADHD symptoms may be associated with certain adverse long-term outcomes, the nature of the relationships between ADHD symptoms and outcomes such as substance use, mental health, offending and economic challenges remains unclear. These associations are likely to be complex, with potential for dynamic and reciprocal influences between some areas of difficulty in adult functioning. Gaining a better understanding of how these patterns develop over time and across different stages of life would help identify the most effective points for intervention. For example, our findings suggest that the poor long-term economic outcomes among people who exhibit ADHD symptomatology are largely explained by early individual factors and the family environment. This suggests that a focus on these distal risk factors may have more impact on those outcomes than treatment for ADHD. On the other hand, data from a large Swedish study of violent reoffending following release from prison showed that stimulant medication dispensing for ADHD was associated with a large reduction in violent reoffending. Reference Chang, Lichtenstein, Långström, Larsson and Fazel29 ADHD is both prevalent among people in prison Reference Fazel and Favril30 and often not identified or treated. Reference Baggio, Heller, Perroud, Buadze, Schleifer and Wolff31 Therefore, if ADHD treatment does reduce violent reoffending, this would have important implications for both health service delivery in prisons and justice policy.
Our findings also have some implications for mental health clinicians. The high prevalence of coexisting substance use disorders, criminal offending and internalising disorders makes it challenging to identify ADHD in adults. ADHD symptoms may be overshadowed by those coexisting problems and, in some cases, there is symptom overlap (for example, heavy use of alcohol or cannabis causing problems with cognitive functioning, including inattention). Reference Foulds and Newton-Howes32 These diagnostic challenges show the importance of having well-qualified clinicians, using multiple information sources and taking a longitudinal rather than cross-sectional approach to diagnostic formulation. The presence of comorbid substance use disorder and antisocial behaviours not only makes ADHD more difficult to diagnose, but it can also complicate treatment. Clinicians are often reluctant to prescribe stimulant medications for these individuals because of concerns about the risk of misuse or diversion of the medication. However, if ADHD symptoms cause a person’s substance use or offending, effective ADHD treatment may help address those coexisting problems.
The limitations of this study are important to consider. This study captures the experience of a group of people who were born in New Zealand in the 1970s, and caution is needed in generalising the findings to other settings. Only 5 participants reported being prescribed stimulant medication for ADHD to age 40 years, which suggests that these findings may not generalise to largely medicated cohorts. Furthermore, we could not examine the extent to which these outcomes may have been influenced by autism spectrum disorders (ASD). At the time when the assessment strategy for the CHDS was developed in the 1980s, ASD were considered uncommon and therefore did not form part of the assessment regime. Also, retention in the study to age 40 years was over 75% but, while our methods were robust on the assumption that data were missing at random, excess dropout among those with ADHD would, if anything, have led to underestimation of the size of effects reported in this study. Similarly, controlling for school performance and cognitive ability, as we did in the present analyses, could be considered a case of covariate over-control; again, however, this would have led to an underestimation of the size of effects. In addition, we chose not to apply corrections for multiple testing in the present analysis. There is considerable debate in the literature regarding the use of corrections for multiple testing, particularly with highly correlated outcomes such as those in the present paper, and there is increasing belief that the corrections that are commonly used in the literature may be too conservative in nature, Reference Rubin33 strongly increasing the risk of type II error. Adult ADHD treatment, while increasing, remains difficult to access in New Zealand. Reference Beaglehole, Jarman and Frampton34 Even once treatment is accessed, adherence rates for stimulant treatment in adults are concerningly low. Reference Ahmed and Aslani35
In summary, ADHD symptoms are associated with a broad range of impairments in early adulthood. Children who have been diagnosed with ADHD, or who display significant ADHD symptomatology, require long-term monitoring in primary care settings, with secondary prevention of mental health and substance use problems and awareness of the common psychosocial problems that these young people are likely to experience. Our findings also suggest that policy-makers should prioritise population prevention strategies that address the long-term burden of morbidity among children who present with symptoms externalising problems such as ADHD.
Supplementary material
The supplementary material is available online at https://doi.org/10.1192/bjp.2025.10475
Data availability
Data are available upon request from the corresponding author.
Author contributions
J.A.F., J.M.B., J.A.K., K.M.D., M.P., J.T.Y., M.R.T., K.D. and R.P. contributed to the conception and design of the work. Data were obtained by J.A.F. and J.M.B.; data analysis was performed by J.M.B. All authors contributed to the drafting and review of the article, and approved of the version to be submitted. All authors are accountable for all aspects of the work.
Funding
This research was supported by the Health Research Council of New Zealand (Programme Grant no. 16/600).
Declaration of interest
The authors declare no conflicts of interest. J.M.B., K.D and R.P. are members of the Editorial Board of BJPsych Open.
Open access
eLetters
No eLetters have been published for this article.