Introduction
Attention deficit hyperactivity disorder (ADHD) is a neurodevelopmental disorder that can have debilitating effects on individuals throughout the lifespan. Clinical and population-based studies have repeatedly shown that both ADHD diagnoses and elevated ADHD symptoms are associated with increased risk of experiencing serious life outcomes, such as educational and occupational difficulties, traffic injuries, criminal convictions and other psychiatric disorders [
1‐
5]. Developing more effective ways of identifying individuals at risk for these serious life events later in life is important to prevent these adverse outcomes from occurring.
Childhood ADHD has an estimated prevalence of 5.3% (95% CI: 5.0–5.6%) worldwide and often persists into adulthood where the prevalence rate is around 2.5% (95% CI: 2.1–3.1%) [
6,
7]. While parents and teachers are used as main sources for establishing diagnoses in children, self-report becomes increasingly important during diagnostic interviews in adolescence and young adulthood in both clinical and research settings.
There is scarcity of research that has studied how well parent- and self-ratings of ADHD symptoms in adolescence predict adverse socioeconomic and health outcomes in adulthood. This is an essential issue to investigate, as prevention of serious outcomes later in life is an important task for clinicians working with ADHD. Results from a study using a clinical adolescent ADHD sample found that low academic achievers, compared to high academic achievers, displayed more ADHD symptoms, although group differences were larger for parent-ratings (medium effect size; Cohen’s
d = 0.60) than self-ratings (small effect size; Cohen’s
d = 0.26) [
8]. Another clinical study of young adults investigated how strongly parent- and self-reports of ADHD during interviews were associated with life events, including academic, occupational and criminal events, after accounting for reports from the other informant [
9]. The study found that parent-reports of ADHD symptoms were significantly associated with all events, while self-reports were only significantly associated to employer-rated ADHD and work performance. The findings suggest that parent-reports of ADHD in young adults are more strongly associated to life outcomes and thus have higher concurrent validity than self-reports of ADHD. Similar results were found in a clinical sample of females with ADHD, where parent-rated ADHD symptoms were significantly associated with a higher number of poor outcomes than self-ratings [
10]. A major limitation of these studies is that the majority of outcomes were rated subjectively, by the individuals, their parents or employers, which may have biased the results as associations may become inflated due to common method variance between predictor and outcome. Further, large-scale and longitudinal population-based studies that compare the predictive value of parent- and self-ratings of ADHD symptoms in adolescence are currently lacking.
Prospective studies have shown that ADHD in childhood predicts later adverse life outcomes [
5,
11,
12], but less is known about the predictive value of ADHD in adolescence and whether it changes across development. To our knowledge, no study has compared the predictive associations between parent- and self-rated ADHD symptoms across adolescence and life outcomes in adulthood. If one of the two source informants would be superior in predicting serious life outcomes, it would suggest that the source informant provides a more accurate assessment of the individual’s ADHD symptomatology and impairment, as ADHD diagnoses predict a range of adverse life outcomes [
1‐
5]. Prospective studies have for example shown that individuals with ADHD are at increased risk of poor educational and occupational performance [
1,
13], traffic injuries [
3], substance and violence related crimes and convictions [
5,
14] and disorders, such as substance use disorders (SUDs) and suicide attempt [
2,
25]. Further, if the predictive strength of ADHD ratings varies across ages, it would be valuable to identify when in development parent- and self-ratings have more or less accurate prognostic values. This in turn could be used to inform clinicians about which informant at which stage in development is more or less valid as a predictor of risk for adverse life outcomes.
In this longitudinal, population-based study, we aimed to examine the predictive associations of parent- and self-ratings of ADHD symptoms in adolescence on assessments of adverse socioeconomic and health outcomes from Swedish national registries. We aimed to (1) compare how well parent- and self-ratings of ADHD symptoms assessed in early and late adolescence predict academic, occupational, social and psychiatric outcomes in young adulthood and (2) examine whether parent- and self-ratings of ADHD symptoms independently predict these outcomes over and above the other informant, to examine whether the source informants provide any unique information. We predicted that parent-ratings of ADHD symptoms would more strongly predict adverse life outcomes than self-ratings based on previous research suggesting that parent-report of ADHD in adolescence has greater construct and concurrent validity.
Results
See Table
1 for the frequency of adverse socioeconomic and health outcomes and the number of parent- and self-rated ADHD symptoms in early and late adolescence. Adolescents rated their levels of ADHD symptoms as more severe than parents at both 13–14 and 16–17 years.
Table 1
Rates of adverse socioeconomic and health outcomes in young adulthood
Rates (%) of adverse socioeconomic outcomes |
No higher education | 1099 (73%) | 1163 (81%) | 2262 (77%) |
Unemployment | 71 (5%) | 61 (4%) | 132 (4%) |
Criminal conviction | 17 (1%) | 80 (6%) | 97 (3%) |
Rates (%) of adverse health outcomes |
Traffic-related injury | 78 (5%) | 79 (6%) | 157 (5%) |
Suicide attempt | 24 (2%) | 11 (1%) | 35 (1%) |
SUD diagnosis | 38 (3%) | 41 (3%) | 79 (3%) |
Mean (SD) ratings of ADHD symptoms at 13–14 years |
Self-rated | 3.86 (2.69) | 3.58 (2.71) | 3.72 (2.70) |
Parent-rated | 1.16 (1.86) | 1.58 (2.19) | 1.36 (2.04) |
Mean (SD) ratings of ADHD symptoms at 16–17 years |
Self-rated | 4.22 (2.80) | 3.38 (2.65) | 3.82 (2.76) |
Parent-rated | 1.12 (1.80) | 1.24 (1.96) | 1.18 (1.88) |
Predictive value of parent- and self-rated ADHD symptoms at 13–14 and 16–17 years on life outcomes in early adulthood
13–14 years
Figure
1 depicts the odds of experiencing each life outcome for individuals that score over the 90th centile on the distribution of ADHD symptoms compared to under the 90th centile at 13–14 years. For parent-ratings, the odds of being in an accident, being criminal, not completing higher education, and having a SUD (OR = 1.80–2.96) were significantly higher for individuals in the top 10% compared to the lower 90% of ADHD symptom distribution. For self-ratings, however, only the odds of having a SUD (OR = 2.56) was significantly higher for individuals in the top 10% compared to the lower 90% of the distribution.
The logistic regression analyses showed that parent-rated ADHD symptoms at 13–14 years significantly predicted not completing higher education, unemployment, criminality, unintentional injuries and SUD (OR = 1.12–1.21; Table
2). The discriminative accuracy (AU-ROC) of parent-ratings on these adverse outcomes ranged between 0.58 and 0.65 (Table
3). Self-rated ADHD symptoms at 13–14 years also significantly predicted not completing higher education, criminality and SUD, with slightly lower point estimates (OR = 1.07–1.17) than parent-ratings, but did not predict unintentional injury or unemployment, or any other outcomes (Table
2). The discriminative accuracy (AU-ROC) of self-ratings on the significantly predicted adverse outcomes ranged between 0.55 and 0.62 (Table
3). The discriminative accuracy when both parent- and self-ratings were included in the logistic regression model ranged between 0.58 and 0.65. The AU-ROC values increased slightly when self-ratings were added in the model compared to when only parent-ratings were included for unemployment and SUDs, however, these changes were not significant as the confidence intervals were overlapping (Table
3).
Table 2
Predictive value of parent- and self-rated ADHD symptoms across adolescence on adverse socioeconomic and health outcomes in early adulthood
13–14 years |
No graduate degree | 1.21 (1.12, 1.31)** | 1.18 (1.09, 1.28)** | 1.07 (1.03, 1.11)** | 1.03 (0.99, 1.08) |
Unemployment | 1.13 (1.05, 1.22)** | 1.16 (1.07, 1.25)** | 1.00 (0.92, 1.08) | 0.97 (0.89, 1.05) |
Criminality | 1.21 (1.11, 1.32)** | 1.20 (1.09, 1.32)** | 1.11 (1.01, 1.23)* | 1.02 (0.91, 1.15) |
Injuries | 1.12 (1.05, 1.20)** | 1.11 (1.02, 1.19)** | 1.07 (1.00, 1.14) | 1.04 (0.96, 1.12) |
Suicide attempts | 1.13 (1.00, 1.28) | 1.13 (0.97, 1.31) | 1.09 (0.95, 1.26) | 1.00 (0.85, 1.18) |
Substance use disorders | 1.15 (1.05, 1.26)** | 1.10 (0.98, 1.22) | 1.17 (1.06, 1.29)** | 1.10 (0.98, 1.23) |
16–17 years |
No graduate degree | 1.49 (1.35, 1.63)** | 1.44 (1.30, 1.60)** | 1.15 (1.10, 1.20)** | 1.06 (1.01, 1.12)* |
Unemployment | 1.16 (1.07, 1.25)** | 1.16 (1.06, 1.27)** | 1.03 (0.96, 1.10) | 0.99 (0.92, 1.07) |
Criminality | 1.29 (1.17, 1.43)** | 1.23 (1.06, 1.42)** | 1.15 (1.06, 1.25)** | 1.09 (0.95, 1.24) |
Injuries | 1.13 (1.04, 1.22)** | 1.11 (1.02, 1.22)* | 1.06 (1.00, 1.13) | 1.01 (0.93, 1.09) |
Suicide attempts | 1.19 (1.03, 1.37)* | 1.11 (0.93, 1.33) | 1.12 (1.00, 1.25)* | 1.10 (0.96, 1.26) |
Substance use disorders | 1.20 (1.07, 1.34)** | 1.12 (0.96, 1.30) | 1.14 (1.05, 1.25)** | 1.13 (1.01, 1.27)* |
Table 3
Discriminatory accuracy of parent- and self-rated ADHD symptoms on adverse socioeconomic and health outcomes estimated by the area under the receiver operating characteristic curve (AU-ROC)
13–14 years |
No graduate degree | 0.59 (0.56, 0.61) | 0.55 (0.52, 0.58) | 0.59 (0.56, 0.62) |
Unemployment | 0.58 (0.52, 0.64) | 0.50 (0.44, 0.56) | 0.60 (0.54, 0.66) |
Criminality | 0.65 (0.56, 0.73) | 0.59 (0.51, 0.68) | 0.65 (0.57, 0.74) |
Injuries | 0.61 (0.56, 0.66) | 0.56 (0.50, 0.61) | 0.61 (0.55, 0.67) |
Suicide attempts | 0.58 (0.47, 0.69) | 0.56 (0.44, 0.69) | 0.58 (0.46, 0.71) |
Substance use disorders | 0.59 (0.51, 0.68) | 0.62 (0.53, 0.70) | 0.60 (0.52, 0.69) |
16–17 years |
No graduate degree | 0.62 (0.60, 0.64) | 0.61 (0.58, 0.63) | 0.64 (0.61, 0.66) |
Unemployment | 0.60 (0.54, 0.65) | 0.52 (0.47, 0.58) | 0.58 (0.51, 0.65) |
Criminality | 0.69 (0.61, 0.77) | 0.63 (0.56, 0.71) | 0.72 (0.64, 0.80) |
Injuries | 0.57 (0.51, 0.62) | 0.55 (0.50, 0.61) | 0.56 (0.50, 0.62) |
Suicide attempts | 0.62 (0.50, 0.73) | 0.61 (0.51, 0.72) | 0.64 (0.53, 0.75) |
Substance use disorders | 0.60 (0.50, 0.69) | 0.62 (0.54, 0.71) | 0.65 (0.56, 0.74) |
16–17 years
Figure
2 depicts the odds of experiencing each life outcome for individuals that score over the 90th centile on the distribution of ADHD symptoms compared to under the 90th centile at 16–17 years. For parent-ratings, the odds of being criminal, not completing higher education and being unemployed (OR = 2.21–5.27) were significantly higher for individuals in the top 10% compared to the lower 90% of the ADHD symptom distribution. For self-report, however, only the odds of not completing higher education (OR = 1.81) was significantly higher for individuals in the top 10% of the distribution compared to the lower 90%.
The logistic regression analyses showed that parent-rated ADHD symptoms at 16–17 years significantly predicted all outcomes in early adulthood (OR = 1.13–1.49; Table
2). The discriminative accuracy (AU-ROC) of parent-ratings on the outcomes ranged between 0.57 and 0.69 (Table
3). Self-rated ADHD symptoms significantly predicted all outcomes except for unintentional injuries and unemployment (OR = 1.035–1.15; Table
2). The discriminative accuracy (AU-ROC) of self-ratings on these adverse outcomes ranged between 0.52 and 0.63. The discriminative accuracy when both parent- and self-ratings were included in the logistic regression model ranged between 0.56 and 0.72. The AU-ROC values increased slightly when self-ratings were added in the model compared to when only parent-ratings were included for all outcomes except for unemployment and accidents, however, these changes were not significant as the confidence intervals were overlapping.
Unique predictive value of parent- and self-rated ADHD symptoms at 13–14 and 16–17 years on life outcomes in early adulthood, over and above the other informant
13–14 years
Parent-rated ADHD symptoms at 13–14 years of age significantly predicted not completing higher education, unemployment, criminality and unintentional injuries when controlling for self-ratings (OR = 1.11–1.20). Self-rated ADHD symptoms did not significantly predict any outcomes in early adulthood when controlling for parent-ratings (Table
2).
16–17 years
Parent-rated ADHD symptoms at 16–17 years significantly predicted not completing higher education, unemployment, criminality and injuries (OR = 1.11–1.44), when controlling for self-ratings. Self-ratings of ADHD symptoms at 16–17 years significantly predicted not completing higher education and SUD (OR = 1.06–1.13) over and above parent-ratings (Table
2).
Sensitivity analyses
The pattern of results did not change when we excluded cases that had missing values of either parent- or self-ratings (Table S1) or when analyses were run on males and females separately (Table S2 and S3). However, higher self-ratings of ADHD symptoms in late adolescence uniquely predicted less unemployment while controlling for parent-ratings only in males. Further, both parent- and self-rated ADHD symptoms showed slightly higher predictive values on life outcomes in males than in females.
Parent-ratings showed similar associations with the outcomes with and without the two additional ADHD rating scale items, suggesting that differences between parent- and self-rated symptoms in their associations with life outcomes is not explained by differences in rating scale items (Table S4). There was, however, one exception; parent-ratings no longer significantly predicted unemployment over and above self-ratings when parent- and self-rating scales included the same items.
We ran additional sensitivity analyses where we dichotomized the ADHD symptom variables using cut-offs at the 95th centile to explore the odds of experiencing life events for individuals over or under the 95th centile. The overall pattern of results remained the same when we used the 5% cut-off as when the 10% cut-off was used (Table S5).
Discussion
In this longitudinal, population-based study, we found that both parent- and self-ratings of ADHD symptoms rated in adolescence predicted several important adverse life outcomes in early adulthood. In general, the associations between ADHD symptoms and outcomes were stronger in late compared to early adolescence, and parent-ratings of ADHD symptoms predicted several outcomes over and above self-ratings in both early and late adolescence. The findings suggest that clinicians can rely on both informant sources in adolescence for predicting risk of future adverse socioeconomic and health outcomes, although some caution should be placed on self-ratings of ADHD in early adolescence.
Past research has repeatedly shown that ADHD symptoms rated by parents in childhood are associated with adverse life outcomes, and our current results extend these findings to show that both parent- and self-rated symptoms in adolescence were significantly associated with adverse outcomes. For clinicians, these findings suggest that both informant sources in adolescence may provide valuable information for risk of serious outcomes in adulthood. However, we found that the discriminatory accuracy of both parent- and self-rated ADHD symptoms were low, suggesting that neither can be used to accurately discriminate between adolescents who will experience adverse outcomes from those who will not. Although the response rate of our twin study is high, non-responders are more likely to be male and have higher rates of ADHD symptoms in childhood [
28,
29]. Given that previous research indicates that associations between ADHD symptoms and behavioral problems increase along with increasing levels of symptom severity, this may suggest that our observed associations between ADHD symptoms and adverse socioeconomic and health outcomes are somewhat underestimated.
Parent-ratings of ADHD symptoms rated in early and late adolescence were generally more strongly associated with the life outcomes compared to self-ratings, although it should be noted that the confidence intervals of the ORs overlapped for most outcomes, suggesting that the difference between parent- and self-ratings in their predictive strength did not significantly differ. Further, parent-ratings predicted several outcomes over and above self-ratings, and the discriminative accuracy did not significantly improve when self-ratings were added to the models compared to when only parent-ratings were used. Our findings suggest that despite parent- and self-ratings of ADHD symptoms only showing modest correlations (0.36–0.37), self-ratings do not have added value beyond parent-ratings for most life outcomes. While multi-informant approaches (e.g., combining informant source ratings) are commonly used based on the belief that each informant provides unique and valuable information, our results suggests that self-ratings are not valuable once parent-ratings have been taken into account in predicting most outcomes, except for SUDs. One possible explanation for the finding that self-ratings significantly predicted SUDs beyond parent-ratings may be that individuals more likely base their ratings of ADHD symptoms on aspects of behavior related to substance use, which parents may not have as much insight into. Our findings are consistent with other lines of research suggesting that parent-ratings have higher concurrent and construct validity than self-ratings in adolescence [
30‐
32] and young adulthood [
10]. For example, one study found higher agreement between parent-ratings of ADHD and clinicians (during interviews with parents) than between self-ratings of ADHD and clinicians (during interviews with children and adolescents) [
30]. Another study found that parent-ratings of ADHD symptoms were more predictive of ADHD diagnostic group than self-ratings [
31]. Parent-ratings of ADHD have also been found to be more strongly associated with underlying objective (cognitive, neurophysiological and movement) indices of ADHD symptomology than self-ratings [
32]. Thus, even though our findings suggest that self-ratings provide some prognostic information, the data indicate that obtaining parent-ratings of ADHD symptoms should be the priority in child and adolescent clinical and research settings.
Both parent- and self-rated ADHD symptoms showed an increase from early to late adolescence in the number of statistically significant associations and also in the effect sizes of the associations with the life outcomes. This overall increase in predictive value is not likely explained by a decrease in endorsed ADHD symptoms with age as parent-rated, but not self-rated, symptoms decreased only slightly (from 1.38 to 1.18). Instead, the findings may reflect the shorter time interval in the predictions with increasing age, or that both parents and adolescents become more accurate in their descriptions of the adolescent’s ADHD symptomatology with age, although this second explanation is unlikely. Future work is needed to extend our findings and examine the predictive validity of parent- and self-rated ADHD in young adulthood. This would be informative in order to explore whether the predictive value of self-ratings increases at a specific point in development from adolescence to adulthood. This could be of great value for both clinicians and researchers when having to decide at which age self-report should be used as the main reporting source of ADHD symptoms.
Sensitivity analyses revealed that differences in results between parent- and self-ratings of ADHD were overall not influenced by (1) differences in missing values or (2) differences in items included in the two rating scales. When we re-ran analyses for sex separately, we found that the pattern of results remained the same, although both parent- and self-rated ADHD symptoms in males showed slightly stronger associations with life outcomes than in females. The stronger associations might reflect that boys tend to display more externalizing features of ADHD than girls [
33,
34], which may be more easily captured by rating scales.
It should be acknowledged that this study used a population-based sample, and therefore findings may not generalize to individuals with clinically diagnosed ADHD. Even though behavioral, clinical and etiological research converge in suggesting that ADHD symptoms are the extreme of a normal continuum of behavior in the general population [
35,
36], it would be informative to examine the predictive validity of parent- and self-ratings in a clinical sample, especially as individuals in this study rated their levels of ADHD symptoms as more severe than their parents, which is in line with population studies [
37] but inconsistent with clinical studies that tend to report higher levels of parent-rated symptoms [
31,
32]. Further, the predictive validity of ratings that we have studied may also reflect concurrent validity, as for some outcomes the time period for which the outcomes were indexed in the registers started one year after the 16/17-years assessment wave, and therefore might already have been present during the assessment. Another possible limitation is that the predictive value of ADHD symptoms on life outcomes should be interpreted with caution due to the modest effect sizes for the majority of associations. Further, as the AU-ROC values were low, we conclude that neither parent- nor self-ratings of ADHD symptoms could accurately discriminate between individuals who experienced adverse outcomes to those who did not.
Conclusion
In conclusion, we found that both parent- and self-ratings of ADHD symptoms in adolescence were significantly associated with adverse socioeconomic and health outcomes in early adulthood. Parent-ratings of ADHD from early to late adolescence were better predictors of serious outcomes, although self-ratings in late adolescence significantly predicted not having a higher educational degree and SUD over and above parent-ratings. Our findings suggest that both informant-ratings of ADHD in adolescence provide valuable prognostic information on risk of future adverse life outcomes. However, clinicians should prioritize parent-ratings over self-rating, especially in younger adolescents.