Background
Attention-deficit/hyperactivity disorder (ADHD) is a neurodevelopmental condition that typically emerges in childhood and is characterized by inattentive and/or hyperactive symptoms that often persist well into adulthood [
1,
2]. While comorbid mental health problems have been well documented in adults with ADHD, less is known on physical health problems, especially in relation to age-related disorders, such as cardiovascular disorders (CVDs) (i.e., ischemic heart disease, heart failure, cerebrovascular disease, venous thromboembolism, tachyarrhythmias, and hypertension), and common metabolic disorders (i.e., type 2 diabetes, obesity, and hyperlipidemia) [
3]. CVDs are the leading cause of death and one of the major causes of disability in adults globally [
4], and the rising burden of metabolic disorders has become a major healthcare concern worldwide [
5,
6].
Emerging research has reported significant associations of ADHD with increased risk for a range of CVDs and metabolic disorders (i.e., cardiometabolic disorders), both in children [
7] and adults [
8‐
16]. Yet, available large-scale studies that have reported an increased risk of cardiometabolic disorders in adults have been limited to definitions of ADHD based on clinical diagnoses using health record data [
9,
10,
12,
14]. This may be problematic in adults, and especially in older adults, in whom ADHD is likely underdiagnosed [
1]. Older individuals may experience adverse health outcomes due to elevated ADHD symptoms without receiving adequate health care [
17], and the population diagnosed with ADHD may not cover all individuals affected by the symptoms. Additionally, there have only been a few longitudinal studies investigating the associations between ADHD and cardiometabolic disorders, and these studies have had follow-up periods limited to early adulthood [
11,
12,
18]. Longitudinal studies that follow individuals until a more advanced age are needed to establish the long-term associations of ADHD with subsequent cardiometabolic disease.
There is also a scarcity of research investigating the underlying mechanisms of the potential associations between ADHD and cardiometabolic disorders. Two potential mechanisms have been proposed. First, ADHD is associated with several adverse life-course outcomes, such as lower socioeconomic status, lifestyle factors, and comorbid psychiatric disorders, which are also risk factors for cardiometabolic disorders [
19,
20]. Indeed, recent studies have pointed towards life-course risk factors (i.e., socio-economic variables, lifestyle factors, and psychiatric comorbidities) that may partially mediate the association between ADHD genetic liability (i.e., a proxy measure of ADHD risk) [
13] or ADHD diagnoses [
10,
14,
21] and the risk of cardiometabolic disorders. However, none of these studies have considered the associations of ADHD symptoms severity with cardiometabolic disorders. Second, a limited number of genetic studies, using family-based and molecular genetic designs, have recently suggested weak-to-moderate genetic correlations between ADHD and several cardiometabolic outcomes [
9,
22,
23]. However, it remains unknown whether the associations between ADHD and cardiometabolic outcomes are confounded by familial or genetic factors. Considering the weak sharing of genetic factors, it seems plausible that the increased risk of cardiometabolic disorders in ADHD may be largely explained by environmental risk factors or bidirectional causal effects. A 2-sample Mendelian randomization study indicated that ADHD may have a causal effect on childhood obesity and coronary artery disease but found limited evidence for inferring causal effects on other cardiometabolic disorders [
24]. Further research is needed to determine the increased risk of cardiometabolic diseases in ADHD and to establish to what extent educational attainment, lifestyle, and psychiatric comorbidities, as well as shared underlying genetic factors, explain the associations. It is crucial to improve our understanding of the increased risk of cardiometabolic diseases in ADHD and to identify potentially modifiable risk factors, in order to guide clinical practice and improve prevention strategies that promote the health of aging individuals with ADHD.
In this study, we employed a longitudinal population-based study to investigate the associations of ADHD symptoms severity in young- and mid-adulthood with subsequent cardiometabolic disorders in later life. We further investigated whether any identified associations were explained by educational attainment, lifestyle factors, and comorbid psychiatric disorders or confounded by familial factors (genetic and environmental factors shared by twins), using a genetically informed co-twin control approach.
Discussion
This is the first large population-based study to investigate associations between ADHD and cardiometabolic disorders using ADHD symptom level data and a longitudinal approach with follow-up of up to 14 years. Additionally, this is the first study to control for unmeasured familial confounding using a robust, co-twin control study design. We found that ADHD symptom severity was associated with increased risk for both CVDs and metabolic disorders; however, these associations were no longer significant after adjusting for educational attainment, lifestyle factors, and comorbid psychiatric disorders. Our results further suggested that the associations between ADHD symptoms and CVDs might be confounded by genetic factors; however, there was not as strong evidence for genetic confounding for metabolic disorders.
Our results confirmed and extended the findings of recent register/health record studies of the increased risks for CVDs and metabolic disorders in adults with clinically diagnosed ADHD [
8‐
10,
13,
14] by showing that the risk remained elevated even at the symptom level of ADHD, in a non-clinical sample. This finding is of importance considering that register/health record data may have biased prevalence estimates since ADHD likely remains underdiagnosed in adults aged 50 years and older [
1]. Some earlier studies did not identify significant associations between ADHD and CVDs and metabolic disorders [
38‐
40]. However, these studies either had small samples, they utilized self-reported medical history of cardiometabolic disorders [
38,
40], or they used medical claims from commercial databases that may not generalize to the general population [
39].
We further found that the strength of the associations attenuated and did not remain significant after adjusting for educational attainment, lifestyle factors (i.e., smoking, body mass index, and physical activity), and comorbid psychiatric disorders (i.e., major depression, generalized anxiety disorder, and alcohol dependence). This indicates that the risk of cardiometabolic disorders associated with ADHD symptoms is at least partly explained by adverse socioeconomic factors, lifestyle, and psychiatric health outcomes [
10,
13,
14]. These variables can be seen as potentially modifiable risk factors of cardiometabolic disorders [
41]—they may be, at least partially, managed through treatment and/or behavioral changes (e.g., psycho/pharmacotherapy for psychiatric comorbidities, smoking cessation, developing healthier eating habits, and increasing physical activity levels), as opposed to non-modifiable risk factors such as age, sex, family history, and ethnicity. Thus, they could be targeted in future preventative strategies to improve long-term health outcomes in individuals with ADHD.
Furthermore, we cannot rule out the additional effects of other potentially relevant early-life factors (e.g., birth weight or preterm birth) or lifestyle factors such as sleep disorders and nutrition [
4,
19,
20]. We did not include these factors in the current study as we did not have access to the relevant data for preterm birth/low birth weight and due to large amounts of missing values on the corresponding scales for sleep disorders and nutrition and the substantial reduction of the statistical power of the analyses which would occur with their inclusion. Future studies using good-quality data on these variables are needed to further investigate the effects of other potentially relevant confounders and mediators of the associations of ADHD symptoms with cardiometabolic disorders.
Additionally, the associations between ADHD symptoms and CVDs seemed to be confounded by familial factors. Our findings also suggested that these familial factors might be, at least to some extent, driven by shared genetic factors that influence both ADHD symptoms and CVDs, as evidenced by the significantly stronger attenuation of the association when controlling for familial factors shared by monozygotic twins compared to dizygotic twins. This is largely in line with a previous register-based sibling study that reported that shared familial factors partly explained the association between ADHD and CVDs [
9]. Yet, the previous study found no evidence of shared genetic factors, possibly due to low statistical power due to few older adults with both clinically diagnosed ADHD and CVDs [
9]. Nevertheless, a recent genome-wide study reported significant genetic correlations between ADHD and coronary heart disease [
22].
For metabolic disorders, on the other hand, our results suggested potential familial confounding of the associations with ADHD, but not strong evidence for genetic confounding, in contrast to previous research [
9,
23]. One reason for this inconsistency may be the different definitions of ADHD and metabolic disorders. This study is the first to investigate familial confounding underlying the association between ADHD and metabolic disorders using symptoms of ADHD as opposed to clinical diagnoses. Furthermore, we used a broad definition of metabolic conditions, including type 2 diabetes, obesity, and hyperlipidemia. A previous register-based sibling study found that results varied for different individual metabolic disorders; there was strong evidence for shared genetic factors between ADHD and obesity, but not ADHD and type 2 diabetes [
9], while genome-wide association studies [
22,
23] have reported small but significant genetic correlations between ADHD and obesity and type 2 diabetes. While it would have been interesting to investigate genetic confounding of the associations between ADHD and the different metabolic disorders, we were not well-powered to study disorders separately within the co-twin framework.
When examining the two ADHD subscales separately, we found that hyperactivity/impulsivity symptoms were significantly associated with metabolic disorders within monozygotic twin pairs, which indicates direct effects via non-shared environmental influences. These findings suggest that symptoms of hyperactivity/impulsivity may lead to cumulative, health-adverse effects throughout adulthood resulting in metabolic disorders. Indeed, it has been argued that hyperactive and impulsive symptoms may lead to abnormal eating behaviors and binge eating, which in turn may lead to obesity and other metabolic disorders [
42]. Also, it is important to note that metabolic disorders are commonly regarded as risk factors for CVDs disorders [
5,
6], and they may stand in a causal pathway between ADHD symptoms and CVDs. Thus, eating behaviors, in addition to other lifestyle factors and psychiatric comorbidity, should be addressed within preventative strategies for cardiometabolic conditions in adults with ADHD.
Overall, we showed that the associations of ADHD with cardiometabolic disorders were mostly independent of the ADHD subscale used (i.e., inattention and hyperactivity/impulsivity subscales), definition of ADHD, and sex, and they remained consistent when we limited the analysis to same-sex dizygotic twin pairs.
Study limitations
The results of the current study need to be considered in light of several potential study limitations. Namely, the utilized ADHD scale covered only currently present symptoms, without addressing whether ADHD symptoms were present since childhood and whether they were accompanied by functional impairments, which are also required as diagnostic criteria by the DSM-IV. This is important, as in adults, ADHD symptoms may resemble the symptoms of other psychiatric disorders, such as anxiety or depression [
43], or neurodegenerative disorders [
44], which may lead to misdiagnosis in both directions. Nevertheless, our sensitivity analyses which investigated the risks of CVDs and metabolic disorders considering only those who were clinically diagnosed with ADHD confirmed the same pattern of the associations identified in the main analysis.
Furthermore, it may be difficult to discern the effects of ADHD symptoms and ADHD medications in relation to the risk of CVDs. ADHD stimulant medications have been shown to affect blood pressure and heart rate and they are prescribed with clinical precautions regarding the cardiovascular risks [
45]. However, only 0.75% of our study population received ADHD medication, and a recent meta-analysis has indicated that there is no statistically significant association between ADHD medication use and CVDs [
46]. It has also been shown that the associations between ADHD genetic liability and cardiometabolic risks remain even in the absence of medical treatment for ADHD [
13].
Furthermore, all variables assessed using the STAGE interview/questionnaire were based on self-assessment. There are clear advantages to this approach. It allows assessment of large study populations at lower costs than assessment by trained professionals and provides wider coverage of affected individuals compared to using data from electronic health registers, which only capture individuals who receive a formal diagnosis and in Sweden only cover specialist healthcare visits (i.e., more severe patients). Furthermore, there is very poor coverage of lifestyle factors in electronic health registers, such as obesity, smoking, and physical activity. Finally, using continuous scales of ADHD symptoms rather than clinical diagnoses allows for more detailed data and in turn provides greater statistical power, which is needed in within-twin analyses. On the other hand, measurements of health-related variables based on self-assessment may arguably be unreliable [
47]. For instance, it has been shown that self-reports in adults may underestimate ADHD symptoms compared to parent reports [
48]. This issue was partially addressed by conducting an additional analysis of the association of clinically diagnosed ADHD with cardiometabolic disorders. Furthermore, it is reassuring that our findings on the link between ADHD and cardiometabolic disorders are largely in line with past findings using electronic health records [
8‐
10,
13,
14]. Future studies using health-related data assessed by trained professionals are, however, still needed to replicate our findings.
Furthermore, about 5% of the study population had self-reported BMI categorized as obese at the time of STAGE, although only 1.9% received a clinical diagnosis of obesity during the follow-up. Thus, individuals diagnosed with obesity were likely already obese at the start of the follow-up and probably only those with associated health problems sought health care and received a diagnosis of obesity. Consequently, the outcome of obesity in the current study probably covers individuals with the most severe clinical presentations of the condition.
In the current paper, the included covariates may likely have mediated the association between ADHD symptoms and cardiometabolic disorders. However, the covariates were assessed at the same time as the exposure (i.e., ADHD symptoms), and we only adjusted our analyses for these covariates to observe whether the associations changed after their inclusion in the model. To more thoroughly investigate whether they have the role of mediators and to provide estimates of the presumed mediators’ effects on the associations, future studies need to consider assessing these variables after the assessment of ADHD symptoms, as well as conducting a formal mediation analysis, such as the causal mediation analysis [
49].
Additionally, the oldest individuals in our cohort were only 59 years old at the end of follow-up (i.e., end of 2018). The incidence of cardiometabolic disorders peaks after the age of 60–65 [
50,
51], and we may need a longer follow-up of these individuals to a more advanced age to fully understand the associations with ADHD symptoms and the underlying mechanisms. Furthermore, as the non-responders in the STAGE were more likely to be male, have at least one parent born outside Sweden, and be diagnosed with a psychiatric disorder, the generalizability of our study findings may be limited, as non-responders might have had higher levels of ADHD symptoms than responders [
26]. Future studies should investigate the associations of ADHD symptoms with cardiometabolic disorders in study samples with a higher representation of these demographic groups.
Finally, due to the discrepancies in the analytical sample sizes for the analysis at the whole study population level, after adjustments for covariates, and within dizygotic/monozygotic twins and considering the overall small effect sizes of the associations with overlapping confidence intervals across analyses, we cannot rule out that the attenuation of the associations after the adjustments for covariates and for familial factors shared between monozygotic twins may be partially due to the loss of statistical power. Nevertheless, when it comes to the associations between ADHD symptoms and CVDs, we found a more pronounced and statistically significant attenuation in the strength of the association within monozygotic twins compared to the analysis within dizygotic twins. This was not the case for the associations with metabolic disorders, although the decrease in the number of individuals with disease outcomes was equivalent for the analyses with both CVDs and metabolic disorders. Thus, our results may indicate genetic confounding of the association between ADHD symptoms and CVDs.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.