Article Text
PDF
Abstract
Background Attention-deficit/hyperactivity disorder (ADHD) is a lifespan neurodevelopmental condition resulting from complex interactions between genetic and environmental risk factors. There is evidence that ADHD is associated with other mental disorders, but it remains unclear whether and in what way a causal relationship exists.
Objective To investigate the direct and indirect causal paths between ADHD and seven common mental disorders.
Methods Two-sample network Mendelian randomisation analysis was performed to identify psychiatric disorders causally related to ADHD. Total and direct effects were estimated in an univariable and multivariable setting, respectively. Robustness of results was ensured in three ways: a range of pleiotropy-robust methods, an iterative approach identifying and excluding outliers, and use of up to two genome-wide association studies per outcome to replicate results and calculate subsequently pooled meta-estimates.
Results Genetic liability to ADHD was independently associated with the risk of anorexia nervosa (OR 1.28 (95% CI 1.11 to 1.47); p=0.001). A bidirectional association was found with major depressive disorder (OR 1.09 (95% CI 1.03 to 1.15); p=0.003 in the forward direction and OR 1.76 (95% CI 1.50 to 2.06); p=4×10−12 in the reverse direction). Moreover, after adjustment for major depression disorder, a direct association with both suicide attempt (OR 1.30 (95% CI 1.16 to 1.547); p=2×10−5) and post-traumatic stress disorder (OR 1.18 (95% CI 1.05 to 1.33); p=0.007) was observed. There was no evidence of a relationship with anxiety, bipolar disorder or schizophrenia.
Conclusions This study suggests that ADHD is an independent risk factor for a number of common psychiatric disorders.
Clinical implications The risk of comorbid psychiatric disorders in individuals with ADHD needs to be considered both in diagnosis and treatment.
- Depression & mood disorders
- Eating disorders
- Schizophrenia & psychotic disorders
- Suicide & self-harm
Data availability statement
Data are available in a public, open access repository. This study is based on freely available summary-level data. Datasets used in the main analyses can be obtained from iPSYCH (https://ipsych.dk/en) project and PGC (https://www.med.unc.edu/pgc), respectively.13–20 Data used in the replication analyses are available at the FinnGen project (https://www.finngen.fi/en).12
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
What is already known on this topic
Despite growing evidence from observational studies, the question of whether and in what way ADHD is causally related to other psychiatric disorders has remained unanswered.
What this study adds
Under consideration of mediation effects, this network Mendelian randomisation study found that ADHD increases the risk of major depression, post-traumatic stress syndrome, anorexia nervosa and suicide attempt.
How this study might affect research, practice or policy
This study opens new insights into the paths between psychiatric disorders. Thus, in clinical practice, patients with ADHD should be monitored for the psychiatric disorders included in this study and preventive measures should be initiated if necessary.
Background
Attention-deficit/hyperactivity disorder (ADHD) is a lifespan neurodevelopmental condition characterised by the presence of inattentive and/or hyperactive/impulsive symptoms in children and adolescents that extends to adulthood in up to 65%.1 Consequently, individuals with ADHD exhibit deficits in executive function, emotional regulation and motivation.2 Worldwide, the prevalence of ADHD is estimated to be 5.3% in children/adolescents and 2.5% in adults.3 Epidemiological studies show that ADHD in childhood is more common in boys than in girls,4 but in adults there seems to be an equal gender distribution.5 Prospective studies including children with the disease observed a decline of symptoms with increasing age. Therapeutic approaches consist mainly of pharmacological and psychological treatments, but to date there is no single or universal treatment for this disorder.6
The aetiology of ADHD is not fully understood. The disorder is thought to result from complex interactions between genetic and environmental risk factors. Furthermore, prenatal and perinatal factors have been discussed as contributing causes.1 However, genetic factors seem to play the most important role, as the heritability rate is 70–80%.7
There are a number of observational studies showing that ADHD is associated with non-mental diseases across the lifespan (eg, neurological disorders, endocrine and metabolic diseases, and autoimmune diseases).8 In addition, there is evidence that ADHD is related to other mental disorders; in particular, ADHD appears to be associated with mood and anxiety disorders.9 10 In a very recent meta-analysis by Nourredine et al, ADHD in childhood was found to be associated with a higher risk of later psychotic disorders.11 However, whether there is a causal link between ADHD and other mental disorders remains unknown because, in observational studies, confounding and reverse causation cannot be excluded.
Objective
In this study we explored whether there is a causal relationship between ADHD and other mental disorders—namely, major depressive disorder (MDD), bipolar disorder, anxiety disorder, schizophrenia, post-traumatic stress disorder (PTSD), anorexia nervosa (AN) and at least one suicide attempt by two-sample network Mendelian randomisation (MR) analysis.
Methods
Study design
To understand the cause–effect relationships between ADHD and psychiatric disorders, our study based on the MR approach consisted of several steps. The MR framework uses genetic variants in an instrumental variable setting to test for a causal effect of an exposure on an outcome. Where possible, we extended this study design with a meta-analysis combining MRs of two different cohorts per phenotype. Our stepwise network design is shown in online supplemental method 1 and can be described as follows.
Supplemental material
As a first step, our aim was to find out which of the seven mentioned disorders were basically related to ADHD by calculating total effects (ie, any path between ADHD and the respective outcome) applying univariable MR. In a second step, we performed further univariable MR analyses to identify ADHD-related disorders that could potentially act as confounders or mediators due to the effects detected in the first stage. Finally, we calculated the so-called direct effects and extracted in this way the effects that could be attributed directly to ADHD using multivariable MR.
Data collection
In this comprehensive MR study we used a range of datasets to obtain the best possible evidence. Regarding the sample sizes and a clear case definition based on diagnoses by psychiatrists (ICD10 and DSM), we used genome-wide association studies (GWASs) from the iPSYCH project and the Psychiatric Genomics Consortium (PGC) in the main analyses. Phenotypes from PGC including combined diagnoses or substance use disorders were not considered in this study. If available, we replicated the results using summary statistics from the FinnGen12 project and calculated meta-analysis estimates applying random-effects models for these outcomes. Phenotype and cohort descriptions are shown in online supplemental method 2 and summarised in table 1.
Description of outcome datasets and the number of genetic instruments used before and after outlier correction
All datasets were restricted to individuals from European ancestry of both sexes. Where available, genetic variants with an imputation info score <0.8 or a minor allele frequency <0.01 were excluded from the analyses. More detailed information can be found in the respective GWASs.13–20
Instrument selection
As relevant genetic variants we chose single nucleotide polymorphisms (SNPs) associated with the respective exposure based on the genome-wide association threshold of p=5×10−8. To ensure independence of instruments, SNPs in linkage disequilibrium (LD) within a 10 000 kb window and a r 2 threshold of 0.001 were pruned using PLINK clumping. If a genetic variant was missing from the outcome dataset, we searched for proxies that were in LD with r 2 ≥0.8 using the 1000 Genomes database. After the harmonisation process, in which we attempted to infer forward strand alleles using allele frequencies, we excluded palindromic SNPs with a minor allele frequency >0.42.
Statistical analyses
Univariable MR
To estimate the lifetime causal effects and investigate potential outliers responsible for horizontal pleiotropy, we iteratively applied the radial regression framework. In each iteration step we performed the inverse-variance weighted (IVW) regression with second order weights as the main method, and excluded variants based on the Cochran’s Q and Rueckers Q statistics, considering a significance threshold of 0.01. The robustness of estimates was assessed by applying a range of pleiotropy robust methods in the context of sensitivity analyses. These included the weighted mean, weighted median, MR-RAPS (many weak instrument analysis) and MR-PRESSO (outlier corrected estimates) methods. Directional and widespread horizontal pleiotropy was tested using the MR-Egger test and the MR-PRESSO global test (based on the residual sum of squares) and tests of the heterogeneity statistics (Q statistics), respectively. To assess whether the assumption that an exposure causes an outcome is valid, the MR-Steiger test was applied.
Meta-analyses
With the exception of the phenotypes suicide attempt and MDD, for which we were not able to find further non-overlapping GWASs, we performed a meta-analysis based on two MR studies per outcome. Using the results from our MR analyses, we calculated pooled estimates by performing IVW random effects models to account for between-study heterogeneity. Weights, heterogeneity statistics and comparison of fixed and random effects models of meta-analyses per outcome are shown in online supplemental figure 1.
Network and multivariable MR
To assess potential confounding and mediation mechanisms, we modelled the reverse directions with ADHD as outcome and estimated the causal effects of ADHD-related psychiatric disorders on each other by performing multiple univariable MR analyses in the context of a network MR.21 This means that, as shown in online supplemental figure 1, we tested the network consisting of all possible paths and their directions between ADHD and related mental disorders (found in the first step) for notable associations in specific univariable MR analyses (ie, each arrow represents one MR analysis). In this way, we aimed to identify phenotypes that either mediate an association or bias it in the sense of horizontal pleiotropy. After finding mental disorders potentially responsible for horizontal pleiotropy, we used multivariable MR analyses to calculate the direct effects of ADHD on one of the associated outcomes. In this context, we estimated the effects that could be directly attributed to the ADHD rather than to another mental disorder, allowing the genetic instruments to be associated with multiple phenotypes. Where appropriate, we assessed mediation mechanisms considering also bidirectional relationships on the log scale using the product-of-coefficient method22 (see online supplemental method 3). More details about the multivariable MR can be found elsewhere.23 The robust IVW method with multiplicative random effects was used as the principal approach, while the weighted median, MR-Egger and MR-Lasso served as sensitivity analyses within multivariable MR. Multivariable heterogeneity statistics and the MR-Egger intercept were calculated to quantify and test for horizontal and directional pleiotropy, respectively.
Regarding multiple testing issues based on seven null hypotheses (outcomes), a Bonferroni adjusted threshold of 0.007 was considered as a type I error for the analyses. ORs presented can be interpreted as the average change in the respective outcome per 2.72-fold increase in the prevalence of the exposure.24
All analyses were performed using the open source statistical software R (v4.1.2). For MR we used mainly the packages TwoSampleMR (v0.5.6), MendelianRandomization (v0.6.0), RadialMR (v1.0), mr.raps (v0.2) and MRPRESSO (v1.0). Figures were created with ggplot2 (v3.3.5).
Reporting follows the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization: The STROBE-MR Statement.
Results
Univariable MR: total effects
Starting with 9–11 potential genetic instruments for ADHD (see online supplemental table 1), the number of SNPs was reduced to 6–11 during the outlier assessment (see online supplemental table 2), depending on the respective outcome (table 1).
Results based on MR and subsequent meta-analyses revealed positive associations between the genetic liability to ADHD and the risk of a suicide attempt (OR 1.34 (95% CI 1.16 to 1.55); p=8×10−5), AN (OR 1.28 (95% CI 1.11 to 1.47); p=0.001), PTSD (OR 1.20 (95% CI 1.08 to 1.34); p=0.001) and MDD (OR 1.09 (95% CI 1.03 to 1.15); p=0.003) (figure 1). No notable associations were observed with anxiety, bipolar disorder and schizophrenia. All of these estimates, except MDD and suicide attempt, resulted from meta-analyses shown in online supplemental figure 1.
Effect estimates in the form of odds ratios (ORs) and 95% confidence intervals representing the impact of attention-deficit/hyperactivity disorder on the risk of psychiatric outcomes. Estimates derived from meta-analyses based on two Mendelian randomisation studies per outcome (except suicide and major depression disorder). The results should be interpreted based on the Bonferroni corrected threshold of 0.007.
Looking at the study-specific MR sensitivity analyses, all findings were supported by the pleiotropy-robust approaches, which led to consistent point estimates, suggesting robust results (see online supplemental figure 2). Comparing the first iterations before outlier exclusion and the outlier-corrected final models, most effect estimates were consistent in terms of direction and magnitude. The only exception was the positive association between ADHD and schizophrenia, which disappeared completely when the identified outliers were removed.
No directional pleiotropy was detected in any of the final MR models regarding the MR-Egger intercept test. Despite the unadjusted p values of the PRESSO global test and the Cochran’s Q statistic (0.03 and 0.02) in the model with AN (see online supplemental table 3), there was no evidence of SNPs driving the effect or of substantial heterogeneity in the plots (leave-one-out analysis, funnel plot and the scatter plot showing genetic associations).
Network MR
To fully understand the paths regarding our findings described above, we aimed to investigate potential bidirectional and indirect effects of ADHD-related psychiatric disorders from our analyses (ie, MDD, PTSD, AN and at least one suicide attempt) on each other. Since an analysis of PTSD and the phenotypes from the FinnGen cohort was not possible due to lack of instruments, we were only able to use MDD and AN as candidates in the network MR.
In this context, genetic liability to MDD was observed to be positively associated with the risk of PTSD (OR 1.84 (95% CI 1.59 to 2.14); p=8×10−16) and suicide attempt (OR 1.61 (95% CI 1.25 to 2.08); p=3×10−4) but not with AN (OR 1.50 (95% CI 0.96 to 2.33); p=0.073) (figure 2). In addition, modelling the relation of MDD to ADHD yielded an OR of 1.76 (95% CI 1.50 to 2.06) (p=4×10–12), implying a bidirectional association (figure 2). The modelling directions were supported by the MR-Steiger tests (see online supplemental table 4). In contrast, no associations could be found with AN as exposure and MDD, PTSD and suicide attempt as outcomes (figure 2). Again, the estimates with PTSD and AN as outcomes were pooled using meta-analyses, which are shown in online supplemental figure 3. All results were supported by the study-specific sensitivity analyses (see online supplemental figure 4 and online supplemental figure 5).
Results from network Mendelian randomisation analyses testing causal effects of major depression disorder or anorexia nervosa on the risk of psychiatric outcomes. Estimates are given in the form of odds ratios (ORs) and 95% confidence intervals. Estimates derived from meta-analyses based on two Mendelian randomisation studies per phenotype (except the outcomes suicide and major depression disorder). Results should be interpreted based on the Bonferroni corrected threshold of 0.007.
Multivariable MR: direct effects
With regard to these results, we used multivariable MR to assess whether and to what extent the relationships between ADHD and both PTSD and suicide attempt are affected by MDD. After adjustment for MDD, ADHD was still positively related to PTSD (OR 1.18 (95% CI 1.05 to 1.33); p=0.007) and suicide attempt (OR 1.30 (95% CI 1.16 to 1.47); p=2×10−5) as shown in figure 3. MDD was also directly related to PTSD (OR 1.67 (95% CI 1.37 to 2.02); p=2×10−7) and suicide attempt (OR 1.42 (5% CI 1.08 to 1.87); p=0.011), suggesting that the associations between ADHD and both psychiatric outcomes were mediated by MDD (figure 3). Considering the bidirectional association with ADHD, MDD mediated 21.1% of the proportion of the AHDH–PTSD association (indirect effect: OR 1.05) and 10.3% of the ADHD–suicide attempt association (indirect effect: OR 1.03).
Summary of the findings of this study based on two-sample Mendelian randomisation (MR) analyses. Effect estimates of the associations between ADHD and both suicide attempt and PTSD resulted from multivariable MR analyses after adjusting for MDD. Estimates are given in the form of odds ratios (ORs) and 95% confidence intervals. ADHD, attention-deficit/hyperactivity disorder; AN, anorexia nervosa; MDD, major depressive disorder; PTSD, post-traumatic stress disorder.
All multivariable results were consistently confirmed by sensitivity analyses, with no evidence of pleiotropy considering the tests due to the multivariable MR-Egger intercept and the Q statistic (see online supplemental figure 6, online supplemental table 5).
Discussion
In this two-sample network MR analysis we evaluated the causal links between ADHD and a range of psychiatric disorders. We identified an impact of ADHD on PTSD, suicide attempt and AN and a bidirectional association between ADHD and MDD. Regarding the bidirectional relationship between ADHD and MDD, a common underlying pathophysiological pathway can be hypothesised, so that both mental disorders may separately and jointly increase the risk of suicide attempts or PTSD. In this context, however, the increased risk of AN can be attributed entirely to ADHD. There was no evidence for a causal link between ADHD and bipolar disorders, anxiety or schizophrenia.
ADHD and MDD
A number of studies consistently found that approximately 35–50% of adults with ADHD experience one or more major depressive episodes during their lifetime; this percentage is significantly higher than the risk for a major depressive episode in the general population (approximately 15%). Data from family studies suggest a familial association and shared genetic risk between ADHD and depression.25 26 The present finding that ADHD is associated with MDD underlines previous results from observational27 28 and a recent MR analysis.29 Imaging studies have shown that patients with ADHD exhibit alterations in prefrontal cortex (PFC) circuits and demonstrate weaker activation of the PFC when attempting to regulate attention and behaviour. The PFC requires optimal levels of norepinephrine and dopamine for proper functioning. Genetic studies have consistently found alterations in genes involved in catecholamine transmission in patients with ADHD.30 The monoamine hypothesis of depression postulates that depression is caused by decreased modulatory function of serotonin, norepinephrine or dopamine in the brain.31 Based on the biological aspects of depression and ADHD, the two disorders might share similar pathophysiological brain regions including decreased activity in the PFC,32 33 amygdala and hippocampus regions.34–37 Anhedonia, the neurobiology of which is complex and involves disruption of pathways regulating reward and motivation,38 represents a major feature of both MDD and ADHD.37 The activity of various components of the limbic–cortical–striatal–pallidal–thalamic pathway correlates with hedonic tone in healthy individuals and is altered in patients with MDD and ADHD.
Note that the observed bidirectional relationship between ADHD and MDD should be seen with caution. ADHD often develops at an early age, which is why the majority of cases considered in the GWAS were diagnosed in childhood (14 878 in childhood vs 6961 in adulthood).39 MR estimates the mean lifetime effect and is not able to model effects that occur within a specific time period. Thus, the reverse direction may be biased.
ADHD and PTSD
PTSD is a mental and behavioural disorder that can be triggered by a traumatic life event. Classic symptoms include hyperarousal, avoidance and intrusion. Prior investigations have suggested that ADHD is a risk factor for PTSD.40 41 In part, studies have questioned whether ADHD and PTSD are really two different disorders.42 However, it is now believed that the two disorders must be considered as two distinct conditions despite sharing common risk factors.43 Both ADHD and PTSD exhibit genetic transmission44 45 but, with a few exceptions, the onset of ADHD significantly precedes the onset of PTSD, suggesting that ADHD is an antecedent risk factor for the development of PTSD. Our results support a positive relationship between ADHD and the risk of PTSD, which are in accordance with a longitudinal observational study46 and a systematic review and meta-analysis.40
ADHD and AN
We also found evidence for a causal link between ADHD and eating disorders, in this case with AN. This finding supports prior results from systematic reviews and a meta-analysis showing a higher risk of eating disorders in individuals with ADHD.47 In the meta-analysis by Nazar et al, the pooled OR of diagnosing any eating disorder in ADHD was 3.82 (95% CI 2.34 to 6.24), and for diagnosing AN in ADHD patients the OR was 4.28 (95% CI 2.24 to 8.16). Both ADHD and eating disorders share neurocognitive deficits; patients with the diseases have difficulties in shifting, action sequences, working memory and inhibitory control.48 Several theories address possible explanations for the association of ADHD with eating disorders. A recent systematic review examining mechanisms underlying comorbid AN and ADHD suggests that emotion dysregulation may mediate the relationship between these two disorders.49
ADHD and suicide attempt
The present study showed that ADHD is related to at least one suicide attempt. Although some previous studies postulated that ADHD was not associated with an increased risk of suicide attempt or completion,50–53 an increased risk of suicide with ADHD has recently been suggested.54 A recent systematic review and meta-analysis investigating the relationship between ADHD and suicidal spectrum behaviours including 57 observational studies reported significant associations between ADHD and suicidal attempts (OR 2.37, 95% CI 1.64 to 3.43).55 One potential explanation for these findings is that ADHD and suicidal behaviour share common genetic factors that may reflect genetic variants associated with impulsivity, a trait that is highly heritable.56 Impulsivity is a core component of ADHD and closely associated with suicidal behaviour.57
ADHD and schizophrenia
Contrary to the previously reported association found by Nourredine et al, we could not confirm a link between ADHD and schizophrenia after outlier removal. Their systematic review and meta-analysis summarised 12 observational studies including 1.85 million participants. Five longitudinal studies were included for the meta-analysis with the outcome schizophrenia yielding an OR of 4.59 (95% CI 3.83 to 5.50) for participants with childhood ADHD.11 However, regarding the overall quality of the studies included in the meta-analysis, a high risk of bias in different areas in several studies was observed.
ADHD and anxiety
Furthermore, we found no evidence for a relationship between ADHD and general anxiety disorder. With a prevalence of about 25–50%, anxiety disorders represent one of the most common comorbidities in patients with ADHD.58 Epidemiological and clinical studies have indicated a high comorbidity of anxiety disorders in children and adolescents59 as well as in adult patients with ADHD.27 60 In longitudinal studies, patients with ADHD were found to develop a higher level of anxiety disorders compared with controls.61 62
ADHD and bipolar disorders
After adjustment for multiple testing, our results regarding the risk of bipolar disorders in patients with ADHD could not clearly confirm the findings from a recent systematic review and meta-analysis including six studies which reported a higher risk of bipolar disorder in individuals with ADHD compared with healthy controls (risk ratio 8.97, 95% CI 4.26 to 18.87).63 Another systematic review and meta-analysis including 71 studies with 646 766 participants from 18 countries found that one in 13 adults with ADHD was also diagnosed with bipolar disorder.10
Implications
The present MR study extends existing research by examining the causal association between ADHD and common psychiatric disorders. There is little doubt that ADHD and other mental disorders share common pathophysiological causes, but the exact causative mechanisms have not yet been fully elucidated.30 36 64 Also, prenatal and perinatal factors are discussed as possible underlying risk factors.65 A further important issue in this connection is that children with ADHD are often treated with psychostimulant drugs which could have an effect on the development of further mental disorders later in life.11 Furthermore, systemic inflammation could in part explain the joint occurrence of ADHD and other mental disorders.66
Due to the association between ADHD and PTSD, clinicians who treat patients with ADHD should be alert to the fact that their patients with this diagnosis are at a higher risk of developing PTSD later in life. Thus, early and effective treatment of ADHD may reduce the risk and a subsequently more severe course of PTSD.45
Strengths and limitations
The strengths of our study, which is less prone to unmeasured confounding and reverse causality than observational studies, include robustness of the results based on an iterative approach focusing on outliers, up to two datasets per outcome, and a range of sensitivity analyses in both the univariable and multivariable design.
The results of our MR analyses partially confirmed most, but not all, of the previously reported associations in a hypothesis-free context. For example, we could not maintain the association between ADHD and schizophrenia after removing outliers. This may be due to the various limitations of observational studies, such as unobserved confounding, reverse causality, publication and selection bias or the limitations of MR. As with any statistical method, it is important to remember that absence of evidence is not equivalent to evidence of absence. MR examines lifelong effects, but some causal factors for the onset of mental disorders may have more acute effects. Further investigations are necessary to address these unresolved issues. Another limitation of MR arises from horizontal pleiotropy—that is, one and the same gene may be associated with different phenotypic traits. The genetic architecture of psychiatric phenotypes is highly polygenic and pleiotropic67 and often the biological function of their associated genetic variants is unknown. However, we were able to confirm our results by using different approaches such as network MR in an iterative way, a series of sensitivity analyses, and multivariable MR accounting for potential confounding and mediation effects. It is therefore unlikely that the observed relationships of ADHD with MDD, AN, suicide attempts and PTSD were apparent only because of horizontal pleiotropy. Furthermore, caution should be taken when interpreting the magnitude of calculated effect estimates and mediation effects since MR does not provide accurate point estimates in case of binary exposures.22 For this reason, the estimators presented should be viewed as tests of causality or mediation rather than true causal effects. Finally, the results are based on data from individuals of European descent, so generalisability to other ethnicities is limited.
Conclusions
The present findings are important for clinicians treating patients with ADHD because they provide a guide as to the mental comorbidities on which to focus preventively and therapeutically during the course of the disease. ADHD serves as an early indicator of other mental disorders due to shared psychopathologies. Therefore, it is necessary to monitor patients with ADHD for early signs of mental comorbidities. In addition, an increased risk of AN in patients with ADHD must be considered. This would allow for early treatment tailored to the patient.
Data availability statement
Data are available in a public, open access repository. This study is based on freely available summary-level data. Datasets used in the main analyses can be obtained from iPSYCH (https://ipsych.dk/en) project and PGC (https://www.med.unc.edu/pgc), respectively.13–20 Data used in the replication analyses are available at the FinnGen project (https://www.finngen.fi/en).12
Ethics statements
Patient consent for publication
Ethics approval
Not applicable since the study is based on summary-level data. In all original studies, ethical approval and participant consent to participate had been obtained
Acknowledgments
The authors acknowledge the participants and investigators of PGC and the FinnGen study.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors CM participated in data collection, supervised the project and drafted the Background and Discussion sections. DF performed the statistical analyses, created all tables and plots and drafted the Methods and Results sections. Both authors participated in the design of the study, investigation and interpretation of results. CM and DF act as guarantors for the study.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors. The open access publication of this article was supported by the Open Access Fund of the Medical Faculty of the University of Augsburg.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.