www.nature.

com/npp

ARTICLE OPEN

Genetic evidence for a potential causal relationship between

insomnia symptoms and suicidal behavior: a Mendelian
randomization study
Malik Nassan1,2,7,9, Iyas Daghlas2,3,8,9, John W. Winkelman4, Hassan S. Dashti5,6 , International Suicide Genetics Consortium and
✉
Richa Saxena5,6

© The Author(s) 2022

Insomnia and restless leg syndrome (RLS) are associated with increased risk for suicidal behavior (SB), which is often comorbid with
mood or thought disorders; however, it is unclear whether these relationships are causal. We performed a two-sample Mendelian
randomization study using summary-level genetic associations with insomnia symptoms and RLS against the outcomes of risk of
major depressive disorder (MDD), bipolar disorder (BP), schizophrenia (SCZ), and SB. The inverse-variance weighted method was
used in the main analysis. We performed replication and sensitivity analyses to examine the robustness of the results. We identified
outcome cohorts for MDD (n = 170,756 cases/329,443 controls), BP (n = 20,352/31,358), SCZ (n = 69,369/236,642), SB-Cohort-2019
1234567890();,:

(n = 6569/14,996 all with MDD, BP or SCZ; and SB within individual disease categories), and SB-Cohort-2020 (n = 29,782/519,961).
Genetically proxied liability to insomnia symptoms significantly associated with increased risk of MDD (odds ratio (OR) = 1.23, 95%
confidence interval (CI) = 1.2–1.26, P = 1.37 × 10–61), BP (OR = 1.15, 95% CI = 1.07–1.23, P = 5.11 × 10–5), SB-Cohort-2019 (OR = 1.17,
95% CI = 1.07–1.27, P = 2.30 × 10–4), SB-Cohort-2019 in depressed patients (OR = 1.34, 95% CI = 1.16–1.54, P = 5.97 × 10–5), and SB-
Cohort-2020 (OR = 1.24, 95% CI = 1.18–1.3, P = 1.47 × 10–18). Genetically proxied liability to RLS did not significantly influence the
risk of any of the outcomes (all corrected P > 0.05). Results were replicated for insomnia with MDD and SB in Mass General Brigham
Biobank and were consistent in multiple lines of sensitivity analyses. In conclusion, human genetic evidence supports for the first
time a potentially independent and causal effect of insomnia on SB and encourages further clinical investigation of treatment of
insomnia for prevention or treatment of SB.

Neuropsychopharmacology (2022) 47:1672–1679; https://doi.org/10.1038/s41386-022-01319-z

INTRODUCTION suicide attempts OR = 1.78, suicide deaths OR = 1.54] [9]. How-

Insomnia and restless leg syndrome (RLS) have emerged as ever, more research is needed to uncover whether these
modifiable risk factors for mood disorders and suicidal behavior associations represent causal relationships and mechanisms
(SB) [1–3]. Insomnia is a clinical diagnosis characterized by underlying the connection between insomnia and suicide.
difficulty falling or staying asleep that is associated with distress Notably, schizophrenia (SCZ) has a known association with
and/or dysfunction [4, 5]. Insomnia disorder has a prevalence of insomnia, mood disorders, and SB that is not fully characterized
10–20.0% [4, 6]. Cross-sectional and longitudinal observational [10–13]. Insomnia is heritable with heritability estimated from twin
studies demonstrate that insomnia is associated with increased studies to be 0.39 [14]. This has motivated genome-wide
risk for psychiatric disorders [1]. Sleep disturbances have been association studies, which have identified over 200 SNPs
associated with an increased longitudinal risk for bipolar disorders associated with insomnia symptoms [15].
(BP) ((odds ratio (OR) = 1.72)) and depressive disorders (OR = 1.62) RLS is a clinical diagnosis in which predominantly nighttime leg
[7]. More specifically, in a meta-analysis of prospective cohort restlessness at rest, relieved by movement, leads to distress and
studies, insomnia was associated with an increased risk of sleep disturbance. The prevalence of clinically significant RLS is
depression (pooled relative risk was 2.27) [8]. Furthermore, in a estimated to be 2.5% [16]. Cross-sectional and longitudinal
more recent meta-analysis, insomnia was associated with observational studies demonstrate that RLS is associated with an
increased longitudinal risk for SB [suicidal ideation OR = 2.10, increased risk for MDD, reduced quality of life, and overall

1
Division of Sleep Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA. 2Center for Genomic Medicine, Massachusetts General Hospital,
Harvard Medical School, Boston, MA, USA. 3Harvard Medical School, Boston, MA, USA. 4Departments of Psychiatry and Neurology, Massachusetts General Hospital, Harvard
Medical School, Boston, MA, USA. 5Center for Genomic Medicine and Department of Anesthesia, Critical Care and Pain Medicine, Massachusetts General Hospital, Harvard
Medical School, Boston, MA, USA. 6Program in Medical and Population Genetics, Broad Institute, Cambridge, MA, USA. 7Present address: Mesulam Center for Cognitive Neurology
and Alzheimer’s Disease, Northwestern University, Chicago, IL, USA. 8Present address: Department of Neurology, University of California, San Francisco, San Francisco, CA, USA.
9
These authors contributed equally: Malik Nassan, Iyas Daghlas. A full list of members and their affiliations appears in the Supplementary Information.
✉email: [email protected]

Received: 28 September 2021 Revised: 27 March 2022 Accepted: 30 March 2022

Published online: 10 May 2022
 M. Nassan et al.
1673
increased mortality [17–24]. RLS is also associated with an
increased risk of suicide and self-harm (adjusted hazard ratio
was 2.66) [2]. Twin studies have estimated RLS heritability to be up
to 70%, with a 19.6% SNP-based heritability [25, 26]. This
establishment of heritability paved the way for genome-wide
association studies (GWAS) that have identified 20 SNPs that are
significantly associated with RLS [26].
Most studies testing the relationships between insomnia, RLS
and mood disorders and SB are observational studies. Observa-
tional studies (even longitudinal cohorts) can establish associa-
tions between risk factors and diseases, but they are insufficient to
establish causal relationships [27, 28]. Mendelian randomization
(MR) is an analytic method that uses genetic proxies of exposures
to test associations with disease outcomes, and has several Fig. 1 The hypothesized relationships between the studied
distinct advantages for causal inference [29, 30]. Recent large exposures and outcomes. Bold arrows represent the direct effect
GWAS for insomnia, RLS, mood disorders, and most recently SB, between exposures and outcomes. Thin arrows represent indirect
now permit the use of MR to investigate causal relationships pathways. BP bipolar disorder, SB suicidal behavior, RLS restless leg
between these associated disorders and behaviors. Of note, SB is a syndrome, SCZ schizophrenia, MDD major depressive disorder.
complex behavior and caused by genetic and environmental
factors; with estimated heritability from twin studies of 30–55% the diagnostic cut-off of Insomnia Severity Index and Pittsburgh Sleep
Quality Index in an independent sample from the Netherlands Sleep
[31, 32].
Registry (sensitivity/specificity: UKB = 98/96%; 23andMe = 84/80%). Of
In this study, we aimed to utilize MR to assess for the first time note, both scales have high reliability and validity in diagnosing insomnia
whether genetically proxied insomnia symptoms and RLS have [35, 36]. Furthermore, in the same independent sample, the UKB insomnia
causal relationships with SB, and to examine this relationship in questions had good sensitivity of 94% and specificity of 89% in
the presence of mood and thought disorders. differentiating cases and controls in comparison with structured interview
[37]. A total of 250 independent genetic variants (low pairwise linkage
disequilibrium (r2 < 0.1)) were associated with insomnia in the meta-
METHODS analysis of the UKB and 23andMe cohorts at genome-wide significance
Ethical approval and patient consent (P < 5 × 10–8) [Supplementary Table 1S].
Deidentified summary statistics and publicly available data were utilized in We selected genome-wide genetic association data from the largest
this study, and thus no IRB approval was required for the analyses. published meta-analysis of RLS GWAS to generate a genetic proxy for RLS
[Table 1] [26]. This study included three cohorts in the meta-analysis (EU-
RLS GENE, INTERVAL consortia, and 23andMe) in which all participants
Study design were of European ancestry. RLS diagnosis varied across the cohorts and
In this study, we applied a two-sample MR study design using summary- included diagnosis by face-to-face interview by an expert neurologist for
level genetic association data [33]. As exposures, we used previously EU-RLS GENE, through the validated Cambridge-Hopkins Restless Legs
identified genetic variants for insomnia symptoms and RLS to test for their Questionnaire for the INTERVAL consortia, and through a single question
potential causal effects on mood disorders (MDD and BP), SCZ, and SB. Our survey for the 23andMe cohort “Have you ever been diagnosed with
main and sensitivity MR analyses were structured to demonstrate the restless legs syndrome?”. The collective discovery and replication samples
following assumptions: (1) genetic variants are robustly associated with the consisted of 45,896 cases and 382,638 controls. Twenty independent
studied exposures (e.g., insomnia symptoms and RLS), (2) associations of genetic variants (low pairwise linkage disequilibrium (r2 < 0.01)) were
the genetic variants with the exposures and with the outcomes are not associated with RLS at genome-wide significance (P < 5 × 10–8) [Supple-
confounded, and (3) the genetic variants are influencing the risk of the mentary Table 2S]. All GWAS were analyzed with standard quality control
outcomes through the exposures, and not through alternative pathways. procedures, including methods to control for population stratification.
The relationships between the studied exposures and outcomes are
illustrated in [Fig. 1].
Genetic associations with the outcomes
We identified publicly available GWAS summary statistics for the
Genetic associations with the exposures psychiatric outcomes of interest (MDD, BP, SCZ, and SB). To limit
We identified genetic proxies for liability to insomnia symptoms (as confounding by ancestral differences, we selected studies limited to
surrogate for insomnia) as genome-wide significant variants from the individuals of European ancestry. These studies are listed in [Table 1]. Beta
largest published insomnia GWAS at the time of analysis [Table 1] [34]. This coefficients of the genetic variants associated with the exposure and
GWAS meta-analysis included insomnia cases of European ancestry from outcome phenotypes were harmonized by matching effect alleles.
the UK Biobank (UKB: cases 109,402/controls 277,131) and 23andMe Summary statistics GWAS were obtained from the Psychiatric Genomics
cohorts (cases 288,557/controls 655,920). In the UKB: insomnia complaints Consortium (PGC) online database for MDD, BP, SCZ [https://www.med.
were evaluated by asking: “Do you have trouble falling asleep at night or unc.edu/pgc/download-results/]. We included two GWAS for SB: (1) SB-
do you wake up in the middle of the night?” Insomnia cases were defined Cohort-2019, available from PGC and included cases with either MDD, BP,
as participants who answered this question with “usually”, while or SCZ; and provided subset cohorts for analyses stratified by disease, (2)
participants answering “never/rarely” or “sometimes” were defined as SB-Cohort-2020 which is the largest GWAS for SB, and included the
controls. For 23andMe, insomnia cases were confirmed by a positive previous SB-Cohort-2019.
response to at least one of these questions: “Have you ever been For MDD, the most recent meta-GWAS included a total of 246,363 cases
diagnosed with, or treated for: Insomnia?”; “Have you ever been diagnosed and 561,190 controls; we had access to summary statistics from combined
with, or treated for, any of the following conditions: Insomnia but not data set from UKB and PGC (n = 170,756 cases/329,443 controls) and not
Narcolepsy, Sleep apnea or Restless leg syndrome”; “Has a doctor ever told from 23andMe [38]. The UKB MDD diagnosis was based on self-reported
you that you have any of these conditions: Insomnia (difficulty getting to help-seeking for “problems with nerves, anxiety, tension or depression”
sleep or staying asleep)?”; “Have you ever been diagnosed by a doctor with (termed ‘broad depression’), while the PGC cohort utilized a range of
any of the following neurological conditions: Sleep disturbance”; “Do you depression phenotypes (including structured clinical interview as well as
routinely have trouble getting to sleep at night?”; “What sleep disorders broader criteria). A total of 102 genome-wide independent variants were
have you been diagnosed with? Please select all that apply: Insomnia, associated with MDD in the meta-analysis.
trouble falling or staying asleep”; “Have you ever taken these medications? For BP, the most recent meta-GWAS included a total of 20,352 cases and
Prescription sleep aids”; “In the last 2 years, have you taken any of these 31,358 controls of European descent (collected from 32 studies); and
medications? Prescription sleep aids”. Insomnia definition had higher replication analysis of 822 variants (P < 1 × 10−4) in 9412 cases and 137,760
accuracy in the UKB cohort than in the 23andMe cohort when compared to controls [39]. The combined analysis identified 30 genome-wide significant

Neuropsychopharmacology (2022) 47:1672 – 1679

 M. Nassan et al.
1674
Table 1. Summary of the GWAS cohorts included in the analyses.
Study Disease Cases Controls Number of genome-wide SNPs
Jansen et al. Nat Genet (2019) Insomnia 397,959 933,051 250
Schormair et al. Lancet Neurol (2017) RLS 45,896 382,638 20
Howard et al. Nat Neurosci (2019) MDD 170,756 329,443 102
Stahl et al. Nat Genet (2019) BP 20,352 31,358 30
Ripke et al. MedRxiv (2020) SCZ 69,369 236,642 270
Mullins et al. American Journal of Psychiatry (2019) SB 2019 6569 14,996 0
Mullins et al. Biol Psychiatry (2021) SB 2020 29,782 519,961 2

genetic variants influencing the risk of BP. Overall, cases needed to meet We then performed multiple sensitivity analyses to assess the
DSM-IV, International Classification of Diseases (ICD)-9, or ICD-10 criteria for robustness of the findings. First, we removed SNPs in LD between the
a lifetime diagnosis of BP (by either structured diagnostic instrument, insomnia symptoms and RLS genetic proxies to reduce any confounding
clinician-administered checklists, or medical record review). Most controls between these exposures (removing SNPs within 10 Mb and with r2 ≥ 0.7).
were examined for the absence of any other lifetime psychiatric disorders. Second, we removed SNPs from the exposure genetic proxies (i.e.,
A total of 30 genome-wide independent variants were associated with insomnia symptoms and RLS) that were in LD (using a more stringent
MDD in the meta-analysis. threshold r2 ≥ 0.01) with an outcome that was significant in the main MR
For SCZ, the most recent meta-GWAS included a total of 69,369 cases IVW analysis (MDD, BP, and SB-Cohort-2020). Third, we performed model-
and 236,642 controls [40]. This was the largest combined cohort yet from based sensitivity analyses that relax various MR assumptions regarding
PGC (included 90 cohorts) and identified 270 independent genome-wide pleiotropy, including: MR-egger regression, weighted median, and MR-
significant genetic loci. Of note, this study included 80% of the sample PRESSO [44, 45]. We also created leave-one-out plots to display the results
from European ancestry and 20% from East Asian ancestry. Cases included from the IVW and Egger regression analyses, to assess for outliers.
diagnoses of SCZ or schizoaffective disorder; details of each of the enrolled We sought to replicate our findings in the Mass General Brigham (MGB)
cohorts are available in the manuscript [40]. A total of 270 genome-wide Biobank (formerly Partners Biobank) for the outcomes of MDD, BP, and SB
independent variants were associated with MDD in the meta-analysis. only (MDD n = 4640/23,849, BP n = 145/28,344, SB n = 1054/27,435) [46].
Two cohorts were used for the outcome of SB (encompassing here both The MGB Biobank is a hospital-based cohort study from the MGB
fatal and non-fatal suicidal attempts). The first cohort is a PGC cohort (SB- healthcare network in Boston, MA with electronic health record (EHR)
Cohort-2019) published in 2019 (cases 6569, controls 14,996) and stratified and genetic data. Recruitment for the Biobank launched in 2010 and
by comorbid psychiatric disorder (MDD, BP, or SCZ) [41]. The subjects were remains ongoing at participating clinics and electronically. Recruitment
obtained from 16 MDD cohorts, 21 BP cohorts, and 9 SCZ cohorts from strategy has been described previously [47]. All recruited patients provided
PGC, where data on suicide attempts had been gathered. Only patients consent written informed upon enrollment. The present study protocol
affected by the three psychiatric disorders were included and all three was approved by the MGB Institutional Review Board (#2018P002276).
psychiatric disorders were defined using structured psychiatric interviews. Effect estimates for MDD, BP, and SB were generated using data for 30,683
All individuals were of European ancestry. Items from structured clinical participants with genetic data and limited to participants of European
interviews offered data on suicidal attempts. Lifetime suicidal attempt was ancestry [48]. Cases of MDD, BP, and SB were determined from EHR using a
characterized across cohorts as an intentional act of self-harm with the validated algorithm based on natural language processing of structured
intent to result in death. Individuals who only endorsed suicidal ideation and unstructured data including coded diagnoses, medications, proce-
were not included as cases. Across the cohorts, there was a sum of 6569 dures, and vital signs [46]. The remaining participants were set as control.
individuals who attempted suicide and 17,232 individuals who had not To determine SNP effects on MDD, BP, and SB, we performed genetic
attempted suicide. No genome-wide significant associations were identi- association analysis in unrelated participants of European ancestry with
fied in the meta-analysis. PLINK logistic regression and an additive genetic model adjusted for age,
The second cohort (SB-Cohort-2020) assessed for SB in 29,782 cases and sex, five principal components, and genotyping array [49].
519,961 controls and is the most recent and largest SB GWAS [pre-print Finally, we ran a reverse MR IVW analysis between MDD and BP (using
released in 2020, and publication was in 2021] [42, 43]. The cohorts included the genome-wide genetic proxies from the same GWAS) as exposures and
samples from European ancestry (the majority of the cases), admixed African insomnia symptoms as outcome (from the same insomnia symptoms
American ancestry (4%), and East Asian ancestry (6%). This GWAS included GWAS but only including the UKB cohort due to lack of public availability
21 cohorts, of which cases were individuals who died by suicide (2 cohorts) of 23andMe data).
or made a non-fatal suicide attempt (19 cohorts) which is defined as a
lifetime act of intentional self-harm with intent to cause one’s own death.
Individuals who only endorsed suicidal ideation were not included as cases. RESULTS
Information on suicidal attempts was obtained via structured clinical Main analyses
interviews for 15 cohorts, self-report questionnaires for 2 cohorts, and ICD
codes or hospital records for 2 cohorts. Cases of death by suicide (2 cohorts) MR analyses showed significant associations of genetically proxied
were obtained from the Medical Examiner’s Office of the Hyogo Prefecture insomnia symptoms with MDD, BP and SB [MDD (OR = 1.23, 95%
and the Division of Legal Medicine, at the Kobe University Graduate School CI = 1.2–1.26, P = 1.37 × 10–61), BP (OR = 1.15, 95% CI = 1.07–1.23,
of Medicine in Japan or the Utah State Office of the Medical Examiner. A P = 5.11 × 10–5), SB-Cohort-2019 (OR = 1.17, 95% CI = 1.07–1.27,
percentage of cases from the Columbia University and iPSYCH cohorts P = 2.30 × 10–4), and SB-Cohort-2020 (OR = 1.24, 95% CI =
included individuals with death by suicide that was established through the 1.18–1.3, P = 1.47 × 10–18)] [Fig. 2]. Analyses for the outcome of
Columbia Classification Algorithm for Suicide Assessment and the Cause of SB in the 2019 cohort stratified by disease status showed that the
Death Register in Denmark, respectively. Two genome-wide significant effect of insomnia symptoms on SB is most robust in the
genetic variants were identified as influencing the risk of SB. depressed population (OR = 1.34, 95% CI = 1.16–1.54, P = 5.97 ×
10–5) [Fig. 2]. The scatterplots for the significant findings are
Mendelian randomization analyses included in Fig. 3 and Supplementary Table 3S. On the other hand,
All analyses were performed in R Version 3.5.3 using the TwoSampleMR genetically proxied liability to RLS was not associated with any of
v0.4.229 package. The inverse-variance weighted (IVW) method was the the study outcomes [Fig. 4 and Supplementary Table 4S].
main MR method used to estimate the effect of genetically proxied liability
to insomnia symptoms or RLS on each of the psychiatric outcomes [44].
The corrected statistical significance threshold is P less than 3.13 × 10–3, Sensitivity analyses
accounting for 16 statistical comparisons across 2 exposures and 8 Results for the effects of genetic liability to insomnia symptoms on
outcomes. the outcomes were unchanged when we removed SNPs in LD

Neuropsychopharmacology (2022) 47:1672 – 1679

 M. Nassan et al.
1675

Fig. 2 Forest plot of inverse-variance weighted MR effects of genetically proxied liability to insomnia symptoms on psychiatric
outcomes. Boxes reflect point estimates and surrounding lines reflect 95% confidence intervals. CI confidence interval, OR odds ratio, BP
bipolar disorder, MDD major depressive disorder, RLS restless leg syndrome, SB suicidal behavior, SCZ schizophrenia.

Fig. 3 Mendelian randomization scatterplots for effects of genetic liability to insomnia on psychiatric outcomes. BP bipolar disorder, MDD
major depressive disorder, SB suicidal behavior, SCZ schizophrenia.

between insomnia symptoms and RLS genetic proxies, and when and may reflect heterogeneity in the insomnia discovery GWAS
we removed SNPs in LD between insomnia symptoms and the based on undiagnosed RLS or pleiotropy at this locus [50]. For BP
significant outcomes (MDD, BP, and SB-cohort-2020) [Supplemen- another single outlier SNP (rs9527083 intergenic in chromosome
tary Tables 5S–9S]. 13) was driving the effect in the opposite direction. For SB-Cohort-
Model-based MR sensitivity analyses were performed to assess 2019 no single outlier was driving in the opposite direction. For
the robustness of effects on potential horizontal pleiotropy comparison, we ran leave-one-out analyses for the IVW method
[Table 2]. Results from the weighted median and MR-PRESSO and found no clear outliers [Supplementary Figs. S6–S10].
were overall similar to results from the IVW analysis. However, the We sought to replicate the findings in the MGB Biobank, an
MR-Egger regression for the effect of genetically proxied liability independent clinical biobank, using available data for MDD, BP
to insomnia symptoms on SB had non-significant findings, and a and SB. These analyses showed replication of the findings for the
point estimate in the opposite direction. We performed a leave- effect of genetic liability to insomnia symptoms on MDD (OR =
one-out MR-Egger analysis to assess whether the results were 1.12, 95% CI = 1.03–1.21, P = 0.0096) and SB (OR = 1.19, 95% CI =
driven by an outlier [Supplementary Figs. S1–S5]. This showed that 1.01–1.40, P = 0.03) but not for BP (P = 0.58). The MGB cohort had
a single outlier SNP (rs113851554 in MEIS1) was driving the effect a small BP case sample size (n = 145) and consequently large
in the opposite direction for MDD, SB in MDD, and SB-Cohort- confidence intervals in the analysis, but the effect was in the same
2020. Notably, this MEIS1 SNP is also a strong risk factor for RLS direction (OR = 1.14) [Supplementary Table 10S]. Finally, the

Neuropsychopharmacology (2022) 47:1672 – 1679

 M. Nassan et al.
1676

Fig. 4 Forest plot of inverse-variance weighted MR effects of genetically proxied liability to RLS on psychiatric outcomes. Boxes reflect
point estimates and surrounding lines reflect 95% confidence intervals. CI confidence interval, OR odds ratio, BP bipolar disorder, MDD major
depressive disorder, RLS restless leg syndrome, SB suicidal behavior, SCZ schizophrenia.

Table 2. MR main and sensitivity analyses results.

Exposure Outcome Outcomes N case/ Sensitivity analyses N SNPsa OR 95% CI P
cont
Insomnia MDD 170,756/329,443 IVW 231 1.23 1.20 1.26 1.37 × 10–61
egger_regression 231 1.09 0.99 1.20 0.07
weighted_median 231 1.18 1.15 1.21 3.49 × 10–35
Pressob na 1.23 1.20 1.26 8.75 × 10–44
Insomnia BP 20,352/31,358 IVW 235 1.15 1.07 1.23 5.11 × 10–5
egger_regression 235 1.33 1.02 1.73 0.03
weighted_median 235 1.13 1.04 1.22 2.39 × 10–3
Pressob
na 1.16 1.09 1.24 6.33 × 10–6
Insomnia SB-cohort- 6569/14,996 IVW 231 1.17 1.07 1.27 0.0002
2019 egger_regression 231 0.85 0.61 1.18 0.34
egger (removing MEIS1 230 0.86 0.59 1.26 0.44
rs113851554)
weighted_median 231 1.13 1.01 1.26 0.04
Presso na na na na na
Insomnia SB in MDD 1622/8786 IVW 231 1.34 1.16 1.54 5.97 × 10–5
egger_regression 231 1.10 0.62 1.93 0.75
weighted_median 231 1.28 1.04 1.58 0.018
Presso na na na na na
Insomnia SB-cohort- 29,782/519,961 IVW 225 1.24 1.18 1.30 1.47 × 10–18
2020 egger_regression 225 0.99 0.81 1.21 0.9
egger (removing MEIS1 224 1.04 0.82 1.31 0.74
rs113851554)
Weighted_median 225 1.18 1.12 1.25 4.61 × 10–10
Pressob
na 1.23 1.18 1.28 4.02 × 10–18
a
Exposure SNPs available in the outcome.
b
Presso outliers: MDD (“rs113851554” “rs2431108” “rs2815757” “rs55772859” “rs73163783”), BP (rs2431108 “rs324017” “rs4090240” “rs521484” “rs6973090”), SB-
cohort-2020 (“rs10502966” “rs12666306” “rs1937447” “rs56133505” “rs73671843”).
Bolded P values indicate significant association.

reverse MR IVW analysis between MDD and BP as exposures and outcomes, indicating that our findings for insomnia were not
insomnia symptoms as outcome showed that genetic liability to driven by RLS. These results were replicated, and consistent across
MDD but not BP is a risk factor for insomnia symptoms several lines of sensitivity analyses. This is the first comprehensive
[Supplementary Table 11S]. study analyzing the causality of insomnia symptoms and RLS for
SB utilizing MR.
We found a robust association between insomnia symptoms
DISCUSSION and MDD (OR = 1.23, P = 1.7 × 10–61), which is consistent with the
In this two-sample MR study we found for the first-time evidence existing literature [34]. In addition, when we sub-classified SB by
for a potentially independent and causal effect of insomnia disorder, the most robust association of insomnia symptoms with
symptoms on SB, and further strengthened the evidence for SB was within the MDD sub-group. However, LD analyses,
insomnia being a potential causal risk factor for MDD and BP. sensitivity MR analyses, and a replication analysis in an
However, genetically proxied RLS (which can be comorbid with independent sample all demonstrated insomnia as an indepen-
insomnia) was not associated with any tested psychiatric dent risk factor for SB independently of MDD. Our results are in

Neuropsychopharmacology (2022) 47:1672 – 1679

 M. Nassan et al.
1677
line with a recent prospective study in which insomnia was an replicate our findings for insomnia as a risk factor for MDD and SB
independent risk factor for SB, with an effect that was more in an independent sample from MGB biobank.
pronounced among patients with MDD (of the total effect, 32% This study has limitations to consider. Although sensitivity
was mediated by MDD) [51]. analyses (including MR-PRESSO) were consistent with the main
Identifying insomnia symptoms as a potentially causal risk results, horizontal pleiotropy cannot be completely excluded [65].
factor SB is important from both mechanistic and clinical On the other hand, the discovery samples for the variants
perspectives. One proposed “stress accumulation” hypothesis is associated with insomnia asked questions more consistent with
that insomnia is associated with the insufficient dissolution of insomnia symptoms rather than an insomnia disorder diagnosis,
emotional distress (due to REM sleep dysfunction and fragmenta- although as mentioned above those insomnia symptoms had a
tion) which can theoretically lead to emotional distress accumula- good correlation with validated scales. Furthermore, as UK
tion. On the other hand, an insomnia effect on the frontal cortex Biobank and 23andMe participants are healthier than the general
might disrupt emotion regulation and lead to disinhibition [52]. population, our findings may not be generalizable to patients with
Collectively, insomnia-induced stress accumulation and behavioral more comorbidities [66]. In addition, there is mixed evidence for
disinhibition might be part of how insomnia causes SB. Another the association between nightmares and suicide [67]. Nonetheless,
proposed hypothesis implicates the hyperarousal state found in further observational and genetic research to test the association
insomnia that has been conceptualized as a biomarker for suicide. between nightmares, insomnia, and SB is warranted. Moreover,
This hypothesis implicates agitation, irritability, and hypervigilance although most of the used cohorts included only European
as mediators between insomnia and SB [53, 54]. ancestry, two cohorts included a minority of other ancestries: SCZ
Our finding that insomnia is a causal risk factor for MDD and SB (included 20% East Asian ancestry), and SB-Cohort-2020 (included
further supports the treatment of insomnia in patients with MDD 4% African and 6% East Asian ancestry). Although only a small
and SB. Treatment of insomnia and depression can be done percentage, and accounted for in the original GWAS; this could
concomitantly or sequentially [55, 56]. Our finding of a bidirec- still introduce a minor limitation in this article. Lastly, as RLS and
tional effect (since reverse MR showed MDD increased the risk for insomnia (and psychiatric disorders) research moves from
insomnia symptoms) supports a concurrent treatment approach subjective symptoms to objective biomarkers to diagnose these
for both disorders. Our results also support the investigation of disorders more accurately, genetic proxies utilizing these
treatment of insomnia for the prevention of SB. A recent clinical biomarker-driven GWAS might create a more homogeneous
trial utilizing the sedative zolpidem—although not meeting the genetic signature and more precise MR results.
primary outcome of reducing the scores on the Scale for Suicide In conclusion, this two-sample MR analysis demonstrated for
Ideation—demonstrated a reduction in suicidal ideation on the the first-time robust evidence for a potentially independent and
Columbia-Suicide Severity Rating Scale in depressed patients with causal effect of insomnia on SB. This finding encourages further
more severe insomnia [57]. On the other hand, an observational clinical trials targeting insomnia for the prevention and
study found that cognitive behavioral therapy for insomnia (CBT-I) treatment of SB.
was associated with a 65% reduction in OR of suicidal thoughts,
independent of changes in depression [58]. However, large
randomized controlled trials evaluating sleep medications and REFERENCES
CBT-I effects on SB are still warranted [59, 60]. 1. Alvaro PK, Roberts RM, Harris JK. A Systematic Review Assessing Bidirectionality
The association of sleep disturbance with BP is well established between Sleep Disturbances, Anxiety, and Depression. Sleep. 2013;36:1059–68.
[61]. Furthermore, treating insomnia in BP patients (via CBT-I) https://doi.org/10.5665/sleep.2810
improved sleep, mood, and functioning [62]. The use of sedative 2. Zhuang S, Na M, Winkelman JW, Ba D, Liu CF, Liu G, et al. Association of Restless
hypnotics to treat mania is a common practice although more Legs Syndrome With Risk of Suicide and Self-harm. JAMA Netw Open. 2019;2:
clinical trials are needed [63]. Although not passing Bonferroni e199966. https://doi.org/10.1001/jamanetworkopen.2019.9966
level of statistical significance, insomnia symptoms were asso- 3. Fernandes SN, Zuckerman E, Miranda R, Baroni A. When Night Falls Fast: Sleep
ciated with a nominal increase in SCZ risk. This is in line with their and Suicidal Behavior Among Adolescents and Young Adults. Child Adolesc
Psychiatr Clin N Am. 2021;30:269–82. https://doi.org/10.1016/j.chc.2020.08.009
known comorbidity and reported increased severity of SCZ and
4. Buysse DJ. Insomnia. JAMA. 2013;309:706–16. https://doi.org/10.1001/jama.2013.
worsening clinical outcomes due to insomnia [11]. 193
Genetically proxied RLS was not significantly associated with 5. Edinger JD, Bonnet MH, Bootzin RR, Doghramji K, Dorsey CM, Espie CA, et al.
any of the studied outcomes. This could be due to multiple Derivation of research diagnostic criteria for insomnia: report of an American
reasons. First, the previous observational studies might have not Academy of Sleep Medicine Work Group. Sleep. 2004;27:1567–96. https://doi.org/
adjusted for relevant variables that were driving the effect (like 10.1093/sleep/27.8.1567
other sleep disturbance for example). One study found that the 6. Bjoroy I, Jorgensen VA, Pallesen S, Bjorvatn B. The prevalence of insomnia sub-
RLS-depression association might be partially explained by sleep types in relation to demographic characteristics, anxiety, depression, alcohol
disturbance and periodic limb movements [21]. Second, the consumption and use of hypnotics. Front Psychol. 2020;11:527. https://doi.org/
10.3389/fpsyg.2020.00527
phenotyping of RLS in previous studies might have created a
7. Scott J, Kallestad H, Vedaa O, Sivertsen B, Etain B. Sleep disturbances and first
heterogeneous group of patients (such as using a survey question onset of major mental disorders in adolescence and early adulthood: a sys-
vs a clinical diagnosis by an expert sub-specialist). The poor tematic review and meta-analysis. Sleep Med Rev. 2021;57:101429. https://doi.
diagnostic performance of survey instruments for RLS has recently org/10.1016/j.smrv.2021.101429
been established [64]. Lastly, larger GWAS in more homogeneous 8. Li L, Wu C, Gan Y, Qu X, Lu Z. Insomnia and the risk of depression: a meta-analysis
phenotype samples may produce a more accurate genetic proxy of prospective cohort studies. BMC Psychiatry. 2016;16:375. https://doi.org/
of RLS that can be utilized to re-test these associations in the 10.1186/s12888-016-1075-3
future. 9. Harris LM, Huang X, Linthicum KP, Bryen CP, Ribeiro JD. Sleep disturbances as risk
The key strength of this study is the use of MR, which reduces factors for suicidal thoughts and behaviours: a meta-analysis of longitudinal
studies. Sci Rep. 2020;10:13888. https://doi.org/10.1038/s41598-020-70866-6
bias due to confounding and assesses causality rather than
10. De Sousa A, Shah B, Shrivastava A. Suicide and schizophrenia: an interplay of
association. Furthermore, the use of large sample sizes (by factors. Curr Psychiatry Rep. 2020;22:65. https://doi.org/10.1007/s11920-020-
selecting the largest published GWAS) leads to more precise 01188-7
estimates of MR effect sizes. This study did not only look at SB in 11. Robertson I, Cheung A, Fan X. Insomnia in patients with schizophrenia: current
general, but also looked at SB classified by diagnosis which is a understanding and treatment options. Prog Neuropsychopharmacol Biol Psy-
rigorous approach. Another strength is that we were able to chiatry. 2019;92:235–42. https://doi.org/10.1016/j.pnpbp.2019.01.016

Neuropsychopharmacology (2022) 47:1672 – 1679

 M. Nassan et al.
1678
12. Miller BJ, Parker CB, Rapaport MH, Buckley PF, McCall WV. Insomnia and suicidal 36. Backhaus J, Junghanns K, Broocks A, Riemann D, Hohagen F. Test-retest reliability
ideation in nonaffective psychosis. Sleep. 2019;42:zsy215. https://doi.org/ and validity of the Pittsburgh Sleep Quality Index in primary insomnia. J Psy-
10.1093/sleep/zsy215 chosom Res. 2002;53:737–40. https://doi.org/10.1016/s0022-3999(02)00330-6
13. Miller BJ, McEvoy JP, McCall WV. Insomnia, suicidal ideation, and suicide attempts 37. Hammerschlag AR, Stringer S, de Leeuw CA, Sniekers S, Taskesen E, Watanabe K,
in the clinical antipsychotic trials of intervention effectiveness. J Clin Psychiatry. et al. Genome-wide association analysis of insomnia complaints identifies risk
2021;82:20m13338. https://doi.org/10.4088/JCP.20m13338 genes and genetic overlap with psychiatric and metabolic traits. Nat Genet.
14. Madrid-Valero JJ, Rubio-Aparicio M, Gregory AM, Sanchez-Meca J, Ordonana JR. 2017;49:1584–92. https://doi.org/10.1038/ng.3888
The heritability of insomnia: systematic review and meta-analysis of twin studies. 38. Howard DM, Adams MJ, Clarke TK, Hafferty JD, Gibson J, Shirali M, et al. Genome-
Sleep Med Rev. 2021;58:101437. https://doi.org/10.1016/j.smrv.2021.101437 wide meta-analysis of depression identifies 102 independent variants and
15. Dashti HS, Ordovas JM. Genetics of sleep and insights into its relationship with highlights the importance of the prefrontal brain regions. Nat Neurosci.
obesity. Annu Rev Nutr. 2021;41:223–52. https://doi.org/10.1146/annurev-nutr- 2019;22:343–52. https://doi.org/10.1038/s41593-018-0326-7
082018-124258 39. Stahl EA, Breen G, Forstner AJ, McQuillin A, Ripke S, Trubetskoy V, et al. Genome-
16. Picchietti DL, Van Den Eeden SK, Inoue Y, Berger K. Achievements, challenges, wide association study identifies 30 loci associated with bipolar disorder. Nat
and future perspectives of epidemiologic research in restless legs syndrome Genet. 2019;51:793–803. https://doi.org/10.1038/s41588-019-0397-8
(RLS). Sleep Med. 2017;31:3–9. https://doi.org/10.1016/j.sleep.2016.06.007 40. Ripke S, Walters JT, O’Donovan MC. Mapping genomic loci prioritises genes and
17. Winkelmann J, Prager M, Lieb R, Pfister H, Spiegel B, Wittchen HU, et al. "Anxietas implicates synaptic biology in schizophrenia. medRxiv: 2020.2009.2012.20192922
tibiarum". Depression and anxiety disorders in patients with restless legs syn- [Preprint]. 2020. Available from: https://doi.org/10.1101/2020.09.12.20192922.
drome. J Neurol. 2005;252:67–71. https://doi.org/10.1007/s00415-005-0604-7 41. Mullins N, Bigdeli TB, Borglum AD, Coleman JRI, Demontis D, Mehta D, et al.
18. Winkelman JW, Redline S, Baldwin CM, Resnick HE, Newman AB, Gottlieb DJ. GWAS of Suicide Attempt in Psychiatric Disorders and Association With Major
Polysomnographic and health-related quality of life correlates of restless legs Depression Polygenic Risk Scores. Am J Psychiatry. 2019;176:651–60. https://doi.
syndrome in the Sleep Heart Health Study. Sleep. 2009;32:772–8. https://doi.org/ org/10.1176/appi.ajp.2019.18080957
10.1093/sleep/32.6.772 42. Mullins N, Kang J, Campos AI, Coleman JRI, Edwards AC, Galfalvy H, et al. Dis-
19. Mallon L, Broman JE, Hetta J. Restless legs symptoms with sleepiness in relation to secting the shared genetic architecture of suicide attempt, psychiatric disorders
mortality: 20-year follow-up study of a middle-aged Swedish population. Psychiatry and known risk factors. medRxiv. 2020:2020.12.01.20241281. https://doi.org/
Clin Neurosci. 2008;62:457–63. https://doi.org/10.1111/j.1440-1819.2008.01831.x 10.1101/2020.12.01.20241281.
20. Li Y, Mirzaei F, O'Reilly EJ, Winkelman J, Malhotra A, Okereke OI, et al. Prospective 43. Mullins N, Kang J, Campos AI, Coleman JRI, Edwards AC, Galfalvy H, et al. Dis-
study of restless legs syndrome and risk of depression in women. Am J Epidemiol. secting the Shared Genetic Architecture of Suicide Attempt, Psychiatric Disorders,
2012;176:279–88. https://doi.org/10.1093/aje/kws016 and Known Risk Factors. Biol Psychiatry. 2021. https://doi.org/10.1016/j.
21. Koo BB, Blackwell T, Lee HB, Stone KL, Louis ED, Redline S, et al. Restless Legs biopsych.2021.05.029
Syndrome and Depression: Effect Mediation by Disturbed Sleep and Periodic 44. Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-
Limb Movements. Am J Geriatr Psychiatry. 2016;24:1105–16. https://doi.org/ Base platform supports systematic causal inference across the human phenome.
10.1016/j.jagp.2016.04.003 Elife. 2018;7. https://doi.org/10.7554/eLife.34408.
22. Zhang C, Li Y, Malhotra A, Ning Y, Gao X. Restless legs syndrome status as a 45. Burgess S, Davey Smith G, Davies NM, Dudbridge F, Gill D, Glymour MM, et al.
predictor for lower physical function. Neurology. 2014;82:1212–8. https://doi.org/ Guidelines for performing Mendelian randomization investigations. Wellcome
10.1212/WNL.0000000000000284 Open Res. 2019;4:186. https://doi.org/10.12688/wellcomeopenres.15555.2
23. Li Y, Wang W, Winkelman JW, Malhotra A, Ma J, Gao X. Prospective study of 46. Dashti HS, Redline S, Saxena R. Polygenic risk score identifies associations
restless legs syndrome and mortality among men. Neurology. 2013;81:52–9. between sleep duration and diseases determined from an electronic medical
https://doi.org/10.1212/WNL.0b013e318297eee0 record biobank. Sleep. 2019;42:zsy247. https://doi.org/10.1093/sleep/zsy247
24. Molnar MZ, Lu JL, Kalantar-Zadeh K, Kovesdy CP. Association of incident restless 47. Dashti HS, Cade BE, Stutaite G, Saxena R, Redline S, Karlson EW. Sleep health,
legs syndrome with outcomes in a large cohort of US veterans. J Sleep Res. diseases, and pain syndromes: findings from an electronic health record biobank.
2016;25:47–56. https://doi.org/10.1111/jsr.12335 Sleep. 2021;44. https://doi.org/10.1093/sleep/zsaa189.
25. Winkelmann J, Schormair B, Xiong L, Dion PA, Rye DB, Rouleau GA. Genetics of 48. Dashti HS, Daghlas I, Lane JM, Huang Y, Udler MS, Wang H, et al. Genetic
restless legs syndrome. Sleep Med. 2017;31:18–22. https://doi.org/10.1016/j. determinants of daytime napping and effects on cardiometabolic health. Nat
sleep.2016.10.012 Commun. 2021;12:900. https://doi.org/10.1038/s41467-020-20585-3
26. Schormair B, Zhao C, Bell S, Tilch E, Salminen AV, Putz B, et al. Identification of 49. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, et al. PLINK: a
novel risk loci for restless legs syndrome in genome-wide association studies in tool set for whole-genome association and population-based linkage analyses.
individuals of European ancestry: a meta-analysis. Lancet Neurol. 2017;16:898– Am J Hum Genet. 2007;81:559–75. https://doi.org/10.1086/519795
907. https://doi.org/10.1016/S1474-4422(17)30327-7 50. El Gewely M, Welman M, Xiong L, Yin S, Catoire H, Rouleau G, et al. Reassessing
27. Kraemer HC, Kazdin AE, Offord DR, Kessler RC, Jensen PS, Kupfer DJ. Coming to GWAS findings for the shared genetic basis of insomnia and restless legs syn-
terms with the terms of risk. Arch Gen Psychiatry. 1997;54:337–43. https://doi.org/ drome. Sleep. 2018;41. https://doi.org/10.1093/sleep/zsy164.
10.1001/archpsyc.1997.01830160065009 51. Anna Karin H, Hossjer O, Bellocco R, Ye W, Trolle LY, Akerstedt T. Insomnia in the
28. Jelinek GA. Determining causation from observational studies: a challenge for context of short sleep increases suicide risk. Sleep. 2021;44. https://doi.org/
modern neuroepidemiology. Front Neurol. 2017;8:265. https://doi.org/10.3389/ 10.1093/sleep/zsaa245.
fneur.2017.00265 52. Van Someren EJW. Brain mechanisms of insomnia: new perspectives on causes
29. Wootton RE, Jones HJ, Sallis HM. Mendelian randomisation for psychiatry: how and consequences. Physiol Rev. 2021;101:995–1046. https://doi.org/10.1152/
does it work, and what can it tell us? Mol Psychiatry. 2022;27:53–57. https://doi. physrev.00046.2019
org/10.1038/s41380-021-01173-3 53. Bonnet MH, Arand DL. Hyperarousal and insomnia: state of the science. Sleep
30. Evans DM, Davey Smith G. Mendelian randomization: new applications in the Med Rev. 2010;14:9–15. https://doi.org/10.1016/j.smrv.2009.05.002
coming age of hypothesis-free causality. Annu Rev Genomics Hum Genet. 54. Calati R, Nemeroff CB, Lopez-Castroman J, Cohen LJ, Galynker I. Candidate bio-
2015;16:327–50. https://doi.org/10.1146/annurev-genom-090314-050016 markers of suicide crisis syndrome: what to test next? A concept paper. Int J
31. Voracek M, Loibl LM. Genetics of suicide: a systematic review of twin studies. Neuropsychopharmacol. 2020;23:192–205. https://doi.org/10.1093/ijnp/pyz063
Wien Klin Wochenschr. 2007;119:463–75. https://doi.org/10.1007/s00508-007- 55. Asarnow LD, Manber R. Cognitive behavioral therapy for insomnia in depression.
0823-2 Sleep Med Clin. 2019;14:177–84. https://doi.org/10.1016/j.jsmc.2019.01.009
32. Roy A, Segal NL. Suicidal behavior in twins: a replication. J Affect Disord. 56. Winkelman JW. How to identify and fix sleep problems: better sleep, better
2001;66:71–74. https://doi.org/10.1016/s0165-0327(00)00275-5 mental health. JAMA Psychiatry. 2020;77:99–100. https://doi.org/10.1001/
33. Davies NM, Holmes MV, Davey Smith G. Reading Mendelian randomisation stu- jamapsychiatry.2019.3832
dies: a guide, glossary, and checklist for clinicians. BMJ. 2018;362:k601. https:// 57. McCall WV, Benca RM, Rosenquist PB, Youssef NA, McCloud L, Newman JC, et al.
doi.org/10.1136/bmj.k601 Reducing Suicidal Ideation Through Insomnia Treatment (REST-IT): A Randomized
34. Jansen PR, Watanabe K, Stringer S, Skene N, Bryois J, Hammerschlag AR, et al. Clinical Trial. Am J Psychiatry. 2019;176:957–65. https://doi.org/10.1176/appi.
Genome-wide analysis of insomnia in 1,331,010 individuals identifies new risk ajp.2019.19030267
loci and functional pathways. Nat Genet. 2019;51:394–403. https://doi.org/ 58. Trockel M, Karlin BE, Taylor CB, Brown GK, Manber R. Effects of cognitive beha-
10.1038/s41588-018-0333-3 vioral therapy for insomnia on suicidal ideation in veterans. Sleep. 2015;38:259–
35. Morin CM, Belleville G, Belanger L, Ivers H. The Insomnia Severity Index: psy- 65. https://doi.org/10.5665/sleep.4410
chometric indicators to detect insomnia cases and evaluate treatment response. 59. Tubbs AS, Fernandez FX, Ghani SB, Karp JF, Patel SI, Parthasarathy S, et al. Pre-
Sleep. 2011;34:601–8. https://doi.org/10.1093/sleep/34.5.601 scription medications for insomnia are associated with suicidal thoughts and

Neuropsychopharmacology (2022) 47:1672 – 1679

 M. Nassan et al.
1679
behaviors in two nationally representative samples. J Clin Sleep Med. FUNDING
2021;17:1025–30. https://doi.org/10.5664/jcsm.9096 RS and JW are partially supported by NIH grant R01HL146751.
60. Pigeon WR, Funderburk JS, Cross W, Bishop TM, Crean HF. Brief CBT for insomnia
delivered in primary care to patients endorsing suicidal ideation: a proof-of-
concept randomized clinical trial. Transl Behav Med. 2019;9:1169–77. https://doi. COMPETING INTERESTS
org/10.1093/tbm/ibz108 None of the individual authors have conflict of interest to disclose. International
61. Kaplan KA. Sleep and sleep treatments in bipolar disorder. Curr Opin Psychol. Suicide Genetics Consortium disclosures for their GWAS of suicidal behavior:
2020;34:117–22. https://doi.org/10.1016/j.copsyc.2020.02.001 statistical analyses were carried out on the NL Genetic Cluster Computer (http://
62. Harvey AG, Soehner AM, Kaplan KA, Hein K, Lee J, Kanady J, et al. Treating www.geneticcluster.org) hosted by SURFsara and the Mount Sinai high performance
insomnia improves mood state, sleep, and functioning in bipolar disorder: a pilot computing cluster (http://hpc.mssm.edu), which is supported by the Office of
randomized controlled trial. J Consult Clin Psychol. 2015;83:564–77. https://doi. Research Infrastructure of the National Institutes of Health under award numbers
org/10.1037/a0038655 S10OD018522 and S10OD026880. This work was conducted in part using the
63. Plante DT, Winkelman JW. Sleep disturbance in bipolar disorder: therapeutic resources of the Advanced Computing Center for Research and Education at
implications. Am J Psychiatry. 2008;165:830–43. https://doi.org/10.1176/appi. Vanderbilt University, Nashville, TN. This work is supported by R01MH116269 (DMR)
ajp.2008.08010077 and R01MH121455 (DMR). Research reported was also supported by NIGMS of the
64. Fulda S, Allen RP, Earley CJ, Hogl B, Garcia-Borreguero D, Inoue Y, et al. We need National Institutes of Health under award number T32GM007347 (JK). The content is
to do better: A systematic review and meta-analysis of diagnostic test accuracy of solely the responsibility of the authors and does not necessarily represent the official
restless legs syndrome screening instruments. Sleep Med Rev. 2021;58:101461. views of the National Institutes of Health.
https://doi.org/10.1016/j.smrv.2021.101461
65. Verbanck M, Chen CY, Neale B, Do R. Detection of widespread horizontal pleio-
tropy in causal relationships inferred from Mendelian randomization between
ADDITIONAL INFORMATION
complex traits and diseases. Nat Genet. 2018;50:693–8. https://doi.org/10.1038/
s41588-018-0099-7 Supplementary information The online version contains supplementary material
66. Fry A, Littlejohns TJ, Sudlow C, Doherty N, Adamska L, Sprosen T, et al. Com- available at https://doi.org/10.1038/s41386-022-01319-z.
parison of Sociodemographic and Health-Related Characteristics of UK Biobank
Participants With Those of the General Population. Am J Epidemiol. Correspondence and requests for materials should be addressed to Richa Saxena.
2017;186:1026–34. https://doi.org/10.1093/aje/kwx246
67. Hedstrom AK, Bellocco R, Hossjer O, Ye W, Trolle Lagerros Y, Akerstedt T. The Reprints and permission information is available at http://www.nature.com/
relationship between nightmares, depression and suicide. Sleep Med. 2021;77:1– reprints
6. https://doi.org/10.1016/j.sleep.2020.11.018
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims
in published maps and institutional affiliations.

ACKNOWLEDGEMENTS
We thank the GWAS consortia that have made their data publicly available. We thank
the International Suicide Genetics Consortium that shared its summary statistics
with us. This article is licensed under a Creative Commons Attribution 4.0
International License, which permits use, sharing, adaptation,
distribution and reproduction in any medium or format, as long as you give appropriate
AUTHOR CONTRIBUTIONS credit to the original author(s) and the source, provide a link to the Creative Commons
All authors contributed to drafting the work or revising it critically for important licence, and indicate if changes were made. The images or other third party material in
intellectual content, gave final approval of the version to be published; and provided this article are included in the article's Creative Commons licence, unless indicated
an agreement to be accountable for all aspects of the work. MN and ID provided otherwise in a credit line to the material. If material is not included in the article's
substantial contributions to the conception or design of the work, and the Creative Commons licence and your intended use is not permitted by statutory
acquisition, analysis, or interpretation of data for the work. JW, HSD, and RS provided regulation or exceeds the permitted use, you will need to obtain permission directly
substantial contributions to the conception or design of the work, and the from the copyright holder. To view a copy of this licence, visit http://creativecommons.
interpretation of data for the work. International Suicide Genetics Consortium org/licenses/by/4.0/.
provided summary statistics, provided substantial contributions to the interpretation
of data for the work, contributed to revising the work critically for important
intellectual content, and gave final approval of the version to be published. © The Author(s) 2022

Neuropsychopharmacology (2022) 47:1672 – 1679