The quintiles of resting heart rate were as follows: 35 to 61 beats per minutes for the first quintile, 62 to 68 beats per minutes for the second quintile, 69 to 74 beats per minutes for the third quintile, 75 to 82 beats per minutes for the fourth quintile, and 83 to 145 beats per minutes for the fifth quintile. Resting heart rate was measured at a mean (SD) age of 18.3 (0.6) years at conscription testing. Follow-up started at conscription and lasted until the first diagnosis of each disorder, emigration, death, or end of register coverage (December 31, 2013), whichever came first.
eTable 1. ICD codes for psychiatric disorders and definition of violent criminality
eTable 2. Rate of treatments for psychiatric disorders and violent convictions by quintiles of resting heart rate
eTable 3. Rate of treatments for psychiatric disorders and violent convictions by quintiles of systolic blood pressure
eTable 4. Rate of treatments for psychiatric disorders and violent convictions by quintiles of diastolic blood pressure
eTable 5. Cox proportional hazard ratios (95% confidence intervals) for psychiatric disorders and violent criminality associated with quintiles of systolic blood pressure
eTable 6. Cox proportional hazard ratios (95% confidence intervals) for psychiatric disorders and violent criminality associated with quintiles of diastolic blood pressure
eTable 7. Cox proportional hazard ratios (95% confidence intervals) from sensitivity analyses for resting heart rate
eTable 8. Cox proportional hazard ratios (95% confidence intervals) from sensitivity analyses for diastolic blood pressure
eFigure 1. Kaplan-Meier Survival Curves by Top (solid lines) and Bottom Quintiles (dashed lines) of Resting Heart Rate
eFigure 2. Kaplan-Meier Survival Curves by Top (solid lines) and Bottom Quintiles (dashed lines) of Systolic Blood Pressure
eFigure 3. Kaplan-Meier Survival Curves by Top (solid lines) and Bottom Quintiles (dashed lines) of Diastolic Blood Pressure
Latvala A, Kuja-Halkola R, Rück C, D’Onofrio BM, Jernberg T, Almqvist C, Mataix-Cols D, Larsson H, Lichtenstein P. Association of Resting Heart Rate and Blood Pressure in Late Adolescence With Subsequent Mental DisordersA Longitudinal Population Study of More Than 1 Million Men in Sweden. JAMA Psychiatry. Published online October 26, 2016. doi:10.1001/jamapsychiatry.2016.2717
Are differences in cardiac autonomic function (as indexed by resting heart rate and blood pressure) associated with subsequent psychiatric disorders?
In this longitudinal cohort study of more than 1 million men in Sweden, a higher resting heart rate in late adolescence was associated with an increased risk for subsequent obsessive-compulsive disorder, schizophrenia, and anxiety disorders, whereas a lower resting heart rate in late adolescence was associated with an increased risk for substance use disorders and violent criminality. Similar associations were observed with systolic/diastolic blood pressure.
Differences in autonomic nervous system functioning may predate or represent an early marker of psychiatric disorders.
Differences in cardiovascular autonomic activity between individuals with psychiatric disorders and healthy controls have been observed, but whether cardiovascular autonomic abnormalities are associated with subsequent psychiatric disorders is unknown.
To investigate whether differences in cardiac autonomic function as indexed by resting heart rate and blood pressure are associated with psychiatric disorders during the lifetime of men in Sweden.
Design, Setting, and Participants
We conducted a longitudinal register-based study of Swedish men whose resting heart rate (n = 1 039 443) and blood pressure (n = 1 555 979) were measured at military conscription at a mean (SD) age of 18.3 (0.6) years during the period from 1969 to 2010, with register-based follow-up data available until the end of 2013. Analyses were performed from November 18, 2015, to June 9, 2016.
Main Outcomes and Measures
Dates of inpatient/outpatient diagnoses of anxiety disorders, obsessive-compulsive disorder, posttraumatic stress disorder, depressive disorders, bipolar disorder, schizophrenia, and substance use disorders and convictions for violent crimes, between 1973 and 2013, were obtained from nationwide registers. Adjustments were made for height, weight, body mass index, cardiorespiratory fitness, cognitive ability, and socioeconomic covariates.
After adjustment for covariates, Cox regression models with up to 45 years of follow-up data showed that men (mean [SD] age of 18.3 [0.6] years at conscription) with resting heart rates above 82 beats per minute had a 69% (95% CI, 46%-94%) increased risk for obsessive-compulsive disorder, a 21% (95% CI, 11%-33%) increased risk for schizophrenia, and an 18% (95% CI, 13%-22%) increased risk for anxiety disorders compared with men with resting heart rates below 62 beats per minute. Similar associations were observed with systolic/diastolic blood pressure. In contrast, lower resting heart rate and lower systolic blood pressure were associated with substance use disorders and violent criminality.
Conclusions and Relevance
Our results suggest that for men, differences in heart rate and blood pressure in late adolescence are associated with lifetime major psychiatric disorders, with higher levels associated with obsessive-compulsive disorder, schizophrenia, and anxiety disorders and lower levels associated with substance use disorders and violent behavior. Differences in autonomic nervous system functioning may predate or represent an early marker of psychiatric disorders.
Abnormalities in cardiac autonomic functioning have been observed in a range of psychiatric disorders.1 For example anxiety disorders,2 depression,3 and schizophrenia4,5 have been associated with reduced beat-to-beat variation in heart rate, indicating reduced parasympathetic activity.
The resting heart rate (RHR) is an easy-to-measure clinical variable indexing cardiac autonomic nervous system activity6 and is known to predict the health and lifespan of an individual.7- 9 In contrast to the confirmed associations with somatic health, the link between RHR and psychiatric problems is less clear. An accelerated heart rate during panic attacks is a core symptom of panic disorder, and an elevated RHR in patients with schizophrenia was suggested already in early studies.10 More recent studies have reported an elevated RHR in patients with major depression,11 posttraumatic stress disorder,12,13 generalized anxiety disorder,14 panic disorder,15 or psychotic disorders.16 Despite these suggestive findings, studies have often been limited by small and selected samples, and the associations between RHR and different psychiatric disorders have not been studied systematically.
Besides RHR, alterations in blood pressure (BP), regulated by the autonomic nervous system as a function of cardiac output and vascular resistance,17 have been observed in patients with schizophrenia,5 anxiety,18,19 or depression.20 However, the results have been inconsistent, with some studies reporting higher BP and others reporting lower BP. Furthermore, the findings may have been partly due to the adverse effects of medication.20
The observed cardiac autonomic abnormalities are often assumed to be induced by psychiatric problems. Psychiatric disorders increase the risk for cardiovascular morbidity and mortality,21- 23 and altered autonomic functioning has been hypothesized to mediate the association.23,24 However, because autonomic functions are intrinsically involved in emotions,25 it is also possible that autonomic activity precedes psychopathology rather than vice versa. While cross-sectional comparisons of psychiatric patients and healthy controls have been typical, some prospective longitudinal studies have been published. For example, depressive symptoms were associated with a higher RHR and a lower heart rate variability in men26 and with a low BP in the elderly,27 and a low heart rate variability before military deployment was associated with posttraumatic stress disorder after exposure to combat.28 In a subset of the current cohort, we found that a lower RHR in late adolescence was associated with subsequent violent and nonviolent antisocial behavior.29
In this large-scale cohort study, we aimed to investigate whether differences in autonomic function as indexed by RHR and BP were associated with psychiatric disorders during the lifetimes of men in Sweden. Using population-based register data of men with a follow-up period of up to 45 years, we studied the associations of RHR and BP in late adolescence with diagnoses of major psychiatric disorders later in life.
We linked several Swedish national registers with longitudinal data available until the end of 2013. A unique personal identification number, given to all citizens at birth, was used for the register linkages. Resting heart rate and BP were measured during a 2-day conscription assessment for the Swedish Armed Forces, mandatory until 2010 for Swedish men at 18 years of age. Only men with severe diseases, handicaps, or intellectual disability were exempt, and more than 95% of the men generally attended the conscription assessment.30 We combined information from the Total Population Register and the Conscription Register to identify men who were born in Sweden and conscripted during the register’s coverage period from 1969 to 2010 (N = 1 840 911). To enable linkage with parental data, we identified the parents of each individual from the Multi-Generation Register,31 and those with missing information on parents’ identity were excluded (n = 28 759). Information about emigrations and deaths was available from the Migration Register and the Cause of Death Register, respectively. We excluded men who had emigrated from Sweden before conscription (n = 17 791). The total study population thus included 1 794 361 conscripted men. The study was approved by the Regional Ethical Review Board of Stockholm, Sweden. No informed consent was required for the register-based study because the participants were not approached personally.
Resting heart rate and BP were measured while the individual was in the supine position, with an appropriate cuff at heart level after 5 to 10 minutes of rest during the conscription examination.32 Similar to our earlier study,29 we excluded as outliers or potential data errors RHR measurements below 35 and above 145 beats per minute (n = 172), systolic BP (SBP) measurements below 80 and above 180 mm Hg (n = 503), and diastolic BP (DBP) measurements below 40 and above 100 mm Hg (n = 1434). Possibly owing to changes in the registering procedures, RHR measurements were almost completely missing from the Conscription Register for the years from 1984 to 1993, and data coverage varied also during other periods. A valid RHR measurement was available for 1 039 443 men, a valid SBP measurement was available for 1 555 979 men, and a valid DBP measurement was available for 1 554 829 men.
The National Patient Register contains information about all psychiatric inpatient admissions in Sweden since 1973 and all outpatient treatments since 2001.33 Diagnoses were made by the attending physician according to Swedish translations of the International Classification of Diseases (ICD) revisions 8 (1969-1986), 9 (1987-1996), or 10 (1997-present). We studied 7 categories of major psychiatric disorders: anxiety disorders (ICD-10 codes F40, F41, F44, F45, and F48), obsessive-compulsive disorder (OCD) (code F42), posttraumatic stress disorder (code F43), depressive disorders (codes F32-F34, F38, and F39), bipolar disorder (codes F30 and F31), schizophrenia (including schizoaffective disorder) (codes F20 and F25), and substance use disorders (SUDs) (codes F10-F19). The validity of psychiatric diagnoses in the National Patient Register is generally good.33,34 We also included violent criminal convictions from the Swedish Crime Register as an indicator of antisocial behavior. Full ICD codes and a description of violent convictions are given in eTable 1 in the Supplement.
We included several potential confounders as covariates. Height, weight, and body mass index were included because they may be associated with both cardiovascular functioning and the risk for mental disorders.35 We adjusted for general cognitive ability (IQ), assessed with the Swedish Enlistment Battery30 as part of the conscription testing, because IQ is associated with psychiatric morbidity,36 although its association with RHR and BP is unknown. Because differences in RHR and BP exist among different ethnic groups, and because immigration may be associated with an altered risk for mental disorders, we adjusted for parental immigrant status. The family’s socioeconomic status during the man’s childhood, available from National Censuses in 1960, 1970, 1980, 1985, and 1990, was also included owing to its possible associations with psychopathology and cardiovascular measures.37 The socioeconomic status variable was derived from the occupation of the head of the household and coded into 3 classes: low (skilled and unskilled workers across all fields), medium (low- and intermediate-position white collar workers), and high (high-position white collar workers and self-employed professionals and entrepreneurs). Because RHR and BP negatively correlate with physical fitness, which may be associated with psychiatric disorders, we adjusted for maximal workload (Wmax, measured in watts) achieved in an exercise test with a cycle ergometer during the conscription testing.38 Wmax was available for a subset of 1 306 369 men (782 545 with an RHR measurement). The Wmax value was divided by weight to take into account body size. Finally, we included birth and conscription years to adjust for potential cohort and period effects.
We constructed Cox proportional hazards regression models to estimate the relative hazards of receiving diagnoses of psychiatric disorders and being convicted of violent crimes during the follow-up period starting from conscription. In separate models, we followed the participants until the first occurrence of each outcome (ie, first diagnosis of each psychiatric disorder and first conviction for a violent offense). Men with psychiatric diagnoses or violent convictions predating conscription were excluded from the respective models. Those for whom the studied outcome did not occur within the study period contributed person-time at risk until the end of follow-up (end of 2013), emigration, or death, whichever occurred first.
As main analyses, we constructed Cox regression models for each outcome using variables denoting 10-unit increases in RHR, SBP, and DBP, as well as with quintile categorizations. We chose quintiles as an illustrative way to analyze the associations while allowing for potential nonlinearity. The first model only adjusted for birth and conscription years. The second model added all covariates with the exception of cardiorespiratory fitness, and the third model also adjusted for fitness in the subset of men with exercise test data. Nonindependence of brothers was adjusted for by using a robust sandwich estimator for standard errors. Graphical inspection of the Schoenfeld residuals for RHR, SBP, and DBP revealed no violations of the proportional hazards assumption.
We conducted several sensitivity analyses to ensure the validity of our results. First, to ensure that the results for the different psychiatric disorders were not due to comorbidities, we included each disorder only if it was the first-occurring psychiatric diagnosis for the participant, censoring men with other predating diagnoses at the date of the first diagnosis. Second, to rule out the potential effects of preexisting cardiovascular conditions, we excluded men with any diagnoses of cardiovascular diseases (ICD-8/9 codes 390-469; ICD-10 codes I00-I99) before conscription (n = 8551). Third, to reduce the likelihood that the observed associations would be caused by acute psychopathology at the time of measurement, we began follow-up 5 years after conscription. Fourth, to reduce bias due to potential unobserved disorders among the older cohorts (ie, men conscripted before the start of the National Patient Register on January 1, 1973), we repeated the models with follow-up starting in 1973. Fifth, because the validity of ICD-8/9 diagnoses in the National Patient Register may be less optimal, we reran the models among men conscripted during the period from 1997 to 2010 (n = 474 750) with ICD-10 diagnoses only. Sixth, to inspect potential bias caused by periods with low coverage of RHR data, we reanalyzed the associations using only data from years with more complete coverage. Finally, we tested whether listwise deletion of missing values in covariates might have biased the results by investigating the crude associations among individuals without missing data on covariates.
Statistical significance was assessed using 95% CIs in Cox regression models. Whenever the 95% CI covered the null, the association was considered nonsignificant; whenever it did not, the association was considered significant.
Descriptive data of the study population are presented in Table 1. eTables 2 to 4 in the Supplement display the person-years at risk and the incidence rates for psychiatric disorders and violent convictions by quintiles of RHR, SBP, and DBP. The unadjusted statistics suggested some associations between the cardiovascular measures and psychiatric disorders; Kaplan-Meier survival curves for selected outcomes are shown in the Figure, and Kaplan-Meier survival curves for all outcomes are shown in eFigures 1 to 3 in the Supplement. The median length of follow-up was 31.8 years (range, 0-45.2 years) in the RHR analyses.
Table 2 gives an overview of the results from the Cox regression models, showing hazard ratios associated with 10-unit increases in RHR, SBP, and DBP. In the base models adjusting only for birth and conscription years, every 10-unit increase in RHR was associated with an 8% (95% CI, 7%-9%) increased risk for anxiety disorders, 18% (95% CI, 15%-21%) increased risk for OCD, 5% (95% CI, 4%-6%) increased risk for depression, and 10% (95% CI, 8%-12%) increased risk for schizophrenia, as well as a 6% (95% CI, 5%-6%) reduced risk for violent criminality. The associations were weakened when adjusting for the covariates but remained significantly different from unity. Similar results emerged for the BP measurements; for example, a 10-unit increase in DBP was associated with elevated risks for anxiety (hazard ratio, 1.06 [95% CI, 1.05-1.07]), OCD (hazard ratio, 1.11 [95% CI, 1.08-1. 15]), and schizophrenia (hazard ratio, 1.11 [95% CI, 1.08-1.13]) in fully adjusted models.
The results for RHR quintiles concurred with the previous analysis; a higher RHR was most strongly associated with an elevated risk for OCD (Table 3). In the fully adjusted analysis, men whose RHR was above 82 beats per minute had a 69% (95% CI, 46%-94%) increased risk for OCD compared with men whose RHR was below 62 beats per minute. Similarly, men in the highest RHR category had a 21% (95% CI, 11%-33%) increased risk for schizophrenia and an 18% (95% CI, 13%-22%) increased risk for anxiety. Substance use disorders and violent convictions were associated with a lower RHR, especially after adjustment for physical fitness. The associations for OCD, schizophrenia, anxiety, SUDs, and violence were generally replicated with the BP measurements (eTables 5 and 6 in the Supplement).
We found no significant bias due to the various tested factors for our main results (eTables 7 and 8 in the Supplement). Excluding men with comorbid psychiatric diagnoses or cardiovascular diseases before conscription did not reduce the associations of higher RHR and DBP with anxiety, OCD, and schizophrenia. Similarly, starting the follow-up only 5 years after conscription had no effect on the associations. When only ICD-10 diagnoses were included, the results were commensurate with the main analyses, although the estimates were not as precise because of the reduced follow-up period and sample size. Finally, the associations among men without missing data on covariates closely resembled those among all men.
In this large-scale longitudinal cohort study, we found men with higher RHR and higher BP in late adolescence to be more likely to have received a diagnosis of OCD, schizophrenia, or anxiety disorder later in life. The strongest associations were seen with OCD; men in the highest fifth of the RHR distribution (RHR > 82 beats per minute) had a 70% to 80% increased risk in analyses with varying adjustments compared with men in the lowest fifth (RHR < 62 beats per minute). Similarly, men with the highest DBP (>77 mm Hg) had a 30% to 40% higher risk for OCD than men with the lowest DBP (<60 mm Hg). For schizophrenia and anxiety disorders, the corresponding risk increases ranged between 20% and 40%. These effect sizes are comparable to the established associations of RHR with cardiovascular disease and mortality; a meta-analysis linked RHR greater than 80 beats per minute with a 45% increased risk for all-cause mortality and a 33% increased risk for cardiovascular mortality,7 and an analysis of 3 prospective cohort studies found RHR above 72 beats per minute to increase the risk of incident heart failure by 30% after adjustments.8 Our findings are novel; there are no previous prospective studies linking these cardiovascular measures to subsequent OCD, schizophrenia, or anxiety disorders.
We also found that the risk for SUDs and antisocial behavior was predicted by lower RHR and SBP, especially when adjusting for the confounding effect of physical fitness. While we are unaware of studies on SUDs, we previously reported in a subset of the present data the association of lower RHR with violent and nonviolent criminality, as well as assaults and unintentional injuries.29 Four additional years of follow-up were available for the present analyses.
Our results from a prospective design excluding men with prior psychiatric diagnoses suggest that altered cardiac autonomic activity may precede or represent an early marker for a range of psychiatric disorders in men. While heart rate and BP are influenced by both branches of the autonomic nervous system, parasympathetic influences dominate in resting conditions.6 Our findings are thus compatible with earlier findings indicating reductions in parasympathetic function in anxiety disorders and schizophrenia.3,6 We extend these earlier observations into a longitudinal design considering autonomic activity as a potential predictor of psychiatric morbidity. While our findings clearly cannot establish causality, they are compatible with an association between altered autonomic activity and subsequent psychiatric problems. This contrasts with the often hypothesized opposite direction.23,24
Our results may also be interpreted as reflecting potential underlying domains of psychopathology. One interpretation is provided by altered autonomic functioning being linked with neuroticism,39 which is an important precursor of psychiatric illness. Similarly, the temperament trait behavioral inhibition is associated with elevated RHR and other autonomic measures in childhood,40 and is predictive of future anxiety disorders.41 Importantly, the cortical and subcortical brain regions involved in autonomic processing include the amygdala, insula, and cingulate cortex, all of which play a role in emotion processing and have been linked to psychiatric morbidity.42 Our results are also potentially interesting for transdiagnostic models of psychopathology such as the Research Domain Criteria.43 In the Research Domain Criteria matrix, heart rate and BP serve as indicators for 2 separate constructs: acute threat (“fear”) and arousal. Our findings prospectively link the elevated levels on these suggested dimensions to the ICD-based diagnoses of OCD, schizophrenia, and anxiety disorders and prospectively link the lower levels to SUDs and antisocial behavior. This fits well with theoretical accounts taking low RHR in antisocial individuals to indicate fearlessness or chronically low arousal.29,44 Similarly, elevated levels of fear and/or arousal in individuals with OCD, anxiety, or schizophrenia are consistent with the clinical presentation of these disorders.
The included physical, cognitive, and socioeconomic covariates had only marginal effects on the associations, whereas poorer cardiorespiratory fitness partly accounted for the links between higher RHR and psychiatric morbidity. Mental disorders are associated with less physical activity and poorer fitness,5 and fitness correlates negatively with RHR.45 Also, in our data, RHR and cardiorespiratory fitness were modestly negatively correlated (ρ = −0.25 [95% CI, −0.26 to −0.25]).
The interpretation of our findings must be considered in light of several limitations. First, our study only included data on men, and it is uncertain whether our findings are generalizable to women. Compared with men, women have a higher heart rate but show relatively greater parasympathetic control of the heart.46 While these differences are poorly understood, they imply that associations between RHR and psychiatric disorders may be different in men and women. Second, many psychiatric disorders have their onset in childhood or early adolescence, including OCD and anxiety disorders.47 We excluded men with registered diagnoses before conscription at an average age of 18 years, but because seeking treatment typically occurs much later than the onset of symptoms, our results may partly reflect symptomatic disorders already present during conscription but invisible in medical register data. Thus, rather than representing true risk factors, we may have uncovered early markers for the persistence of mental disorders. However, we tried to exclude potential acute psychopathology by starting the follow-up period only 5 years after conscription, which had no effect on the associations. Third, while we controlled for several important covariates, it is possible that residual confounding due to unmeasured factors contributed to the results. Finally, our measures of autonomic activity, RHR and BP, were not optimal, and we did not have more detailed measures (such as heart rate variability) available. However, our results with the relatively simple measures should motivate future prospective longitudinal studies with more informative indicators to investigate autonomic precursors of psychiatric disorders.
In a large and representative sample of men, higher RHR and BP in late adolescence were associated with subsequent diagnoses of OCD, schizophrenia, and anxiety disorders, whereas lower RHR and BP were associated with subsequent diagnoses of SUDs and violent criminality. These associations should be confirmed in other longitudinal studies, and the underlying mechanisms should be studied with more detailed measures of autonomic functioning and designs that can more clearly elucidate causal processes.
Accepted for Publication: September 1, 2016.
Corresponding Author: Antti Latvala, PhD, Department of Public Health, University of Helsinki, PO Box 20, Helsinki, Finland FI-00014 (email@example.com).
Published Online: October 26, 2016. doi:10.1001/jamapsychiatry.2016.2717
Author Contributions: Dr Latvala had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Latvala, Kuja-Halkola, Rück, Larsson, Lichtenstein.
Acquisition, analysis, or interpretation of data: Latvala, Kuja-Halkola, D’Onofrio, Jernberg, Almqvist, Mataix-Cols, Larsson, Lichtenstein.
Drafting of the manuscript: Latvala.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Latvala.
Obtained funding: Latvala, D’Onofrio, Almqvist, Larsson, Lichtenstein.
Administrative, technical, or material support: Kuja-Halkola, D’Onofrio, Lichtenstein.
Study supervision: Larsson, Lichtenstein.
No additional contributions: Rück, Jernberg, Mataix-Cols.
Conflict of Interest Disclosures: Dr Larsson has served as a speaker for Eli Lilly and Shire and has received a research grant from Shire, all outside the submitted work. No other disclosures are reported.
Funding/Support: We acknowledge financial support from the Academy of Finland (grant 277209, Dr Latvala); the Swedish Research Council for Health, Working Life and Welfare; the Swedish Research Council; and the Swedish Research Council through the Swedish Initiative for Research on Microdata in the Social and Medical Sciences framework (grant 340-2013-5867). Dr Rück is supported by the Swedish Research Council (grant K2013-61P-22168).
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.