Mean levels of inflammatory markers in participants with a history of psychiatric admissions adjusted for age and sex. A, Levels of C-reactive protein in patients with psychiatric history and the general population. B, Levels of fibrinogen. Error bars indicate SEM.
Hamer M, Stamatakis E, Steptoe A. Psychiatric Hospital Admissions, Behavioral Risk Factors, and All-Cause MortalityThe Scottish Health Survey. Arch Intern Med. 2008;168(22):2474-2479. doi:10.1001/archinte.168.22.2474
Emerging research is beginning to uncover a high prevalence of modifiable risk factors in patients with severe mental illness. We examined whether the association between episodes of psychiatric illness that involved hospitalization and all-cause mortality is mediated by behavioral risk factors.
Data were collected from a nationally representative sample of 19 898 men and women from the Scottish Health Surveys (1995, 1998, and 2003) that were linked to a patient-based database of hospital admissions and deaths up to September 2006. There were 597 participants with at least 1 hospital admission for a psychiatric episode.
During a mean follow-up of 8.5 years, participants with a history of a psychiatric episode had a higher risk of all-cause mortality (age- and sex-adjusted hazard ratio, 3.25; 95% confidence interval, 2.63-4.02). In addition, they were more likely to be heavy smokers (odds ratio, 4.69; 95% confidence interval, 3.79-5.82), have low physical activity levels (2.24; 1.75-2.87), come from a lower socioeconomic group (2.17; 1.72-2.72), and be separated or divorced from their partner (2.02; 1.63-2.52). In multivariate analyses, the association between history of a psychiatric episode and all-cause mortality was attenuated but remained statistically significant (hazard ratio, 2.02; 95% confidence interval, 1.62-2.52; P<.001) after adjustments for psychosocial factors, smoking, physical activity, body mass index, and current mental health.
Behavioral risk factors may partly mediate the association between psychiatric episodes and mortality. These data highlight the need for prevention and treatment strategies among individuals with a history of severe mental illness.
Individuals with severe mental illness are at a higher risk of premature mortality than the general population, with most excess deaths attributable to cardiovascular disease, not suicide.1 Emerging research is beginning to uncover a high prevalence of modifiable risk factors, such as diabetes mellitus, dyslipidemia, hypertension, and obesity, in patients with severe mental illness that may partly explain their higher levels of mortality and morbidity.2 Furthermore, poor health behaviors appear to be highly prevalent in severe mental illness. For example, low levels of physical activity are consistently reported,3 and in a previous analysis of the Scottish Health Survey (SHS), schizophrenic participants demonstrated a high prevalence of smoking and poorer dietary intake.4 A strong association between smoking and depression also exists.5 In addition, antipsychotic medication has been associated with weight gain that presents an additional risk factor.6 Given that physical activity has been shown to improve mental health7 and to reduce cardiovascular risk factors,8 this may play a key role in prevention and treatment strategies. However, the extent to which behavioral risk factors might mediate the association between mental illness and mortality has not been established. Two previous studies9,10 examined whether physical inactivity could be an intermediate factor in the association between depression and cardiovascular mortality, but the results were somewhat conflicting.
The aims of the present study were to (1) examine the association between history of psychiatric illness involving hospitalization and physical activity and smoking, (2) examine the association between psychiatric episodes and all-cause and cardiovascular mortality, and (3) determine whether physical inactivity and smoking could be an intermediate factor in the association between psychiatric episodes and mortality. In the present analyses, we used a representative sample of men and women from the SHSs (1995, 1998, and 2003) in which detailed data on self-reported physical activity were available, together with medical history of psychiatric hospital admissions and follow-up of all-cause mortality for a mean (SD) of 8.5 (3.3) years.
The SHS is a periodic survey (typically every 3-5 years) that draws a nationally representative sample of the general population living in households. The main aims of the survey are to estimate the prevalence of a range of health conditions and to monitor progress toward Scottish health targets. The sample was drawn using multistage stratified probability sampling with postcode sectors selected at the first stage and household addresses selected at the second stage. Stratification was based on geographic areas and not on individual characteristics of the population. Different samples were drawn for each survey. The present analyses combined data from 19 898 respondents (54.7% female) 16 years and older (mean [SD] age, 45.1 [15.5] years) measured in the 1995, 1998, or 2003 SHS. Participants gave full informed consent to participate in the study, and ethical approval was obtained from the London Research Ethics Council.
Data were collected during 2 household visits. During the first visit, trained interviewers collected self-reported data and measured height and weight. During the second visit, nurses collected nonfasting blood samples from consenting adults and medical history, including long-standing illness (physician-diagnosed cardiovascular disease; hypertension; diabetes mellitus; cancer; neuromuscular conditions; endocrine or metabolic conditions; epilepsy; bronchitis, asthma, and other respiratory disorders; and complaints related to the stomach, digestive system, and bowel). The overall response rate ranged from 60% to 76% for the different survey years, with approximately 40% of all eligible participants seeing a nurse. Detailed information on the survey method can be found elsewhere.11 The surveys were linked to a patient-based database of hospital admissions and deaths up to September 2006 (Information Services Division, Edinburgh, Scotland). Mortality from cardiovascular causes was coded according to International Classification of Diseases, Ninth Revision (ICD-9) (codes 390-459)12 and International Statistical Classification of Diseases, 10th Revision (ICD-10)13 (codes I01-I99). From these records, we also obtained respective information documenting admissions for psychiatric episodes (from 1980 onward) with the length of hospitalization and a diagnosis based on ICD-10.
Interviewers took the height and weight measurements and asked demographic (eg, marital status, social class) and health-related questions (eg, smoking, physical activity). Social-occupational class was defined using the Registrar General classification (I/II, professional/intermediate; III, skilled nonmanual or manual; IV/V, part skilled or unskilled). Current mental health was assessed from the 12-item General Health Questionnaire (GHQ-12), which is a measure of psychological distress devised for population studies.14 Physical activity interviews inquired about participation in the 4 weeks before the interview (1998 and 2003) or during a typical week (1995). Frequency of participation (for at least 20 minutes per occasion) was assessed across 3 domains of activity: leisure-time sports (eg, cycling, swimming, running, aerobics, dancing, and ball sports, such as football and tennis), walking for any purpose, and domestic physical activity (eg, heavy housework, do-it-yourself activities, manual work, and gardening work). Participants were also asked whether participation in sports made them feel out of breath or sweaty to assess intensity. The validity of the physical activity questions is supported by the results of an unpublished study (Nicholas Wareham, PhD, and Kirsten Rennie, PhD, 1995) of 174 British adults in which the output of individually calibrated heart rate monitors (4 times during a year for 4 consecutive days on each occasion) was compared against an early version of the questions. The SHS questionnaire appeared to be a valid measure of energy expenditure for total physical activity score in men (P = .03) and women (P = .02). A statistically significant correlation was also observed between self-reported activity from the questionnaire and aerobic fitness in men (P = .001) and women (P = .04).
In a subgroup of participants, analysis of inflammatory markers and blood cholesterol was performed. Analysis of C-reactive protein (CRP) levels from serum was performed using the N Latex high-sensitivity CRP mono-immunoassay on the Behring Nephelometer II analyzer (Dade Behring, Deerfield, Illinois). The limit of detection was 0.17 mg/L, and the coefficient of variation was less than 6% for this assay. Fibrinogen levels were determined using the Organon Teknika MDA 180 analyzer (Organon Teknika Corp, Durham, North Carolina), using a modification of the Clauss thrombin clotting method, with a coefficient of variation of less than 10%. The total cholesterol level was also measured using the DAX cholesterol oxidase assay method on an Olympus 640 analyzer (Olympus Diagnostics, Tokyo, Japan).
We used logistic regression to examine the association between history of a psychiatric episode and a number of behavioral risk factors. Cox proportional hazards models were used with months as the time scale to estimate the risk of all-cause and cardiovascular mortality in relation to history of a psychiatric episode (yes/no). For participants who survived, the data were censored to September 2006. The proportional hazards assumption was examined by comparing the cumulative hazard plots grouped on exposure, although no appreciable violations were noted. In multivariate models, we adjusted for age, sex, socioeconomic group (professional or intermediate [reference], skilled nonmanual, skilled manual, part skilled or unskilled), marital status (single or never married [reference], married, widowed, separated or divorced), body mass index category (calculated as weight in kilograms divided by height in meters squared; underweight, <18.5; normal weight, 18.5-25.0 [reference]; overweight, 25.01-30.0; obese, 30.1-40.0; morbidly obese, >40.0), long-standing illness (yes/no), smoking (never [reference]; previous; present light smokers, <10 cigarettes per day; heavy smokers, ≥10 cigarettes per day), physical activity category for total physical activity sessions per week of any intensity lasting at least 20 minutes (≤3.5 [reference], 3.51-5.0, 5.01-6.75, or >6.75), and current mental health (GHQ-12 score, <4 [reference] or ≥4). In a subgroup analysis, we compared cardiovascular risk factors (CRP, fibrinogen, and total cholesterol levels) between participants with psychiatric episodes and the general population using general linear models adjusting for age and sex. A log transformation was performed for CRP because of the skewed distribution. All analyses were conducted using a commercially available software program (SPSS, version 14; SPSS Inc, Chicago, Illinois).
A total of 597 participants had at least 1 psychiatric episode that involved hospitalization, with nearly half (n = 262) having 2 or more continuous inpatient stays. The mean (SD) length of stay for the first admission was 5.9 (26.0) days. Approximately 35% (34.9%) of admissions were depressive mood disorder, 25.2% were related to substance abuse, 18.3% were schizophrenia and related psychoses, and 17.2% were anxiety and stress related, with the remainder associated with eating and personality disorders. At the time of the survey, 39.0% of participants with a history of psychiatric hospitalization demonstrated significant psychological distress, as evidenced by a score of 4 or more on the GHQ-12.
The participants with a history of psychiatric episodes were more likely to be smokers, have low physical activity levels, come from a lower socioeconomic group, be separated or divorced from their partner, have long-standing illness, and demonstrate psychological distress (Table 1). In particular, of the participants with a psychiatric history, 69.6% did not undertake any domestic activity, 59.1% did no walking of any type, and 85.1% participated in no sports activities, compared with 67.0%, 54.3%, and 76.5%, respectively, of respondents from the general population. Taken together, 36.7% of the participants with a psychiatric history undertook fewer than 3 sessions per week of any type of activity compared with 24.2% of participants from the general population. In relation to the different types of psychiatric admissions, those related to substance abuse were the least physically active. A total of 62.5% of the participants with a psychiatric history were current smokers compared with 32.5% of the general population, and in particular, admissions of substance abuse and psychotic disorders demonstrated the highest smoking rates. Interestingly, psychiatric history was also associated with underweight or morbid obesity but not with overweight or obesity categories. In fact, psychiatric patients were less likely to be overweight and obese. In a subgroup analysis of participants with available data for inflammatory markers and cholesterol level (n = 8341), psychiatric patients (n = 190) demonstrated higher levels of CRP (P = .03) and fibrinogen (P = .08) but not cholesterol (P = .18) (Figure).
The mean length of follow-up was 8.5 years, ranging from 3 to 11 years. One thousand forty-eight deaths were recorded up to the date of censorship. A strong relationship was seen between a history of psychiatric episodes and risk of all-cause and cardiovascular mortality (Table 2). Other independent predictors of death included smoking (hazard ratio for smoking ≥10 cigarettes per day compared with referent of never smoked, 2.55; 95% confidence interval, 2.13-3.04), physical activity (those reporting daily activity compared with sedentary behavior, 0.58; 0.47-0.73), social occupational class (manual class compared with professional, 1.20; 0.99-1.46), marital status (married compared with single, 0.51; 0.42-0.63), and psychological distress (GHQ-12 score ≥4 compared with GHQ-12 score <4, 1.30; 1.09-1.53). When we adjusted for socioeconomic group and marital status, the association between psychiatric episodes and all-cause mortality was attenuated (approximately 18%), although this was largely mediated through marital status (approximately 15.5%). Further adjustments for physical activity and smoking reduced the hazard ratio by an additional 29%, and smoking was largely responsible for these effects, independently explaining nearly 24% of the association between psychiatric episode and death. In the fully adjusted model, the association between psychiatric history and death was attenuated but remained statistically significant (P < .001). A similar pattern was observed in the case of cardiovascular death. In further analyses stratified by psychiatric diagnosis, the highest risks of death were observed for admissions related to dementia (age- and sex-adjusted hazard ratio, 4.65; 95% confidence interval, 2.78-7.76), substance abuse (2.46; 0.79-7.66), depressive mood disorder (2.14; 0.89-5.16), and schizophrenia (1.70; 0.72-5.01). Similar analyses relating to cardiovascular death were limited by statistical power. When we excluded deaths from suicide or psychoactive substance abuse (n = 26), this did not alter the overall results.
We also performed additional analyses excluding participants with existing cardiovascular disease at baseline (n = 734, clinically confirmed from hospital records). In these analyses, there were a total of 897 all-cause deaths (238 from cardiovascular causes), and the associations between psychiatric episodes and all-cause mortality (age- and sex-adjusted hazard ratio, 3.55; 95% confidence interval, 2.85-4.43) and cardiovascular mortality (3.36; 2.17-5.21) were slightly strengthened.
In the present study, smoking and low physical activity were highly prevalent among individuals with a history of psychiatric illness that involved hospitalization, and partly mediated the association between psychiatric episodes and greater risk of all-cause and cardiovascular mortality. Psychiatric patients also demonstrated elevated inflammatory markers of long-term disease risk. Although our study did not directly examine the associations of mental health with mortality, the findings are consistent with a previous cohort study9 of patients with coronary artery disease that showed that smoking and sedentary behavior partially mediated the association between depressive symptoms and mortality. However, in a small cohort of 2285 healthy elderly men, physical inactivity was not an intermediate factor in the association between depression and cardiovascular mortality during 10 years of follow-up,10 although depression and inactivity combined to produce an additive risk increase of 33%. Given that the association between psychiatric episode and mortality persisted after adjustments for psychosocial factors, health behaviors, obesity, chronic illness, and current mental health, other unmeasured intermediate factors must also be involved. Potential mechanisms include stress-induced inflammation,15 and inflammatory markers were elevated in this study in participants with a psychiatric history that involved hospitalization.
Our data confirm previous reports3 demonstrating that patients with mental illness appear to be significantly less active than the general population. Feelings of low confidence and social support toward exercising may partly contribute to this association.16 In addition, psychiatric symptoms, such as poor attention, memory loss, apathy, and amotivation, might be a barrier to behavioral modification. Comorbidities that cause functional impairment are also likely to play a role. Recent evidence suggests that patients with severe mental illness have a high prevalence of modifiable cardiovascular risk factors but are less likely to receive cardioprotective drug therapies.2 Given that physical activity can reduce cardiovascular risk factors, such as insulin resistance, inflammation, hypertension, and dyslipidemia, this may play a key role in prevention and treatment strategies for reducing morbidity and mortality. Interestingly, given the potential effects of antipsychotic medication on weight gain and sedentary lifestyle, we might have expected psychiatric episodes to be associated with obesity, although this trend was not consistently observed. Results from other studies17,18 are also mixed, with some showing increased prevalence of obesity with mental illness and others not showing this effect. It was not possible to take into account the effects of antipsychotic medications in the present analyses, although recent data suggest that the excess death from cardiovascular disease in severe mental illness is independent of such medications.1
Participants with a history of psychiatric illness that involved hospitalization were less likely to be married or have a partner, which further mediated the greater risk of all-cause and cardiovascular mortality. A considerable body of research supports direct influences of marriage on a range of health markers.19 Prospective evidence suggests that low social support confers a greater risk of coronary heart disease,20 although in a large trial of myocardial infarction patients with depression and low perceived social support, treatment with cognitive behavior therapy had no effect on event-free survival after 29 months of follow-up despite significant improvement in psychosocial outcomes.21 Nevertheless, it is likely that marriage, social support, and health behaviors are closely related, and physical activity interventions can be designed to help improve social networks and enhance self-efficacy and coping.
The limitations of the present study should be recognized. Given that much psychiatric illness is managed in primary care or outpatient clinics, this study only addresses severe cases that could not be managed outside the hospital. Thus, the presence of prevalent but unidentified psychiatric illness that never led to hospitalization may have introduced biases into our analyses. In addition, because the sample was drawn from adults living in the community, people who were psychiatric inpatients at the time of the survey were excluded. Our study may therefore have underestimated the strength of associations between psychiatric illness, health behavior, and mortality. The distribution of primary diagnoses for hospital admissions was, however, broadly comparable with recent national data from England.22 Because of the retrospective nature of information on psychiatric admissions, we do not know the extent to which changes in mental health may have influenced the outcomes. Nevertheless, we attempted to control for current mental health at the time of the survey using the GHQ-12, which is a validated measure of psychological distress. Several of the measures, including smoking and physical activity, were assessed by self-report; thus, it is possible that more precise assessment of these factors may have affected the results. It is difficult to comment on the causal chain of events in relation to mental illness, various health behaviors, and disease. For example, it is possible that physical inactivity results from mental disorders or is a risk factor in the development of such conditions, and that patients may smoke as a method of coping with their distress. It is also possible that mental illness and poor health behaviors are markers of subclinical disease, thus reflecting reverse causality, although we attempted to adjust for long-standing illness in our analyses.
In summary, behavioral risk factors, such as physical inactivity and smoking, may partly mediate the association between psychiatric episodes and all-cause mortality. These data highlight the need for prevention and treatment strategies among individuals with a history of severe mental illness.
Correspondence: Mark Hamer, PhD, Department of Epidemiology and Public Health, University College London, 1-19 Torrington Pl, London WC1E 6BT, England (firstname.lastname@example.org).
Accepted for Publication: May 22, 2008.
Author Contributions:Study concept and design: Hamer and Steptoe. Acquisition of data: Stamatakis. Analysis and interpretation of data: Hamer and Stamatakis. Drafting of the manuscript: Hamer and Stamatakis. Critical revision of the manuscript for important intellectual content: Hamer and Steptoe. Statistical analysis: Hamer and Stamatakis. Study supervision: Steptoe.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grant funding from the British Heart Foundation (Drs Hamer and Steptoe) and the National Institute for Health Research (Dr Stamatakis). The Scottish Health Survey is funded by the Scottish Executive.
Disclaimer: The views expressed in this article are those of the authors and not necessarily of the funding bodies.