Social Integration and Suicide Mortality Among Men: 24-Year Cohort Study of U.S. Health Professionals

Alexander C. Tsai; Michel Lucas; Ayesha Sania; Daniel Kim; Ichiro Kawachi

doi:10.7326/M13-1291

Original Research |15 July 2014

Social Integration and Suicide Mortality Among Men: 24-Year Cohort Study of U.S. Health Professionals Free

Alexander C. Tsai, MD, PhD *; Michel Lucas, PhD, RD *; Ayesha Sania, PhD; Daniel Kim, MD, DrPH; Ichiro Kawachi, MD, PhD

Alexander C. Tsai, MD, PhD

From Massachusetts General Hospital, Harvard Medical School, Harvard School of Public Health, and Bouvé College of Health Sciences at Northeastern University, Boston, Massachusetts; Mbarara University of Science and Technology, Mbarara, Uganda; and Centre de Recherche du Centre Hospitalier Universitaire de Québec, Université Laval, Québec City, Québec, Canada.

Michel Lucas, PhD, RD

Ayesha Sania, PhD

Daniel Kim, MD, DrPH

Ichiro Kawachi, MD, PhD

Article, Author, and Disclosure Information

* Drs. Tsai and Lucas contributed equally to this work.

From Massachusetts General Hospital, Harvard Medical School, Harvard School of Public Health, and Bouvé College of Health Sciences at Northeastern University, Boston, Massachusetts; Mbarara University of Science and Technology, Mbarara, Uganda; and Centre de Recherche du Centre Hospitalier Universitaire de Québec, Université Laval, Québec City, Québec, Canada.
Note: All authors had full access to all of the data in the study and agreed with the decision to submit the manuscript for publication.
Acknowledgment: The authors thank their colleagues in the Channing Division of Network Medicine, Department of Medicine, Brigham and Women's Hospital and at Harvard Medical School for their assistance with the HPFS data and for reviewing the analysis protocol, findings, statistical code, and manuscript.
Grant Support: By the NIH (P01CA055075). Dr. Tsai was supported by NIH K23MH096620 and, with Dr. Kawachi, by a seed grant from the Robert Wood Johnson Foundation Health and Society Scholars Program. Dr. Kim was supported by NIH R00HL089459.
Disclosures: Authors have disclosed no conflicts of interest. Forms can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M13-1291.
Reproducible Research Statement:Study protocol: Available at www.hsph.harvard.edu/hpfs. Statistical code: Available from Dr. Tsai (e-mail, actsai@partners.org). Data set: External collaborator information is available at www.hsph.harvard.edu/hpfs.
Requests for Single Reprints: Alexander C. Tsai, MD, PhD, Center for Global Health, Room 1529-E3, Massachusetts General Hospital, 100 Cambridge Street, 15th Floor, Boston, MA 02114; e-mail, actsai@partners.org.
Current Author Addresses: Dr. Tsai: Center for Global Health, Room 1529-E3, Massachusetts General Hospital, 100 Cambridge Street, 15th Floor, Boston, MA 02114.
Dr. Lucas: Centre de Recherche du Centre Hospitalier Universitaire de Québec, Université Laval, 2875, Boulevard Laurier, Edifice Delta II, Bureau 600, Québec City, Québec G1V 2M2, Canada.
Dr. Sania: Room 911, Building 3, Harvard School of Public Health, 677 Huntington Avenue, Boston, MA 02115.
Dr. Kim: Department of Health Sciences, Bouvé College of Health Sciences at Northeastern University, 360 Huntington Avenue, Robinson Hall Room 316J, Boston, MA 02115.
Dr. Kawachi: Department of Social and Behavioral Sciences, Harvard School of Public Health, 677 Huntington Avenue, Kresge Building, 7th Floor, Boston, MA 02115.
Author Contributions: Conception and design: A.C. Tsai, I. Kawachi.
Analysis and interpretation of the data: A.C. Tsai, M. Lucas, A. Sania, D. Kim, I. Kawachi.
Drafting of the article: A.C. Tsai.
Critical revision of the article for important intellectual content: A.C. Tsai, M. Lucas, A. Sania, D. Kim, I. Kawachi.
Final approval of the article: A.C. Tsai, M. Lucas, A. Sania, D. Kim, I. Kawachi.
Provision of study materials or patients: I. Kawachi.
Statistical expertise: A.C. Tsai, M. Lucas, A. Sania, I. Kawachi.
Obtaining of funding: A.C. Tsai, I. Kawachi.
Administrative, technical, or logistical support: I. Kawachi.
Collection and assembly of data: I. Kawachi.

Abstract

Background:

Suicide is a major public health problem. Current thinking about suicide emphasizes the study of psychiatric, psychological, or biological determinants. Previous work in this area has largely relied on surrogate outcomes or samples enriched for psychiatric morbidity.

Objective:

To evaluate the relationship between social integration and suicide mortality.

Design:

Prospective cohort study initiated in 1988.

Setting:

United States.

Participants:

34 901 men aged 40 to 75 years.

Measurements:

Social integration was measured with a 7-item index that included marital status, social network size, frequency of contact, religious participation, and participation in other social groups. Vital status of study participants was ascertained through 1 February 2012. The primary outcome of interest was suicide mortality, defined as deaths classified with codes E950 to E959 from the International Classification of Diseases, Ninth Revision.

Results:

Over 708 945 person-years of follow-up, there were 147 suicides. The incidence of suicide decreased with increasing social integration. In a multivariable Cox proportional hazards regression model, the relative hazard of suicide was lowest among participants in the highest (adjusted hazard ratio [AHR], 0.41 [95% CI, 0.24 to 0.69]) and second-highest (AHR, 0.52 [CI, 0.30 to 0.91]) categories of social integration. Three components (marital status, social network size, and religious service attendance) showed the strongest protective associations. Social integration was also inversely associated with all-cause and cardiovascular-related mortality, but accounting for competing causes of death did not substantively alter the findings.

Limitations:

The study lacked information on participants’ mental well-being. Some suicides could have been misclassified as accidental deaths.

Conclusion:

Men who were socially well-integrated had a more than 2-fold reduced risk for suicide over 24 years of follow-up.

Primary Funding Source:

National Institutes of Health and Robert Wood Johnson Foundation.

Editors’ Notes

Context

Suicide is a major cause of death and is increasing in incidence, especially among middle-aged men. Identify-ing modifiable risk factors might lead to effective interventions.

Contribution

In a 24-year prospective cohort study, middle-aged men were less likely to commit suicide if they scored highly on measures of social integration. Marriage, frequent attendance at religious services, and having a large social network seemed especially protective.

Caution

Participants were born between 1913 and 1948. Types and amounts of social interaction among men born more recently may differ from those of earlier generations.

Implication

The degree of social integration should be considered in studies of suicide.

—The Editors

Suicide is a major cause of death in the United States, particularly among men, for whom it is one of the 10 leading causes (1). Suicide rates have increased substantially over the past decade, particularly among middle-aged persons (2). Worldwide, the number of deaths resulting from self-harm increased by more than 30% between 1990 and 2010 (3), with current projections showing a continued increase in its contribution to global mortality and disease burden (4). In 2000, the most recent year for which data are available, the estimated value of lost productivity due to suicide in the United States exceeded $60 billion (5).

Much of the discourse around suicide prevention has emphasized the study of its psychiatric, psychological, or biological determinants (6). Yet, as noted in an Institute of Medicine report (7), “If ever a condition begged for an integrated understanding that takes into account biological, clinical, subjective, and social factors, this [suicide] is it.” Durkheim (8) advanced an explicitly sociological understanding of suicide, postulating that social integration would be inversely related to suicide. However, because of the lack of high-quality, individual-level, national data, most findings in support of this theory have been derived from the modeling of social integration and suicide at the aggregate level (9).

There have been few prospective studies of social integration and suicide. Although suicide is a leading contributor to premature mortality, especially among men, it is rare relative to the number of persons exposed to its potential risk factors (for example, 36 000 suicides in the United States in a population exceeding 300 million). Individual-level studies, lacking sufficient statistical power to investigate suicide mortality in general population samples, have instead used proxy outcomes, such as suicidal thinking (10) or suicide attempts (11). Other researchers have studied these proxy outcomes in samples enriched for psychiatric morbidity, such as persons receiving psychiatric care (12). However, in the 21-country World Health Organization World Mental Health Surveys, a substantial proportion of suicidal behaviors occurred in the absence of formally diagnosed mental disorder (13), suggesting that a deeper understanding of factors driving suicide even in generally healthy populations is needed. To address these limitations, we used data from the HPFS (Health Professionals Follow-up Study) to estimate the association between social integration and suicide mortality over 24 years of follow-up.

Methods

Study Design and Setting

The HPFS is an ongoing prospective cohort study of men in the United States who were aged 40 to 75 years when the study was initiated in 1986 (14). Women were not eligible for participation. Of the potentially eligible dentists, optometrists, osteopathic physicians, pharmacists, podiatrists, and veterinarians who were contacted, 51 529 men (33%) responded to the baseline questionnaire. Every 2 years, follow-up questionnaires are mailed to participants to obtain updated information on medical history, diet, lifestyle habits, and other health behaviors. During each survey cycle, up to 7 repeated mailings are used, with the option of a short-form questionnaire for serial nonresponders (15). The combined (long form plus short form) survey follow-up rate for participants enrolled in the baseline cohort exceeds 90% overall, with a response rate to the long-form questionnaire of approximately 70% in any given survey cycle. All participants provided written informed consent. The HPFS and the analysis described here were approved by the Office of Human Research Administration at the Harvard School of Public Health.

Measures

The primary outcome of interest was suicide mortality, assessed continuously during the course of the study between the return of the baseline questionnaire and 1 February 2012. Outcome ascertainment was largely independent of survey administration. In most instances, HPFS staff members were informed of a participant's death by next of kin, through questionnaires serially returned by the U.S. Postal Service, or through reports from participants’ professional organizations. The vital status of serial nonresponders was ascertained through the U.S. National Death Index, a method that has been shown to have 98% sensitivity and 100% specificity for ascertainment of deaths (16, 17). Physicians blinded to exposure status reviewed death certificates and hospital or pathology reports to classify individual causes of death. Deaths caused by self-inflicted injuries were classified according to the underlying causes listed on the death certificate. For this study, we specifically examined deaths in the “Suicide and Self-Inflicted Injury” cluster (codes E950 through E959) of the International Classification of Diseases, Ninth Revision (ICD-9). The specific categories of deaths included suicide and self-inflicted poisoning by solid or liquid substances (E950), gases in domestic use (E951), or other gases and vapors (E952); suicide and self-inflicted injury by hanging, strangulation, and suffocation (E953), submersion (E954), firearms and explosives (E955), cutting and piercing instruments (E956), jumping (E957), or other and unspecified means (E958); and late effects of self-inflicted injury (E959).

The primary exposure of interest was social integration, measured with a 7-item index that included questions about marital status, social network size, frequency of contact with social ties, religious participation, and participation in other social groups (18, 19). The responses to these items yield a total score from 1 to 12 that is typically analyzed as a 4-level categorical variable. Specific procedural details on the construction of the social integration index are available in the Appendix and Appendix Table 1. At baseline, the index showed acceptable reliability (Cronbach α, 0.72). Additional psychometric analyses (20) and evidence of its construct validity (21, 22) have been described elsewhere. Because the index was not added to the HPFS survey instrument until 1988 (and again in 1996) (23, 24), we considered 1988 as the baseline year in our analysis. The short-form questionnaires mailed to serial nonresponders in 1988 and 1996 were limited to questions about height, weight, smoking status, medical diagnoses, and procedures and did not contain the social integration questions.

Appendix Table 1. Social Integration Index

Statistical Analysis

All statistical analyses were done with SAS, version 9.2 (SAS Institute, Cary, North Carolina). Each participant contributed person-time beginning from return of the 1988 questionnaire (or, for the analysis of change in social integration described later, return of the 1996 questionnaire) until death or the end of follow-up (1 February 2012), whichever occurred first. To estimate the association between social integration and suicide mortality, we fit a multivariable Cox proportional hazards model to the data (25), using the independent increment model (also described as the Andersen–Gill model) to handle time-updated variables efficiently (26). Under this data structure, a new data record was created for every questionnaire cycle at which a participant was at risk for suicide, with independent variables set to their values at the time the questionnaire was returned. To adjust for potential confounding by age, calendar time, and any potential 2-way interactions between these time scales, we stratified the analysis jointly by age in months at the start of follow-up and calendar year of the questionnaire cycle. The time scale for the analysis was then measured in months since the start of the current questionnaire cycle. We also adjusted our estimates for baseline occupational category (27, 28) and the following time-updated variables: employment status (29); smoking status (30); alcohol intake (31); caffeine intake (32); antidepressant medication use (13, 33); body mass index (34, 35); self-report of a routine physical examination within the previous 2 years; weekly physical activity (36); and history of hypertension, diabetes, hypercholesterolemia, or renal failure (37, 38). To evaluate for departures from the proportional hazards assumption, we used likelihood ratio tests comparing models with and without the interaction terms of age by the specific exposures of interest. To test for linear trends across categories of exposure, we modeled the medians within each category of exposure.

At baseline, the percentage of observations with missing values on the covariates ranged from 0.1% (employment status) to 2.1% (body mass index). After the baseline survey, the percentage of observations with missing values on the covariates ranged from 3.8% (smoking status) to 19.0% (physical activity). We assumed missingness at random (that is, missingness that did not depend on unobserved data but could depend on observed data) and used the method of multiple imputation to handle missing data. We used the SAS macros PROC MI to impute missing data and PROC MIANALYZE for estimation over 20 imputed data sets.

To assess the extent to which different trajectories of social integration (39) may affect suicide mortality, we fit multivariable Cox proportional hazards regression models with participants classified into 1 of 5 groups on the basis of their levels of social integration in 1988 and 1996: no change in social integration (lowest category [category 1] in 1988 and 1996), decrease in social integration from 1988 to 1996, no change in social integration (intermediate category [category 2 or 3] in 1988 and 1996), increase in social integration from 1988 to 1996, or no change in social integration (highest category [category 4] in 1988 and 1996). In this analysis, 1996 was specified as the baseline year and suicide mortality was assessed during the 16-year period between the return of the 1996 questionnaire and 1 February 2012.

To determine whether our findings were robust to the specific form of the social integration index that was used, we conducted 2 sensitivity analyses. First, assuming the relationship between social integration and suicide mortality was linear across the range of social integration values, we fit a multivariable Cox proportional hazards regression model specifying social integration as a continuous variable. Second, making no assumptions about the specific form of the index, we fit a multivariable Cox proportional hazards regression model with each component of the social integration index specified as a separate variable in order to assess the extent to which specific components of social integration were protective against suicide mortality.

We conducted 2 additional sensitivity analyses to evaluate our findings for potential misclassification or confounding. First, for manner-of-death determinations by the medical examiner for the death certificate, some suicides could have been misclassified as deaths from undetermined causes or as accidental deaths. In the United States, this may be due partly to the requirement of strong corroborative evidence (40) and partly to social pressures, such as familial stigma or concerns about life insurance reimbursement (41). We therefore examined firearm-related suicides (ICD-9 code E955) separately for 2 reasons: Firearms are generally more lethal than other means (42, 43), and firearm-related deaths are less likely to be classified as being from undetermined causes than deaths related to more passive means (44–46), which is the category most susceptible to misclassification (47). Consistent with the latter, although only 3% of deaths due to injuries in the United States are described as deaths from undetermined causes, the percentage of firearm-related deaths described as being from undetermined causes is smaller by an order of magnitude (0.2%) (48). Second, despite our attempts to adjust for important medical history and health behavior variables, our estimates could still be confounded by differences in comorbidity. Severe medical conditions, such as cancer and cardiovascular disease, have been shown to have a positive association with suicide risk (49–51), and poor health has been shown to adversely affect social relationships (39, 52, 53). We therefore refitted the multivariable regression models after excluding participants who reported a history of cancer (specifically, any type of cancer except for nonmelanoma skin cancer) or a serious cardiovascular condition (specifically, acute myocardial infarction, coronary artery bypass graft surgery, percutaneous transluminal coronary angioplasty, or stroke) at baseline.

Finally, to assess the extent to which competing causes of death could affect our results, we conducted parallel analyses to estimate the association between social integration and 3 major competing outcomes: all-cause mortality, cardiovascular-related mortality (ICD-9 codes 390 to 459), and cancer-related mortality (ICD-9 codes 140 to 208). To formally account for competing causes of death in our analysis, we used the proportional subdistribution hazards regression model proposed by Fine and Gray (54) to estimate the cumulative incidence of suicide. We plotted the cumulative incidence of suicide across different categories of social integration, using a test statistic based on the nonparametric maximum likelihood estimator of the subdistribution hazard to compare cumulative incidence across categories (55). We then used the Fine and Gray model (54) to estimate the differences in cumulative incidence in the presence of covariates, and we expressed these as subdistribution hazard ratios. The results of these sensitivity analyses are reported in the Appendix.

Role of the Funding Source

The study was supported by the National Institutes of Health (NIH) and the Robert Wood Johnson Foundation. The sponsors had no role in study design; collection, analysis, or interpretation of data; or writing of the report.

Results

Of 51 529 men initially enrolled into the HPFS cohort, 34 901 (67.7%) responded to the social integration questions in the 1988 survey. Responders had a lower incidence of suicide than nonresponders over the 24 years of follow-up (21 vs. 30 per 100 000 person-years; Mantel–Haenszel rate ratio, 0.69 [95% CI, 0.52 to 0.92]). Baseline summary statistics for the sample are provided in Table 1. The mean age was 56.6 years (SD, 9.8). Most survey respondents were or had been engaged in dentistry (19 757 [56.6%]) and veterinary (7237 [20.7%]) occupations, and most (26 453 [75.8%]) were working full-time at baseline. The mean social integration index value was 6.6 (SD, 3.1). Most participants were classified into the highest (14 476 [41.5%]) and second-highest (7508 [21.5%]) categories of social integration. Socially isolated (that is, less socially integrated) men were more likely to smoke, consumed more alcohol and caffeine, and were less physically active than socially integrated men.

Table 1. Baseline Sample Characteristics, by Social Integration Category Measured in 1988*

Over 708 945 person-years of follow-up, there were 147 suicides (Table 2). The most frequent means were firearms and explosives (89 [60.5%]); poisoning by solid or liquid substances (20 [13.6%]); and hanging, strangulation, or suffocation (15 [10.2%]). When competing causes of death were accounted for, the cumulative incidence of suicide was highest among the most socially isolated men (Figure). The Gray test (55) rejected the null hypothesis that the cumulative incidence functions were equal across categories (P < 0.001). The same pattern was observed in estimates from the multivariable Cox proportional hazards regression models (Table 3). The hazard of suicide was lowest among participants in the highest (adjusted hazard ratio [AHR], 0.41 [CI, 0.24 to 0.69]) and second-highest (AHR, 0.52 [CI, 0.30 to 0.91]) categories of social integration. A decreasing trend was observed across the categories (P for trend < 0.001). We did not identify any violations of the proportional hazards assumption (P values ranged from 0.21 to 0.82).

Table 2. Suicides During 1988–2012

Figure.

Cumulative incidence of suicide, by social integration category measured in 1988.

Table 3. Relative Hazard Ratios for Suicide During 1988–2012, by Social Integration Category Measured in 1988

For the analysis of changes in social integration from 1988 to 1996, a total of 26 526 participants were included (Table 2). Over these 8 years, the mean change in the social integration index was −0.35 (SD, 2.7), a 5.3% difference relative to the baseline mean value; 1100 participants (4.1%) remained in the lowest category of social integration, 7174 (27.0%) experienced a decrease in social integration, 6299 (23.7%) remained in intermediate categories of social integration, 4707 (17.7%) had an increase in social integration, and 7246 (27.3%) remained in the highest category of social integration. Participants who were categorized as having the highest level of social integration at both time points had a reduced hazard of suicide over subsequent follow-up (AHR, 0.36 [CI, 0.13 to 0.99]), whereas other trajectories of social integration were less protective; a decreasing trend across trajectory categories was observed (P for trend = 0.008) (Table 4).

Table 4. Relative Hazard Ratios for Suicide During 1996–2012, by Social Integration Trajectory From 1988 to 1996

The sensitivity analyses did not substantively alter our findings. When the social integration index was specified as a continuous variable, each 1-point difference in the index was associated with a 10% lower relative hazard of suicide (AHR, 0.90 [CI, 0.85 to 0.95]). When the components of social integration were examined separately, marital status, social network size, and religious service attendance were identified as having statistically significant protective influences (Table 5). The estimates remained qualitatively similar when we restricted the analysis of outcomes to firearm-related suicides only and when we excluded participants with a baseline history of cancer or a serious cardiovascular condition (Appendix Table 2).

Table 5. Relative Hazard Ratios for Suicide During 1988–2012, by Social Integration Components Measured in 1988

Appendix Table 2. Relative Hazard Ratios for Suicide During 1988–2012, by Social Integration Category Measured in 1988: Sensitivity Analysis

The incidence of nonsuicide mortality greatly exceeded the incidence of suicide mortality in this cohort (Appendix Table 3). To investigate the extent to which competing risks could bias our findings, we examined the associations between social integration and competing causes of death. Participants in higher categories of social integration had lower hazards of all-cause mortality and cardiovascular-related mortality, with decreasing trends across categories (P for trend = 0.018 for all-cause mortality and 0.002 for cardiovascular-related mortality) (Appendix Table 4). When we excluded participants who reported any history of cancer or serious cardiovascular condition at baseline from estimation, the results were qualitatively similar but only the trend test for lower hazards of cardiovascular-related mortality across categories of increasing social integration remained statistically significant (P for trend = 0.004) (Appendix Table 5). Finally, the estimates from the Fine and Gray (54) proportional subdistribution hazards regression models are reported in Appendix Table 6. Comparison of these subdistribution relative hazard ratios with those reported in Tables 3, 4, and 5 does not reveal substantive differences.

Appendix Table 3. Deaths During 1988–2012, by Social Integration Category Measured in 1988

Appendix Table 4. Relative Hazard Ratios for Competing Mortality Outcomes During 1988–2012, by Social Integration Category Measured in 1988

Appendix Table 5. Relative Hazard Ratios for Competing Mortality Outcomes During 1988–2012, by Social Integration Category Measured in 1988: Participants With Any History of Cancer or Serious Cardiovascular Condition at Baseline Excluded

Appendix Table 5. Relative Hazard Ratios for Competing Mortality Outcomes During 1988–2012, by Social Integration Category Measured in 1988: Participants With Any History of Cancer or Serious Cardiovascular Condition at Baseline Excluded

Appendix Table 6. Subdistribution Relative Hazard Ratios for Suicide During 1988–2012, Accounting for Competing Mortality Risks

Discussion

In this longitudinal study of 34 901 men in the United States, we found that social integration at baseline was associated with a lower risk for suicide over 24 years of follow-up. The association was statistically significant, large in magnitude, unlikely to be driven by competing mortality risks, and robust to several sensitivity analyses. Previous studies have shown that socially isolated men have a greater incidence of death (23, 24, 56). These studies, however, lacked sufficient follow-up and accrual of events to separate accidental death from suicide, even though their patterning by various sociodemographic indicators is known to diverge substantially (57). Kposowa (58) had access to individual-level, 10-year follow-up data on suicide mortality in a general population sample but was limited to using marital status as the sole measure of social integration. Our study overcame these and other substantive data limitations encountered by previous researchers (10–12) while providing further evidence in support of the importance of social integration to both overall and mental health (59, 60).

Different aspects of social integration may matter for suicide risk. Because the empirical weighting scheme of the original social integration index has been criticized (61, 62), we disaggregated the index into its components. We found that elements of both family and religious integration were associated with a lower risk for suicide. Our finding about the protective effects of religious integration is a notable contribution to the literature, given the observation by Koenig and colleagues (63) that only 2 of the 141 studies in their systematic review about religious or spiritual involvement and suicide were based on longitudinal data. Overall, our findings are consistent with those of the study by Breault (64), which showed ecological associations among state- and county-level suicide, divorce, and church membership rates, lending further support to the sociological notion that family and religious integration may influence suicide risk.

Our study also uniquely analyzes how suicide mortality is related to different trajectories of social integration, which are important to consider given that changes in social integration can result in exposure misclassification during follow-up (potentially leading to underestimates of the true association with suicide). A few studies have attempted to understand how these changes affect health (65–69), but none have studied suicide mortality. Our analysis suggests that different trajectories of social integration were associated with suicide, namely that persistently high and increasing levels of social integration had a protective influence. However, the long duration between assessments of social integration could also have resulted in misclassification. For example, if a study participant was married in 1988, subsequently divorced, and then remarried before 1996, he would have been considered continuously married in our analysis. Depending on the extent of differential misclassification in relation to the outcome, the direction of potential bias is unpredictable.

An important analytic contribution of our analysis is the exclusion of competing mortality risks as a potential explanation for our findings. For example, if socially integrated men were also more likely to die of cardiovascular disease, such early deaths could have depleted the pool of participants at risk for subsequent suicide. Consideration of this possibility is analytically important given that, in the HPFS, the incidence of nonsuicide mortality exceeded that of suicide mortality by an order of magnitude. However, our analysis suggests that competing risks are unlikely to explain the findings. First, we found that social integration was associated with a reduced hazard of cardiovascular-related mortality, which is consistent with what has previously been reported (23, 24, 70). Therefore, a naive analysis disregarding censoring from competing causes of death would be expected to be biased toward, not away from, the null. Second, our competing-risks analyses confirmed that the cumulative incidence of suicide is lower among more socially integrated men.

Taken together, our findings provide an empirical bridge between a sociological understanding of suicide (8, 9) and prevailing clinical and neurobiological models of suicide and suicidal behaviors (6). In the prototypical “stress-diathesis model” (71), a diathesis is conceptualized as a predisposition toward mental illness that is influenced by genetic, biological, or psychological traits and is activated by stressful life events (72, 73). Acute changes in social integration, such as marital dissolution (74), are frequently viewed as psychosocial stressors that activate a predisposition to suicide, but social integration may also be conceptualized more broadly as contributing to the diathesis itself. The association between social integration and suicide mortality observed in our study suggests that this component of the diathesis may not be entirely latent (73). A deeper investigation of social integration would require information on the exchange of social support as well as more detailed measures of reciprocity, geographic proximity, social network density, or relationship quality (75–77). Regarding relationship quality, we were unable to study the phenomenon of loneliness, a construct closely related to social isolation but defined differently as the subjective perception of inconsistencies between one's desired and actual social relationships (78). Our analysis was also limited to studying the effects of net, rather than compositional (79, 80), changes in social network size. All of these factors are potentially relevant to a substantive sociological understanding of suicide. Nonetheless, our study extends prior work in its focus on suicide mortality, an unambiguous, terminal outcome that has been largely neglected in psychiatric treatment studies (81).

Interpretation of our findings is subject to several limitations in addition to those already mentioned. First, we had limited ability to adjust for differences in mental health status. The protective effects of social integration on suicide mortality are plausibly mediated by mental health status, given the well-known associations between social integration and emotional well-being (60) and between mood disorders and suicide (13, 33). However, mental health status may also confound the observed relationship between social integration and suicide. For example, as was conceptualized by Coyne (82, 83) in his interactional theory of depression, persons with depressed mood may engage in certain interpersonal behaviors (such as excessive reassurance seeking) that elicit rejection by significant others. Because data on depressed mood were not added to the HPFS until 2002, we had limited ability to account for potential differences in mental health status. However, our regression models were adjusted for antidepressant medication use, which is likely to be a specific indicator of mental health status given the extent to which major depressive disorder is undertreated in the United States (84–86). Adjustment for antidepressant medication use did not alter our findings, which strengthens the plausibility of the observed estimates.

Second, our findings may not be generalizable to other populations. Extending the analysis to more socioeconomically diverse samples of men, to women, or to men or women from a different generation may have yielded different results. That the sample is limited to men employed or previously employed in medically related professions clearly limits the scope of our study's generalizability in many ways. Our estimate of the suicide mortality rate among men in the HPFS over the 24-year follow-up period (21 per 100 000 person-years) is lower than the 2010 national suicide mortality rate of 30.0 to 30.7 per 100 000 person-years among men aged 45 to 59 years (2), suggesting a lower overall level of psychiatric morbidity. If we had access to data on men from a more diverse range of socioeconomic backgrounds, including them in the analysis could have introduced greater possibilities for unobserved confounding, potentially biasing our estimates away from the null. The lack of representation of women in our sample is also a limitation. Not only is the burden of suicide in the United States lower among women (1), the effects of social integration on suicide mortality may also be different. Kposowa (58) found that being married and having a higher education level or higher income exerted strongly protective influences against suicide among men but not among women, although Kawachi and Berkman (60) suggested that the quality of relationships and the exchange of social support may be critical effect modifiers to consider. If we had access to data on women (whether in medically related professions or not), our analysis may have shown a null association. Finally, the men enrolled in the study were raised during a period characterized by relative stability in family structure and membership in religious institutions. Since then, the function and authority of family and religious institutions have changed: The median age at first marriage (87) and the rate of divorce (88) have increased, the fertility rate (89) and church attendance among younger generations (90) have decreased, and general population surveys reveal more voluntary and less obligatory attitudes toward marriage and childbearing (91). Parallel analyses of data on men or women from a different generation could yield different results.

Third, as mentioned, some suicides could have been misclassified as deaths from undetermined causes or as accidental deaths if incorrect information had been provided by the certifying physician, coroner, or medical examiner for the death certificate. Because ascertainment of suicide based on death certificates is likely to be highly specific but relatively less sensitive, we may have underestimated the number of suicides in the HPFS. This phenomenon is consistent with a more general problem of underreporting of suicides in the United States (7, 92). When we limited our analyses to firearm-related suicides, the estimated associations were nearly identical, suggesting that misclassification was unlikely to have biased our estimates away from the null. In addition, there is no reason to believe that underreporting could have produced the pattern of associations observed in our study. For misclassification to bias our estimates of the association between social integration and suicide away from the null, it would have had to occur disproportionately among socially well-integrated decedents. However, physicians employed by the HPFS to review death certificates and hospital or pathology reports to classify individual causes of death were blinded to exposure status.

In summary, in this 24-year longitudinal study of men in the United States, we found evidence to support the hypothesis that social integration is protective against suicide mortality. Despite the limitations noted, our study lends further support to the development of targeted social integration interventions aimed at strengthening existing social network structures or creating new ones. Examples include universal prevention strategies, such as those adopted by the U.S. Air Force (93), as well as selective prevention strategies, such as church- and community-based programs for contacting and supporting isolated persons (7). Our use of a general, nonenriched population sample provides important evidence to directly support population-level policies and programs targeting social integration in order to achieve primary suicide prevention (94), especially in light of prior work showing that a sizeable minority of persons who engage in suicidal behaviors do not meet criteria for a mental disorder (13). Given the strength of the observed relationship between social integration and suicide, research to identify its mechanisms may provide important insights into addressing this important public health issue.

Appendix: Supplementary Information on Methods

Details of the construction of the social integration index were described by Berkman (18) and were ultimately published in peer-reviewed form by Berkman and Syme (19). The index comprises 7 questions about marital status, social network size, frequency of contact with social ties, religious participation, and participation in other social groups. Marital status was a dichotomous variable that was equal to 1 if the participant reported currently being married and 0 otherwise (widowed, divorced, or never married). The numbers of close friends and relatives were elicited in 2 separate questions and grouped into 5 categories (none, 1 to 2, 3 to 5, 6 to 9, or ≥10). Similarly, frequency of contact with friends and relatives was elicited in 2 separate questions and grouped into 3 categories of contacts seen at least once a month (none, 1 to 2, or ≥3). Frequency of attendance at religious services was a dichotomous variable that was equal to 1 if the participant reported attendance at least once per week and 0 otherwise. Participation in other social groups was also a dichotomous variable and was equal to 1 if the participant reported spending any number of hours on group activities and 0 otherwise.

The variables on marital status, social network size, and frequency of contact were combined to classify study participants into 3 categories of intimate ties. Persons in the lowest category were either married with few friends or relatives or unmarried with a moderate number of friends or relatives seen frequently. Persons in the intermediate category were either married with a moderate number of friends or relatives seen frequently or unmarried with a large number of friends or relatives seen frequently. Finally, persons in the highest category were married with a large number of friends or relatives seen frequently. The variables on religious service attendance and group membership were combined into a single index of extended ties, with categories determined by participation in any social or recreational group, labor union, commercial or professional association, community group, service group, or charity, as well as frequent attendance (vs. infrequent or no attendance) at religious services. Berkman (18) combined the indices of intimate and extended ties to create the final social integration index scores shown in Appendix Table 1. The specific form of the social integration index was informed by conceptual considerations and empirically observed patterning in mortality rates in the original Alameda County Study.

Appendix Table 2 corresponds to sensitivity analyses, described in the text, that were designed to assess the extent to which the findings were affected by outcome misclassification (second and third columns) or confounding by comorbidity (fourth through seventh columns). Appendix Table 3 presents data on competing causes of death. Appendix Table 4 corresponds to supplemental analyses designed to assess the extent to which social integration was associated with these competing causes of death. The analyses in Appendix Table 5 are identical to those described in Appendix Table 4, except estimation was restricted to participants who did not report any history of cancer (specifically, any type of cancer except nonmelanoma skin cancer) or serious cardiovascular condition (specifically, acute myocardial infarction, coronary artery bypass graft surgery, percutaneous transluminal coronary angioplasty, or stroke) at baseline. Appendix Table 6 presents the results from 3 separate proportional subdistribution hazards regression models fitted to the data to account for competing mortality risks.

References

Heron
M
.
Deaths: leading causes for 2008.
Natl Vital Stat Rep
2012
60
1
94
PubMed
PubMed
Centers for Disease Control and Prevention (CDC)
Suicide among adults aged 35–64 years—United States, 1999–2010.
MMWR Morb Mortal Wkly Rep
2013
62
321
5
PubMed
PubMed
Lozano
R
,
Naghavi
M
,
Foreman
K
,
Lim
S
,
Shibuya
K
,
Aboyans
V
,
et al
Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010.
Lancet
2012
380
2095
128
PubMed
CrossRef

PubMed
Mathers
CD
,
Loncar
D
.
Projections of global mortality and burden of disease from 2002 to 2030.
PLoS Med
2006
3
e442
PubMed
CrossRef

PubMed
Corso
PS
,
Mercy
JA
,
Simon
TR
,
Finkelstein
EA
,
Miller
TR
.
Medical costs and productivity losses due to interpersonal and self-directed violence in the United States.
Am J Prev Med
2007
32
474
82
PubMed
CrossRef

PubMed
Mann
JJ
.
Neurobiology of suicidal behaviour.
Nat Rev Neurosci
2003
4
819
28
PubMed
CrossRef

PubMed
Goldsmith
SK
,
Pellmar
TC
,
Kleinman
AM
,
Bunney
WE
.
Reducing Suicide: A National Imperative.
Washington, DC
National Academies Pr
2001
ix
Durkheim
E
.
Le Suicide: Etude de Sociologie. 2nd ed.
Paris
Prs Univ France
1967
Wray
M
,
Colen
C
,
Pescosolido
B
.
The sociology of suicide.
Annu Rev Sociol
2011
37
505
28
CrossRef
Thorlindsson
T
,
Bjarnason
T
.
Modeling Durkheim on the micro level: a study of youth suicidality.
Am Sociol Rev
1998
63
94
110
CrossRef
Bearman
PS
,
Moody
J
.
Suicide and friendships among American adolescents.
Am J Public Health
2004
94
89
95
PubMed
CrossRef

PubMed
Holma
KM
,
Melartin
TK
,
Haukka
J
,
Holma
IA
,
Sokero
TP
,
Isometsä
ET
.
Incidence and predictors of suicide attempts in DSM-IV major depressive disorder: a five-year prospective study.
Am J Psychiatry
2010
167
801
8
PubMed
CrossRef

PubMed
Nock
MK
,
Hwang
I
,
Sampson
N
,
Kessler
RC
,
Angermeyer
M
,
Beautrais
A
,
et al
Cross-national analysis of the associations among mental disorders and suicidal behavior: findings from the WHO World Mental Health Surveys.
PLoS Med
2009
6
e1000123
PubMed
CrossRef

PubMed
Grobbee
DE
,
Rimm
EB
,
Giovannucci
E
,
Colditz
G
,
Stampfer
M
,
Willett
W
.
Coffee, caffeine, and cardiovascular disease in men.
N Engl J Med
1990
323
1026
32
PubMed
CrossRef

PubMed
Rimm
EB
,
Stampfer
MJ
,
Colditz
GA
,
Giovannucci
E
,
Willett
WC
.
Effectiveness of various mailing strategies among nonrespondents in a prospective cohort study.
Am J Epidemiol
1990
131
1068
71
PubMed
PubMed
Stampfer
MJ
,
Willett
WC
,
Speizer
FE
,
Dysert
DC
,
Lipnick
R
,
Rosner
B
,
et al
Test of the National Death Index.
Am J Epidemiol
1984
119
837
9
PubMed
PubMed
Rich-Edwards
JW
,
Corsano
KA
,
Stampfer
MJ
.
Test of the National Death Index and Equifax Nationwide Death Search.
Am J Epidemiol
1994
140
1016
9
PubMed
PubMed
Berkman
LF
.
Social networks, host resistance, and mortality: a follow-up study of Alameda County residents [Dissertation].
Berkeley
University of California at Berkeley
1977
Berkman
LF
,
Syme
SL
.
Social networks, host resistance, and mortality: a nine-year follow-up study of Alameda County residents.
Am J Epidemiol
1979
109
186
204
PubMed
PubMed
Berkman
LF
,
Breslow
L
.
Health and Ways of Living.
New York
Oxford Univ Pr
1983
Schaefer
C
,
Coyne
JC
,
Lazarus
RS
.
The health-related functions of social support.
J Behav Med
1981
4
381
406
PubMed
CrossRef

PubMed
Goodenow
C
,
Reisine
ST
,
Grady
KE
.
Quality of social support and associated social and psychological functioning in women with rheumatoid arthritis.
Health Psychol
1990
9
266
84
PubMed
CrossRef

PubMed
Kawachi
I
,
Colditz
GA
,
Ascherio
A
,
Rimm
EB
,
Giovannucci
E
,
Stampfer
MJ
,
et al
A prospective study of social networks in relation to total mortality and cardiovascular disease in men in the USA.
J Epidemiol Community Health
1996
50
245
51
PubMed
CrossRef

PubMed
Eng
PM
,
Rimm
EB
,
Fitzmaurice
G
,
Kawachi
I
.
Social ties and change in social ties in relation to subsequent total and cause-specific mortality and coronary heart disease incidence in men.
Am J Epidemiol
2002
155
700
9
PubMed
CrossRef

PubMed
Cox
DR
,
Oakes
D
.
Analysis of Survival Data.
London
Chapman and Hall
1984
Therneau
TM
,
Grambsch
PM
.
Modeling Survival Data: Extending the Cox Model.
New York
Springer-Verlag
2000
Hawton
K
,
Agerbo
E
,
Simkin
S
,
Platt
B
,
Mellanby
RJ
.
Risk of suicide in medical and related occupational groups: a national study based on Danish case population-based registers.
J Affect Disord
2011
134
320
6
PubMed
CrossRef

PubMed
Roberts
SE
,
Jaremin
B
,
Lloyd
K
.
High-risk occupations for suicide.
Psychol Med
2013
43
1231
40
PubMed
CrossRef

PubMed
Lewis
G
,
Sloggett
A
.
Suicide, deprivation, and unemployment: record linkage study.
BMJ
1998
317
1283
6
PubMed
CrossRef

PubMed
Miller
M
,
Hemenway
D
,
Rimm
E
.
Cigarettes and suicide: a prospective study of 50,000 men.
Am J Public Health
2000
90
768
73
PubMed
CrossRef

PubMed
Mukamal
KJ
,
Kawachi
I
,
Miller
M
,
Rimm
EB
.
Drinking frequency and quantity and risk of suicide among men.
Soc Psychiatry Psychiatr Epidemiol
2007
42
153
60
PubMed
CrossRef

PubMed
Kawachi
I
,
Willett
WC
,
Colditz
GA
,
Stampfer
MJ
,
Speizer
FE
.
A prospective study of coffee drinking and suicide in women.
Arch Intern Med
1996
156
521
5
PubMed
CrossRef

PubMed
Nock
MK
,
Hwang
I
,
Sampson
NA
,
Kessler
RC
.
Mental disorders, comorbidity and suicidal behavior: results from the National Comorbidity Survey Replication.
Mol Psychiatry
2010
15
868
76
PubMed
CrossRef

PubMed
Rimm
EB
,
Stampfer
MJ
,
Colditz
GA
,
Chute
CG
,
Litin
LB
,
Willett
WC
.
Validity of self-reported waist and hip circumferences in men and women.
Epidemiology
1990
1
466
73
PubMed
CrossRef

PubMed
Mukamal
KJ
,
Kawachi
I
,
Miller
M
,
Rimm
EB
.
Body mass index and risk of suicide among men.
Arch Intern Med
2007
167
468
75
PubMed
CrossRef

PubMed
Chasan-Taber
S
,
Rimm
EB
,
Stampfer
MJ
,
Spiegelman
D
,
Colditz
GA
,
Giovannucci
E
,
et al
Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals.
Epidemiology
1996
7
81
6
PubMed
CrossRef

PubMed
Colditz
GA
,
Martin
P
,
Stampfer
MJ
,
Willett
WC
,
Sampson
L
,
Rosner
B
,
et al
Validation of questionnaire information on risk factors and disease outcomes in a prospective cohort study of women.
Am J Epidemiol
1986
123
894
900
PubMed
PubMed
Druss
B
,
Pincus
H
.
Suicidal ideation and suicide attempts in general medical illnesses.
Arch Intern Med
2000
160
1522
6
PubMed
CrossRef

PubMed
Cerhan
JR
,
Wallace
RB
.
Predictors of decline in social relationships in the rural elderly.
Am J Epidemiol
1993
137
870
80
PubMed
PubMed
Breiding
MJ
,
Wiersema
B
.
Variability of undetermined manner of death classification in the US.
Inj Prev
2006
12
Suppl 2
ii49
ii54
PubMed
CrossRef

PubMed
Pescosolido
BA
,
Mendelsohn
R
.
Social causation or social construction of suicide? An investigation into the social organization of official rates.
Am Sociol Rev
1986
51
80
100
CrossRef
Spicer
RS
,
Miller
TR
.
Suicide acts in 8 states: incidence and case fatality rates by demographics and method.
Am J Public Health
2000
90
1885
91
PubMed
CrossRef

PubMed
Shenassa
ED
,
Catlin
SN
,
Buka
SL
.
Lethality of firearms relative to other suicide methods: a population based study.
J Epidemiol Community Health
2003
57
120
4
PubMed
CrossRef

PubMed
Platt
S
,
Backett
S
,
Kreitman
N
.
Social construction or causal ascription: distinguishing suicide from undetermined deaths.
Soc Psychiatry Psychiatr Epidemiol
1988
23
217
21
PubMed
CrossRef

PubMed
Linsley
KR
,
Schapira
K
,
Kelly
TP
.
Open verdict v. suicide - importance to research.
Br J Psychiatry
2001
178
465
8
PubMed
CrossRef

PubMed
Lindqvist
P
,
Gustafsson
L
.
Suicide classification—clues and their use. a study of 122 cases of suicide and undetermined manner of death.
Forensic Sci Int
2002
128
136
40
PubMed
CrossRef

PubMed
Rockett
IR
,
Wang
S
,
Stack
S
,
De Leo
D
,
Frost
JL
,
Ducatman
AM
,
et al
Race/ethnicity and potential suicide misclassification: window on a minority suicide paradox?
BMC Psychiatry
2010
10
35
PubMed
CrossRef

PubMed
Miniño
AM
,
Anderson
RN
,
Fingerhut
LA
,
Boudreault
MA
,
Warner
M
.
Deaths: injuries, 2002.
Natl Vital Stat Rep
2006
54
1
124
PubMed
Fang
F
,
Fall
K
,
Mittleman
MA
,
Sparén
P
,
Ye
W
,
Adami
HO
,
et al
Suicide and cardiovascular death after a cancer diagnosis.
N Engl J Med
2012
366
1310
8
PubMed
CrossRef

PubMed
Stenager
EN
,
Madsen
C
,
Stenager
E
,
Boldsen
J
.
Suicide in patients with stroke: epidemiological study.
BMJ
1998
316
1206
PubMed
CrossRef

PubMed
Larsen
KK
,
Agerbo
E
,
Christensen
B
,
Søndergaard
J
,
Vestergaard
M
.
Myocardial infarction and risk of suicide: a population-based case-control study.
Circulation
2010
122
2388
93
PubMed
CrossRef

PubMed
Cornwell
B
.
Good health and the bridging of structural holes.
Soc Networks
2009
31
92
103
PubMed
CrossRef

PubMed
Schafer
MH
.
Structural advantages of good health in old age: investigating the health-begets-position hypothesis with a full social network.
Res Aging
2013
35
348
70
CrossRef
Fine
JP
,
Gray
RJ
.
A proportional hazards model for the subdistribution of a competing risk.
J Am Stat Assoc
1999
94
496
509
CrossRef
Gray
RJ
.
A class of k-sample tests for comparing the cumulative incidence of a competing risk.
Ann Stat
1988
16
1141
54
CrossRef
Cerhan
JR
,
Wallace
RB
.
Change in social ties and subsequent mortality in rural elders.
Epidemiology
1997
8
475
81
PubMed
CrossRef

PubMed
Rockett
IR
,
Regier
MD
,
Kapusta
ND
,
Coben
JH
,
Miller
TR
,
Hanzlick
RL
,
et al
Leading causes of unintentional and intentional injury mortality: United States, 2000-2009.
Am J Public Health
2012
102
e84
92
PubMed
CrossRef

PubMed
Kposowa
AJ
.
Marital status and suicide in the National Longitudinal Mortality Study.
J Epidemiol Community Health
2000
54
254
61
PubMed
CrossRef

PubMed
Berkman
LF
,
Glass
T
,
Brissette
I
,
Seeman
TE
.
From social integration to health: Durkheim in the new millennium.
Soc Sci Med
2000
51
843
57
PubMed
CrossRef

PubMed
Kawachi
I
,
Berkman
LF
.
Social ties and mental health.
J Urban Health
2001
78
458
67
PubMed
CrossRef

PubMed
Brissette
I
,
Cohen
S
,
Seeman
TE
.
Social Support Measurement and Intervention: A Guide for Health and Social Scientists.
New York
Oxford Univ Pr
2000
53
85
Dean
K
,
Holst
E
,
Kreiner
S
,
Schoenborn
C
,
Wilson
R
.
Measurement issues in research on social support and health.
J Epidemiol Community Health
1994
48
201
6
PubMed
CrossRef

PubMed
Koenig
HG
,
King
DE
,
Carson
VB
.
Handbook of Religion and Health. 2nd ed.
New York
Oxford Univ Pr
2012
Breault
KD
.
Suicide in America: a test of Durkheim's theory of religious and family integration, 1933–1980.
Am J Sociol
1986
92
628
56
CrossRef
Giordano
GN
,
Lindstrom
M
.
The impact of changes in different aspects of social capital and material conditions on self-rated health over time: a longitudinal cohort study.
Soc Sci Med
2010
70
700
10
PubMed
CrossRef

PubMed
Thomas
PA
.
Trajectories of social engagement and limitations in late life.
J Health Soc Behav
2011
52
430
43
PubMed
CrossRef

PubMed
Seeman
TE
,
Miller-Martinez
DM
,
Stein Merkin
S
,
Lachman
ME
,
Tun
PA
,
Karlamangla
AS
.
Histories of social engagement and adult cognition: midlife in the U.S. study.
J Gerontol B Psychol Sci Soc Sci
2011
66
Suppl 1
i141
52
PubMed
CrossRef

PubMed
Thomas
PA
.
Gender, social engagement, and limitations in late life.
Soc Sci Med
2011
73
1428
35
PubMed
CrossRef

PubMed
Thomas
PA
.
Trajectories of social engagement and mortality in late life.
J Aging Health
2012
24
547
68
PubMed
CrossRef

PubMed
Ikeda
A
,
Kawachi
I
.
Handbook of Behavioral Medicine.
New York
Springer
2010
237
61
deCatanzaro
D
.
Human suicide: toward a diathesis-stress hypothesis.
Behav Brain Sci
1980
3
283
90
CrossRef
Meehl
PE
.
Schizotaxia, schizotypy, schizophrenia.
Am Psychol
1962
17
827
38
CrossRef
Monroe
SM
,
Simons
AD
.
Diathesis-stress theories in the context of life stress research: implications for the depressive disorders.
Psychol Bull
1991
110
406
25
PubMed
CrossRef

PubMed
Cantor
CH
,
Slater
PJ
.
Marital breakdown, parenthood, and suicide.
Journal of Family Studies
1995
1
91
102
CrossRef
Berkman
LF
.
Assessing the physical health effects of social networks and social support.
Annu Rev Public Health
1984
5
413
32
PubMed
CrossRef

PubMed
Berkman
LF
.
Social networks, support, and health: taking the next step forward.
Am J Epidemiol
1986
123
559
62
PubMed
PubMed
Berkman
LF
.
The role of social relations in health promotion.
Psychosom Med
1995
57
245
54
PubMed
CrossRef

PubMed
Peplau
LA
,
Perlman
D
.
Love and Attraction: An International Conference.
Oxford, United Kingdom
Pergamon Pr
1979
101
10
Feld
SL
,
Suitor
JJ
,
Hoegh
JG
.
Describing changes in personal networks over time.
Field Methods
2007
19
218
36
CrossRef
Cornwell
B
,
Laumann
EO
.
The health benefits of network growth: New evidence from a national survey of older adults.
Soc Sci Med
2013
PubMed
Perlis
RH
.
Hard outcomes: clinical trials to reduce suicide [Editorial].
Am J Psychiatry
2011
168
1009
11
PubMed
CrossRef

PubMed
Coyne
JC
.
Toward an interactional description of depression.
Psychiatry
1976
39
28
40
PubMed
PubMed
Coyne
JC
.
Depression and the response of others.
J Abnorm Psychol
1976
85
186
93
PubMed
CrossRef

PubMed
Hirschfeld
RM
,
Keller
MB
,
Panico
S
,
Arons
BS
,
Barlow
D
,
Davidoff
F
,
et al
The National Depressive and Manic-Depressive Association consensus statement on the undertreatment of depression.
JAMA
1997
277
333
40
PubMed
CrossRef

PubMed
McGlynn
EA
,
Asch
SM
,
Adams
J
,
Keesey
J
,
Hicks
J
,
DeCristofaro
A
,
et al
The quality of health care delivered to adults in the United States.
N Engl J Med
2003
348
2635
45
PubMed
CrossRef

PubMed
Kessler
RC
,
Demler
O
,
Frank
RG
,
Olfson
M
,
Pincus
HA
,
Walters
EE
,
et al
Prevalence and treatment of mental disorders, 1990 to 2003.
N Engl J Med
2005
352
2515
23
PubMed
CrossRef

PubMed
Goldstein
JR
,
Kenney
CT
.
Marriage delayed or marriage forgone? New cohort forecasts of first marriage for U.S. women.
Am Sociol Rev
2001
66
506
19
CrossRef
Goldstein
JR
.
The leveling of divorce in the United States.
Demography
1999
36
409
14
PubMed
CrossRef

PubMed
Cherlin
A
.
Recent changes in American fertility, marriage, and divorce.
Ann Am Acad Pol Soc Sci
1990
510
145
54
CrossRef

PubMed
Miller
AS
,
Nakamura
T
.
On the stability of church attendance patterns during a time of demographic change: 1965–1988.
J Sci Study Relig
1996
35
275
84
CrossRef
Thornton
A
,
Young-DeMarco
L
.
Four decades of trends in attitudes toward family issues in the United States: the 1960s through the 1990s.
J Marriage Fam
2001
63
1009
37
CrossRef
Phillips
DP
,
Ruth
TE
.
Adequacy of official suicide statistics for scientific research and public policy.
Suicide Life Threat Behav
1993
23
307
19
PubMed
PubMed
Knox
KL
,
Litts
DA
,
Talcott
GW
,
Feig
JC
,
Caine
ED
.
Risk of suicide and related adverse outcomes after exposure to a suicide prevention programme in the US Air Force: cohort study.
BMJ
2003
327
1376
PubMed
CrossRef

PubMed
Ertel
KA
,
Glymour
MM
,
Berkman
LF
.
Social networks and health: a life course perspective integrating observational and experimental evidence.
J Soc Pers Relat
2009
26
73
92
CrossRef

Appendix Table 1. Social Integration Index

Table 1. Baseline Sample Characteristics, by Social Integration Category Measured in 1988*

Table 2. Suicides During 1988–2012

Table 3. Relative Hazard Ratios for Suicide During 1988–2012, by Social Integration Category Measured in 1988

Table 4. Relative Hazard Ratios for Suicide During 1996–2012, by Social Integration Trajectory From 1988 to 1996

Table 5. Relative Hazard Ratios for Suicide During 1988–2012, by Social Integration Components Measured in 1988

Appendix Table 2. Relative Hazard Ratios for Suicide During 1988–2012, by Social Integration Category Measured in 1988: Sensitivity Analysis

Appendix Table 3. Deaths During 1988–2012, by Social Integration Category Measured in 1988

Appendix Table 4. Relative Hazard Ratios for Competing Mortality Outcomes During 1988–2012, by Social Integration Category Measured in 1988

Appendix Table 5. Relative Hazard Ratios for Competing Mortality Outcomes During 1988–2012, by Social Integration Category Measured in 1988: Participants With Any History of Cancer or Serious Cardiovascular Condition at Baseline Excluded

Appendix Table 6. Subdistribution Relative Hazard Ratios for Suicide During 1988–2012, Accounting for Competing Mortality Risks

This feature is available only to Registered Users

Subscribe/Learn More

PDF

Not Available

Original Research |15 July 2014

Social Integration and Suicide Mortality Among Men: 24-Year Cohort Study of U.S. Health Professionals Free

Abstract

Editors’ Notes

Context

Contribution

Caution

Implication

Methods

Study Design and Setting

Measures

Appendix Table 1. Social Integration Index

Statistical Analysis

Role of the Funding Source

Results

Table 1. Baseline Sample Characteristics, by Social Integration Category Measured in 1988*

Table 2. Suicides During 1988–2012

Figure.

Table 3. Relative Hazard Ratios for Suicide During 1988–2012, by Social Integration Category Measured in 1988

Table 4. Relative Hazard Ratios for Suicide During 1996–2012, by Social Integration Trajectory From 1988 to 1996

Table 5. Relative Hazard Ratios for Suicide During 1988–2012, by Social Integration Components Measured in 1988

Appendix Table 2. Relative Hazard Ratios for Suicide During 1988–2012, by Social Integration Category Measured in 1988: Sensitivity Analysis

Appendix Table 3. Deaths During 1988–2012, by Social Integration Category Measured in 1988

Appendix Table 4. Relative Hazard Ratios for Competing Mortality Outcomes During 1988–2012, by Social Integration Category Measured in 1988

Appendix Table 5. Relative Hazard Ratios for Competing Mortality Outcomes During 1988–2012, by Social Integration Category Measured in 1988: Participants With Any History of Cancer or Serious Cardiovascular Condition at Baseline Excluded

Appendix Table 6. Subdistribution Relative Hazard Ratios for Suicide During 1988–2012, Accounting for Competing Mortality Risks

Discussion

Appendix: Supplementary Information on Methods

References

Figure.

Appendix Table 1. Social Integration Index

Table 1. Baseline Sample Characteristics, by Social Integration Category Measured in 1988*

Table 2. Suicides During 1988–2012

Table 3. Relative Hazard Ratios for Suicide During 1988–2012, by Social Integration Category Measured in 1988

Table 4. Relative Hazard Ratios for Suicide During 1996–2012, by Social Integration Trajectory From 1988 to 1996

Table 5. Relative Hazard Ratios for Suicide During 1988–2012, by Social Integration Components Measured in 1988

Appendix Table 2. Relative Hazard Ratios for Suicide During 1988–2012, by Social Integration Category Measured in 1988: Sensitivity Analysis

Appendix Table 3. Deaths During 1988–2012, by Social Integration Category Measured in 1988

Appendix Table 4. Relative Hazard Ratios for Competing Mortality Outcomes During 1988–2012, by Social Integration Category Measured in 1988

Appendix Table 5. Relative Hazard Ratios for Competing Mortality Outcomes During 1988–2012, by Social Integration Category Measured in 1988: Participants With Any History of Cancer or Serious Cardiovascular Condition at Baseline Excluded

Appendix Table 6. Subdistribution Relative Hazard Ratios for Suicide During 1988–2012, Accounting for Competing Mortality Risks

This feature is available only to Registered Users

0 Comments

See Also

Related Articles

Journal Club

Related Topics

PubMed Articles

Content

Information For

Services

Awards and Cover

Other Resources

Follow Annals On

You need a subscription to this content to use this feature.