Figure 1. Study flow diagram.
Figure 2. Four-year event curves for all-cause mortality by living arrangement status. Kaplan-Meier cumulative incidence curves for all-cause mortality within the entire population with 4-year follow-up by living arrangement status. Unadjusted hazard ratio (HR), 1.27 (95% CI, 1.19-1.37); adjusted HR, 1.11 (95% CI, 1.01-1.23) (P < .01, log-rank test). The HR represents the risk associated with living alone compared with living with others (reference). Adjusted for age, sex, employment, education, ethnic origin, geographic region, history of smoking, diabetes mellitus, body mass index, atrial fibrillation/flutter, heart failure, vascular disease status (polyvascular disease, single vascular disease, or risk factors only), prior ischemic event (≤1 year, >1 year, or no ischemic event), statins (at baseline), aspirin (at baseline).
Figure 3. Four-year hazard ratios (HRs) for mortality associated with living alone stratified by baseline sociodemographic categories. A, Unadjusted HRs for mortality associated with living alone compared with living with others (reference) stratified by baseline sociodemographic categories. Horizontal lines indicate 95% CIs. B, Adjusted HRs for mortality associated with living alone compared with living with others (reference) stratified by baseline sociodemographic categories. Horizontal lines indicate 95% CIs. *Adjusted for significant risk factors including age, sex, employment, education, race/ethnicity, geographic region, history of smoking, diabetes mellitus, body mass index, atrial fibrillation/flutter, heart failure, vascular disease status (polyvascular disease, single vascular disease, or risk factors only), prior ischemic event (≤1 year, >1 year, or no ischemic event), statins (at baseline), aspirin (at baseline). A and B, Age 65 years was rounded from 65.5 years.
Udell JA, Steg PG, Scirica BM, Smith SC, Ohman EM, Eagle KA, Goto S, Cho JI, Bhatt DL, REduction of Atherothrombosis for Continued Health (REACH) Registry Investigators FT. Living Alone and Cardiovascular Risk in Outpatients at Risk of or With Atherothrombosis. Arch Intern Med. 2012;172(14):1086-1095. doi:10.1001/archinternmed.2012.2782
Author Affiliations: Thrombolysis In Myocardial Infarction (TIMI) Study Group, and Cardiovascular Division, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts (Drs Udell, Scirica, and Bhatt and Mr Cho); Cardiovascular Division, Department of Medicine, Women's College Hospital, University of Toronto, Toronto, Ontario, Canada (Dr Udell); INSERM U-698, Paris, France (Dr Steg); Université Paris-Diderot, Sorbonne Paris Cité, Paris (Dr Steg); Department of Cardiology, Assistance Publique–Hôpitaux de Paris, Hôpital Bichat, Paris (Dr Steg); Department of Medicine, University of North Carolina at Chapel Hill, Chapel Hill (Dr Smith); Division of Cardiology, Department of Medicine, Duke University, Durham, North Carolina (Dr Ohman); Department of Medicine, University of Michigan Cardiovascular Center, Ann Arbor (Dr Eagle); Department of Medicine, Tokai University School of Medicine, Kanagawa, Japan (Dr Goto); and Department of Medicine, VA Boston Healthcare System, Boston (Dr Bhatt).
Group Information: A complete list of the REACH Registry Investigators was published in JAMA. 2006;295(2):180-189.
Background Living alone, a proxy for social support, has been inconsistently linked with cardiovascular risk.
Methods We investigated whether living alone was associated with increased mortality and cardiovascular risk in the global REduction of Atherothrombosis for Continued Health (REACH) Registry. Stable outpatients at risk of or with atherothrombosis were recruited from December 1, 2003, through December 31, 2004, and followed up to 4 years for cardiovascular events. Events were examined by living arrangement with risk adjustment for age, sex, clinical risk factors, therapy, preexisting vascular disease, and sociodemographic factors. Effect modification was tested by age, sex, employment, ethnicity, education, and geography.
Results Among the 44 573 REACH participants, 8594 (19%) were living alone. Living alone was associated with higher 4-year mortality (14.1% vs 11.1%) and cardiovascular death (8.6% vs 6.8%; log-rank P < .01 for both comparisons); however, there was significant effect modification by age (P value for interaction = .03). Specifically, among younger participants, living alone compared with those living with others was associated with higher mortality (age 45-65 years: 7.7% vs 5.7%; adjusted hazard ratio [HR], 1.24 [95% CI, 1.01-1.51]; age 66-80 years: 13.2% vs 12.3%; adjusted HR, 1.12 [95% CI, 1.01-1.26]), but this was not observed among older participants (age > 80 years: 24.6% vs 28.4%; adjusted HR, 0.92 [95% CI, 0.79-1.06]). A similar trend was observed for the risk of cardiovascular death.
Conclusions In an international outpatient population with atherothrombosis aged 45 years or older, living alone was associated with increased mortality among all but the most elderly patients, although this observation warrants confirmation.
Social isolation, or an absence of real or perceived social support, may be associated with poor health consequences, including cardiovascular (CV) disease.1- 15 The degree to which people are integrated with others may be as strongly predictive of incident myocardial infarction (MI) and coronary heart disease (CHD) survival as smoking, elevated cholesterol levels, and hypertension.16- 18 Moreover, the risk associated with living alone is relevant given that approximately 1 in 7 American adults are living alone.19,20 To some, living alone may have associated social and economic benefits but also potential drawbacks, including social mobility and isolation,20- 22 as predicted by Sorokin23 in the early 20th century. Increasingly, individuals in developed and developing countries are living away from extended family or friends, participating in mobile employment, and communicating remotely. There is substantial epidemiological evidence that social isolation may alter neurohormonal-mediated emotional stress, influence health behavior, and effect access to health care, resulting in association with, or acquisition of, CV risk.1- 3,8- 10,15,24- 29 Prior CV risk estimates associated with social isolation vary considerably owing to the difficulty in reliably and precisely estimating multidimensional psychosocial measures among diverse settings. To our knowledge, no prior report has comparatively studied the risk associated with social isolation among an internationally diverse population of outpatients with varying CV and sociodemographic risk factors, including employment, education, and geographic region of dwelling.16,17 The global REduction of Atherothrombosis for Continued Health (REACH) Registry provided a unique opportunity to prospectively investigate whether a simple measure of social isolation, living alone, is associated with increased CV risk and mortality across international populations of outpatients at risk of, or with, atherothrombosis.
The methods of the REACH Registry have previously been described in detail.30- 33 Briefly, the REACH Registry recruited 67 888 outpatients 45 years or older from 44 countries. The current report includes all eligible patients enrolled in centers that participated in the entire 4-year follow-up study (Figure 1). Patients with prior coronary, cerebrovascular, or peripheral artery disease (PAD) (herein referred to collectively as established atherothrombosis), and participants without established atherothrombosis but at least 3 CV risk factors were recruited from December 1, 2003, through December 31, 2004, and followed until 2008. Those with multiple risk factors consisted of patients with diabetes mellitus (DM), microalbuminuria, ankle-brachial index (ABI) less than 0.9, asymptomatic carotid stenosis of at least 70%, carotid intima media thickness, hypertension, dyslipidemia, current smoking, or age 65 years or older for men or 70 years or older for women. Established coronary artery disease was defined as a history of stable angina, unstable angina, percutaneous coronary intervention (PCI), coronary artery bypass grafting, or MI. Established cerebrovascular disease consisted of patients with a neurologist or hospital report documenting a history of ischemic stroke or transient ischemic attack. Those with established PAD had prior or current symptomatic claudication plus at least 1 supporting diagnostic test or vascular intervention, either an ABI less than 0.9, prior revascularization, or amputation owing to peripheral vascular disease. Polyvascular disease was defined as coexistent established arterial disease in at least 1 arterial territory (any combination of coronary, cerebral, peripheral vascular, or all 3).
A broad representation of physician clinics, including a balance of generalist and specialist practices; rural, suburban, and urban settings; and office-based and hospital-based practices enrolled participants in the REACH Registry.30,31 Sociodemographic variables collected include race/ethnicity, number of years of formal education completed (0-8 years or no high school, 9-12 years or high school, trade or technical school, and university or college) and employment status (full-time employed, part-time employed, unemployed, retired, incapacitated for work, or other status). Race/ethnicity was self-reported. If certain countries or local rules did not allow race/ethnicity recording, the data for those patients were considered missing. Participants were also categorized by region from where they were enrolled, including North America, Latin America, Western Europe, Eastern Europe, the Middle East, Japan, and other Asian countries. For our main exposure of interest, participants were asked at baseline whether they lived alone (yes/no).
At annual follow-up, physicians reported patients' clinical outcomes determined from available data according to local clinical practice. The primary end point of this analysis was all-cause mortality. The secondary end point was CV death, and a composite of CV death, nonfatal MI, and nonfatal stroke. Cardiovascular death was defined as fatal stroke, MI, or other CV death. Any MI or stroke followed by death in the subsequent 28 days was considered fatal. Individual CV events were also evaluated. Although CV events were not adjudicated, data quality was audited in person at a sample of 10% of all sites to confirm source documentation and accuracy for all of their submitted case report forms. These sites were chosen randomly 6% of the time, and an additional 4% were chosen because of the number of queries and missing data from the sites.
The prevalence of CV risk factors, established atherothrombosis, and number of diseased vascular beds were compared between participants who were and were not living alone at baseline with the χ2 test for categorical variables and the t test for continuous variables.
Cumulative event rates through 48 months were determined for those participants who were and were not living alone. Curves for event-free survival were constructed with the Kaplan-Meier approach with event rate comparisons of living arrangement status calculated using the log-rank statistic. Survival time was calculated according to the date of the outcome as collected by the enrolling physicians. For patients who did not experience the event of interest, data were censored at the time of the last visit with available information.
To account for potential confounders, we constructed Cox models to compare the hazard ratios (HRs) and 95% CIs for living alone associated with mortality and CV events in the overall population. These models were additionally adjusted for age, sex, employment, education, race or ethnic origin, geographic region, history of smoking, DM, body mass index, atrial fibrillation or flutter, heart failure, vascular disease status at baseline (polyvascular disease, single vascular disease, or risk factors only), prior ischemic event (≤1 year, >1 year, or no ischemic event), baseline statin use and baseline aspirin use. The covariates were chosen because they all predicted 4-year CV events in a stepwise-selection Cox model constructed in the overall REACH population.33 The homogeneity of the effect of living alone and CV events across sociodemographic subgroups was tested by adding interaction terms for living alone by sociodemographic categories, including terms for age, sex, employment status, race/ethnicity, geographic region, and education, given prior research suggesting an association between sociodemographic factors and likelihood of living alone.5,6 Two-sided P values comparing HRs by living alone status were calculated with the Wald χ2 test with P < .05 considered significant. Statistical analyses were performed with SAS software (version 9; SAS Institute Inc).
The study was prepared in compliance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.34 Data analysis was conducted independent of the registry sponsors by 2 of us (J.A.U. and J.I.C.), who planned the data analysis and had access to the entire raw data set. The study was approved by the institutional review boards at each site according to local requirements, and signed informed consent was required for all patients.
Of the 44 573 participants eligible for inclusion in the 4-year REACH Registry in whom living arrangement status was collected, 8594 were living alone (19.3%). Table 1 presents the prevalence rates of living alone within groups of patients categorized by degree of atherothrombosis. The prevalence of living alone ranged from 24.3% in those with multiple risk factors to 20.9% in those with stable atherosclerosis without a prior ischemic event, and 18.2% in those with a prior ischemic event.
The baseline characteristics of the population according to living arrangement are presented in Table 2. Participants who lived alone, compared with those living with others, more frequently were older than 80 years, female, living in North America or Western Europe, of black or white race, had less than 12 years of formal education, and were retired from employment (P < .001 for all comparisons). The prevalence of living alone increased with age. Among participants aged 45 to 65 years, 66 to 80 years, and older than 80 years, 13.6%, 19.8%, and 34.6%, were living alone, respectively. Living alone was also associated with an increased prevalence of CV risk factors, including hypertension, hyperlipidemia, renal impairment, obesity, heart failure, atrial fibrillation, and smoking.
During the follow-up period, a total of 4338 deaths and 2612 CV deaths were recorded. Living alone was associated with a higher 4-year rate of all-cause mortality (14.1% vs 11.1%; HR, 1.27 [95% CI, 1.19-1.37]; P < .001; Figure 2), CV death (8.6% vs 6.8%; HR, 1.25 [95% CI, 1.14-1.37]; P < .001), and the composite of CV death, nonfatal MI, and nonfatal stroke (Table 3).
After adjustment for clinical and sociodemographic factors associated with CV events, living alone remained independently associated with higher all-cause mortality (P = .04) and CV death (P = .04) compared with those living with others; however, there was significant heterogeneity in the mortality risk associated with living alone when participants were stratified by age (P value for interaction = .03). An increased risk of mortality was consistent among those participants 45 to 65 years of age living alone compared with those living with others (7.7% vs 5.7%; unadjusted HR, 1.31 [95% CI, 1.09-1.57]; adjusted HR, 1.24 [95% CI, 1.01-1.51]) and among those participants 66 to 80 years old living alone compared with those living with others (13.2% vs 12.3%; unadjusted HR, 1.07 [95% CI, 0.97-1.18]; adjusted HR, 1.12 [95% CI, 1.01-1.26]). In contrast, individuals older than 80 years living alone were not at increased risk of mortality compared with those living with others (24.6% vs 28.4%; unadjusted HR, 0.86 [95% CI, 0.76-0.97]; adjusted HR, 0.92 [95% CI, 0.79-1.06]). A similar trend was observed between the association of age and living alone for the risk of CV death (Table 3). Table 4 shows the event rates and hazard associated with living alone according to baseline CV risk category. Among participants with a history of risk factors only, we observed no increased risk associated with living alone after multivariate adjustment. In contrast, among participants with a history of atherothrombosis, living alone remained associated with an increased risk for CV death after multivariate adjustment with consistent risk modification by age. The effect of living alone among other sociodemographic subgroups is presented in Table 5 and Table 6 and Figure 3. Other sociodemographic factors were not significant modifiers of the risk associated with living alone.
Among this large contemporary international cohort with or at risk of atherothrombosis, living alone was common and associated with an increased risk of mortality and CV death among middle-aged participants. To our knowledge, our report is the first to prospectively compare the CV risk associated with living alone in an international outpatient cohort with wide-ranging CV and sociodemographic risk factors.
Prior reports have shown mixed results for an association between social isolation and mortality or recurrent CV events after MI, predominantly among North American populations. Consistent with our findings, within the Multicenter Diltiazem Post-Infarction Trial, living alone was an independent risk factor for recurrent cardiac events (nonfatal MI or CV death) at 6 months following an acute MI.5 Whereas the Global Use of STrategies to Open occluded coronary arteries (GUSTO)-III investigators reported that living alone was associated with an increased crude rate of 30-day and 1-year mortality after MI, but after multivariate adjustment the mortality risk of living alone was attenuated.6 Similarly, within the US-based Prospective Registry Evaluating Myocardial Infarction: Events and Recovery, living alone was not independently associated with an increased mortality risk 1 year after MI.14 A population-based cohort from Western Canada found that living alone increased mortality risk after MI in men but not women.11
Other cohort studies have recorded detailed psychosocial factors at the time of MI in order to capture a patient's social situation more accurately.4,7,8 In these studies, the risk associated with living alone after an MI was predominantly attenuated by the presence or lack of emotional support4,7 or was a significant modifier of another psychological risk factor, such as depression.8 However, these studies took place over a 3-decade period with differences in baseline patient characteristics and inclusion criteria, quality of in- and out-patient CV care, and ascertainment of follow-up events, among other methodological factors. Notably, the magnitude of the mortality risk associated with social integration in prior studies has also varied according to the degree to which measures of structural and functional aspects of social support were captured.18 For instance, social integration was most highly associated with mortality risk when evaluated in a multidimensional orthogonal manner and least likely when considered as a binary metric, such as living alone (yes/no). This suggests that our observed increased risk of living alone for CV mortality is likely a conservative estimate.
Within the REACH registry, there was significant heterogeneity in the mortality risk associated with living alone and age. In participants 45 to 65 years of age, among whom living with others was more common, living alone was associated with a significantly higher risk of all-cause and CV mortality compared with participants living with others. In comparison, although the prevalence of living alone nearly tripled among those older than 80 years, living alone was not associated with a higher risk of mortality in the oldest participants. Thus, living alone at a younger age may be a marker of a stressful psychosocial situation, such as job strain or isolation, with adverse neurohormonal effects on the CV system, that may have more impact in youth to influence health behavior and wellness.3 In contrast, living alone in elderly individuals seems to be a more normative behavior and potentially associated with independence and better health rather than a maladaptive environment associated with CV risk. Our findings seem plausible if a participant was otherwise independent and healthy to survive until 80 years of age.
Our observation that the risk of living alone was significant only among middle-aged participants with established atherothrombosis may suggest that living alone is associated not with an increased risk of incident CV events but rather with lethal recurrent events, perhaps by amplifying susceptibility to sudden cardiac death. This observation seems in keeping with a reported association between living with a spouse and seeking prompt medical attention for a serious CV event, such as new chest pain,29 before a lethal complication ensues. In contrast, those individuals who live alone may delay seeking medical attention for an ischemic symptom and increase their risk for a fatal event.
If living arrangement is an independent prognostic factor in middle-aged patients with established atherothrombosis, then this simple measure of social isolation may provide health care practitioners a unique opportunity to screen their patients after an acute ischemic event to identify those at higher risk for CV mortality.35 Physicians may counsel patients regarding their risk, emphasize the need to seek immediate medical attention without delay for recurrent symptoms, and refer them to rehabilitation programs with a psychosocial intervention. However, to our knowledge, no prospective study of intensive cardiac rehabilitation support has demonstrated an effective strategy to overcome the mortality risk associated with social isolation post-MI.36
Despite varying social prevalence and cultural influence associated with the likelihood of living alone, there was no other factor that significantly modified the effect of living alone, aside from age. The lack of a significant interaction among other sociodemographic factors may be related to there being fewer participants in certain subgroups. Nevertheless, the risk associated with living alone demonstrated a consistent trend in men and women, and most geographic regions, races/ethnicities, and employment status. The heterogeneity observed between age and the risk associated with living alone was independent of other socioeconomic factors in our multivariate model.
The limitations of our study are worth noting and include bias and unmeasured confounding. As with any registry, volunteer bias may have influenced the type of patients recruited by physicians; however, the sampling from 44 countries constituting numerous ethnicities and cultures makes this threat less likely. Our exposure of interest, living alone, was dichotomized at one time point. Because we did not ascertain participant living status longitudinally, we cannot comment on whether the risk is dynamic over time. Furthermore, we did not collect information on other potential confounders, such as self-reported depressive symptoms, quality of life, psychological diagnoses, marital status, or type of dwelling. Finally, the lack of a significant effect of living alone among older participants after multivariate adjustment may be related to the low number of participants in this age group.
In conclusion, living alone was independently associated with an increased risk of mortality and CV death in an international cohort of stable middle-aged outpatients with or at risk of atherothrombosis. Younger individuals who live alone may have a less favorable course than all but the most elderly individuals following development of CV disease, and this observation warrants confirmation in further studies. Younger patients living alone after an atherothrombotic ischemic event may be an additional risk group on which to focus efforts to improve prognosis.
Correspondence: Deepak L. Bhatt, MD, MPH, Department of Medicine, VA Boston Healthcare System, 1400 VFW Pkwy, Boston, MA 02132 (firstname.lastname@example.org).
Accepted for Publication: April 9, 2012.
Published Online: June 18, 2012. doi:10.1001/archinternmed.2012.2782
Author Contributions: Drs Udell and Bhatt had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Udell, Steg, Scirica, Ohman, and Bhatt. Acquisition of data: Udell, Steg, and Eagle. Analysis and interpretation of data: Udell, Steg, Scirica, Smith, Eagle, Cho, and Bhatt. Drafting of the manuscript: Udell. Critical revision of the manuscript for important intellectual content: Udell, Steg, Scirica, Smith, Ohman, Eagle, Eagle, Cho, and Bhatt. Statistical analysis: Udell and Cho. Obtained funding: Udell and Steg. Administrative, technical, and material support: Ohman, Eagle, and Bhatt. Study supervision: Steg, Scirica, Eagle, and Bhatt.
Financial Disclosure: Dr Steg has been a consultant or on advisory boards for Ablynx, Astellas, AstraZeneca, Bayer, Boehringer-Ingelheim, BMS, Daiichi-Sankyo-Lilly, Eisai, GlaxoSmithKline, Medtronic, MSD, Pfizer, Roche, sanofi-aventis, Servier, and The Medicines Co, and is a stockholder and cofounder of Aterovax. Dr Scirica has received grants for clinical research via the TIMI Study Group and Brigham and Women's Hospital from AstraZeneca Pharmaceuticals LP, Daiichi-Sankyo Inc, Merck & Co Inc, Johnson & Johnson Pharmaceutical Research and Development LLC, Bayer HealthCare Pharmaceuticals, Bristol-Myers-Squibb Co, and Gilead Science, and has served as a consultant for Gilead Sciences Inc, Arena Pharmaceuticals, and Lexicon. Dr Ohman's university salary is partially supported by Daiichi-Sankyo and Maquet, and he has been a consultant for AstraZeneca, Boehringer-Ingelheim, Bristol Myers Squibb, Gilead Sciences, Liposcience, Merck, Pozen Inc, Roche, sanofi-aventis, The Medicines Co, and WebMD. Dr Goto has received research grants from sanofi-aventis, Eisai, and Boehringer-Ingelheim, and has been a consultant or on advisory boards for Eisai, sanofi-aventis, and Otsuka. Dr Bhatt has been on the advisory board for Medscape Cardiology; on the board of directors for Boston VA Research Institute and the Society of Chest Pain Centers; the chair for the American Heart Association Get With The Guidelines Science Subcommittee; received honoraria from the American College of Cardiology (editor, Clinical Trials, Cardiosource), Duke Clinical Research Institute (clinical trial steering committees) and Slack Publications (chief medical editor, Cardiology Today Intervention), WebMD (CME steering committees); has received research grants from Amarin, AstraZeneca, Bristol-Myers Squibb, Eisai, Ethicon, Medtronic, sanofi aventis, and The Medicines Company; and has done unfunded research for PLx Pharma and Takeda.
Funding/Support: The REACH Registry is sponsored by sanofi-aventis, Bristol-Myers Squibb, and the Waksman Foundation (Tokyo, Japan). The REACH Registry is endorsed by the World Heart Federation. Dr Udell is a recipient of a Postdoctoral Research Fellowship from the Canadian Institutes for Health Research and Canadian Foundation for Women's Health.
Role of the Sponsors: The design and conduct of the REACH study were done by the academic executive committee in collaboration with the sponsors; the collection and management of the data were done by the sponsors under the direction of the academic executive committee.
Disclaimer: The funding sources had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.
Previous Presentation: This study was presented as an abstract at the 2011 AHA Scientific Sessions; November 16, 2011; Orlando, Florida.
Online Only Material: An audio author interview is available here.