Article Text
PDF
Abstract
Objective To explore whether living alone and loneliness 1) are associated with poor patient-reported outcomes at hospital discharge and 2) predict cardiac events and mortality 1 year after hospital discharge in women and men with ischaemic heart disease, arrhythmia, heart failure or heart valve disease.
Methods A national cross-sectional survey including patients with known cardiac disease at hospital discharge combined with national register data at baseline and 1-year follow-up. Loneliness was evaluated using one self-reported question, and information on cohabitation was available from national registers. Patient-reported outcomes were Short Form-12, Hospital Anxiety and Depression Scale and HeartQoL. Clinical outcomes were 1-year cardiac events (myocardial infarction, stroke, cardiac arrest, ventricular tachycardia/fibrillation) and all-cause mortality from national registers.
Results A total of 13 443 patients (53%) with ischaemic heart disease, arrhythmia, heart failure or heart valve disease completed the survey. Of these, 70% were male, and mean age was 66.1 among women and 64.9 among men. Across cardiac diagnoses, loneliness was associated with significantly poorer patient-reported outcomes in men and women. Loneliness predicted all-cause mortality among women and men (HR 2.92 (95% CI 1.55 to 5.49) and HR 2.14 (95% CI 1.43 to 3.22), respectively). Living alone predicted cardiac events in men only (HR 1.39 (95% CI 1.05 to 1.85)).
Conclusions A strong association between loneliness and poor patient-reported outcomes and 1-year mortality was found in both men and women across cardiac diagnoses. The results suggest that loneliness should be a priority for public health initiatives, and should also be included in clinical risk assessment in cardiac patients.
- epidemiology
- coronary artery disease
- heart failure
- valvular heart disease
Statistics from Altmetric.com
Introduction
Previous research has pointed to the importance of social relationships to health as loneliness and lack of social support is associated with an increased risk of ischaemic morbidity and mortality.1 2 There are numerous theoretical definitions of loneliness, but most researchers within the field agree on three elements: 1) loneliness is a result of perceived deficiencies in a person’s social relationships, 2) it is a subjective experience and is not the same as social isolation and 3) it is unpleasant and distressing.3 A systematic review concluded that poor social relationships were associated with a 29% increase in risk of coronary heart disease and a 32% increase in the risk of stroke,1 and a recent study links loneliness to an increased risk of incident cardiovascular disease of 27%.4 Similarly, lack of social support appears to play a role in the progression of cardiovascular disease.5 6 A meta-analysis found that poor social support negatively affected cardiac and all-cause mortality in prognostic studies (relative risk (RR) range 1.59–1.71).5 Living alone is related to loneliness, but living alone does not equal feeling lonely. However, living alone has also been linked to increased risk of mortality and cardiac events, but evidence is not consistent.7 8 Recently, a study showed that men living alone had a significantly increased risk of both all-cause and cardiovascular mortality compared with men not living alone after 32 years of follow-up.9
The explanation for the link between loneliness and health includes different pathways. One pathway is behavioural as the existence of social relationships has a direct health effect because it promotes healthy behaviour such as exercise, healthy eating, not smoking and greater adherence to medical regimens.6 10 Feeling lonely on the other hand impairs the capacity to self-regulate, minimises the likelihood of performing physical activity and is a risk factor for obesity and excessive alcohol abuse.11 These differences become more evident with age, as the effects of poor health behaviour start to show.10
Other pathways are psychological and biological. The existence of social relationships is beneficial because it increases feelings of safety and trust and ‘buffers’ the potentially harmful influences of stress-induced cardiovascular reactivity.6 Furthermore, it is possible that social support helps reduce the exposure to stressful events.6 The pathways may impact each other and affect biological processes with an impact on surrogate biological markers.10
Research on the health effects of loneliness and living alone within cardiology has often focused on ischaemic heart disease (IHD). Furthermore, studies rarely include patient-reported outcomes (PROs), which are important independent predictors of health outcomes.12 Two previous studies including patients with implantable cardioverter defibrillators and congenital heart disease, respectively found low perceived social support and chronic loneliness to be associated with perceived health status and depression symptoms.13 14 Therefore, the objectives of this study were to explore:
Whether living alone and loneliness are associated with poor PROs at hospital discharge in women and men with IHD, arrhythmia, heart failure, or heart valve disease.
Whether living alone and loneliness predict cardiac events and mortality 1 year after hospital discharge in women and men with IHD, arrhythmia, heart failure or heart valve disease.
Methods
The methods are described in more detail in the prepublished protocol.15 In the following, a brief overview of the DenHeart study is presented.
Study design
The DenHeart study is a cross-sectional survey combined with data from national registers at baseline and 1-year follow-up. All patients hospitalised at a heart centre were invited to answer a self-reported questionnaire at hospital discharge to evaluate PROs across cardiac diagnostic groups. The present study includes only patients with IHD, arrhythmia, heart failure and heart valve disease.
Setting and participants
Over 1 year (15 April 2013 to 15 April 2014), all patients discharged or transferred from the five Danish heart centres were invited to participate in the study.
Eligibility criteria
All patients were consecutively included. Patients who were under 18 years of age, who did not have a Danish civil registration number or who did not understand Danish were excluded from the study. For ethical reasons, patients who were unconscious at the time of transfer from a heart centre were also excluded.
Recruitment
To avoid recall bias, patients were asked to complete and return the questionnaire before they left the hospital or complete it at home within 3 days of discharge and return it by mail in a prepaid envelope.
Variables
Baseline
Survey data were combined with data from the following Danish national registers at baseline: The Danish Civil Registration System (gender, age, marital status),16 The Danish National Patient Register (cardiac diagnosis at discharge, comorbidity)17 and Danish Education Registers (educational level).18
To combine patient-reported data with demographic and clinical variables from the national registers, all responders were matched to a hospital discharge in The Danish National Patient Register.
Responders were divided into diagnostic groups based on their primary ICD-10 action diagnosis obtained from The Danish National Patient Register. Four diagnostic groups were included in the analyses, and they were defined as follows: IHD: I20-I25, T82.3D, Z95.1, Z95.5; arrhythmia: I44-I49, Z95.0, R00.0, R00.1, R00.2, R00.8A, T75.0, T75.4, T82.1, T82.8; heart failure: I11.0, I42.0-I43.8, I50, I51.7, R57.0 and heart valve disease: I05.0-I06.0, I34.0-I37.2, I39.1, I39.2, I51.1A, Z95.2-Z95.4.
Information on comorbidity was obtained from The Danish National Patient Register and calculated 10 years back, not including the index discharge. The Tu comorbidity index score was calculated19 with information on primary and secondary diagnoses for all patients. The following diseases were included: congestive heart failure, cardiogenic shock, arrhythmia, pulmonary oedema, malignancy, diabetes, cerebrovascular disease, acute/chronic renal failure, chronic obstructive pulmonary disease. A Tu comorbidity score of zero equals no comorbidities, a score of one means one of the included comorbidities, etc. All diagnoses were weighted equally.
Exposure
The objective information on cohabitation was obtained from national registers. Living alone was defined as a man or woman not in an identifiable cohabitation. Cohabitation was defined by the following categories: married couple, other couple and household consisting of several people from more than one family (eg, nursing home or collective).
As a subjective measure of loneliness patients answered a question about feeling lonely from the Danish National Health Survey20:
“Does it ever happen that you are alone even though you would prefer to be with other people?” Loneliness was defined by the answers: “yes, often” and “yes, sometimes”. Other possible responses were “yes, but rarely” and “no”.
Patient-reported outcomes
The following questionnaire instruments were included in the DenHeart questionnaire:
The Short Form-12 (SF-12), a brief measure of health-related quality of life that generates both a physical component score (PCS) and a mental component score (MCS). Higher scores indicate a better health status.21 The Hospital Anxiety and Depression Scale (HADS), a 14-item questionnaire that assesses levels of anxiety and depression symptoms in medically ill patients. Scores of 8–10 suggest the presence of a mood disorder. Scores ≥11 indicate the probable presence of a mood disorder.22 HeartQoL, a disease-specific questionnaire that measures quality of life in cardiac patients and produces a global score and two subscales: a physical and an emotional scale ranging from 0 to 3 with higher scores indicating better quality of life status.23
Furthermore, patients answered questions about health behaviour (current and previous smoking behaviour, alcohol intake during a typical week and medicine compliance), height and weight.
Prognostic outcomes
From The Danish Civil Registration System information on all-cause mortality during the first year after the index admission was obtained. Information on cardiac events during the first year was obtained from The Danish National Patient Register. Cardiac events were defined as: myocardial infarction: I21; stroke: I60-I64, I67; cardiac arrest: I46; ventricular tachycardia/ventricular fibrillation: I49.
Study size
The study population consisted of all patients discharged from the five heart centres during the project period. To avoid selection bias, all patients were included consecutively. Hospitalised patients with lung disease and other non-cardiac diseases were excluded. The analyses in this paper are based on 13 446 patients with IHD (n=7169, 53%), arrhythmia (n=4316, 32%), heart failure (n=987, 7%) and heart valve disease (n=974, 7%).
Statistical methods
A total of 17 respondents were lost to follow-up in the registers as they did not have an address in Denmark. These subjects were excluded from the present analyses. Follow-up was continued until the first cardiac event or until death, emigration or end of follow-up.
Baseline differences in demographic, clinical and behavioural variables were tested using χ2 tests. For continuous variables t-tests were used.
To explore the association between loneliness and PROs, linear regression models were used for continuous outcomes, and logistic regression models were used for binary outcomes. All analyses were adjusted for living alone (when loneliness is the independent variable), loneliness (when living alone is the independent variable), age, educational level, cardiac diagnosis, comorbidity (Tu comorbidity index), body mass index, smoking behaviour, alcohol intake and medicine compliance.
Multivariate Cox proportional hazards regression models, with age as the time scale, were used to explore the predictive value of loneliness at baseline, and cardiac events and all-cause mortality after 1 year. Unadjusted and adjusted results are presented. The proportional hazards assumption was checked graphically using Kaplan-Meier curves and log-minus-log plots. The resulting curves were found to be parallel, indicating that the proportional hazards assumption was met. Results are presented as hazard ratios (HRs) with 95% confidence intervals (CIs).
Differences between men and women and diagnostic groups were tested by including interactions. Based on the results, analyses were conducted for women and men separately. Furthermore, interactions between living alone and loneliness were tested in all models.
All analyses were conducted using SAS V.9.4.
Patient and public involvement
In designing the study focus was on including PROs as an important contribution to health assessment and further development of care. Patients were involved in survey preparation. The questionnaire was pretested for feasibility by 12 (10 males, 2 females) patients in the age range 52–81 years (mean 65.9) from medical and surgical wards at three of the heart centres. The questionnaire was adjusted accordingly.
Ethics
The study complies with the Declaration of Helsinki. DenHeart is registered at ClinicalTrials.gov (NCT01926145) and approved by the institutional boards of the heart centres. Patients signed informed consent.
Results
Inclusion of patients is presented in the flow chart (figure 1). A total of 13 446 (53%) patients were included in the analyses. Demographic and clinical profile is presented in table 1 for women and men stratified by loneliness. Significant baseline differences were seen between those feeling lonely and those not feeling lonely.
Flow chart.
Demographic and clinical profile at baseline for women and men stratified by loneliness
Loneliness and patient-reported outcomes
For both women and men living alone was associated with a reduced risk of anxiety symptoms (odds ratio (OR)=0.71 (95% CI 0.59 to 0.85) and OR 0.80 (95% CI 0.69 to 0.93)), respectively (table 2). Women feeling lonely had a 6.15 (95% CI −7.20 to −5.09) point lower MCS score and a 1.81 (95% CI −2.71 to −0.83) point lower PCS score compared with women not feeling lonely. Furthermore, women feeling lonely had approximately 2.7 times higher odds of reporting symptoms of anxiety and depression and reported a significantly lower quality of life score compared with women not feeling lonely. Likewise, men feeling lonely reported significantly poorer MCS, PCS and quality of life scores and had almost three times higher odds of reporting symptoms of anxiety and depression compared with men not feeling lonely (table 2). Mean values and proportions for PROs and prognostic outcomes are presented in online supplementary table 1.
Supplemental material
Association between living alone, loneliness and patient-reported outcomes at hospital discharge
No statistically significant interaction was found between living alone and loneliness.
Loneliness and cardiac events and mortality
After adjustment for potential confounders, women feeling lonely had an almost tripled risk of all-cause mortality compared with women who did not feel lonely (HR=2.92 (95% CI 1.55 to 5.49)) (table 3).
Living alone and loneliness as a predictor of cardiac events and mortality after 1 year
After adjustment for potential confounders, men living alone had an increased risk of cardiac events (HR 1.39 (95% CI 1.05 to 1.85)) and men feeling lonely had a doubled risk of all-cause mortality (HR 2.14 (95% CI 1.43 to 3.22)) (table 3).
No statistically significant interaction was found between living alone and loneliness.
Discussion
The results of this study show that women and men who feel lonely report significantly poorer mental and physical health, quality of life and are more likely to report anxiety and depression symptoms. In both women and men living alone was associated with reduced risk of anxiety symptoms. After adjustment for potential confounders loneliness was associated with a significantly increased risk of all-cause mortality among women and men. Among men only, living alone predicted increased risk of cardiac events.
Interpretation
The present results show that loneliness is significantly associated with PROs. This highlights the possible impairing effects on health associated with loneliness as experienced and reported by the patient. The findings of the predictive value of loneliness on mortality in this study are in line with previous research in cardiovascular disease with loneliness significantly influencing the prognosis.1 2
Looking at unadjusted estimates for cardiac events and mortality there were significant risk differences between those who felt lonely compared with those who did not, and those who lived alone compared with those who lived with others, both among men and women. After adjustment some significant differences remain. Behavioural risk factors and comorbidity therefore does not seem to explain the association between loneliness and morbidity and mortality. This indicates that the behavioural pathway cannot, in this study, fully explain the association. Conventional risk factors were found in another study to explain most of the risk of acute myocardial infarction and stroke.2 Furthermore, a large cohort study found the excess mortality among lonely people to be attributed to both unhealthy lifestyle, and also socioeconomic conditions and lower mental well-being.24
In the present study, the subjective measure of loneliness was a much stronger predictor of both PROs and mortality compared with the objective measure of living alone. Differing definitions and measures of loneliness and cohabitation can make it difficult to compare results across studies directly.6 25 However, the fact that the subjective feeling of loneliness is associated with poor health outcomes is well established,10 while the findings of health effects associated with living alone point in different directions. Previous studies show an increased mortality risk among patients living alone,9 but in one study this is only the case among younger patients.8 Yet another study does not find any significant associations with risk of mortality or readmission.7
This is also evident in the finding that both men and women living alone had a decreased risk of anxiety symptoms compared with those living with someone. There can be both positive and negative aspects of close relationships. Stressful social relations are associated with higher risk of incident IHD and can increase mortality.26 Similarly, in an elderly population there is greater risk of living with a sick spouse or partner which might cause worry and anxiety. These are factors that patients living alone would not be facing to the same extent as those living with someone and might help explain who they report a better mental health.
Loneliness was a strong predictor of poor health outcomes in both men and women but living alone was only a predictor of cardiac events in men. Previous studies found that women have larger networks than men and being separated, divorced or single has a greater impact on men’s social networks compared with women’s.27
Because of the design of this study it is not possible to make conclusions about causal mechanisms. There is a possibility of reverse causality, as it is unknown whether loneliness or disease came first.28 Furthermore, the feeling of loneliness can change within the first year after hospital discharge. However, the findings are in line with previous research suggesting that loneliness is associated with changes in cardiovascular, neuroendocrine and immune function as well as unhealthy lifestyle choices which can impact negative health outcomes.
Generalisability
This is a national Danish study and international differences may exist in treatment as well as culture and social behaviour. The response rate was 53%, which is not unexpected in a population of severely ill patients. However, this may raise concerns regarding representativeness. The proportion of patients in each diagnostic group is comparable among responders and non-responders and they have similar sociodemographic and clinical characteristics. However, a higher mortality rate was detected among non-responders.29
Implications
There are indications that the burden of loneliness and social isolation is growing.30 Furthermore, increasing evidence points to their influence on poor health outcomes being equivalent to the risk associated with severe obesity.11 30 Public health initiatives should therefore aim at reducing loneliness.
Strengths and limitations
This study includes a large sample of patients with IHD, arrhythmia, heart failure and heart valve disease and found no significant differences between diagnostic groups. This builds on existing research primarily focused on IHD and indicates that the association between loneliness and health outcomes may apply to other diagnostic groups as well. We included both an objective measure (living alone) and a subjective measure (loneliness) in the analyses. Furthermore, in addition to objective outcomes such as mortality, we included PRO measures to further unfold the effects of loneliness on health.
The presence of non-response bias is a possibility in survey-based research. The response rate alone is often a poor indicator of non-response bias. Survey estimates are only affected if respondents and non-respondents differ on particular indicators of interest. A larger proportion of patients were living alone among non-responders. Furthermore, non-responders seemed to be a little older, not married and have a lower educational level compared with responders (online supplementary table 2). Thus, we cannot rule out non-response bias in the present survey. Self-reported outcomes are by nature subjective and therefore, sources of bias may exist. Recall bias is the most serious problem in epidemiological surveys. Social desirability bias can be an issue in self-reporting of lifestyle factors or a sensitive subject such as loneliness. However, there is little reason to suspect that such possible bias should differ systematically according to, for example, loneliness. Finally, we do not have information about physical activity or clinical cardiac risk factors such as serum cholesterol, blood pressure, life expectancy or cardiac medication. If such variables can explain some of the association between loneliness and health outcomes, we cannot rule out that inclusion in the present analyses could have affected the findings.
Supplemental material
Key messages
What is already known on this subject?
Loneliness and poor social support have been linked to both the development of ischaemic heart disease and cardiovascular mortality.
Loneliness affects health outcomes through behavioural, psychological and biological pathways.
What might this study add?
At hospital discharge, loneliness is associated with poor patient-reported mental and physical health and is a strong predictor of mortality after 1 year in both men and women across cardiac diagnoses.
A subjective measure of loneliness is a stronger predictor of poor outcomes than the objective measure of living alone.
How might this impact on clinical practice?
Loneliness should be included in clinical risk assessment of cardiac patients.
Acknowledgments
The authors would like to thank the patients who took the time to participate in the survey, the 800 cardiac nurses involved in data collection and the heart centres for prioritising this study in a busy clinic. The authors would also like to thank the DenHeart research expert committee.
References
Footnotes
Contributors SKB conceived the overall idea for the DenHeart study and all authors designed the study. AVC performed the statistical analyses and wrote the first draft of the manuscript. All revised the manuscript critically. All have given their final approval of the version to be published.
Funding This work was supported by Helsefonden; the Danish heart centres; the Novo Nordisk Foundation, Familien Hede Nielsens Fond and The Danish Heart Association.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval According to Danish legislation, surveys should only be approved by the Danish Data Protection Agency (2007-58-0015/30-0937). Use of register data were permitted by The Danish National Board of Health.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement No data are available.