Association of psychosocial factors with all‐cause hospitalizations in patients with atrial fibrillation

Abstract Background A high burden of cardiovascular comorbidities puts patients with atrial fibrillation (AF) at high risk for hospitalizations, but the role of other factors is less clear. Hypothesis To determine the relationship between psychosocial factors and the risk of unplanned hospitalizations in AF patients. Methods Prospective observational cohort study of 2378 patients aged 65 or older with previously diagnosed AF across 14 centers in Switzerland. Marital status and education level were defined as social factors, depression and health perception were psychological components. The pre‐defined outcome was unplanned all‐cause hospitalization. Results During a median follow‐up of 2.0 years, a total of 1713 hospitalizations occurred in 37% of patients. Compared to patients who were married, adjusted rate ratios (aRR) for all‐cause hospitalizations were 1.28 (95% confidence interval [CI], 0.97‐1.69) for singles, 1.31 (95%CI, 1.06‐1.62) for divorced patients, and 1.02 (95%CI, 0.82‐1.25) for widowed patients. The aRRs for all‐cause hospitalizations across increasing quartiles of health perception were 1.0 (highest health perception), 1.15 (95%CI, 0.84‐1.59), 1.25 (95%CI, 1.03‐1.53), and 1.66 (95%CI, 1.34‐2.07). No different hospitalization rates were observed in patients with a secondary or primary or less education as compared to patients with a college degree (aRR, 1.06; 95%CI, 0.91‐1.23 and 1.05; 95%CI, 0.83‐1.33, respectively). Presence of depression was not associated with higher hospitalization rates (aRR, 0.94; 95%CI, 0.68‐1.29). Conclusions The findings suggest that psychosocial factors, including marital status and health perception, are strongly associated with the occurrence of hospitalizations in AF patients. Targeted psychosocial support interventions may help to avoid unnecessary hospitalizations. Trial registration ClinicalTrials.gov Identifier NCT02105844.

Conclusions: The findings suggest that psychosocial factors, including marital status and health perception, are strongly associated with the occurrence of hospitalizations in AF patients. Targeted psychosocial support interventions may help to avoid unnecessary hospitalizations.

| INTRODUCTION
Atrial fibrillation (AF) is expected to affect nearly 18 million Europeans in the future. 1 Patients with AF have multiple comorbidities and a high risk of complications, [2][3][4] which puts them at increased risk of being admitted to the hospital. 5 Although many hospitalizations are likely triggered by medical conditions, nonmedical factors may also be crucial.
It is well-established that social and psychological conditions (eg, marital status, education, mental health) play an important role in determining an individual's health. 6 These psychosocial factors have been associated with the risk of cardiovascular adverse events, 7,8 and evidence suggests that the effects are comparable in strength to those associated with physical activity, smoking, or alcohol use. 9,10 Prior studies addressed the relationships of psychosocial risk factors with incident AF and heart failure hospitalizations. [11][12][13] Among AF patients, those with a low social status, low education, or low household income had a higher risk of death as compared to individuals without such psychosocial constraints. 14 However, only little is known whether psychosocial factors affect the risk of hospitalizations in AF patients. For instance, patients with low social support may be less able to cope with serious health conditions and life crises, which may increase their tendency to seek medical advice and hospital care. Given that hospitalizations are strong drivers of healthcare expenditures, more evidence on this topic may help to establish new preventive strategies. We therefore aimed to investigate the prevalence of psychosocial factors and their associations with all-cause hospitalizations in a large cohort of wellcharacterized patients with AF.

| Study Population
The Swiss Atrial Fibrillation Cohort (Swiss-AF) is a large prospective cohort study of patients who had previously diagnosed AF enrolled across 14 centers in Switzerland. Details of the study design and first results have been published previously. 15,16 Patients were enrolled if they had documented AF and were aged 65 years or older. Exclusion criteria were short, reversible AF episodes (ie, AF occurring after cardiac surgery) or inability to give informed consent. The study protocol was approved by the local ethics committees, and written informed consent was obtained from all participants.

| Assessments
Demographic and clinical information were collected using standardized case report forms and validated questionnaires. Yearly follow-up visits were performed by local study personnel to collect patient characteristics, clinical measures and outcome events. Marital status and education level were social factors captured by the case report forms; depression and health perception were available psychological components. Participants were asked if they were married, single, divorced or widowed. Education level was evaluated using the sum of completed years at school, high school or college, and defined as primary or less (≤6 years), secondary (high school or similar: 6 to ≤12 years) and college or university (college or university degree: >12 years of education). Depression and depressive symptoms were measured using the Geriatric Depression Scale (GDS), 17 with a total point score ranging from 0 to 15, and a total score of >5 points was used to indicate depression. 18 Health perception was self-assessed by patients indicating their current state of health using a visual analogue scale (VAS) ranging from 0 (worst) to 100 (best). The VAS used in this study was based on the EuroQol VAS and has been validated for AF patients. 19,20 For the purpose of the present analyses, we divided patients into quartiles of total VAS; the first quartile was defined as the reference (highest health perception).

| Outcome
The outcome of this study was all-cause hospitalization, defined as any unplanned admission leading to at least one overnight stay.
Elective hospitalizations or emergency department evaluations were not counted. The occurrence of events was assessed at yearly followup examinations through on-site visit, phone call, or information gathered from the family physician.

| Statistical analysis
Baseline characteristics are presented as means ± standard deviations (±SD) for continuous variables and as counts (percentages) for categorical variables. To account for the repeated occurrence of hospitalizations within patients, we used the total number of all-cause hospitalizations as the primary outcome and applied negative binomial regression models to calculate rate ratios and 95% confidence intervals (CI). Models were adjusted for a predefined set of cardiovascular and noncardiovascular variables known to be associated with hospitalizations. 21 These variables consisted of age, sex, body mass index (BMI), history of hypertension, diabetes, coronary heart disease, prior stroke or transient ischemic attack (TIA), heart failure, peripheral vascular disease, renal failure, cancer, and previous falls. We then constructed a combined multivariable model including all psychosocial factors and covariates to determine the strongest predictors for allcause hospitalizations.
In a next step, we conducted time-to-event analyses to find out how psychosocial factors influence the risk of first all-cause hospitalization. We used Kaplan-Meier methods to estimate the cumulative incidence of first all-cause hospitalization across psychosocial factors and curves were compared by the log-rank test. Incidence rates were calculated per 100 patient-years of follow-up. We constructed multivariable Cox proportional hazards models to test the association of psychosocial factors with the risk of first all-cause hospitalization, adjusted for the same variable set as described above, and calculated the hazard ratios with corresponding 95% CIs. As indicated above, we also built a combined multivariable model including all psychosocial factors and covariates in a single model.  Baseline characteristics are shown in Table 1   Model was combined and adjusted for age, sex, body mass index, hypertension, diabetes, coronary heart disease, prior stroke/TIA, heart failure, peripheral vascular disease, renal failure, cancer, and previous falls.

| DISCUSSION
The present study investigated relationships between psychosocial factors and the risk of unplanned hospitalizations in patients with AF. Several important findings emerged. First, the rate of first and recurrent hospitalizations was high. Second, being divorced or having a low health perception was associated with a higher risk for unplanned hospitalizations. Third, depression and low education level were not associated with first or recurrent all-cause hospitalizations.
Our study showed that patients who were single or divorced revealed a higher risk of hospitalizations relative to those who were married. Consistently, previous studies from nonAF populations showed higher hospitalization rates for unmarried compared to married individuals. 22,23 These findings are in line with the notion that compared to those living alone (single, divorced), patients who have close relationships to others can rely on better social support, 24 while those with lacking support show increased needs for hospital care. 25 Our results further indicated that the risk of hospitalizations was closely associated with the patients' subjective perception of health.
Specifically, patients who felt in good health conditions were less likely to be admitted to the hospital. Evidence from studies of nonAF populations showed that patients who reported poor or fair health conditions exhibited an up to five times higher risk of hospitalization or death as compared to those reporting excellent or good health. 26,27 Previous studies suggested that self-efficacy is a key predictor of heart failure hospitalization and all-cause death. 28 One may assume that social support and help of close others strengthen self-efficacy beliefs, acting as a buffer of distress due to medical illness, which prevents patients with high social support from striving for hospital care.
This view also corresponds to the high rates of hospitalization in unmarried patients observed in the present study. Moreover, low health perception has often been reported in AF populations. 29 The clinical implication of our findings is that a better awareness of the patients' psychosocial conditions may help clinicians to intervene more sensitively and to be more responsive in offering specific support. Such interventions may include to improve the patient's social relations and to strengthen their self-efficacy in face of illness, which may imply psychosocial counseling, self-helping group assign-