Author + information
- Received June 4, 2015
- Revision received December 7, 2015
- Accepted December 9, 2015
- Published online March 1, 2016.
- aCenter of Research on Psychology in Somatic Diseases (CoRPS), Tilburg University, Tilburg, the Netherlands
- bDepartment of Cardiology, Elisabeth-TweeSteden Hospital, Tilburg/Waalwijk, the Netherlands
- ↵∗Reprint requests and correspondence:
Dr. Nina Kupper, Department of Medical and Clinical Psychology, Tilburg University, Center of Research on Psychology in Somatic Diseases (CoRPS), Warandelaan 2, 5037 AB / P.O. Box 90153 5000 LE, Tilburg, the Netherlands.
Objectives This study examined the association of self-care with all-cause mortality in a cohort of patients with chronic heart failure (HF).
Background Although self-care is crucial to maintain health in patients with chronic HF, studies examining an association with clinical outcomes are scarce.
Methods Consecutive patients with chronic HF (n = 559, mean age 66.3 ± 9.5 years, 78% men) completed the 9-item European Heart Failure Self-care Behaviour scale. Our endpoint was all-cause mortality. Associations between self-care and all-cause mortality were assessed with Kaplan-Meier analyses and multivariable Cox regression accounting for standard sociodemographic and clinical covariates, psychological distress, and self-rated health.
Results After a median follow-up of 5.5 ± 2.4 years (range 16 weeks to 9.9 years), 221 deaths (40%) from any cause were recorded. There was no evidence of a mortality benefit in patients high over those low in global self-care (p = 0.71). In post hoc analyses, low self-reported sodium intake was associated with increased mortality (adjusted hazard ratio: 1.47; 95% confidence interval: 1.10 to 1.96; p = 0.01). Other significant predictors of mortality were: male sex, lack of a partner, New York Heart Association functional class III to IV, and increasing comorbid conditions.
Conclusions Global self-care was not associated with long-term mortality whereas low self-reported sodium intake independently predicted increased all-cause mortality beyond parameters of disease severity. Replication of findings is needed as well as studies examining the correspondence of subjectively and objectively measured sodium intake and its effects on long-term prognosis in patients with chronic HF.
Chronic heart failure (HF) is a leading cause of (re-)hospitalization and death (1), and a significant clinical and economic burden for health care systems of developed countries (2). Besides established risk factors, poor HF outcomes have been attributed to poor HF self-care (3). Self-care refers to the complex regimen patients with chronic HF are instructed to undertake to maintain their health, for example, sodium and fluid restriction, weight monitoring, and medication adherence. Although it is assumed that self-care is associated with improved outcomes, the available evidence is less clear.
Preliminary findings suggest a potentially beneficial relationship between self-care and event-free survival that is predominantly due to fewer hospitalizations (4,5). Methodological limitations leave uncertainties about a relationship with mortality rates. Previous studies comprised mostly small samples and were limited in follow-up period (<2 years). Studies that focused on distinct self-care behaviors showed that medication nonadherence was associated with increased risk for adverse cardiac events in patients with HF (6), whereas low sodium intake was associated with increased event-free survival (i.e., hospitalization or death) (7). Until now, HF-targeted disease management programs that emphasize self-care have repeatedly been associated with decreased hospital readmission, but evidence on mortality rates is lacking (8).
Given the dearth of evidence in terms of long-term mortality, this study examines the association of self-reported self-care with all-cause mortality in a cohort of patients with chronic HF. This study hypothesized that self-care was associated with a greater benefit in long-term prognosis. Post hoc analyses were performed to examine the association of each individual self-care behavior (e.g., low sodium intake) with mortality.
Participants and procedure
Eligibility requirements and inclusion procedures have been described previously (9,10). Between 2003 and 2008, patients with chronic HF were consecutively recruited from three cardiology outpatient clinics from hospitals in the Netherlands. Inclusion criteria comprised a diagnosis of chronic systolic HF, left ventricular ejection fraction (LVEF) ≤40%, age ≤80 years, stable on oral HF medication for ≥1 month, and absence of a myocardial infarction (MI) or hospital admission in the month before inclusion. Patients were treated optimally according to prevailing HF guidelines (11). Patients were excluded in the case of other life-threatening comorbidities with a life expectancy <1 year, psychiatric comorbidity except for mood disorders, severe cognitive impairment (e.g., documented dementia or Alzheimer’s, extracted from medical records), or insufficient Dutch linguistic competence.
In total, 709 eligible patients were approached for participation by their treating cardiologist or nurse during their outpatient visit to the cardiology department. If willing to participate, patients were called within 2 weeks by an independent investigator to schedule a study appointment in which additional verbal and written information about the study was provided. Patients signed informed consent before participation. To guarantee anonymity, study numbers were assigned to each participant that were only available for members of the research team. Patients completed a questionnaire at home to assess socio-demographic, psychological variables, and HF self-care that was returned in a stamped and pre-addressed envelope. We checked for missing items accordingly and patients were contacted when a questionnaire was not returned within 2 weeks or in case of missing items. Our final sample consisted of 559 patients (response rate = 79%).
This is a secondary analysis of data originated from two observational prospective studies (9,10) for which ethics approval was obtained from the medical ethics committees of all hospitals. The investigation conforms the principles outlined in the Declaration of Helsinki (2013).
HF self-care was assessed using the 9-item version of the European Heart Failure Self-care Behaviour (EHFScB-9) scale (Figure 1) (12,13). Items were rated on a 5-point Likert scale ranging from 1 (I completely agree) to 5 (I do not agree at all). Scores ranged from 9 to 45 with higher scores reflecting worse self-care. The total scale consists of a 4-item consulting behavior subscale (10,12) that assesses whether patients contact their physician when HF-specific symptoms (e.g., shortness of breath, weight, fatigue) increase. Psychometric properties of the EHFScB-9 have been shown reliable for the total scale and consultation behavior subscale previously (14). Cronbach’s α was 0.80 for the total scale, indicating good internal consistency in this dataset.
Median split was used to distinguish between patients high versus low in global self-care because no validated cut-off score is available for the EHFScB-9. Scores of ≥21 and ≥9 were used to define low (i.e., worse) global self-care and consulting behavior, respectively. Low performance of each separate self-care behavior was categorized with a score of ≥3. A sensitivity analysis was performed with a standardized cut-off score of ≥70 indicating better self-care, which corresponds with the recommended cut-off point for the other well-known self-care measures (i.e., the Self-Care Heart Failure Index scale) (15).
We used purpose-designed items in the questionnaires to assess socio-demographic variables including educational level, current smoking status, partner status, and employment status. From patients’ medical records, information was obtained on age, sex, disease characteristics (i.e., etiology, LVEF, New York Heart Association (NYHA) functional class I to II vs. III to IV), cardiac history (i.e., previous MI, percutaneous coronary intervention, or coronary artery bypass graft surgery), serum sodium, and pharmaceutical treatment (e.g., beta-blockers, angiotensin-converting enzyme inhibitors, diuretic agents, psychotropic medication). Body mass index was calculated as weight (kilograms) divided by height (meters) squared.
Comorbidity at baseline was identified from searching patients’ medical records. An abbreviated version of the Charlson comorbidity index (CCI) (16) was used which included MI, cerebrovascular disease, chronic obstructive pulmonary disease, diabetes mellitus, peripheral arterial disease, renal dysfunction (i.e., estimated glomerular filtration rate <60 ml/min/1.73 m2 or a formal diagnosis of renal failure in the medical record), and cancer. The index was calculated by assigning a weight of 2 to renal dysfunction and cancer, and a weight of 1 to the other comorbid conditions. The final comorbidity score was the arithmetic sum of the value assigned to each identified condition. To minimize the number of predictors in the multivariable model, we accounted for increasing age by adding 1 point to the score for each decade of life >50 years at baseline.
Psychological distress was assessed with the short 4-item Symptoms of Anxiety-Depression index (SAD4) that has been validated in a post-MI population against generally accepted methods to assess psychological distress (17). The SAD4 assesses anxiety symptoms (i.e., tension and restlessness) and depressive symptoms (i.e., feeling blue and hopelessness). Items are answered on a 5-point Likert scale ranging from 0 (not at all) to 4 (very much). Cronbach’s α was .86 in a previous study (17) and .90 in the current sample indicating excellent internal consistency. Elevated psychological distress was defined according to a previously defined cut-off score of ≥3 (17).
Self-rated health was assessed with the general health item from the RAND 36-item Short-Form Health Survey (18): “In general, would you say your health is […]”. The item was scored on a 5-point Likert scale ranging from 1 (excellent) to 5 (poor). A score of ≥4 was indicative of poor self-rated health.
Our endpoint was death from any cause (i.e. all-cause mortality). Information on mortality including date of death was collected in the first week of March 2013 by checking patients’ medical records and/or by contacting the general practitioner. The median time of follow-up was 5.5 ± 2.4 years. Death was recorded in 221 patients (40%). We classified cause of death into cardiac, HF-related, or multisystem failure (N = 126) versus non−HF-related (N = 95) cause of death. The non−HF-related group, however, substantially consisted of patients with an unknown cause of death. Accordingly, it was chosen to use all-cause mortality as our endpoint.
Patients were classified into high or low global self-care performance groups and compared for baseline demographic and clinical characteristics using independent samples’ Student t tests for normally distributed continuous variables and a chi-squared test or Fisher exact tests for nominal variables. Person-years of follow-up were calculated from date of inclusion (i.e., baseline) until death, last follow-up visit, or March 1, 2013. To assess the impact of self-care at baseline on all-cause mortality, Kaplan-Meier survival analysis was performed and survival in patients performing high versus low self-care was compared using the log-rank (Mantel-Cox) test. The proportional hazards assumption was violated for self-care after inspection of the Log-minus-Log plots indicating that the effect of self-care on survival changed over time. When splitting the sample on HF-related (log-rank chi-square = 0.009; p = 0.925) versus non−HF-related (log-rank chi-square = 0.539; p = 0.463) mortality, similar findings were found including a violation of the hazards assumption. In post hoc analyses, cumulative Kaplan-Meier plots were constructed for the association of each individual key element of self-care with mortality. If the hazards assumption was held, Cox proportional hazards models were used to evaluate the relationship with all-cause mortality. Unadjusted and adjusted models were performed for relevant covariates that were selected a priori (19) and entered using forced entry in a second block: sex, partner status, NYHA function class (III-IV), CCI, prescribed medication (i.e., beta-blockers and diuretic agents), and current smoking status. In a third block, we controlled for elevated symptoms of anxiety/depression as well as for self-rated health given its predictive value in terms of mortality (20) and significant correlations with self-care in the current study (r = 0.14, p = 0.001). A 2-tailed p value <0.05 was considered to indicate statistical significance for all analyses. Analyses were performed with SPSS 20.0 for Windows (IBM SPSS Statistics for Windows, Version 20.0, IBM Corp USA, Armonk, New York).
In Table 1, we present baseline characteristics for the total cohort and stratified for mortality status. Those deceased at follow-up were older, more often male, unemployed, and lower educated. With respect to clinical characteristics, those deceased had a lower LVEF, higher NYHA functional class, a higher comorbidity score, and lower levels of serum sodium. Deceased patients were less often prescribed with beta-blockers but more often with diuretic agents. Self-rated general health was poorer in patients who were deceased at follow-up. Elevated anxiety/depression was present in 33% of patients and 55% rated their general health as poor. In Table 2, baseline characteristics are compared for patients reporting high versus low self-care. The incidence of mortality in patients performing high (40%) versus low (39%) self-care did not differ significantly. The prevalence rate of poor self-rated health was significantly higher among those performing low self-care. High self-care performers were more likely to have a partner, lower body mass index scores, to be prescribed with angiotensin-converting enzyme inhibitors/angiotensin receptor blockers, beta-blockers, and diuretic agents, and not smoking compared with low self-care performers (Table 2).
Self-care and all-cause mortality
The proportional hazards assumption for self-care was violated given the crossing lines at multiple time points, indicating that the effect of global self-care on survival changed over time (Figure 2A). For all covariates, the hazards assumption was met. No significant difference in survival was found between those performing high self-care (mean survival 6.7 years, 95% confidence interval [CI]: 6.4 to 7.2) and those performing low self-care (mean survival 6.9 years, 95% CI: 6.5 to 7.3) (log-rank chi-square = 0.139, p = 0.710) (Figure 2A). Sensitivity analyses with a standardized score of ≥70 indicating better global self-care did not lead to different results (high self-care mean survival = 6.7 years versus 6.9 years in those with low self-care; log-rank chi-square = 1.46; p = 0.23).
No differences were found between those high versus low in self-reported daily weight monitoring, fluid restriction, consultation behavior, regular exercise, and medication adherence. With respect to low sodium intake though, the cumulative hazard survival curve differed significantly between patients highly limiting their sodium intake (mean survival time = 6.6 years; 95% CI: 6.3 to 6.9) versus those low (mean survival time = 7.2 years; 95% CI: 6.8 to 7.6) in limiting their sodium intake (log-rank chi-square = 4.95; p = 0.026) (Figure 2B). In unadjusted Cox regression analysis, low self-reported sodium intake (hazard ratio [HR]: 1.38; 95% CI: 1.04 to 1.84; p = 0.03) was significantly associated with an increased mortality risk. After controlling for demographic and clinical covariates, low sodium intake (HR: 1.47; 95% CI: 1.10 to 1.97; p = 0.01) remained associated with an increased risk for mortality. In the final block controlling for psychological distress and self-rated health, low sodium intake still predicted mortality (Table 3). Other multivariable predictors of all-cause mortality were male sex, NYHA function class III to IV, CCI, and the lack of a partner (Table 3).
This is one of the first studies investigating the association of self-reported self-care with long-term all-cause mortality in a fairly large cohort of patients with chronic HF. However, neither global self-care nor any of the individual self-care behaviors predicted all-cause mortality except for low sodium intake. Patients reporting high self-care differed from those low in self-care in baseline characteristics but this did not seem to explain the lack of an association with mortality. No differences were found with respect to HF disease severity or other comorbid conditions, but those high in self-care were more likely to be prescribed with HF medication. They were also more likely to have a healthier life(-style) in general, to have a partner, and to have a higher educational level compared with those low in self-care, reflecting that social support and education level contribute to effective self-care (21). Other predictors of all-cause mortality in the current study were male sex, NYHA function class, comorbid conditions, and lack of a partner, which is in accordance with the current published data (19).
Despite assumed benefits of self-care on prognosis, our finding is in line with the available published data; none of the prior studies have been successful in showing an association between self-care and mortality, only with fewer hospital readmissions (4,22). Systematic reviews documented that HF-targeted and self-care education interventions seemed effective in decreasing hospital readmissions, but not in mortality rates (8,23). Similar findings have been reported for the effects of exercise-based cardiac rehabilitation programs. Recent meta-analyses supported associations with fewer hospital readmissions, but not with mortality (e.g., Lewinter et al. ). Only one randomized controlled clinical trial has documented an all-cause mortality benefit of a nurse-directed comprehensive chronic HF management program that was administered in a cardiac rehabilitation setting. The sample was small (N = 105) with a follow-up of 12 months, stressing the need for larger clinical trials with longer follow-ups replicating these, yet promising results.
The absence of proof does not imply that self-care is not crucial in prolonging life expectancy. Although patients being poor in self-care are at increased risk for being hospitalized, medical treatment may ultimately prevent them from early deceasing. A recent study (25) showed that poor self-rated health predicted increased health care utilization. In our study, those low in self-care also tended to have poor self-rated health. Therefore, effective self-care remains important in the treatment of patients with chronic HF to decrease the clinical and economic burden of HF on the health care systems (23).
We found that low self-reported sodium intake was robustly associated with an increased mortality risk in post-hoc analyses, irrespective of important risk factors. A clear explanation why low sodium intake was adversely associated with prognosis cannot be provided, mainly because of the use of a single item to assess sodium restriction as well as the generally poor correspondence between self-report and objective measures of adherence (26–29). This is further complicated by (potentially circulating) mechanisms underlying the association of self-reported sodium intake and medical outcome including treatment effects including self-care education and knowledge, and disease progression (30,31). In advancing HF, both the physician and the patient may have an increased awareness of the importance of sodium restriction to prevent further decompensation. Future research is needed to replicate our findings and to examine the: 1) ecological validity of self-reported sodium intake; and 2) correspondence between self-reported low sodium intake with objective measures (e.g., urinary sodium) and its actual relationship with long-term mortality.
Several limitations should be considered when interpreting these results. Foremost, self-care was assessed with a self-report measure, which has limitations such as recall bias and over- and underestimation (27–29). The EHFScB-9 does not reckon with the disease stage of the individual patient nor is every patient recommended to each self-care behavior. It could be argued that the items lack specificity as they ask for general level of adherence that is greatly influenced by the subjective interpretation of the patient. In general, it is recommended to use multiple types of self-care measures to assess “real” self-care (30). However, the EHFScB-9 has been recommended as a reliable and valid self-care instrument (32). Other limitations were the lack of a validated cut-off score to compare high versus low self-care and the use of single items to assess the separate elements of self-care. Nevertheless, changing the cut-off score to 70 using a standardized self-care score did not lead to different results. We can only carefully hypothesize that (too) low sodium intake may be adversely associated with HF prognosis due to the psychometric limitations as well as potential multiple testing effects. Further, no conclusions can be drawn with respect to the effect of change in self-care over time. However, self-care does not tend to change greatly over time (33). Finally, no information was available on rehospitalization preventing us from running secondary analyses on cardiac events and HF-related hospitalization.
This study is among the first that examined self-reported self-care in relation to mortality in a relatively large sample of >500 patients with a significant median length of follow-up of >5 years. Although global HF self-care was not associated with all-cause mortality, low self-reported sodium intake predicted increased all-cause mortality beyond clinical and self-reported parameters of disease severity.
COMPETENCY IN MEDICAL KNOWLEDGE: Our findings indicate no association between self-reported self-care and long-term all-cause mortality. On the contrary, a negative association was found for low self-reported sodium intake and long-term prognosis. More awareness and research is therefore warranted on the clinical use and (psychometric) meaning of self-reported self-care in chronic HF in terms of prognosis, especially with respect to sodium intake.
TRANSLATIONAL OUTLOOK: Based on our findings, more studies are needed to examine the correspondence between subjectively and objectively measured self-care, such as sodium intake, and its actual relationship with long-term prognosis. Large scale clinical trials of patients with newly diagnosed HF that are being followed over time in which subjective and objective measures of self-care are assessed would provide better insight into the long-term effects of self-care on prognosis.
Part of this work was supported with a VICI grant (453-04-004) from the Dutch Organization for Scientific Research (NWO) awarded to Prof. Dr. Denollet. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- Charlson comorbidity index
- European Heart Failure Self-care Behaviour scale
- heart failure
- left ventricular ejection fraction
- New York Heart Association
- Received June 4, 2015.
- Revision received December 7, 2015.
- Accepted December 9, 2015.
- American College of Cardiology Foundation
- Roger V.L.,
- Go A.S.,
- Lloyd-Jones D.M.,
- et al.
- Pelle A.J.,
- Pedersen S.S.,
- Schiffer A.A.,
- Szabo B.,
- Widdershoven J.W.,
- Denollet J.
- Schiffer A.A.,
- Denollet J.,
- Widdershoven J.W.,
- Hendriks E.H.,
- Smith O.R.
- McMurray J.J.,
- Adamopoulos S.,
- Anker S.D.,
- et al.
- Vellone E.,
- Jaarsma T.,
- Stromberg A.,
- et al.
- Theuns D.A.,
- Schaer B.A.,
- Soliman O.I.,
- et al.
- Pocock S.J.,
- Ariti C.A.,
- McMurray J.J.,
- et al.
- Lewinter C.,
- Doherty P.,
- Gale C.P.,
- et al.
- Chamberlain A.M.,
- Manemann S.M.,
- Dunlay S.M.,
- et al.
- ↵Kessing D, Denollet J, Widdershoven J, Kupper N. Psychological determinants of heart failure self-care: systematic review and meta-analysis. Psychosom Med; in press.
- Lee C.S.,
- Mudd J.O.,
- Hiatt S.O.,
- Gelow J.M.,
- Chien C.,
- Riegel B.
- Kessing D.,
- Denollet J.,
- Widdershoven J.,
- Kupper N.