Association of depression and anxiety with adherence in primary care patients with heart failure—cross-sectional results of the observational RECODE-HF cohort study

Abstract

Background

Psychological distress has a negative impact on the prognosis and quality of life for patients with heart failure. We investigated the association between psychological distress and the patients’ adherence to medical treatment (medication adherence) and self-care advice (lifestyle adherence) in heart failure. We further examined whether there are different factors associated with low medication compared to low lifestyle adherence.

Method

This secondary analysis of the RECODE-HF cohort study analyzed baseline data of 3099 primary care heart failure patients aged 74 ± 10 years, 44.5 % female. Using multivariable regression, factors relating to medication and lifestyle adherence were investigated in order to estimate the extent to which these factors confound the association between psychological distress and adherence.

Results

Psychological distress was significantly associated with poorer medication adherence but not with lifestyle adherence after controlling for confounders. We identified different factors associated with medication compared to lifestyle adherence. A higher body mass index, a less developed social network, living alone, fewer chronic co-morbidities and unawareness of the heart failure diagnosis were only related to lower lifestyle adherence. Higher education was associated with poorer medication adherence. Male sex, younger age, lower self-efficacy and less familiar relation with the general practitioner were common factors associated with both lower medication and lifestyle adherence.

Conclusion

Promising factors for increasing medication adherence (reduction of psychological distress) and lifestyle adherence (explaining the patient his/her heart failure diagnosis more than once and increase in the patients’ self-efficacy), which were found in this cross-sectional study, must be further investigated in longitudinal studies.

Anxiety, depression, health services research, heart failure, patient adherence, primary care

Key Messages

• Psychological distress is associated with worse medication adherence.

• Factors associated with worse medication and lifestyle adherence differ.

• To raise adherence, the strategy needs to fit the type of adherence addressed.

• Some modifiable factors could be easily addressed in clinical practice.

Background

Heart failure (HF) affects about 1–4% of the adult population in developed countries and remains a fatal clinical syndrome with 5-year mortality of 45–59% (1), even though temporal trends showed a better survival over the past decades (1). This likely reflects improved detection and positive effects of pharmacological treatment and clinical management of the disease. Also, lifestyle changes are indicated in most patients with HF. Although there is little evidence that isolated lifestyle advice positively affects quality of life or prognosis, the American College of Cardiology Foundation/American Heart Association (ACCF/AHA) Guideline for the Management of Heart Failure (2) considers education about self-care recommendations to be ‘necessary albeit not always sufficient, to significantly improve outcomes’. Clinical guidelines, therefore, address self-care strategies and medical treatment as aspects of interdisciplinary HF management, for example (3), which should be tailored to the individual patient (4).

A major European study (5) showed limited adherence to self-care recommendations and the need to identify risk groups with poor HF lifestyle adherence has been concluded by the Heart Failure Association of the European Society of Cardiology (6).

HF patients with depression and anxiety have a worse prognosis and quality of life than HF patients without these co-morbid conditions, for example (7–9). A worse prognosis in cardiac disease is even evident for sub-threshold depressive symptoms (10). Despite this high burden, a recent European study found a treatment gap in mental health service use of 79% for late-life depression in older adults (11). Symptoms of anxiety and depression (psychosocial distress) may be directly related to adherence in HF patients. They may be less able to adhere to either recommended lifestyle behaviour or prescribed HF medication due to psychological barriers. Depression treatment could improve patient adherence (12). Even though depression and anxiety have been reported to reduce patients’ ability to adhere to medical treatment (13), only one study known to the authors investigated the association between psychological distress and HF patients’ ability to adhere to medical treatment (medication adherence) and self-care behaviour (lifestyle adherence) (14) within a small cohort of 115 patients. The results regarding lifestyle adherence were inconclusive as each aspect was investigated separately; a general conclusion could not be drawn.

Objectives

Our aim was to investigate the association between psychological distress and adherence to medication and lifestyle recommendations within a cohort of 3099 HF patients using a comprehensive concept of lifestyle adherence. We hypothesized that psychological distress is associated with lower medication and lifestyle adherence. In further analysis, we investigated whether different factors are associated with low medication adherence compared to low lifestyle adherence.

Methods

This is a secondary analysis of the baseline data of the prospective RECODE-HF cohort study conducted in Germany (15). We invited 4220 GPs between February 2012 and June 2014 to participate in the study; 293 were willing to participate (6.6%). Practice staff identified 13 830 eligible HF patients from electronic patient records, who received a written invitation to participate in the study by their GPs. Five thousand three hundred eighty-five (38.9%) patients consented to participate. A structured telephone interview was conducted with the GPs of 3821 of the 5385 consenting HF patients and 3387 patients met the inclusion criteria. A flow chart with detailed information was provided earlier (16). The data of 3099 HF patients with valid information on psychological distress and medication adherence, as well as lifestyle adherence, could be included in the analyses.

Data were collected between August 2012 and November 2014. Eligibility criteria were: age ≥18 years, last consultation within the last 6 months, diagnosis of HF within the last 5 years. Patients who have died since the last consultation, suffered from dementia or were not regular patients of the surgery were excluded. Patients received written invitations and study information by their general practitioner (GP) by mail. If interested, they returned the consent form to the study centre. In return, the patients received the baseline questionnaire by mail. No reminder was sent in case of non-responding to either study invitation or questionnaire. All incoming questionnaires were screened for psychological distress. GPs were interviewed by telephone about their patients with psychological distress and—due to limited financial resources—about a randomly selected sample of 80% of their patients who had no psychological distress. The interview focused on somatic/psychological co-morbidities and GP–patient relationship (15).

The study was conducted in compliance with the Declaration of Helsinki, adheres to the Belmont Report principles and was approved by the local ethics committees, and study participants gave written informed consent prior to participating in the study.

Psychological distress

Psychological distress was measured using the Patient Health Questionnaire Depression Scale (PHQ-9) (17), the Hospital Anxiety and Depression Scale (HADS) (18), HADS-D and HADS-A subscales and selected items of the PROMIS Anxiety Scale (19,20). Psychological distress was defined as either symptom of depression/adjustment disorder (PHQ-9 score >8 and HADS-D score >8) or anxiety (PROMIS Anxiety score >18) or both (16).

Lifestyle adherence

We defined lifestyle adherence as adherence to self-care behaviour according to the recommendations of the German College of General Practitioners and Family Physicians (21). The nine items (Supplementary material 1) for measuring lifestyle adherence were adapted from the patient information of the clinical HF guideline (21) with a score of 0 (low) to 9 (excellent) lifestyle adherence. The score was normally distributed.

Medication adherence

Medication adherence was assessed using the Morisky Scale (©MMAS-8) (22). Its score ranges from 0 (poor) to 8 (excellent) medication adherence. Due to a ceiling effect of the scale in our data (high medication adherence), we combined the low (score <6) and medium (score 6 to <8) adherence groups proposed by the authors into one group of moderate adherence and retained the group with high medication adherence (score 8). This led to a dichotomized score with moderate (0 to <8) and high (8) medication adherence (items listed in (22)).

Potential risk factors

To identify factors associated with low adherence, we included potentially relevant factors based on literature research and clinical reasoning. Education was evaluated using the Comparative Analysis of Social Mobility in Industrial Nations (CASMIN) criteria (23) and migrant background was defined in accordance with Schenk et al. (24). The patients’ self-efficacy was assessed by the General Self-Efficacy Scale (25) and social network was measured by the Lubben Social Network Scale (26). Further questions addressed the form of housing, body height and weight, type of health insurance and the patients’ awareness of their own HF diagnosis (‘Have you been told to suffer from heart failure by a physician?’ [yes/no]). The severity of HF New York Heart Association class I-IV (NYHA I–IV) and co-morbidities were assessed in the GP-interview. The interviewer read out a list of all diseases listed in the Charlson Index (27) followed by the question ‘What additional diseases does the patient have?’. All co-morbidities were grouped according to the MultiCare List of chronic conditions (28). Lastly, the GP was asked to state his/her years of practice as GP and to rate his/her familiarity with the patient: ‘On a scale of one (unfamiliar) to ten (very familiar): How well do you know the patient?’

Statistical analysis

Lifestyle adherence was investigated by linear regression analyses, with a higher score in the outcome variable indicating better lifestyle adherence. Medication adherence was assessed by logistic regression analyses, with the outcome of high medication adherence. Both models were controlled for age, sex, educational status and HF severity. Since we recruited more than one patient per GP, we included the GP as a random effect using the GENLINMIXED procedure in SPSS Version 23. To analyze potential risk factors associated with low adherence, we added the following factors to the models: body mass index, number of co-morbidities, type of health insurance, form of housing, migrant background, social network status, self-efficacy, patients’ awareness of their own HF diagnosis and familiarity between GP and patient. No general data imputation strategies were used, but missing data on individual items were imputed according to the instructions of the validated questionnaires. Significance level was set at P < 0.05.

Results

Association between psychological distress and adherence

Table 1 shows that 900 (29%) patients screened positive for psychological distress and 2199 (71%) negative. Medication adherence was generally high with a median of 8 (IQR 1) out of 8 points (mean 7.2, standard deviation [SD] 1.1). After dichotomization, 1606 (52%) patients showed high and 1493 (48%) moderate medication adherence. Fewer patients with psychological distress (40%) reported high medication adherence compared to patients without psychological distress (57%; chi-square = 68.370; P < 0.001).The patients showed a mean lifestyle adherence of 5.4 out of 9 points. Supplementary material 1 shows the aspects of lifestyle adherence and the percentage of patients who were adherent to each lifestyle aspect. Patients with psychological distress showed significantly lower scores for lifestyle adherence (5.3 versus 5.5, t(3097) = 2.473; P = 0.013) than patients without psychological distress.

Table 1.

Patient characteristics of primary care patients with heart failure in Germany (2012–14)

	Patients without psychological distress	Patients with psychological distress	Total n	Total valid n
Age in years, mean (SD)	74.4 (9.8)	72.6 (11.0)	73.9 (10.2)	2950
Male sex, n (%)	1255 (57.1)	422 (46.9)	1677 (54.1)	3055
Educational level (CASMIN 3), n (%)				3048
Low	1313 (59.7)	638 (70.9)	1951 (63.0)
Middle	627 (28.5)	201(22.3)	828 (26.7)
High	222 (10.1)	47(5.2)	269 (8.7)
Health insurance, n (%)				3014
Statutory	1933 (87.9)	817 (90.8)	2750 (88.7)
Private	196 (8.9)	46 (5.1)	242 (7.8)
By social welfare office	11 (0.5)	11 (1.2)	22 (0.7)
Form of housing				3038
Private household with others	1497 (68.1)	578 (64.2)	2075 (67.0)
Alone in private household	632 (28.7)	281 (31.2)	913 (29.5)
Home for the elderly/nursing home	29 (1.3)	21 (2.3)	50 (1.6)
Migrant background (yes), n (%)	98 (4.5)	73 (8.1)	171 (5.5)	2912
NYHA functional class, n (%)				3045
Class 1/2	1675 (76.2)	595 (66.1)	2270 (73.2)
Class 3/4	486 (22.1)	289 (32.1)	775 (25.0)
Patient’s awareness of their heart failure diagnosis, n (%)	1573 (71.5)	689 (76.6)	2262 (73.0)	2837
Number of co-morbidities (MultiCare List), mean (SD)	4.4 (2.4)	4.8 (2.5)	4.5 (2.4)	3042
Body mass index, mean (SD)	28.8 (5.6)	29.8 (6.7)	29.1 (6.0)	2788
Medication adherence high (Score 8), n (%)	1244 (56.6)	362 (40.2)	1606 (51.8)	3099
Lifestyle adherence score, mean (SD)	5.5 (1.5)	5.3 (1.5)	5.4 (1.5)	3099
Self-efficacy score, mean (SD)	32.9 (5.7)	26.8 (6.8)	31.1 (6.6)	3059
Lubben Social Network Scale, mean (SD)	15.7 (5.7)	13.6 (5.9)	15.0 (5.9)	2995
GP’s estimate of his/her familiarity with the patient, mean (SD)	7.2 (2.2)	7.5 (2.1)	7.3 (2.2)	3099
Total number of participants, n (%)	2199 (71)	900 (29)	3099 (100)

Table 2 shows the models that investigate the association between psychological distress and higher lifestyle adherence (Model 1) and high medication adherence (Model 2). After controlling for socio-demographic factors, HF severity, and GP cluster, psychological distress was significantly associated with poorer medication adherence (Model 2). The association between psychological distress, and lifestyle adherence was only marginally significant (P = 0.054, Model 1).

Table 2.

Results of the linear regression examining the association between psychological distress and lifestyle adherence (Model 1) and logistic regression examining the association between psychological distress and medication adherence (Model 2) in primary care patients with heart failure (2012–14)

	Model 1				Model 2
	Coefficient B	Sig.	95% confidence interval		Odds ratio	Sig.	95% confidence interval
			Lower value	Upper value			Lower value	Upper value
Psychological distress	−0.117	0.054	−0.236	0.002	0.501	<0.001	0.422	0.595
Male sex	−0.169	0.002	−0.278	−0.060	0.875	0.094	0.748	1.023
Age	0.025	<0.001	0.019	0.030	1.016	<0.001	1.008	1.023
Education (reference: low)
Middle	0.014	0.815	−0.106	0.135	0.998	0.986	0.839	1.187
High	−0.026	0.791	−0.217	0.165	0.533	<0.001	0.404	0.703
NYHA Class 3/4 (reference: Class 1/2)	−0.032	0.612	−0.155	0.091	1.052	0.574	0.881	1.256
Constant	3.749	<0.001	3.329	4.169	0.476	0.016	0.260	0.871
Akaike information criterion (corrected)	10 254.505					12 299.405

n = 2874; models are controlled for the GP cluster effect.

To identify factors associated with poorer adherence; these models were recalculated with additional variables (as described above). Figure 1 gives an overview of correlates for both forms of adherence. The set of associated factors shows a partial overlap.

Figure 1.

Potential risk and protective factors associated with higher medication and lifestyle adherence in 2051 primary care heart failure patients in Germany (2012–14).

Tables 3 (Model 3—lifestyle adherence) and 4 (Model 4—medication adherence) show the extended models. Both show a better model fit (lower values of Akaike corrected) than the original models.

Table 3.

Model 3—results of the linear regression examining the association between potential risk and protective factors and higher lifestyle adherence in primary care patients with heart failure (2012–14)

Model 3	Coefficient B	Sig.	95% confidence interval
			Lower value	Upper value
Psychological distress	0.025	0.752	−0.131	0.181
Age	0.023	<0.001	0.017	0.030
Self-efficacy	0.021	<0.001	0.010	0.031
GP’s estimate of his/her familiarity with the patient	0.062	<0.001	0.030	0.093
Lubben Social Network Score	0.024	<0.001	0.013	0.035
Number of co-morbidities (MultiCare List)	0.027	0.048	0.000	0.054
Male sex	−0.331	<0.001	−0.464	−0.198
Patient’s unawareness of heart failure diagnosis	−0.236	0.003	−0.391	−0.080
Body mass index	−0.019	0.001	−0.030	−0.008
Form of housing (reference: private household with others)
Alone in private household	−0.219	0.002	−0.361	−0.077
Home for the elderly/nursing home	−0.358	0.169	−0.867	0.152
Education (reference: low)
Middle	0.044	0.546	−0.098	0.185
High	0.056	0.634	−0.175	0.287
Migrant background (reference: no)	0.098	0.495	−0.184	0.380
NYHA Class 3/4 (reference: Class 1/2)	−0.028	0.712	−0.175	0.120
Health insurance (reference: statutory)
Private	0.017	0.887	−0.217	0.251
By social welfare office	0.318	0.424	−0.463	1.099
GPs’ male sex	−0.147	0.068	−0.305	0.011
GPs’ years of practice	−0.004	0.278	−0.011	0.003
Constant	3.143	<0.001	2.368	3.918

Akaike information criterion (corrected): 7329.273; n = 2051; model is controlled for the general practitioner cluster effect; higher scores in the outcome variable represent better lifestyle adherence.

Factors associated with lifestyle adherence

Male sex, higher body mass index, living alone in a private household and the patients’ unawareness of their HF diagnosis were associated with poorer lifestyle adherence (Table 3). Higher age, greater familiarity between GP and patient and higher Lubben Social Network Score, as well as higher self-efficacy scores and a higher number of co-morbidities were associated with better lifestyle adherence. In this model, psychological distress even lost the trend for significance.

Factors associated with medication adherence

The expanded model in Table 4 shows potential risk factors for medication adherence: psychological distress remains significantly associated with poorer medication adherence. The patients’ male sex and a high level of education were also associated with reduced medication adherence (odds ratio <1). Higher age, higher self-efficacy and greater familiarity between GP and patient were associated with better medication adherence (odds ratio >1).

Table 4.

Model 4—results of the logistic regression to investigate the association between potential risk and protective factors and high medication adherence in primary care patients with heart failure (2012–14)

Model 4	Odds ratio	Sig.	95% confidence interval
			Lower value	Upper value
Psychological distress	0.635	<0.001	0.506	0.797
Age	1.012	0.016	1.002	1.022
Self-efficacy	1.049	<0.001	1.033	1.066
GP’s estimate of his/her familiarity with the patient	1.060	0.013	1.013	1.110
Lubben Social Network Score	1.008	0.354	0.991	1.024
Number of co-morbidities (MultiCare List)	1.002	0.936	0.963	1.042
Male sex	0.816	0.041	0.671	0.991
Patient’s unawareness of their heart failure diagnosis	0.885	0.294	0.705	1.112
Body mass index	0.987	0.090	0.971	1.002
Form of housing (reference: private household with others)
Alone in private household	1.040	0.711	0.845	1.281
Home for the elderly/nursing home	1.448	0.339	0.678	3.092
Education (reference: low)
Middle	0.947	0.607	0.770	1.165
High	0.671	0.021	0.479	0.941
Migrant background	0.914	0.672	0.604	1.384
NYHA Class 3/4 (reference: Class 1/2)	1.188	0.120	0.956	1.475
Health insurance (reference: statutory)
Private	0.850	0.353	0.604	1.197
By social welfare office	0.730	0.602	0.224	2.379
GPs’ male sex	1.094	0.436	0.872	1.374
GPs’ years of practice	0.990	0.063	0.980	1.001
Constant	0.126	< 0.001	0.041	0.395

Akaike information criterion (corrected) = 8892.165; n = 2051; model is controlled for the general practitioner cluster effect.

Discussion

Psychological distress was significantly associated with decreased medication adherence. There was no independent association between psychological distress and lifestyle adherence. Male sex, younger age, lower self-efficacy and lower familiarity between GP and patient were common factors associated with reduced medication and lifestyle adherence. A high level of education was associated with lower medication adherence only. Higher body mass index, weaker social network, living alone and fewer chronic co-morbidities were only associated with less lifestyle adherence

Comparison with existing literature

We found a significant association between psychological distress and reduced medication adherence. DiMatteo et al. reported similar findings in their meta-analysis (29). Contrary to our hypothesis, psychological distress was not independently associated with reduced lifestyle adherence. This is consistent with the findings of Schweitzer et al., who found that depression could not predict lifestyle adherence, while anxiety explained only minimal variability in two aspects of lifestyle adherence (14). However, in several studies (30), depression was consistent and significantly associated with self-care behaviour measured by the European Heart Failure Self-Care Behaviour Scale (EHFScBS). The different ways of operationalization (depression versus depressive and/or anxious symptomatology) might lead to discrepancies between the studies. In addition, our results are in line with a study of Gallagher et al. (31), which found a significant association between social support and self-care management in HF patients but no association between depression and self-care management. We also found a significant association between a weaker social network and reduced lifestyle adherence so that the association between the social network and lifestyle adherence seems to be stronger in our analyses than the association between psychological distress and lifestyle adherence.We found an association between high education and medication adherence that could reflect the autonomy of highly educated patients in deciding about their own medical treatment.

In our study, 27% of the patients were unaware of their HF diagnosis. This does not necessarily mean that the diagnosis was never communicated to the patient but rather that the patients did not remember it. This is consistent with previous studies that showed that HF is a diagnosis that is not well understood by patients (32). Lainščak et al. found a better recall of more than four HF self-care items when a health care professional had clearly explained the health condition to the patient (5). In our sample, remembrance of the HF diagnosis was positively associated with lifestyle but not with medication adherence. The knowledge about the disease seems to be more important for lifestyle adherence than for medication adherence. If the HF diagnosis is addressed and self-management is explained more than once, adherence to lifestyle could be improved.

The association between self-efficacy and lifestyle adherence is consistent with previous studies (14,33,34), which suggests a promising approach to intervention programs. In contrast to our findings, Schweitzer et al. found no significant association between medication adherence and self-efficacy (14). Since the sample in our study was 30-times larger, we assume that our findings are more reliable.

A higher physician-rated familiarity between GP and patient was associated with both medication adherence and lifestyle adherence. Therefore, getting to know a patient and building a more trusting relationship could increase patients’ adherence. Marx et al. also concluded that seeking dialogue directly after diagnosis before starting medical treatment might strengthen medication adherence in hypertensive patients as this increases the acceptance of diagnosis and therapy (35).

Factors associated with adherence and impact for clinical practice

Patients who are less familiar with their GP and patients with lower self-efficacy are at risk of both poorer lifestyle and medication adherence. GPs should keep a special eye on them. Patients with a weaker social network, those who live alone, patients who are unaware of their HF diagnosis or patients with a higher body mass index have a particular risk of poorer lifestyle adherence. Patients with psychological distress or high education are more at risk for reduced medication adherence.

Self-rated medication adherence was generally high in our patients. Gjesing et al. also found high medication adherence of HF patients in specialized clinics and after referral back to their GPs (36). Our data do not suggest an urgent need for action, but it might be helpful to address medication adherence iteratively in patients with psychological distress; they might be more at risk to forget regular medication intake. The development of reminder strategies could improve medication adherence.

Wallbach et al. showed by urine analysis a non-adherence rate in patients with a hypertensive crisis of 24% (37). It cannot be ruled out that the objectively measured rates of medication adherence are lower than the self-assessed rates and that further measures are needed. There is a need for studies that investigate medication adherence using laboratory values in HF patients.

Our data suggest a higher potential for improving lifestyle adherence—a few simple actions could increase it: asking the patients what they know about their disease, explaining their medical condition and refreshing related lifestyle recommendations could help to understand those and increase motivation to follow them. Addressing why weight control and influenza vaccination are recommended could offer the greatest potential for improving lifestyle adherence as adherence rates in these items are moderate, but the recommendations are easy for patients to implement. Improving the patient’s self-efficacy and a higher familiarity with the patient (by seeking intense dialogue right after diagnosis) could improve both medication and lifestyle adherence. The effectiveness of these strategies needs to be investigated in further studies.

Impact for research

Further intervention studies must demonstrate whether increasing the familiarity between patient and GP, refreshing patient’s knowledge of the disease, addressing the patients’ self-efficacy and reducing psychological distress could improve patient’s adherence to HF treatment. Our findings suggest that interventions in future clinical trials need to be tailored to the type of adherence that is investigated.

Strengths and limitations

This is the first study that simultaneously investigates the association between psychological distress and lifestyle and medication adherence in a large sample of primary care patients with HF. Some limitations should be taken into account: Psychological distress was defined as the presence of symptoms of depression and/or anxiety (not depression diagnosis/anxiety disorder). At the time of planning the study, no validated instrument to measure lifestyle adherence was available in German; the German version of the European Heart Failure Self-care Behaviour Scale was first validated in 2013 (38). However, the scale used in this study to measure lifestyle adherence was based on the patient information provided in the clinical guideline for patients with HF of the German College of General Practitioners and Family Physicians (21) and is, therefore, considered clinically relevant.

The patients were recruited via GPs. In Germany, 94% of the HF patients receive their initial prescription for HF treatment from their GP (39)—the patients are considered representative. We did not check for cognitive impairment, which sometimes co-occurs with HF, but excluded patients with dementia; the potential bias is estimated to be small.

Conclusion

Attention should be paid to patients who have any of the following factors: male, lower age, lower self-efficacy and who are not familiar with the GP. These potential risk factors should be validated in a different sample. Both the ©MMAS-8 scale and the lifestyle adherence scale are easy to assess. They could be used orally during the consultation; this could strengthen the physician–patient relationship, which, in turn, could have a positive impact on lifestyle and medication adherence.

Identifying and reducing psychological distress (e.g. by cognitive behaviour therapy) in future interventional research to increase adherence could be more promising when addressing medication adherence than lifestyle adherence. In order to improve lifestyle adherence, research should focus on measures aiming at the patient’s own understanding of the HF diagnosis. We hypothesize that it is a promising factor to explain patients their HF diagnosis more than once and to help patients to increase their self-efficacy in managing their illness in order to increase lifestyle adherence.

Acknowledgements

Members of the RECODE-HF Study Group: Winfried Adam, Cassandra Behrens, Eva Blozik, Sigrid Boczor, Marion Eisele, Malte Harder, Christoph Herrmann-Lingen, Agata Kazek, Dagmar Lühmann, Anja Rakebrandt, Koosje Roeper, Martin Scherer, Stefan Störk, Jens-Martin Träder.

Declarations

Funding: this study was funded by the German Federal Ministry of Education and Research (grant numbers 01GY1150 and 01EO1004).

Ethical approval: Medical Association of Hamburg, Approval No. PV3889; Ethics Committee of the Medical Faculty of the University of Würzburg, Approval No. 125/12.

Conflict of interest: ME, GM and EB are members and MS is the President of the German College of General Practitioners and Family Physicians. SB received fees for lecturers/statistical consulting of Asklepios Medical School GmbH. EB is employed by Helsana Health Insurances, Switzerland. SS is a member of the German Cardiac Society and the writing group of the National Guideline Heart Failure Care. CHL receives royalties for the German HADS from Hogrefe Huber publishers, is the President of the German College of Psychosomatic Medicine, chairs its working group on Psychosomatics in Cardiology, is a member of the German Society for Cardiology and of other scientific societies for psychosomatic/behavioural medicine. All other authors declare that they have no conflicts of interest.

Author notes

These authors are shared first authors.