The association of general practitioners’ awareness of depression and anxiety with change in quality of life in heart failure patients: results of the prospective observational RECODE-HF cohort study

Abstract

Background

Depression and anxiety are more prevalent in patients with heart failure (HF) than in the general population and reduce quality of life (QoL); therefore, clinical guidelines recommend screening HF patients for depression/anxiety.

Objective

We investigated, whether the general practitioners’ (GPs) awareness of patients’ symptoms of depression and/or anxiety (psychosocial distress) was associated with a change in QoL.

Methods

In this prospective observational study, we recruited 3,129 primary care HF patients in Germany. Patients completed baseline and 12-month follow-up questionnaires. Their GPs were interviewed. We identified 666 patients with psychosocial distress and compared 2 groups by analysis of covariance: 235 patients with psychosocial distress whose GP was aware of the psychosocial distress and 431 patients with psychosocial distress whose GP was unaware of such distress. Primary outcome was the change in QoL, assessed by the EQ-5D visual analogue scale.

Results

Patients with psychosocial distress showed lower baseline QoL than those without (45.9 vs 64.1; P < 0.001). Within the patients with psychosocial distress, the GPs’ awareness of psychosocial distress was not associated with improvement of QoL (F = 1.285; P = 0.258) or remission of psychosocial distress (odds ratio = 0.887; P = 0.608).

Conclusion

We found no association between the GPs’ awareness of psychosocial distress and change in QoL. Although data for effective treatments of depression in HF are currently insufficient, psychosocial distress strongly impairs the QoL in HF patients. These findings might influence the development of clinical practice guidelines in HF.

Introduction

Depression and anxiety disorders are more prevalent in patients with heart failure (HF)^1,2 than in the general population, and adversely affect quality of life (QoL)^3,4 and prognosis.^5,6 Shared pathophysiological pathways between psychologic and somatic disorder have been discussed.⁷ Although evidence is scarce,⁸ clinical guidelines on the treatment of HF recommend screening for depression^9–11 and anxiety,¹¹ and a guideline-supported screening pathway has been published.¹²

Interventional trials suggested that antidepressant treatment appears to be a save treatment option for depression in HF patients, but results are mixed for reducing depression and improving QoL, morbidity, and mortality.¹³ Beneficial effects on alleviation of depressive symptoms in HF patients have been described for physical exercise and¹⁴ cognitive behavioural therapy.¹⁵ However, so far there is no evidence that screening for depression/anxiety and the treatment following a positive screening result will improve health-related quality of life (HRQoL) or the prognosis of HF patients.

General practitioners (GPs) act as gate keepers to the health system. In HF cardiologists are involved, but patients usually still consult their GP on a regular basis. When routine screening for anxiety and depression is recommended in HF guidelines, the GP’s awareness of symptoms of depression and/or anxiety (psychosocial distress) should impact the patient’s HRQoL. RECODE-HF was designed as a prospective noninterventional cohort study investigating (1) whether psychosocial distress is associated with HRQoL, and (2) whether the GP’s awareness of a HF patient’s psychosocial distress is associated with a change in HRQoL (primary endpoint of the RECODE-HF study). We hypothesized that (1) psychosocial distress is associated with a reduced HRQoL, and (2) that the HRQoL worsens less over time in HF patients, if the GP is aware of the patients’ comorbid psychosocial distress, compared with HF patients whose GP is unaware of the comorbid condition.

Methods

RECODE-HF was a longitudinal, observational follow-up study in a primary care setting in Germany.¹⁶

Recruitment

We invited all 4,220 GPs of 2 German regions (South: Würzburg and surrounding areas; North: Hamburg and surrounding areas) to participate. In the 293 practices willing to participate, eligible HF patients were selected from practice software and invited by the respective GP. Eligibility criteria were: (i) age ≥18 years, (ii) last consultation within the last 6 months, and (iii) received a diagnosis of HF (GP documentation) within the last 5 years. Patients who passed away since their last consultation, were suffering from dementia, or patient who did not belong to the regular patient base of the practice were excluded. Recruitment period was 2/2012 to 6/2014.

Data ascertainment

Data were collected at baseline and after 12-month follow-up. Patients received a self-administered baseline and follow-up questionnaire. A telephone interview with the GP to assess the patients’ comorbidities was scheduled for 3 weeks after sending out the baseline questionnaire to the patients. Data were collected between 8/2012 and 11/2015.

Questionnaires collected at baseline and follow-up included the Hospital Anxiety and Depression Scales (HADS-A and HADS-D),¹⁷ the Patient Health Questionnaire depression subscale (PHQ-9),¹⁸ and items from the PROMIS Depression and Anxiety item pool (e.g. during the last 7 days my worries have overwhelmed me)¹⁹ to assess psychosocial distress. HRQoL was measured by the EQ-5D visual analogue scale (VAS)²⁰ ranging from 0 (worst health state) to 100 (best health state).

Lifestyle adherence (LA) was assessed in the patient questionnaire and defined as adherence to self-care behaviour according to the recommendations of the German College of General Practitioners and Family Physicians.²¹ The 9 items to measure LA (e.g. influenza vaccination within the last year, alcohol intake of no more than 1 drink daily) were adapted from the patient information of the clinical HF guideline²¹ and resulted in a score ranging from 0 (low LA) to 9 (excellent LA). Self-efficacy was measured by the General Self-Efficacy Scale.²² Education was assessed by the CASMIN criteria.²³

In the interviews, the GP was asked to confirm the diagnosis of HF, to rate the HF severity by the New York Heart Association functional class (NYHA class), and to name the patient’s comorbidities. Depression/anxiety was assessed by asking “Does the patient actually display symptoms of depression/anxiety? If yes, would you classify this as diagnosis or symptomatology?” During the interview the GP had access to the patient record. The comorbidity index was calculated based on the number of chronic conditions covered by the MultiCare list.²⁴

Definition of psychosocial distress and GP awareness of psychosocial distress

All baseline questionnaires were analysed for psychosocial distress by a hierarchical algorithm that has been validated in a substudy with different HF patients.²⁵ All patients with a PHQ-9 score >8 and HADS-D score >8 (positive predictive value of 68.8% for prevalent depression or adjustment disorder in the validation sample) and those patients with PROMIS Anxiety score >18 (positive predictive value of 42.9% to have depression/adjustment disorder or anxiety disorder in the validation sample) were defined as displaying psychosocial distress. As the case finding properties do not allow diagnosing depression/adjustment disorder or anxiety disorder, presence of psychosocial distress, i.e. PsySocD(+), is defined as showing symptoms of depression/adjustment disorder and/or symptoms of anxiety. According to the sample size calculation, patients with psychosocial distress were included in the study with a ratio 1:3¹⁶: Due to the studies focus on patients with psychosocial distress [P(+) patients], all P(+) patients and a random sample of 80% of the patients without psychosocial distress [P(−)] (acting as control group) were selected for telephone interviews with the GP and remained in the study. Further, PsySocD(+) patients were subdivided into 2 subgroups of patients in whom the GP stated that the patient displays symptoms/diagnosis of depression and/or anxiety [referred as PsySocD(+/+)] or not [referred as PsySocD(+/−)].

Data analysis

T-Test and chi-square test were used to compare groups depending on the data level, as appropriate. The nominal level of significance was set at P = 0.05 for all analyses. Missing data were imputed according to the manuals of the instruments used. The statistical analyses were performed with SPSS Version 25. A power calculation was performed prior to the study,¹⁶ and post hoc.

To investigate hypothesis 1, we compared HRQoL at baseline (EQ-5D VAS) between PsySocD(+) and PsySocD(−) patients using analysis of covariance. Due to their potential influence on HRQoL in HF patients (literature and clinical reasoning), following potential confounders were included: age, gender, educational level, comorbidity index, NYHA functional class, level of adherence to HF lifestyle recommendations, and patients’ self-efficacy. We also controlled for the random GP cluster effect.

To investigate hypothesis 2, we compared the change in HRQoL between follow-up and baseline between PsySocD(+/+) and PsySocD(+/−) patients. We included the same set of confounders as for hypothesis 1 and added the baseline level of psychosocial distress (HADS-D and HADS-A) and HRQoL (EQ-5D VAS). We re-run the model and added “remission of psychosocial distress” and the interaction “remission of psychosocial distress” × “GP aware of psychosocial distress.” Remission of psychosocial distress was reached, when a patient screened positive for psychosocial distress at baseline PsySocD(+) and negative at follow-up PsySocD(−).

Post hoc analysis

We calculated logistic regression to predict the remission of psychosocial distress between baseline and follow-up. We entered the same set of potential confounders included for the analysis of hypothesis 2 by backwards selection. A post hoc power analysis was calculated with G*Power Version 3.1.5.

Results

Out of 4,420 invited GPs, 293 were willing to participate (6.6%) of those 44.5% were single practices, 47.6% joint practices, and 5.8% group practices with seperate settlement.²⁵ Compared with the total of German GPs in 2013, participating GPs were less frequent female (27.1% in our study vs 40.4% in Germany) but in comparable age (53.1 vs 54.2 years). In total, 13,800 patients received written invitations by their GP, of whom 5,385 (39.0%) returned the written consent to the study centre. The baseline questionnaire was returned by 4,909 patients (91.2%) of whom 3,387 were included into the RECODE-HF study (for details of the study flow see Eisele et al.²⁶). Of those, n = 3,129 patients providing valid information on HRQoL at baseline were included in the analysis of hypothesis 1: n = 934 PsySocD(+); n = 2,195 PsySocD(−). Hypothesis 2 was investigated in all patients with psychosocial distress (PsySocD(+) patients) and valid follow-up information on HRQoL (n = 666): n= 235 PsySocD(+/+) and n= 431 PsySocD(+/−) patients (see Fig. 1). Post hoc power analysis revealed a power of 87% to detect a minimal difference of 5 points between PsySocD(+/+) and PsySocD(+/−) patients. Median time difference between baseline and follow-up was 52 weeks.

Patient characteristics entering analyses for hypotheses 1 and 2 are displayed in Tables 1 and 2. Compared with German Heart failure patients,²⁷ study participants were slightly younger (74 vs 78 years) and rather male (53.9% vs 43.3%).

Hypothesis 1

PsySocD(+) patients displayed a significantly lower HRQoL at baseline than PsySocD(−) patients (see Table 1). This association remained after controlling for potential confounders (F = 248.998; P < 0.001; adjusted means PsySocD(+) = 49.6 and PsySocD(−) = 63.2).

Hypothesis 2

GPs’ awareness of psychosocial distress was not associated with changes in HRQoL between baseline and follow-up (Table 3, model 1), even though GPs initiated any treatment in 85.5% and pharmacologic treatment in 63% of all PsySocD (+/+) patients (Supplement 1). When this model was refined by omitting all nonsignificant variables these effects remained unchanged (F = 2.090; P = 0.149). Between baseline and follow-up, psychosocial distress remitted in 183 (27.5%) and persisted in 409 patients (61.4%). When “remission of psychosocial distress” and its interaction with “GP aware of psychosocial distress” was added to the model (Table 3, model 2), a significant association between “remission of psychosocial distress” and the change in HRQoL (7.661; P = 0.006) became apparent. The improvement (adjusted mean) in HRQoL in patients with remitting psychosocial distress was 6.3 points. By contrast, HRQoL in patients with persisting psychosocial distress worsened by 0.2 points. The interaction term between “GP awareness of psychosocial distress” and “remission of psychosocial distress” missed significance (F = 1.061, P = 0.304).

Post hoc analyses

In logistic regression analyses, GP awareness of psychosocial distress was inversely associated with remission of psychosocial distress (odds ratio 0.628 [0.431; 0.915]; P = 0.015). This association was lost after including potential confounders (HADS-A and HADS-D as marker for the severity of psychosocial distress at baseline) into the model (Table 4).

Discussion

The present study yielded 2 major findings: HF patients with psychosocial distress exhibited significantly lower HRQoL at baseline than HF patients without psychosocial distress. The GPs’ awareness of psychosocial distress was not associated with a change in HRQoL, even though the remission of psychosocial distress between baseline and follow-up was associated with a change in HRQoL.

Strengths and limitations

This is the first study investigating the association between GPs’ awareness of psychosocial distress and associated changes in HRQoL in patients with HF cared for in primary care. Psychosocial distress was identified by a validated algorithm,²⁶ which accounts for overlapping symptoms of depression/anxiety and HF. A power calculation was performed prior to the study,¹⁶ and was confirmed by a post hoc power of 87% to detect a minimal difference of 5 points in the EQ-5D VAS.

Some limitations need to be considered. We might not have been able to detect between-group differences <5 points in the EQ-5D VAS. However, differences below 5 points might miss clinical relevance of the results. The GP response rate was rather low (6.6%) and participating GPs were more frequent male compared with all German GPs. Our results might therefore be slightly biased in favour of male GPs. The patient response rate was 39% and patients were slightly younger and rather male compared with the German HF population. It is therefore likely that we included a higher number of less severely impaired patients. To control for this bias we included the NYHA class and baseline level of anxiety and depression in the models and controlled for this potential bias. Nevertheless, generalization of our results might be limited to slightly younger and less affected HF patients. HF patients identified by the algorithm displayed a symptomatology of depression/adjustment disorder and/or anxiety assessed by questionnaire, but did not always carry an established diagnosis examined by diagnostic interview. However, evidence revealed individuals with subthreshold forms of anxiety/depression showing reduced QoL^28,29 and higher morbidity and mortality after myocardial infarction.³⁰ Nevertheless, our results are limited to the GPs’ awareness of symptoms of depression/anxiety and are not representative for the GP awareness of anxiety/depression disorder.

Comparison with existing literature

Our results are in line with previous studies indicating that HRQoL is reduced in HF patients with depression/anxiety.^3,31–33 Remission of psychosocial distress between baseline and follow-up was associated with improved HRQoL. This is in accordance with the finding of the SADHEART-CHF study, where remission of depression in HF patients was associated with improved health status. Furthermore, the SADHEART-CHF study revealed no effect of sertraline treatment on depression remission compared with placebo.³⁴ This corresponds to our finding that awareness of psychosocial distress (followed by treatment as usual including pharmacologic treatment in 65% of the PsySocD(+/+) patients) was not associated with improvement in HRQoL. A pilot study investigating paroxetine treatment found an effect of paroxetine on both reductions of depressive symptoms and psychological aspects of QoL but not on physical QoL measures.³⁵ Our study focussed on HRQoL, which might be one reason for a missing association between GP awareness of psychosocial distress and QoL.

Nonpharmacologic interventional studies revealed that nurse-based monitoring,³⁶ cognitive behavioural therapy¹⁵ and a “walking with controlled breathing”—program³⁷ improved psychological status and QoL in HF patients compared with usual care. Treatment options except the prescription of antidepressants are not yet available in primary care due to the fact that the referral of elderly HF patients to psychotherapy/exercise programmes is often demanding in everyday practice. We generally have fewer good treatment strategies for HF and comorbid depression/anxiety compared with other diseases like HIV or some cancers and individual strategies focussed on patients’ personal goals might be needed. Therefore, the awareness of psychosocial distress in our study may not have resulted in efficient reduction of psychosocial distress. In the MOOD-HF and SPIRR-CAD studies targeted therapy of the psychological comorbidity (by tablets or psychotherapy) did not resolve the depressive mood in cardiac patients,^38,39 but optimal somatic treatment of cardiac patients may lead to a reduction of psychosocial distress. If so, there is a need to optimize both somatic and psychologic treatment to reduce psychosocial distress in patients with HF. Lastly, due to the observational design, the potential effect of GP initiated measures (with unique starting points) could not be specified; this allows for selection bias in favour of chronic cases in our data, which seems likely by the high rate of >60% of patients exhibiting persistent psychosocial distress. Patients with psychosocial distress and adequate ongoing treatment might have not been identified by our algorithm due to a lack of current symptoms.

Post hoc analyses revealed an inverse association between GPs’ awareness and remission of psychosocial distress: Patients, whose GP was aware of psychosocial distress, had a lower chance for remission than patients, whose GP was unaware. This effect disappeared when baseline levels of anxiety/depression were included in the model. This seems to indicate that GPs are more likely to be aware of psychosocial distress if the patient is more severely affected. Those patients, by definition, have a lower chance for remission of psychosocial distress.

Implications for research and practice

Because GPs’ awareness of psychosocial distress was not associated with changes in HRQoL we can neither support nor negate the recommendation to routinely screen HF patients for anxiety and depression in primary care. Two scenarios of beneficial effects are possible: First the GP becomes aware of a higher number of affected patients, including those with less severe symptoms. In this case, a possible positive effect of routine screening for psychosocial distress on HRQoL might be detectable, because less impaired patients might benefit more from GP-provided psychosomatic care. Second, patients are routinely screened, but the effect of the resulting care path is insufficient to improve HRQoL, e.g. because treatment of psychosocial distress in HF patients is not powerful to reduce HRQoL as long as there is no optimal somatic treatment of the underlying HF. Here, it might be crucial to inform the patients about the screening results, to enable them to differentiate between symptoms of depression and HF symptoms. The DEPSCREEN-INFO study showed a positive effect of depression reduction after patient feedback in cardiac patients.⁴⁰

Conclusion

Depression screenings are not generally part of clinical practice recommendation in primary care but are recommended for patients with HF. We found no evidence for an association between the GP’s awareness of psychosocial distress and the change in HRQoL within 1 year. This missing effect may result from methodological limitations of our observational study or the fact that antidepressants, easily to administer in primary care, might fail to show a beneficial effect on reducing of depressive symptoms. Interventional studies, designed as randomized controlled trials, are needed to shed light on the effect of screening in patients with HF. The effects of optimal somatic management and the feedback of the screening results directly to the patients should be considered in these studies.

Although the data for effective treatments of depression in HF are currently insufficient, psychosocial distress strongly impairs the HRQoL in patients with HF. These findings might influence the development of clinical practice guidelines in HF.

Acknowledgements

We thank all GPs and patients for their good collaboration. 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, and Jens-Martin Träder.

Funding

This work was supported by the German Federal Ministry of Education and Research grant numbers 01GY1150 and 01EO1004.

Ethical approval

The study was conducted in compliance with the Declaration of Helsinki and was approved by the local ethics committees (main study: Medical Association of Hamburg, Approval No. PV3889; Ethics Committee of the Medical Faculty of the University of Würzburg, Approval No. 125/12; substudy: Ethics Committee at the University of Göttingen Medical Center, Approval No. 19/8/11). All study participants gave written informed consent prior to participation in the study.

Conflict of interest

Financial: SB received fees for lecturer/statistical consulting of Asklepios Medical School GmbH. EB works at Helsana Health Insurances, Zürich. CHL receives royalties from Hogrefe Huber publishers for the German HADS. Nonfinancial: ME and EB are members of the German College of General Practitioners and Family Physicians (DEGAM). MS is president of the DEGAM. CHL is president of the German College of Psychosomatic Medicine, chairs its working group on Psychosomatics in Cardiology, is member of the German Society for Cardiology and other scientific societies for psychosomatic/behavioural medicine. SS is member of the German Cardiac Society and the writing group of the National Guideline Heart Failure Care.

Data availability

Data are available upon reasonable request.

References