Generalized Anxiety is a Predictor of Impaired Quality of Life in Patients with Atrial Fibrillation: Findings from the Prospective Observational ARENA Study

Abstract

Objective

Atrial fibrillation (AF) is associated with impaired health-related quality of life (HRQoL), an increased risk of morbidity, and mortality. Traditional AF-related outcomes (e.g., AF recurrence) primarily demonstrate the physiological benefits of AF management but do not focus on the benefits experienced subjectively by the patient (i.e., patient-reported outcomes), which have been suggested as optimal endpoints in AF intervention studies. The aim of this study is to identify medical and psychological factors associated with impaired HRQoL at 1-year follow-up.

Methods

Using data from the prospective observational multicenter ARENA study in patients with AF, we analyzed associations between medical factors, anxiety, and HRQoL at 1-year follow-up assessed using 5-level EuroQoL-5D.

Results

In 1,353 AF patients (mean age 71.4±10.3 years, 33.8% female), none of the medical predictors (e.g., heart disease) or history of cardioversion were associated with HRQoL at the 1-year follow-up. Higher generalized anxiety (ß=−.114, p<.001) but not cardiac anxiety (ß=−0.006, p=.809) at baseline predicted decreased HRQoL, independent of confounding variables and patients’ medical status. Furthermore, the worsening of patients’ generalized anxiety was associated with decreased HRQoL (ß=−.091, p<.001). In contrast, the improvement of generalized anxiety over time predicted higher HRQoL (ß=.097, p<.001). Finally, the worsening of patients’ cardiac anxiety over time was associated with decreased HRQoL (ß=−0.081, p<.001).

Conclusion

Our results highlight the importance of anxiety as a predictor of future HRQoL in patients with AF. Additional studies to examine the impact of anxiety treatment on HRQoL in this population are needed.

Introduction

Atrial fibrillation (AF) is the most frequent cardiac arrhythmia in Western countries [1,2], and it has been estimated that nearly 18 million people in Europe will suffer this medical condition by 2060 [3]. AF is associated with impaired health-related quality of life (HRQoL) [4–6], and an increased risk of mortality and morbidity (e.g., stroke, heart failure, cognitive impairment, sudden cardiac arrest), leading to a loss of around 6 million disability-adjusted life-years in patients with AF worldwide [3,7].

AF can lead to a broad range of symptoms (e.g., palpitations, dizziness, shortness of breath) [2] that are associated with higher AF-related symptom severity and impaired HRQoL [6]. AF-related symptom severity can be assessed either by duration and frequency of AF or by the European Heart Rhythm Association (EHRA) symptom classification [8]. Both AF symptom severity and impaired HRQoL are associated with higher risk of hospitalization [6]. Therefore, the goals of health-care providers should be to alleviate AF symptoms and improve HRQoL with minimal treatment burden, which may reduce the risk of long-term AF-associated complications (e.g., embolism, stroke) as well as mortality [1,5]. Traditional electrical-, rhythm-, and clinical indication-related outcomes (e.g., AF recurrence, re-interventions, AF burden, hospitalizations, thromboembolic events) primarily demonstrate the physiological benefits of AF management but do not focus on the benefits experienced subjectively by the patient (i.e., patient-reported outcomes [PROs]), which have been suggested as optimal endpoints in AF intervention studies [9]. The European Society of Cardiology (ESC), in collaboration with the European Association for Cardio-Thoracic Surgery (EACTS) and the EHRA, have identified the following PROs as important to measure in patients with AF [2,10]: HRQoL, patient-reported symptom status, physical functioning, emotional well-being, cognitive function, exercise tolerance, and ability to work. Of these PROs, HRQoL may be particularly important as a quality indicator for care and outcome in adults with AF, and the EHRA recommends its assessment at baseline and follow-up visits [10].

Psychological factors, AF-related symptoms, biological factors, functional status, and personal characteristics all appear to influence HRQoL in patients with AF [4]. Psychological symptoms, particularly trait anxiety [11,12], AF symptom severity, and AF frequency are associated with impaired HRQoL [6,13]. Additionally, age [4], female gender [14–16], cardiac dysfunction, coronary artery disease (CAD) [17], heart failure [18], higher New York Heart Association (NYHA) class [13], renal insufficiency [18], and diabetes [18–20] are associated with poorer HRQoL in AF patients. In contrast, both successful cardioversion (i.e., maintenance of sinus rhythm by 3 month) and ablation (potentially regardless of whether AF recurs) is associated with improved HRQoL [21–23].

Research studies performed to date have several limitations. Most have been cross-sectional in design, have focused either on medical or psychological symptoms (but not both simultaneously), have not focused on cardiac-specific anxiety, and have had limited sample sizes. Additional studies investigating both medical and psychological predictors of impaired HRQoL in sufficiently large samples of AF patients are necessary.

To close this gap, the aim of this study is to identify medical and patient-reported psychological factors prospectively associated with impaired HRQoL at 1-year follow-up in a large AF population. Based on previous research, and controlling for HRQoL at baseline, age, sex, and EHRA classification, we hypothesize (i) medical conditions (e.g., chronic heart disease, diabetes; hypothesis 1a) and female gender (hypothesis 1b) will be associated with impaired HRQoL at 1-year follow-up, while a history of cardioversion (hypothesis 1c) will be associated with greater HRQoL at 1-year follow-up, (ii) elevated generalized and cardiac-related anxiety levels at baseline will be associated with impaired HRQoL at 1-year follow-up (hypothesis 2), and (iii) an increase in anxiety from baseline to follow-up will be associated with impaired HRQoL at 1-year follow-up (hypothesis 3). The findings of this analysis may help to identify patients with AF at elevated risk for impaired long-term HRQoL, and identify opportunities for anxiety screening and/or treatment recommendations that may improve HRQoL and the multidisciplinary AF approach including psychosocial management (e.g., anxiety reduction) in this population.

Methods

Study design and patient population

As mentioned in the publication on baseline data including predictors of AF symptom severity and HRQoL [24], the ARENA is a prospective observational study in patients with AF of the Foundation Institute for Cardiac Infarction Research (IHF) in cooperation with the Departments of Cardiology of the Hospitals in Ludwigshafen, Heidelberg, and Mannheim, as well as local resident cardiologists and the Heidelberg University Hospital (Department of Clinical Pharmacology and Pharmacological Epidemiology) [25]. Inclusion criteria were age ≥ 18 years, confirmed diagnosis of atrial fibrillation, residence in the area, and informed consent [25]. Recruitment started in August 2016. More than 5,000 patients were screened, and 2,777 patients have been included in the ARENA study between August 2016 and December 2018. We analyzed data of a subset of 1,353 patients with complete baseline and 1-year follow-up data for HRQoL. All research was performed in accordance with relevant guidelines (e.g., STROBE [26]) and regulations. Written informed consent was obtained from all the participants. The study was approved by the Ethics Committee of the Rhineland-Palatine state Medical Association and by the Ethical Review Committee of the University of Heidelberg, and Mannheim Medical Center in Germany. The investigators registered on ClinicalTrials.gov (NCT02978248) on November, 30th 2016 https://clinicaltrials.gov/ct2/show/NCT02978248.

Assessments

Demographic and clinical variables (e.g., sex, age, organic heart diagnoses, diabetes [DM], AF type, history of cardioversion) were assessed at baseline. AF-related symptom severity was assessed using the European Heart Rhythm Association (EHRA) symptom classification [8,27]. HRQoL and generalized and cardiac anxiety were assessed at baseline and at 1-year follow-up. Generalized anxiety (GAD) was assessed using the GAD-2 questionnaire [28] (Items: Over the last 2 weeks, how often have you been bothered by the following problems?: 1. Feeling nervous, anxious or on edge; 2. Not being able to stop or control worrying) with answers including ‘not at all’ (=0), ‘on several days’ (1), ‘on more than half days’ (2) or ‘almost every day’ (3). Cardiac anxiety was assessed using the two-item version of the German Cardiac Anxiety Questionnaire (CAQ-2) [29] (Items: 1. When I have chest discomfort or when my heart is beating fast, I worry that I may have a heart attack.; 2. I avoid activities that make my heart go faster.) with answers including ‘never’ (=0), rarely (1); ‘sometimes’ (2), ‘ifte’ (3) or ‘always’ (4).

HRQoL was assessed using the German version of the EQ-5D-5L [30], a questionnaire for assessing QoL describing five dimensions (mobility, self-care, usual activities, pain/discomfort, anxiety/depression). Each dimension has five levels (no problems, slight problems, moderate problems, severe problems, and extreme problems) [30]. The baseline and follow-up values were converted into index based values (utilities) using the German Value Set for the EQ-5D-5L [31], ranging from −0.661 to 1; higher scores indicate a higher HRQoL.

The two items of GAD-2 and CAQ-2 were each aggregated to sum scores ranging from 0 to 6 for GAD-2 and from 0 to 8 for CAQ-2, with higher scores indicating more anxiety. Reliability analyses indicated the following values for the instruments applied: baseline GAD-2 (alpha = .782), baseline CAQ-2 (alpha = .548), GAD-2 at follow-up (alpha = .821), CAQ-2 at follow-up (alpha = .616), baseline EQ-5D-5L (omega = .855), EQ-5D-5L at follow-up (omega = .869).

Statistical analyses

As in the previous publication [24], we combined EHRA class 2a and 2b (i.e., mild to moderate symptoms) and EHRA class 3 and 4 (severe to disabling symptoms) into a single category each to avoid too many categories with only small numbers of observations. Absolute numbers and percentages are displayed for categorical variables, while differences among EHRA class categories were tested with chi-square tests. For dimensional variables, means and standard deviations are presented; differences among EHRA class categories were tested with Kruskal-Wallis rank sum tests since not all variables were normally distributed.

To test our hypotheses, we used hierarchical linear multiple regression analyses with the HRQoL index-based values at 1-year follow-up as the dependent variable. Baseline HRQoL index-based values, age, sex (hypothesis 1b/model 1, Table 2), and EHRA class were considered as potential confounders; these confounding variables were added to the regression model. EHRA class was dummy coded (EHRA class 2 vs. 1 and EHRA class 3 or 4 vs. 1). To test whether patients’ baseline medical status would predict the HRQoL at 1-year follow-up, we added organic heart diagnoses, diabetes mellitus, stroke/transient ischemic attack, and renal insufficiency as (medical) predictors of worsening, and history of cardioversion (hypothesis 1a and 1c/model 1, Table 2) as a predictor of improvement to the regression (coded as 0 = “no”, 1 = “yes”).

Table 2.

Parameters of multiple regression model predicting patients’ quality of life at 1-year follow-up by medical status at baseline (model 1).

Hypothesis	Predictor	Estimate	SE	95 % CI		Beta	t-Value	p-Value
Confounder	(Intercept)	0.329	0.046	0.239	0.418		7.208	<0.001
	HRQoL Baseline	0.752	0.023	0.707	0.796	0.698	33.192	<0.001
	Age	−0.002	0.000	−0.003	−0.001	−0.068	−3.188	<0.001
	Sex (female)	−0.005	0.010	−0.026	0.015	−0.011	−0.520	0.603
	EHRA class 2 (vs.1)	−0.005	0.010	−0.025	0.016	−0.010	−0.473	0.637
	EHRA class 3/4 (vs. 1)	0.018	0.015	−0.012	0.048	0.026	1.171	0.242
Medical Status	Organic heart diagnoses	−0.012	0.011	−0.034	0.010	−0.022	−1.057	0.291
	DM	−0.021	0.012	−0.045	0.002	−0.036	−1.774	0.076
	Stroke/TIA	−0.013	0.014	−0.041	0.015	−0.019	−0.923	0.356
	Cardioversion	−0.002	0.011	−0.023	0.018	−0.005	−0.228	0.819
	Renal Insufficiency	−0.017	0.014	−0.044	0.011	−0.025	−1.207	0.228

Note. HRQoL = Health-related quality of life; EHRA = European Heart Rhythm Association; DM = Diabetes mellitus; TIA = Transient ischemic attack. Statistically significant results are displayed in bold (p <.01); corrected R² for step 1 = .527 (confounder), corrected R² for step 2 = .528 (medical status; change in R² from step 1 to step 2 p = .140).

To test whether patients’ baseline anxiety would predict patients’ HRQoL at follow-up (hypothesis/model 2, Table 3), we added the baseline sum scores of the GAD-2 and the CAQ-2 screeners (in a sensitivity analysis, we also analyzed the baseline GAD-2 sum scores and the CAQ-2 sum scores in separate analyses; Supplementary File S1a–b). To test whether the change of patients’ anxiety from baseline to follow-up assessment would predict patients’ HRQoL at follow-up (hypothesis/model 3, Table 4), categorical predictors for increased anxiety (worsening) from baseline to follow-up assessment (compared to no change) and for decreased anxiety (improvement) from baseline to follow-up assessment (compared to no change) were added. Since we were interested in the differential effects of the two anxiety screening instruments, and whether worsening or improvement would each be a relevant contributor for HRQoL at follow-up this resulted in four categorical predictors indicating anxiety improvement or worsening separately for the GAD-2 and the CAQ-2. We also analyzed whether patients’ change in anxiety from baseline to follow-up would predict patients’ HRQoL at follow-up when the change of patients’ anxiety would be considered as continuous predictors (again separately for GAD-2 and CAQ-2). These continuous predictors which were calculated by subtracting the baseline score from the follow-up score (model 4, Supplementary File S2a). Collinearity and variance inflation factor were checked, but these did not indicate that any predictors had to be excluded from analyses. Correlations between variables considered in the regression analyses are also displayed in a heatmap (Figure 1). The reliability of the GAD-2 and CAQ-2 were assessed by calculating Cronbach’s alpha. For the EQ-5D-5L, we instead analyzed McDonald’s omega, since this is thought to be more robust and does not assume unidimensionality of the items.

Table 3.

Parameters of multiple regression model predicting patients’ quality of life at 1-year follow-up by baseline anxiety independent from patients’ medical status (model 2).

Hypothesis	Predictor	Estimate	SE	95 % CI		Beta	t-Value	p-Value
Confounder	(Intercept)	0.431	0.051	0.331	0.530		8.469	<.001
	HRQoL Baseline	0.691	0.026	0.640	0.742	0.642	26.550	<.001
	Age	−0.002	0.000	−0.003	−0.001	−0.086	−4.013	<.001
	Sex (female)	0.000	0.010	−0.020	0.021	0.000	0.015	0.988
	EHRA class 2 (vs.1)	0.001	0.011	−0.020	0.022	0.002	0.080	0.936
	EHRA class 3/4 (vs. 1)	0.023	0.016	−0.008	0.053	0.033	1.475	0.141
Medical Status	Organic heart diagnoses	−0.012	0.011	−0.035	0.010	−0.023	−1.079	0.281
	DM	−0.023	0.012	−0.046	0.001	−0.038	−1.880	0.060
	Stroke/TIA	−0.014	0.014	−0.042	0.014	−0.019	−0.956	0.339
	Cardioversion	−0.005	0.011	−0.026	0.016	−0.010	−0.490	0.624
	Renal Insufficiency	−0.020	0.014	−0.047	0.008	−0.029	−1.415	0.157
Baseline Anxiety	GAD-2 Baseline	−0.020	0.004	−0.029	−0.012	−0.114	−4.674	<.001
	CAQ-2 Baseline	−0.001	0.003	−0.007	0.005	−0.006	−0.242	0.809

Note. HRQoL = Health-related quality of life; EHRA = European Heart Rhythm Association; DM = Diabetes mellitus; TIA = Transient ischemic attack; CAQ-2 = Cardiac Anxiety Questionnaire 2-item screener; GAD-2 = Generalized Anxiety Disorder 2-item screener. Statistically significant results are displayed in bold (p <.01); corrected R² for step 1 = .527 (confounder), corrected R² for step 2 = .528 (medical status; change in R² from step 1 to step 2 p = .148), corrected R² for step 3 = .537 (baseline anxiety; change in R² from step 2 to step 3: p < .001).

Table 4.

Parameters of multiple regression model predicting patients’ quality of life at 1-year follow-up by the change in anxiety from baseline to follow-up independent from medical status (model 3).

Hypothesis	Predictor	Estimate	SE	95 % CI		Beta	t-Value	p-Value
Confounder	(Intercept)	0.482	0.052	0.381	0.583		9.353	<.001
	HRQoL Baseline	0.672	0.026	0.621	0.724	0.625	25.692	<.001
	Age	−0.002	0.000	−0.003	−0.001	−0.097	−4.488	<.001
	Sex (female)	0.007	0.011	−0.014	0.028	0.014	0.668	0.504
	EHRA class 2 (vs.1)	−0.001	0.011	−0.022	0.020	−0.003	−0.126	0.900
	EHRA class 3/4 (vs. 1)	0.015	0.016	−0.016	0.045	0.021	0.938	0.348
Medical Status	Organic heart diagnoses	−0.006	0.011	−0.028	0.017	−0.010	−0.482	0.630
	DM	−0.023	0.012	−0.047	0.000	−0.040	−1.944	0.052
	Stroke/TIA	−0.015	0.014	−0.043	0.013	−0.021	−1.024	0.306
	Cardioversion	−0.002	0.011	−0.023	0.018	−0.005	−0.233	0.816
	Renal Insufficiency	−0.020	0.014	−0.048	0.007	−0.030	−1.452	0.147
Baseline Anxiety	GAD-2 Baseline	−0.032	0.005	−0.041	−0.022	−0.176	−6.516	<.001
	CAQ-2 Baseline	−0.003	0.003	−0.010	0.003	−0.026	−1.004	0.316
Anxiety Change	GAD-2: worsening from pre to post (vs. no change)	−0.050	0.012	−0.073	−0.027	−0.091	−4.194	<.001
	GAD-2: improvement from pre to post (vs. no change)	0.051	0.013	0.027	0.076	0.097	4.103	<.001
	CAQ-2: worsening from pre to post (vs. no change)	−0.042	0.012	−0.066	−0.018	−0.081	−3.431	<.001
	CAQ-2: improvement from pre to post (vs. no change)	0.004	0.012	−0.019	0.027	0.008	0.318	0.750

Note. HRQoL = Health-related quality of life; EHRA = European Heart Rhythm Association; DM = Diabetes mellitus; TIA = Transient ischemic attack; GAD-2 = Generalized Anxiety Disorder 2-item screener; CAQ-2 = Cardiac Anxiety Questionnaire 2-item screener. Statistically significant results are displayed in bold (p <.01); corrected R² for step 1 = .527 (confounder), corrected R² for step 2 = .528 (medical status; change in R² from step 1 to step 2 p = .148), corrected R² for step 3 = .537 (baseline anxiety; change in R² from step 2 to step 3: p < .001), corrected R² for step 4 = .568 (anxiety change; change in R² from step 3 to step 4: p < .001).

Figure 1.

Heatmap of correlations between variables. The correlation coefficients indicate significant correlations (p<.01), while the color indicates the direction and size of the association between two variables. Blank fields indicate non-significant correlations between variables (p > .01). Sex, EHRA class, organic heart diagnoses, Diabetes Mellitus, Stroke/TIA, Cardioversion, Renal Insufficiency are binary variables (coded as 0 = “male” and 1 = “female” or 0 = “no”, 1 = “yes”, respectively). All other variables are continuous.

Statistical analyses were performed with the Statistical Package for Social Sciences (SPSS) 29; the heatmap was built with R version 4.2.1.

Results

Sample characteristics

The study population’s baseline characteristics are presented in Table 1 for the total sample and split up by EHRA class. We included N = 1,353 patients with complete data for baseline and follow-up quality of life values (mean age 71.3 ± 10.3 years, 33.8 % female). A history of cardioversion was reported in 26.7 % of the study participants, and 17.3 % of patients had a history of catheter ablation. Analyses indicated differences among EHRA class categories for the following variables: sex, AF type, DM, history of cardioversion, history of ablation, history of stroke, age, ejection fraction, HAS-BLED score, generalized and cardiac anxiety, sleep disturbance, stress at home, and noise, as well as HRQoL. Compared to individuals not included in the analyses, included individuals were younger, less likely to be female, had lower generalized and cardiac anxiety, and were more likely to have undergone cardioversion or ablation. Further details regarding these and additional differences can be found in Supplementary File S0.

Table 1.

Sample baseline characteristics of the study population, overall and split up by EHRA class.

			EHRA class (MD=95)
	Total (n = 1353)		1 (n=515)		2 (n=572)		3 or 4 (n=171)
Categorical Variables	n	%	n	%	n	%	n	%	χ 2	df	p
Sex (female) (MD=4)	457	33.8	118	23	224	39.3	87	50.9	56.602	2	<0.001
Single (MD=41)	267	19.7	84	16.9	122	22	39	23.1	5.253	2	0.072
University entrance qualification (MD=45)	714	52.8	273	54.7	302	54.8	93	55.4	0.439	4	0.979
Organic heart diagnoses (MD=54)	970	71.7	370	73.7	416	74.7	121	73.3	0.191	2	0.909
DM (MD=28)	271	20	107	21.1	112	19.7	31	18.3	0.670	2	0.715
AF type (MD =188)									12.354	5	0.015
Paroxysmal	740	54.7	278	62.8	329	62.9	111	66.1
Persistent	280	20.7	93	21	137	26.2	44	26.2
Permanent	145	10.7	72	16.3	57	10.9	13	7.7
History of cardioversion (MD=87)	361	26.7	122	24.8	168	31.2	60	36.1	9.4 20	2	0.009
History of CA (MD=88)	234	17.3	68	13.9	115	21.4	41	24.7	13.920	2	<0.001
Chronic kidney disease (MD=28)	189	14	61	12	85	15	31	18.3	4.653	2	0.098
Stroke/TIA (MD=28)	170	12.6	64	12.6	81	14.3	12	7.1	6.059	2	0.048
New-onset AF (MD=71)	142	10.5	45	9.1	72	12.8	19	11.2	3.698	2	0.157
Continuous Variables	M	SD	M	SD	M	SD	M	SD	H	df	p
Age (MD=19)	71.36	10.3	73.1	9.2	70.4	10.9	68.6	10.7	32.512	2	<0.001
EF (MD=699)	53.71	12.4	52.1	12.9	54.1	12.7	55.7	10.3	7.169	2	0.028
BMI (MD=28)	28.2	5.3	27.6	4.4	28.6	6	28.7	5.8	3.897	2	0.143
CHA2DS2-VASc (MD=83)	3.25	1.7	3.26	1.6	3.3	1.7	3	1.6	3.236	2	0.198
HAS-BLED Score (MD=107)	1.98	1	2.02	1	2	1.1	1.8	1	6.526	2	0.038
GAD-2 (MD=92)	1.4	1.3	1.1	1.2	1.5	1.3	1.6	1.4	40.123	2	<0.001
CAQ-2 (MD=28)	2.9	1.8	2.4	1.8	3.2	1.7	3.6	1.8	90.699	2	<0.001
Sleep disturbance (MD=25)	1.1	0.9	0.9	0.8	1.2	0.9	1.4	0.9	41.73	2	<0.001
Stress at work (MD=978)	1.5	0.7	1.5	0.7	1.5	0.7	1.6	0.8	0.869	2	0.648
Stress at home (MD=32)	1	0.6	0.9	0.5	1	0.6	1.1	0.5	14.134	2	<0.001
Financial stress (MD=29)	0.3	0.5	0.3	0.5	0.3	0.5	0.3	0.6	2.034	2	0.362
Stress from noise (MD=33)	0.5	0.6	0.4	0.6	0.5	0.6	0.5	0.7	26.458	2	<0.001
EQ-5D-5L Index score	0.8	0.2	0.9	0.2	0.8	0.2	0.8	0.2	26.694	2	<0.001

Note. MD = missing data; AF = atrial fibrillation; BMI = body mass index; CA = catheter ablation; CAQ-2 = Cardiac Anxiety Questionnaire 2-item screener; DM = diabetes mellitus; EF = ejection fraction; EQ5D-5L = EuroQoL-5D; GAD-2 = Generalized Anxiety Disorder 2-item screener; TIA = transient ischemic attack. Differences by EHRA class in categorical variables were tested with χ² tests. Differences in continuous variables were tested with Kruskal–Wallis rank sum tests.

Baseline medical status as predictors of HRQoL at follow-up

The total model significance for all models was p < .001. Corrected R² was .528 for model 1 (Table 2), .535 for model 2 (Table 3), .574 for model 3 (Table 4), and .585 for model 4 (Supplementary File S2a; detailed information about corrected R² building each model can be found in the notes of Table 2, 3, 4, Supplementary File 1a–b, and Supplementary File S2a). All regression analyses indicated that baseline HRQoL (p <.001) was positively and age (p < .001) was negatively associated with patients’ HRQoL at follow-up.

Testing the first hypothesis (Table 2, model 1 testing the hypothesis 1a, 1b, 1c), neither the medical predictors at baseline nor female gender were associated with HRQoL at the 1-year-follow-up (for all p > .05).

Baseline anxiety as a predictor of HRQoL at follow-up

Testing hypothesis 2, higher generalized anxiety (GAD-2) at baseline but not cardiac anxiety (CAQ-2) at baseline predicted decreased HRQoL at follow-up (GAD-2: p <.001, CAQ-2: p = .809), independent from confounding variables and patients’ medical status (Table 3, model 2). When analyzing baseline generalized and cardiac anxiety separately, higher baseline generalized anxiety predicted decreased HRQoL at follow-up (GAD-2: p <.001, Supplementary File S1a), while higher baseline cardiac anxiety did not predict decreased HRQoL at follow-up (p = .092, Supplementary File S1b).

Change of anxiety as a predictor of HRQoL at follow-up

Testing hypothesis 3, the regression analysis indicated that the worsening of patients’ generalized anxiety (GAD-2) from baseline to follow-up was associated with decreased HRQoL at follow-up (p < .001). In contrast, the improvement of generalized anxiety over time predicted higher HRQoL at follow-up (p<.001, Table 4, model 3). Of note, these changes over time significantly impacted HRQoL at follow-up, while higher baseline generalized anxiety still was a significant predictor for lower HRQoL at follow-up (p<.001). For cardiac anxiety (CAQ-2), only the worsening of patients’ cardiac anxiety over time was associated with decreased HRQoL at follow-up (p < .001), while the improvement over time (p = .750) and baseline cardiac anxiety (p = .316) were not.

These results were confirmed when we repeated the analyses with the change of patients’ anxiety considered as dimensional variables: change scores for generalized (GAD-2, p<.001) and cardiac anxiety (CAQ-2, p<.001) both indicated that an increase of anxiety over time predicted a decrease in HRQoL at follow-up (Supplementary File S2a, model 4).

Correlations between variables are displayed with a heatmap (Figure 1). Scatterplots showing the association between patients’ HRQoL at 1-year follow-up and GAD-2 at baseline, CAQ-2 at baseline, change from baseline to follow-up for GAD-2 and CAQ-2 can be found in the Supplementary Files S2b–e.

Discussion

We investigated the prospective associations between medical factors, history of cardioversion, and generalized and cardiac anxiety with HRQoL one year later in 1,353 patients with AF. Surprisingly, neither medical factors (e.g., chronic heart disease, diabetes mellitus) at baseline nor a history of cardioversion were associated with impaired HRQoL at 1-year follow-up after controlling for baseline HRQoL. In contrast, elevated levels of generalized anxiety at baseline were independently associated with impaired HRQoL at 1-year follow-up, and changes in generalized anxiety over time were positively associated with HRQoL at follow-up. Finally, worsening cardiac anxiety was associated with impaired HRQoL at follow-up, but improvements in cardiac anxiety were not associated with HRQoL.

The findings that medical factors were not associated with HRQoL at follow-up are consistent with those of Walters et al. [14], who found that among patients with AF and preserved ejection fraction, diastolic function, AF burden, and ventricular rate were not associated with HRQoL. However, our findings are contrary to those of other studies, which identified relationships between impaired HRQoL and diabetes [18,20], renal insufficiency [18], stroke [18], and chronic heart disease (e.g., heart failure, CAD) [17,18]. There are several potential explanations of the discrepant findings. Firstly, existing studies have used a variety of assessment tools for HRQoL, including the SF-36, Atrial Fibrillation Effect on QualiTy-of-life [AFEQT], and Arrhythmia-Specific Questionnaire in Tachycardia and Arrhythmia [ASTA] HRQoL [14,17,20]. Secondly, nearly all analyses that have examined the relationship between medical factors and HRQoL have been cross-sectional [14,17,18,20] in contrast to our current prospective analysis. Cross-sectional analyses often reach different conclusions regarding the association between key explanatory variables and outcomes than those of longitudinal approaches [32]. In our previously published cross-sectional analysis of the ARENA baseline data [24], we also identified associations between CAD and diabetes with impaired HRQoL, the direction of which cannot be determined from cross-sectional data. Finally, a potential explanation could be related to the relatively low physical symptom burden ARENA participants reported (i.e., over 85% reported low AF-related symptom severity), as both frailty [20] and physical burden [33] have been linked to reduced HRQoL in other studies involving AF patients. Furthermore, it may also be related to the fact that we adjusted for baseline HRQoL which appeared to be the strongest predictor for future HRQoL, indicating that early assessment of HRQoL may yield substantially more information on future HRQoL than any of the medical variables.

In contrast to our hypothesis, a history of cardioversion at baseline was not associated with improved HRQoL at 1-year follow-up. In previous studies, successful cardioversion (i.e., maintenance of sinus rhythm by 3 month) was associated with improved HRQoL [21]. However, AF interventions of the presented study were reported retrospectively and their benefits may have vanished over time or been already represented in the baseline HRQoL scores.

In contrast to medical factors, both baseline and changes in generalized anxiety were independently associated with HRQoL at follow-up, even on top of baseline HRQoL. This is consistent with previous studies, which have found that anxiety is prospectively associated with impaired QoL at 6- or 12-month follow-up [34–36]. Additionally, reductions in cardiac anxiety were associated with higher HRQoL at follow-up. This finding is notable, given the lack of research examining the links between cardiac-specific anxiety and HRQoL in patients with AF. In the current study, change in anxiety accounted for 3.9% of the variance in the regression model, while baseline anxiety accounted only for 0.7%, suggesting that change in anxiety over time may have a greater impact on future HRQoL than a single measure. This highlights the importance of regular anxiety assessments at follow-up visits, as this may better identify those individuals at highest risk for poor HRQoL.

Our finding that anxiety is associated with future HRQoL highlights the importance of anxiety in AF. In this population, 28–38% of patients report symptoms of anxiety [34,36], and anxiety persists in a substantial portion (30–36%) of AF patients [34–36]. In addition to affecting HRQoL [36,37], anxiety is associated with an increased risk of AF occurrence after treatment procedures (e.g., cardioversion or circumferential pulmonary vein ablation) [38–40], and an increased risk of ischemic stroke as well as intracranial hemorrhage in AF patients initiating warfarin [41]. These relationships may be explained, at least in part, by psychological [11,42], physiological (e.g., inflammation, sympathetic hyperactivity) [43], and behavioral [42,44] factors.

These findings highlight the importance of early screening for PROs (e.g., anxiety, HRQoL) in AF patients without a separation of psychiatric and somatic healthcare services (e.g., use of screening questionnaires in outpatient clinics). Given a consistent trend towards a higher risk of impaired HRQoL and adverse events in AF patients with comorbid anxiety, future research should investigate whether early identification and management of anxiety and areas of poor HRQoL lead to improvements in long-term HRQoL, health behavior adherence, and medical outcomes.

Considering our findings in context of ESC, EACTS, and EHRA recommendations [2,10], this analysis identified generalized and cardiac-specific anxiety as important predictors of HRQoL in adults with AF. Accordingly, in addition to supporting the measurement of HRQoL at clinical visits, we would recommend screening for anxiety and other clinically relevant symptoms/disorders, such as depression. Though depression was not measured in the ARENA study, it is highly prevalent in cardiac patients and is associated with poor cardiovascular outcomes. Brief, well-validated screening tools (e.g., GAD-2, PHQ-2) are available for both anxiety and depression screening and could easily be incorporated into clinical care. If someone has elevated anxiety or symptoms, we would consider the following: education with standardized materials focused on anxiety or depression in AF patients, psychocardiological consultation from primary physicians/cardiologists (with special psychocardiological training, e.g., integrated into the German residency/fellowship training of physicians), and if the anxiety or depressive symptoms are severe or recurrent, a psychological/psychiatric or psychosomatic consultation is indicated.

ARENA had several notable strengths and limitations. Its most important strengths are its longitudinal study design, large sample size, and inclusion of a heterogenous population of patients with paroxysmal, persistent, or permanent AF, which promotes generalizability. Limitations of ARENA include the absence of a comparator group of participants without AF, lack of a depression measure, and use of a generic measure of HRQoL (EQ-5D-5L), rather than an AF-specific assessment of QoL. Furthermore, we were not able to control for some important PROs (e.g., physical functioning, emotional well-being, cognitive function, exercise tolerance, and ability to work) recommended by ESC, EACTS, and EHRA to be measured in patients with AF as some of these PROs may confound the relation between anxiety and HRQoL by affecting both baseline anxiety symptoms as well as 1-year HRQoL. Additionally, compared to individuals not included in the analyses, individuals included differed significantly regarding several variables. Finally, the reliability estimate of the CAQ-2 was considerably lower than the GAD-2 estimate, and this may have attenuated the associations between 1-year HRQoL and the CAQ-2 baseline/change score. Future research should investigate the predictive value of cardiac anxiety using a more reliable measurement instrument.

In conclusion, our analyses suggest that anxiety is a robust and independent predictor of future HRQoL in individuals with AF, and that reductions in both general and cardiac-specific anxiety are associated with better future HRQoL as well. These findings suggest that measuring anxiety can help to identify individuals at risk for poor future HRQoL and that reducing both general and cardiac-specific anxiety has the potential to improve HRQoL. Additional studies are needed to better clarify these relationships and determine whether anxiety treatment is able to affect HQRoL in this high-risk population.

Supplementary Material

1

Supplementary File S0. Sample baseline characteristics for patients included and not included in the analysis.

2

Supplementary File S1a. Parameters of multiple regression model predicting patients’ quality of life at 1-year follow-up by baseline generalized anxiety independent from patients’ medical status (model 2).

Supplementary File S1b. Parameters of multiple regression model predicting patients’ quality of life at 1-year follow-up by baseline cardiac anxiety independent from patients’ medical status (model 2).

3

Supplementary File S2a. Parameters of multiple regression model predicting patients’ quality of life at 1-year follow-up by the change in anxiety from baseline to follow-up independent from medical status (model 4).

Supplementary File S2b. Scatterplot showing patients’ quality of life at 1-year follow-up and GAD-2 baseline scores.

Supplementary File S2c. Scatterplot showing patients’ quality of life at 1-year follow-up and change in GAD-2 scores from baseline to follow-up.

Supplementary File S2d. Scatterplot showing patients’ quality of life at 1-year follow-up and CAQ-2 baseline scores.

Supplementary File S2e. Scatterplot showing patients’ quality of life at 1-year follow-up and change in CAQ-2 scores from baseline to follow-up.

Highlights.

Screening of anxiety might identify AF patients at elevated risk for impaired HRQoL

Anxiety is a robust and independent predictor of future HRQoL in individuals with AF

None of the medical predictors were associated with HRQoL at 1-year FU in AF patients

Treatment of anxiety symptoms might lead to improvements of HRQoL in AF patients

Data Availability Statement

The ARENA dataset generated and/or analyzed for the present article is available upon request to the corresponding author.