The study included 113 HF patients {median age 64 years [interquartile range (IQR) 58–69]; 57.52% male}. Patients were divided into two groups:

In the next step, propensity score matching (PSM) was used to construct an artificial control group by matching AF participants one-to-one to participants without AF. Data were collected from 23 AF patients and 23 age-matched non-AF patients with HF.

The study's inclusion and exclusion criteria are defined below.

Every patient underwent a clinical assessment. The following data were included in this analysis:

In all patients, the following diagnostic procedures were also performed:

Statistical analysis was conducted using the R language (4.02). All continuous variables were expressed as median and IQR (Q25–Q75). Categorical variables were expressed as the number of observations (N) and the corresponding percentage (%). The Mann–Whitney U test was used to compare continuous variables. The χ² test was used to compare the qualitative data between the groups. Fisher's exact test for independence (in lower numbers N < 5) was used. All statistical tests were two-sided. Results were considered statistically significant when p < 0.05. The tables present only variables that differ or are selected in the context of the article.

Logistic regression was used to evaluate the impact of the factors on AF occurrence, in both univariate and multivariate designs. The odds ratio (OR) and 95% confidence interval (95% CI) were reported for each factor in univariate and multivariate analyses. All variables significantly associated (p < 0.05) with AF occurrence in the univariate model were included in the multivariate regression analysis to determine the independent predictors of the AF occurrence. Tables with the results of univariate and multivariate analyses show only statistically significant data. PSM was used to construct an artificial control group by matching each AF unit with a non-AF unit of similar characteristics. Due to a high imbalance between genders in group with AF on admission, correctly performing PSM on both sex and age was not possible. Therefore, we have decided to use only age in PSM. Data were collected from 23 AF patients and 23 age-matched non-AF patients. Using these matches, the researchers could estimate the impact of the factors on AF occurrence.

Based on the available echocardiography results (N = 113), the mean LVEF was 44.0% (IQR 39.0–50.0). Of those patients, 26.55% (N = 30) had HFpEF, 46.90% (N = 53) had HFmrEF, and 26.55% (N = 30) had HFrEF. Ischemic etiology of HF was present in 43.36% of all patients (N = 49). Of all patients with baseline rhythm data, 23 (20.35%) had AF, and 90 (79.65%) had SR in ECG. The studied population in both groups was burdened with the most common comorbidities including arterial hypertension, coronary artery disease, diabetes mellitus, valve diseases, and hyperlipidemia. Detailed characteristics of the study population are presented in Table 1. Prior to PS matching, there were no statistically significant differences in sex between AF and SR groups [15 men (65.22%) vs. 50 men (55.56%), p = 0.403]. No significant differences were also observed with regard to systolic and diastolic blood pressure between the SR and AF groups (Table 1). Patients with AF were older (med. 66 vs. 64 years; p = 0.039), with higher NYHA class (III vs. II; p < 0.001) and lower KCCQ overall summary score (52.60 vs. 73.96 points; p = 0.002) (Table 1). In ECG, the AF group had a higher HR (med. 80 vs. 70 bpm, p = 0.004), wider QRS (med. 112 vs. 98 ms; p = 0.004), and longer QTc (med. 467.5 vs. 437 ms; p = 0.005) duration (Table 1).

After PS matching, the 23 matched pairs were analyzed for differences in the baseline characteristics (Table 2). There were no significant discrepancies in the overall characteristics of the groups after PS matching compared with the previous analysis (Table 2).

Regarding the cardiovascular background, the proportion of patients with coronary artery disease (52.2 vs. 26.09%; p = 0.025) and ischemic etiology of HF (47.78 vs. 21.74%; p = 0.024) was higher in the SR group, whereas the proportion of patients with hypertension (95.65 vs. 67.78%; p = 0.007) and with valvular disease, i.e., mitral regurgitation (69.57 vs. 43.33%; p = 0.025) and tricuspid regurgitation (65.22 vs. 27.78%; p = 0.001), was higher in the AF group (Table 1). There were no significant differences in the use of renin–angiotensin system blockers (ACEI and ARB), BB, or MRA between AF and SR groups (Table 1). However, diuretics were used more frequently in the AF group (78.26 vs. 52.22%; p = 0.033) (Table 1).

AF group had greater LAVI (med. 65 vs. 40 mL/m²; p < 0.001) and LVDD (med. 61 vs. 54 mm; p < 0.001) than had the SR group. TAPSE was lower (med. 18 vs. 22 mm; p < 0.001) and RVPs were higher in the AF group than in the SR group (med. 38 vs. 28 mmHg; p < 0.001) (Table 3). There were no statistically significant differences in LVEF between the AF and SR groups (Table 3).

Median VO₂max (0.92 vs. 1.26 L/min, p = 0.016), VO₂max/kg (11 vs. 15 mL/min/kg; p < 0.001), and VO₂%Norm (62.5 vs. 70.0%; p = 0.010) were significantly lower in the AF group vs. the non-AF group (Table 4). RER in the study population was 1.080 (IQR 0.98–1.14) and was lower in the AF than in the non-AF group (p = 0.001) (Table 4). VO₂ at AT was also lower in the AF group (0.735 vs. 1.00 L/min; p = 0.033) (Table 4). We also found VE/VCO₂ to be higher in patients with AF vs. patients without AF (med. 33.85 vs. 32.20; p = 0.049) (Table 4). Moreover, VO₂/HR at peak (8.5 vs. 11 mL/beat; p = 0.038) was significantly impaired in the AF group (Table 4). The other differences in characteristics were not statistically significant and are presented in Table 4.

Furthermore, compared with patients in SR, patients with AF had significantly lower exercise capacity measured as 6MWTD (med. 292.35 vs. 378.4 m; p = 0.001) (Table 1).

Patients with AF had a significantly higher BMI (med. 32.02 vs. 28.51 kg/m², p = 0.017) than their counterparts (Table 5). Moreover, compared with the SR group, the patients with AF had significantly higher BIA parameters, including FM (med. 30.07 vs. 22.50 kg; p = 0.007), FM% (med. 37.0 vs. 27.9%; p = 0.014), ECW (med. 18.5 vs. 18.2 kg; p = 0.031), and MA (med. 66 vs. 59 years; p = 0.001), and a lower H₂O-TBW% (45.7 vs. 50.0%; p = 0.022) (Table 5). The other differences in characteristics were not statistically significant and are presented in Table 5.

Patients with AF showed a higher concentration of sodium (med. 142 vs. 141 mmol/L, p = 0.020) (Table 6) and uric acid (med. 7.10 vs. 6.05 mg/dl, p = 0.001) and a significantly higher level of biomarkers associated with inflammation (CRP and hs-CRP) and myocyte stress and injury (NT-proBNP and hs-cTnT) than non-AF individuals (Table 6). The other characteristics were similar between the groups (Table 6).

Prior to matching, the univariate regression analysis revealed the significant variables affecting AF occurrence in HF patients (Table 7). A few significantly associated parameters were then used in the multivariate regression analysis. A multiple logistic regression model showed that increased body mass, with higher BMI (OR 1.23 per unit increase; 95% CI: 1.09–1.46; p < 0.001), enlargement of the left atrium, greater LAVI (OR 1.07 per unit increase; 95% CI: 1.04–1.14; p = 0.03), lower TAPSE (OR 0.74 per unit increase; 95% CI: 0.46–0.81; p = 0.03), and lower H₂O-TBW% (OR 0.90 per unit increase; 95% CI: 0.81–0.98; p = 0.03) were independently related to AF in patients with HF (Table 7).

We estimated the impact of these factors on AF occurrence in univariate and multivariate regression models also after PS matching (Table 8). The univariate regression analysis revealed similar variables affecting AF occurrence in HF patients compared with results before PS matching (Table 8). Only the laboratory results showed differences, and the uric acid and troponin T were the only parameters significantly associated with AF after PS matching. In a multiple logistic regression model after PS matching, only an increase of AF occurrence was significantly associated with greater LAVI (OR 1.091; 95% CI: 1.036–1.172; p = 0.004). The other variables were similar compared with results before PS-matching analysis, but not statistically significant (Table 8).

The 20% incidence of AF in our population is similar to that previously reported in HF population (29, 30). Numerous previous studies have demonstrated that AF is associated with exercise intolerance, mainly in patients with associated heart disease (31–33). In our study, compared with patients in SR, patients with AF had significantly lower exercise capacity measured by 6MWTD (med. 292.35 vs. 378.4 m; p = 0.001). Similarly, results were obtained by Fung et al. (33) (mean 6MWTD in AF 279 ± 66 vs. 338 ± 86 m in SR) and in the RELAX study (32) (mean 6MWTD in AF 283 ± 107 vs. 313 ± 109 in SR).

Other studies reported that peak VO₂ is 10–20% lower in patients with HFrEF (14, 31) and significantly reduced in HFpEF (16, 32) patients with AF. Our research work confirmed these results. In our analysis, HF patients with AF had impaired peak exercise tolerance, reflected in lower VO₂, lower VO₂/kg, and lower VO₂/HR at peak exercise as compared with those in the SR group. The lower cardiac output during AF due to loss of atrial contraction, irregular ventricular rhythm, and impaired ventricular filling is the most important during the peak of exercise when maximum systolic reserve is used (34). Linde-Edelstam et al. (35) reported that the significance of atrial contraction in LV filling decreases with increasing blood flow velocity and with increasing workload. Moreover, the pressure in the left atrium determines the share of atrial contraction in the LV filling, and an increase in the wedge pressure in pulmonary capillaries reduces the power of atrial contraction on the cardiac output (35).

Moreover, AF was associated in our study with impaired ventilatory efficiency (higher VE/VCO₂). The mean VE/VCO₂ slope was significantly higher in HF patients with AF compared with patients in SR efficiency also in previous studies (31, 32, 36).

Unexpectedly, we observed in our HF patients with AF compared with SR patients lower peak VO₂ at AT, which is in contrast with a few previous reports (31, 36). The paradox of a lower VO₂ at AT in our study could be related to the lower chronotropic response to exercise in our patients resulting from chronic use of BBs, also during hospitalization. This observation raises uncertainties about the use of AT data to define performance and prognosis of HF patients with AF supporting the need of multiparametric evaluation of CPET in this disease. In a study of Agostoni et al. (31), among all variables examined at AT, only HR was independently associated with AF. In AF, the increased HR response during exercises is necessary to maintain cardiac output; and the correlation between VO₂ and cardiac output in HF is particularly strong at AT (31). Lewis et al. (37) demonstrated that a reduction in the ventricular rate in AF was associated with a small increase in stroke volume but was offset by rate-related reductions in cardiac output during exercise.

Compared with the SR group, the patients with AF in our study had a significantly greater BMI and higher body FM in BIA. The links between higher BMI and the risk of the new onset AF have been evidenced previously (overweight increased risk by 18–54% and obesity increased risk by 1.6- to 2.4-fold) (38, 39). There was a 28% increase in the relative risk per 5-unit increase in BMI, a 9% increase per 5-kg increase in body FM, and a 10% increase per 5-kg increase in body weight (39). In addition, REVERSE-AF (39) study has found that weight loss in overweight and obese individuals is associated with a reduced burden and reversal of AF. The mechanisms by which obesity increases the risk of cardiovascular diseases (CVDs) include changes in body composition [increases the percentage of epicardial adipose tissue (EAT)], which can affect hemodynamics and changes in the structure of the heart (left size enlargement and fibrosis, atrial inflammation and lipid infiltration, and changes in atrial electrophysiological properties) (40). Pro-inflammatory cytokines produced by adipose tissue could induce cardiac dysfunction and promote the formation of atherosclerotic plaques. When obesity and coronary heart disease (CHD) coexist, people with class I obesity have a better prognosis than those with normal weight or underweight. This phenomenon has been dubbed the “obesity paradox” (41). But this effect on prognosis in AF patients is controversial. In a study of elderly outpatients with AF, being underweight has been related to worse outcomes (42). On the other hand, in FANTASIIA Registry (43), obesity was associated with more comorbidities, but not with poor prognosis. Further studies are needed to determine the importance of obesity for the AF prognosis in patients with HF.

In our study, patients with AF were more likely to be older and more often were treated with diuretics, and in consequence, they were more dehydrated, with lower H₂O-TBW% (p = 0.022), higher sodium level (p = 0.02), and higher creatinine concentration, but not statistically significant (p = 0.130). The lower H₂O-TBW% was also independently related to AF presentation in our HF patients. Although AF has mainly been described with hyponatremia in patients with systolic HF in previous study (44, 45), one study by Timilsina et al. (46) described AF in association with the correction of chronic hypernatremia. Higher concentration of sodium is seen more often, for example, in patients treated with diuretics (47). Dehydration may be from gastrointestinal losses (gastroenteritis), renal losses (commonly due to diuretic therapy), and insufficient fluid intake, where behavior may play an important role (48). According to National Health and Nutrition Examination Survey (NHANES) survey data (49), US adults are drinking less water than is recommended. The evaluation of hydration status using body mass analysis and concentration of sodium and creatinine should be the rule in HF patient with AF. In a study of Swerdel et al. (48), the rate of dehydration in AF patients was ~9%. In this study, 18- to 80-year-old patients with AF and with comorbid dehydration had a 60% higher risk of ischemic stroke compared with AF patients without comorbid dehydration [absolute risk reduction (ARR) 1.60, 95% CI 1.28–2.00] within 10 days post-AF discharge and 34% higher risk of ischemic stroke in days 11–20 post-AF discharge (ARR 1.34, 95% CI 1.04, 1.74) (48). In a study of Groh et al. (50), the most commonly reported triggers of AF were alcohol (35%), caffeine (28%), exercise (23%), and dehydration (21%). Further studies are needed to determine the importance of various BIA parameters, including hydration status, for the occurrence of AF and prognosis in patients with HF.

The presented analysis is to our best knowledge the first analysis of HF patients to evaluate the overall impact of AF on exercise capacity and quality of life and to assess the relations between body mass compartments and AF occurrence in HF patients. This analysis pays attention to the fact that HF patients with AF are older, less thirsty, and more often treated with diuretics, and as a result, they are more dehydrated, with lower percentage of H₂O-TBW, higher creatinine concentration, and higher sodium level. We demonstrated that dehydration also played an important role in HF patients with AF, and hydration status should be considered in overall HF patient care.

There are several limitations of the present study, which included a small study population without healthy controls. Of 113 patients, 23 had AF, the remaining 90 were in SR, and paired analysis was possible in 23 cases only. This was also a non-randomized study. Clinical background and sample sizes were dissimilar for AF and SR, because of the natural incidence of AF in HF population. However, the data were prospectively collected from the patients. Moreover, we conducted only ECG rhythm-related statistical analysis but did not perform analysis by EF value (HFrEF vs. HFpEF, and HFmrEF). In addition, the patient population was limited to those hospitalized due to HF in our department, which could have caused referral bias because patients referred for hospitalization are not representative of the general AF and HF population. Because the disease severity in our patients was mild or moderate, without HF and AF, the results should be carefully interpreted when applied to different populations. In addition, the study design was limited regarding the evaluation of the effect of medications. Among them, the medication of BBs should be carefully considered because it might significantly affect the behavior of HR. Finally, the present study only included stable patients who were able to undergo CPET. Therefore, the present results cannot be extrapolated to unstable AF patients or with decompensated HF. To summarize, the present results need to be treated with caution and should be reproduced in a larger HF population with AF, preferably in a prospective controlled clinical trial, to confirm its results.

Despite these limitations, the presented analysis is to our best knowledge the first analysis of HF patients to evaluate the overall impact of AF on exercise capacity and quality of life and to assess the relations between body mass compartments and AF occurrence in HF patients. We demonstrated that dehydration also played an important role in HF patients with AF and hydration status should be considered in overall HF patient care.

Moreover, the paradox of a lower VO₂ at AT in AF patients in our study raises uncertainties about the use of AT data to define performance and prognosis of HF patients with AF, supporting the need of multiparametric evaluation of CPET in this disease.