Frailty and comorbidity burden in Atrial Fibrillation

Background With the aging of the population, the characterization of frailty and comorbidity burden is increasingly taking on particular importance. The aims of the present study are to analyze such conditions in a population affected by Atrial Fibrillation (AF), matching it with a population without AF, and to recognize potential independent factors associated with such common cardiovascular disease. Methods This study included subjects consecutively evaluated over 5 years at the Geriatric Outpatient Service, University Hospital of Monserrato, Cagliari, Italy. A sum of 1981 subjects met the inclusion criteria. The AF-group was made up of 330 people, and another 330 people were randomly selected to made up the non-AF-group. The sample was subjected to Comprehensive Geriatric Assessment (CGA). Results In our sample, severe comorbidity burden (p = 0.01) and frailty status (p = 0.04) were significantly more common in patients with AF than without AF, independently on gender and age. Furthermore, the 5-years follow-up demonstrated that survival probability was significantly higher in AF-group (p = 0.03). The multivariate analysis (AUC: 0.808) showed that the presence of AF was independently positively associated with a history of coronary heart disease (OR: 2.12) and cerebrovascular disease (OR: 1.64), with the assumption of Beta Blockers (OR: 3.39), and with the number of drugs taken (OR: 1.12), and negatively associated with the assumption of antiplatelets (OR: 0.09). Conclusions Elderly people with AF are frailer, have more severe comorbidities, and take more drugs, in particular beta blockers, than people without AF, who conversely have a higher survival probability. Furthermore, it is necessary to pay attention to antiplatelets, especially in AF-group, in order to avoid dangerous under- or over-prescriptions.

/fpubh. . Nowadays, in particular, it is used to assess "frailty", a common medical word, whose interpretation is yet not univocal (10)(11)(12). It continues the literature continues to discuss "phenotypes" rather than "definitions": the phenotypic model proposed by Fried et al. (13,14) characterize frailty as a clinical syndrome with 3 or more criteria among weight loss, exhaustion, reduced grip strength, reduced walking speed and physical activity. Anyway, it is known that a categorization of pre-frail and frail people is necessary to stratify different needs for intervention (15). The concept of frailty is led to the concept of multimorbidity, which does not have to be considered as a "long list of illnesses", but rather an indicator of burden (16), mortality (17), reduced quality of life (18) for elderly people. Anyways, multimorbidity is indeed associated with the most common geriatric syndromes, such as cognitive impairment and sarcopenia, and age-related pathologies (19-21). Among them, in cardiovascular medicine, one of the most represented in elderly is Atrial Fibrillation (AF). This common condition is an arrythmia which can be due to a number of factors including genetics, but also aging and lifestyle (22). It is associated with higher risk of hospitalization and higher mortality in elderly (23,24

Aim of the study
The primary aim of this study is to compare the frailty status and the comorbidity burden with the presence/absence of AF in a population of subjects aged 65 years or older, and to verify their impact on total mortality.
The secondary aim of this study is to consider which CGA domains, comorbidities and drugs are independently associated with AF.

Design of the study
This observational cross-sectional study included subjects consecutively evaluated at the Geriatric Outpatient Service of the University Hospital of Monserrato, Cagliari, Italy, over a 5years period.

Inclusion criteria
Age ≥ 65 years; having been subjected to CGA.

Exclusion criteria
Age < 65 years; age ≥ 65 years with acute conditions that contraindicated the CGA's execution; informed consent not provided.
One thousand nine hundred and eighty-one subjects met the inclusion criteria.
Non-valvular AF was present in 330 subjects (AF group): we performed a propensity score model to randomly match them with 330 non-AF controls (non-AF group) based on gender and age (see Section Statistical analysis).
We obtained a final sample of 660 subjects, who were followedup for a 5-years period.

Assessment
The enrolled subjects were evaluated with:  • Charlson Comorbidity Index (CCI) (32) for comorbidity burden's assessment • FRAIL scale (14) for the categorization of the frailty level.
The abovementioned tests were administered by trained geriatricians in outpatient setting.

Statistical analysis
Variables were expressed as means and standard deviations (SDs) or in percentages (%), were appropriate. In order to randomize cases and controls we used the propensity score: we firstly identified AF as "classification variable", then, given a set of covariates, namely age and gender, we performed the test, obtaining neglectable 95% C.I. Kolmogorov-Smirnov test was used to check the distribution of quantitative data. Student's t-test test was used to compare continuous variables; chi-squared test (χ 2 ) was used to compare qualitative variables. Correlations between variables were expressed using Pearson's rho (r). Multivariate analysis was performed with a logistic regression-stepwise (p-values > 0.1 were excluded by the model): its results were expressed as Odds Ratios (ORs) and confidence intervals (C.I.). Kaplan-Meier curves were designed in order to estimate the survival probability: in particular, mean survival times, expressed as Areas Under the survival Curves (AUC) from 0 to 5 years, were reported with their 95% C.I. The .
comparison of survival curves between the two groups was studied with the Logrank test, and expressed as χ 2 and C.I., while the differences in time of occurring event were expressed as Hazard Ratios (HRs). The results are reported indicating p-values in reference to 95% C.I.
MedCalc software (Version 19.5, Ostend, Belgium) was used for the statistical analysis.

Results
The study included 1981 people aged 65 years or more. The propensity score model brought a final sample of 660 subjects, divided in two groups (AF and non-AF), of whom 414 women (62.7%), with average age of 81.2 years (SD: 6.5) (Figure 1). Table 1 summarizes the scores achieved by the two groups in every CGA tests, and the most common co-morbidities and drugs taken.
AF was then considered as dependent variable in a multivariate logistic regression; CGA domains (cognitive status, mood, autonomy, physical performances, nutritional status, comorbidity burden, and frailty), co-morbidities and drugs taken were considered independent variables (Table 3). The Area Under the ROC Curve (AUC) was 0.808, with a standard error of 0.0203 and a 95% C.I. from 0.769 to 0.844. The regression model demonstrated that the presence of AF was independently associated

Discussion
The increasing elderly population is often frail and multimorbid (10), and CGA (6) can early recognize and categorize such common conditions. Among age-related pathologies, one of the most represented is AF, associated with higher risk of hospitalization and mortality (23,24).
The primary aim of our study was to compare the frailty status and the comorbidity burden with the presence/absence of AF in a population of subjects aged 65 years or older. The secondary aim was to consider which domains, comorbidities and drugs were independently associated with AF.
Our data demonstrated that severe comorbidity burden (p = 0.01) and frailty status (p = 0.04) were significantly more common in patients with AF than without AF, although their poor collinearity, and such difference did not depend on gender and age, according to the case-control matching performed in our sample. Moreover, non-AF patients were more likely to survive (HR: 1.27) than AFs. These results are consistent with the literature (33)(34)(35)(36), and show AF being a disease of serious impact on global health status in elderly patients. In our sample, overall mortality was higher compared to the literature (37), in accordance with our inclusion criteria, and the ensuing abovementioned burden.
Then, we performed a multivariate analysis to characterize the weight of different co-variates on AF. We did not include anticoagulant drugs assumption because of the obvious association with AF, as can be also seen by χ 2 analysis (p < 0.0001). The regression model showed an independent association with coronary and cerebrovascular diseases (ORs: 2.12 and 1.64, respectively), consistently with the literature and with AF's pathophysiology, likewise to Beta Blockers intake (OR: 3.39). A data so far never emerged in the scientific literature (38,39) was the inverse association between AF and antiplatelets: our data showed that patients without AF have 91% more chance of taking such drugs. To deepen this result, we must consider that, in AF group, 34% of the patients with a history of coronary or cerebrovascular disease did not take any antiplatelet drugs, while only 9.5% of non-AF group did not take them. Moreover, 10% of AF group and 23.5% of non-AF group was taking antiplatelets in primary prevention (40). We can thus highlight the possible tendency to overprescribe (41,42) such drugs in primary prevention slightly more in absence than in presence of AF, and, on the other hand, .
/fpubh. . the reduced tendency to prescribe them in presence of AF, because of the known increased risk of bleeding (43,44). Our study demonstrates that elderly people with AF are frailer, have more severe comorbidities, and take more drugs, in particular beta blockers, and besides have a higher probability to die. Furthermore, it denounces a dangerous antiplatelets underprescription, that might be carefully considered in such population with additional thrombotic risk factors (45).
Obviously, we recognize the study presents some limitation: firstly, its design did not allow to explore causality among the outcomes and the independent variables; moreover, it is monocentric, and it did not take into account potential geographical differences on FA management; lastly, it did not consider the causes of death, which would have been useful in order to enrich the strength of the results.

Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethics statement
The studies involving human participants were reviewed and approved by Institutional Review Board (or Ethics Committee) of the University of Cagliari. The patients/participants provided their written informed consent to participate in this study.

Author contributions
FS, AP, and AM were principal investigators and contributed to the study design and data analyses. FS, AP, GD, and MS contributed to data collection. FS and AM contributed to the interpretation of the findings and wrote the manuscript. All authors read and approved the final version of the manuscript.