In patients with chronic heart failure (HF), cognitive impairment (CI) is associated with poorer treatment adherence and higher readmission and mortality rates. Previous studies suggest that atrial fibrillation (AF) could impair cognitive function. This study sought to assess the association between permanent AF (permAF) and CI in patients with HF. We evaluated cognitive function in 881 patients with stable HF (73 ± 11 years, 44% women, 48% with preserved ejection fraction) using the Mini-Mental State Examination test (n = 876) and the Pfeiffer’s Short Portable Mental Status Questionnaire (n = 848). CI was defined as a Mini-Mental State Examination score <24 or Short Portable Mental Status Questionnaire (errors) >2. The independent association between permAF and CI was assessed by binary logistic regression analysis. A total of 295 patients (33.5%) had CI, in 5.1% of cases moderate/severe. Patients with permAF had more frequently any degree of CI (43% vs 31%), and moderate/severe CI (8% vs 5%). In the multivariate analysis, CI was associated with permAF (odds ratio 1.54, 95% C.I. 1.05 to 2.28), an older age, female gender, diabetes mellitus, chronic kidney disease, previous stroke, New York Heart Association class III/IV, and lower systolic blood pressure. No interaction was found for AF and CI between patients with reduced and preserved ejection fraction. In conclusion, the presence of permAF is independently associated with CI in patients with HF, both with reduced and preserved ejection fraction. Given the clinical impact of CI in the HF population, active assessment of cognitive function is particularly warranted in patients with HF with permAF.
Chronic heart failure (HF) is a major health problem affecting 1% to 2% of the adult population and >8 to 10% of patients aged ≥75 years. Given its increasing prevalence with age, HF is often associated with other geriatric conditions. Cognitive impairment (CI) seems to be particularly common, with a prevalence that ranges from 25% to 70% among different series depending on the characteristics of the HF populations and the methods used to assess cognition. Loss of cognitive performance interferes with patient self-care and treatment adherence, which ultimately translates into higher rates of hospital admission and increased mortality in patients with HF. Atrial fibrillation (AF), the most common sustained arrhythmia, is also particularly prevalent in the elderly and in the HF population, where it may be present in up to 25% to 40% of patients. Although with discordant results among series, AF itself has also been related to cognitive decrease in the general population, with a potential 1.4-fold increased risk for CI compared to patients with no history of the arrhythmia. A very recent work indicates that AF seems to be associated with worse global cognitive performance also in patients with HF. However, whether this translates into higher rates of actual CI, and whether this is extensible for both populations with reduced ejection fraction (HF-REF) and preserved ejection fraction (HF-PEF) remains to be established. Therefore, we sought to determine the overall prevalence of CI in a large population including both subtypes of HF and the potential incremental risk added by the presence of AF.
Methods
The study comprised a population of 1,236 patients recruited prospectively in our institution from January 2004 to December 2013, with confirmed diagnosis of HF, either HF-REF or HF-PEF (defined as left ventricular ejection fraction ≥45%), according to the European Society of Cardiology criteria. All patients were in stable condition. Patients with significant primary heart valve disease, infiltrative myocardial or pericardial disease, hypertrophic cardiomyopathy, and those with hemoglobin levels <8.5 g/dl or severe co-morbidities, including active malignancies, chronic liver disease, or extra-cardiac disease with a life expectancy <1 year were excluded. Additionally, for the purpose of this study, patients with serious psychiatric illness or inability to undergo neuropsychological testing were also excluded. The study was approved by the local institutional review board and followed the national and international guidelines (1975 Declaration of Helsinki) and national legal regulations on Personal Data Privacy (15/1999). All participants provided written informed consent to participate in the study.
Data on demographic, medical past history, and clinical parameters, including those regarding HF characteristics and status, were collected for all patients. Previous AF was classified as absent, paroxysmal/persistent, or permanent. A preliminary analysis with our cohort indicated a major potential association between permanent AF (permAF) and CI, with no such effect seen for paroxysmal or persistent AF. Consequently, this study was designed to assess the potential association between permAF and CI in patients with stable HF.
Cognitive performance was determined using 2 neuropsychological tests, previously validated for the Spanish language : the Folstein’s Mini-Mental State Examination (MMSE) test, completed in 876 patients (99.4%), and the Pfeiffer’s Short Portable Mental Status Questionnaire (SPMSQ), completed in 848 patients (96.3%). Total MMSE score (0- to 30) was corrected for age and educational level as previously described ; CI and moderate/severe CI were defined by a corrected MMSE score <24 and <10, respectively. Total number of errors in the SPMSQ (0 to 10) were also corrected for educational level, and CI and moderate/severe CI defined by ≥3 and ≥5 corrected errors, respectively. CI and moderate/severe CI were established when criteria were met with either one of the 2 tests. All tests were conducted by 2 specialized nurses under the supervision of the neuropsychological team.
Results throughout the manuscript are reported as mean (SD) or frequency (%) as appropriate. Comparisons between groups were performed with the Student t test or chi-square analysis. The association between CI and clinical variables, including permAF, was evaluated using binary logistic regression models. Variables were included in the multivariate analysis with the use of a backward (conditional) stepwise procedure with a criteria of p <0.05 for inclusion and p >0.10 for removal from the model and also according to clinical relevance. The results were subsequently confirmed by manually removing from the initial model the least significant potential confounder at a time. The potential interaction of permAF and HF subtype or ejection fraction was assessed by introducing the permAF × HF subtype and the permAF × ejection fraction factors in the model. Predicted probabilities by binary regression models were constructed according to permAF and other variables found to be remarkably related to CI (age and gender). A p value of <0.05 was considered statistically significant. Statistical analyses were performed using the SPSS 18.0 software package (SPSS Inc., Chicago, Illinois).
Results
The final population consisted of 881 patients with stable HF and available information on cognitive performance, educational level, and AF ( Figure 1 ). General characteristics of the study population are summarized in Table 1 (left). Approximately half of the patients (48.1%) had HF-PEF. A total of 187 patients (21.2%) had a history of permAF before inclusion. As presented in Table 1 (right), patients with CI differed with respect to those without CI in that they were older, more frequently women and had higher rates of diabetes mellitus, previous stroke, chronic kidney disease (CKD), and permAF. No differences were observed between patients with and without CI with regards to HF subtype, although patients with CI had overall worse functional status, manifested by higher rates of advanced (III to IV) New York Heart Association (NYHA) class.
| General characteristics | Overall (n = 881) | Cognitive Impairment (n = 295) | No cognitive Impairment (n = 586) | p-value |
|---|---|---|---|---|
| Women | 384 (43.6%) | 181 (61.4%) | 203 (34.6%) | <0.001 |
| Age (years) | 72.6 (11.1) | 76.3 (9.2) | 70.3 (11.7) | <0.001 |
| Body mass index (Kg/m2) | 28.3 (5.8) | 28.5 (6.2) | 28.1 (5.6) | 0.204 |
| Smoker | 459 (52.1%) | 117 (39.8%) | 342 (58.6%) | <0.001 |
| Hypertension | 716 (81.3%) | 243 (82.4%) | 473 (80.7%) | 0.309 |
| Hypercholesterolemia | 506 (57.4%) | 181 (61.4%) | 325 (55.5%) | 0.055 |
| Diabetes mellitus | 430 (48.8%) | 161 (54.6%) | 269 (46.0%) | 0.010 |
| Prior myocardial infarction | 226 (25.7%) | 65 (22.9%) | 161 (28.1%) | 0.060 |
| Prior stroke | 106 (12.0%) | 49 (16.6%) | 57 (9.7%) | 0.003 |
| Chronic obstructive pulmonary disease | 204 (23.2%) | 65 (22.0%) | 139 (23.7%) | 0.319 |
| Chronic kidney disease ∗ | 502 (57.0%) | 201 (68.1%) | 301 (51.4%) | <0.001 |
| Permanent atrial fibrillation | 187 (21.2%) | 80 (27.1%) | 107 (18.3%) | 0.002 |
| Atrial fibrillation duration (months) | 34.1 (51.5) | 32.8 (53.0) | 36.3 (48.8) | 0.503 |
| Heart failure subtype | 0.176 | |||
| Reduced ejection fraction | 457 (51.9%) | 146 (49.5%) | 311 (53.1%) | |
| Preserved ejection fraction | 424 (48.1%) | 149 (50.5%) | 275 (46.9%) | |
| Heart failure etiology | 0.042 | |||
| Ischemic | 340 (39.0%) | 114 (39.9%) | 236 (40.6%) | |
| Hypertensive † | 302 (34.7%) | 113 (38.7%) | 188 (32.4%) | |
| Idiopathic | 85 (9.8%) | 17 (5.8%) | 66 (11.4%) | |
| Other | 144 (16.5%) | 48 (16.3%) | 91 (15.5%) | |
| New York Heart Association class | <0.001 | |||
| I | 111 (13.3%) | 14 (5.1%) | 97 (17.4%) | |
| II | 383 (45.9%) | 108 (39.3%) | 275 (49.2%) | |
| III | 278 (33.3%) | 119 (43.3%) | 159 (28.4%) | |
| IV | 62 (7.4%) | 34 (12.4%) | 28 (5.0%) | |
| Treatment, | ||||
| Angiotensin converting enzyme inhibitors or angiotensin receptor blockers | 626 (71.1%) | 201 (68.4%) | 425 (72.6%) | 0.107 |
| Beta-blockers | 765 (86.8%) | 244 (82.7%) | 521 (88.9%) | 0.008 |
| Diuretics | 802 (91.0%) | 277 (93.9%) | 525 (89.6%) | 0.021 |
| Statins | 522 (59.3%) | 184 (64.8%) | 338 (59.0%) | 0.059 |
| Digoxin | 112 (12.7%) | 48 (16.3%) | 64 (10.9%) | 0.017 |
| Amiodarone | 89 (10.1%) | 27 (9.2%) | 62 (10.6%) | 0.295 |
| Oral anticoagulants | 346 (39.3%) | 129 (44.5%) | 217 (37.5%) | 0.028 |
∗ Chronic kidney disease was defined by an estimated glomerular filtration rate <60 ml/min/1.73m 2 .
† Hypertensive etiology was considered when heart failure was attributable exclusively to chronic high blood pressure with no other potential causes of heart failure being identifiable.
Figure 2 summarizes the global cognitive performance of the entire population. A total of 295 of 881 patients (33.5%) met criteria for CI. CI was similarly prevalent in patients with HF-REF (146 of 457, 31.9%) and HF-PEF (149 of 424, 35.1%, p not significant [NS]). Notably, the rate of moderate/severe CI was low (45 of 881, 5.1%), both in the HF-REF (19 of 457, 4.2%) and in the HF-PEF subgroups (26 of 424, 6.1%, p NS, Figure 2 ). Table 2 highlights the different diagnostic performance of both neuropsychological tests. Compared with the SPMSQ, the MMSE test was more prone to establish a diagnosis of CI, with 128 additional patients falling into the CI category, although being more restrictive to define moderate/severe CI (0 cases with the MMSE test vs 45 cases with the SPMSQ test).
| Cognitive impairment criteria | |
|---|---|
| By MMSE or SPMSQ* (n=881) | 295/881 (33.5%) |
| Moderate/severe | 45/881 (5.1%) |
| By MMSE (n=876, score < 24) | 249/876 (28.4%) |
| Moderate/severe CI (score < 10) | 0/876 (0%) |
| By SPMSQ (n=848, errors > 2) | 133/848 (15.7%) |
| Moderate/severe (errors > 4) | 45/848 (5.1%) |
Patient characteristics according to the presence or absence of permAF are presented in Table 3 . Compared to patients without permAF, patients with permAF were older and had more frequently HF-PEF. Of note, a history of previous stroke was similarly present in patients with permAF and in those without. As expected, patients with permAF were more frequently under treatment with digoxin and oral anticoagulants (OACs), whereas the use of amiodarone in this group was anecdotic given the permanent status of the arrhythmia ( Table 3 ).
| General characteristics | Permanent atrial fibrillation | p-value | |
|---|---|---|---|
| No (n= 686) | Yes (n= 187) | ||
| Women | 294 (42.4%) | 90 (48.1%) | 0.092 |
| Age (years) | 71.7 (11.5) | 75.9 (8.5) | <0.001 |
| Body mass index (Kg/m2) | 28.2 (5.9) | 28.3 (5.7) | 0.957 |
| Smoker | 377 (54.6%) | 82 (43.9%) | 0.006 |
| Hypertension | 556 (80.1%) | 160 (85.6%) | 0.054 |
| Hypercholesterolemia | 400 (57.6%) | 106 (56.7%) | 0.439 |
| Diabetes mellitus | 344 (49.6%) | 86 (46.0%) | 0.211 |
| Prior myocardial infarction | 194 (28.8%) | 32 (17.4%) | 0.001 |
| Prior stroke | 80 (11.5%) | 26 (14.0%) | 0.214 |
| Chronic obstructive pulmonary disease | 166 (23.9%) | 38 (20.3%) | 0.175 |
| Chronic kidney disease ∗ | 398 (57.3%) | 104 (55.6%) | 0.366 |
| Heart failure subtype | <0.001 | ||
| Reduced ejection fraction | 384 (55.3%) | 73 (39.0%) | |
| Preserved ejection fraction | 310 (44.7%) | 114 (61.0%) | |
| Treatment, | |||
| Angiotensin converting enzyme inhibitors or angiotensin receptor blockers | 494 (71.4%) | 132 (70.6%) | 0.448 |
| Betablockers | 611 (88.0%) | 154 (82.4%) | 0.030 |
| Diuretics | 625 (90.1%) | 177 (94.7%) | 0.031 |
| Digoxin | 66 (9.5%) | 46 (24.6%) | <0.001 |
| Amiodarone | 78 (11.2%) | 11 (5.9%) | 0.018 |
| Oral anticoagulants | 189 (27.7%) | 157 (84.4%) | <0.001 |
∗ Chronic kidney disease was defined by an estimated glomerular filtration rate ml/min/1.73 m 2 .
As shown in Figure 3 , patients with permAF had higher rates of CI (42.8%) compared to patients without permAF (31.0%, p = 0.002). Patients with PermAF had also higher rates of moderate/severe CI (8.3% vs 4.5%, p = 0.039). Notably, the overall performance in both tests was worse for patients in the permAF group, who presented a trend to lower MMSE scores and higher number of errors in the SPMSQ test ( Figure 3 ). Among patients with permAF, the rates of CI were similar among those with and without OAC treatment (44.8% vs 42.7%, p = 0.493).
The univariate analysis showed that the history of permAF was significantly associated with the presence of CI, together with other variables such as female gender, age, previous history of tobacco use, diabetes mellitus, CKD, stroke, low hemoglobin levels, and variables reflecting a worse functional status such as NYHA class III/IV, low systolic blood pressure, or increased levels of pro-Brain Natriuretic Peptide (BNP) ( Table 4 ). In the multivariate analysis, permAF was confirmed to be independently associated with the presence of CI, with an odds ratio (OR) of 1.54 (95% C.I. 1.05 to 2.28, p = 0.029, Table 4 ). The independent association with CI was also confirmed for female gender, an older age, diabetes mellitus, CKD, previous stroke, NYHA class III/IV, and lower systolic blood pressure values ( Table 4 ). Notably, both the history of permAF and previous stroke showed an independent association with CI even when the 2 variables were included in the model. Several regression models were constructed including other factors considered to have a potential effect on cognitive functioning, such as treatment with OACs and the CHA2DS2-VASc score. None of these variables showed a significant association with CI in the multivariate analyses (p NS for both). Additionally, to assess the potential differential effect of permAF in patients with HF-PEF and HF-REF, interaction analyses were performed. No significant interaction was found between permAF, HF subtype, and CI when the interaction variable (permAF × HF subtype) was introduced in the model (OR 0.72, 95% C.I. 0.33 to 1.57, p = 0.410). Similarly, no significant interaction was obtained between permAF and ejection fraction when the variable permAF × ejection fraction was included in the model (OR 0.98, 95% C.I. 0.96 to 1.01, p = 0.150). These findings indicate that the effect of permAF was similarly relevant in both HF subtypes and through all ejection fraction ranges in our population. Importantly, in all the models tested, permAF persisted as a factor independently associated with CI (OR ranging from 1.63 to 3.65), which further strengthens the association between both conditions.
