The subjects of this study were 101 patients who were admitted to the rehabilitation ward in our hospital for comprehensive treatment after onset of cerebral infarction. The rehabilitation continued for approximately 3 months although the rehabilitation schedule differed depending on the condition of the patients (Af group=91.1±22.6 days vs. non-Af group=80.6±22.0 days).

The final diagnosis of all patients was made with computed tomography or magnetic resonance imaging. The inclusion criteria were 1) man or woman aged 40 or more, 2) patient with first onset of cerebral infarction, and 3) patients who were transferred to our unit from the Neurology Department after an acute phase treatment. Among them, 14 patients could not finish the rehabilitation program due to associated systemic diseases, such as pneumonia, sepsis, acute myocardial infarction, acute renal failure, and shock. Except these 14 patients, 87 patients (77 men, 10 women) with stroke were included. Their average age was 72.6 years.

The exclusion criteria were 1) previous stroke history, 2) previous rehabilitation history, 3) concomitant disease that could prevent the rehabilitation program, 4) patients with cerebral hemorrhage, such as subarachnoid and epidural hemorrhages, 5) history of neurological or mental disease or alcohol addiction, and 6) history of paroxysmal atrial fibrillation.

This study was a retrospective study and study data were obtained through medical records. It was conducted with the informed consent of all subjects.

The rehabilitation program was done for 30 minutes per session, 5 days per week, for 12 weeks. It consisted of neurodevelopmental treatment, therapeutic exercise, functional electrical stimulation, occupation therapy, dysphagia rehabilitation, physical modalities, art therapy, and psychological counseling.

At the first assessment, subjects were examined regarding risk factors of cerebrovascular disease, diabetes, hypertension, dyslipidemia, smoking, coronary artery disease, congestive heart failure, and valvular heart disease. Atrial fibrillation was defined as having an irregular QRS wave with no visible P wave on a 12-lead ECG or atrial flutter concomitant with irregular ventricular response at the time of admission [15,16].

Motor function was evaluated in terms of the manual muscle test (MMT). Cognitive function was assessed in terms of Mini-Mental State Examination (MMSE-K) scores. The National Institutes of Health Stroke Scale (NIHSS) was used to evaluate the patients' neurological scale. The infarcted lesions were also classified by their locations.

The extent and the type of the cerebral infarction were assessed using the Chaudhuri classification [17]. This system is described as follows:

The functional activity level was assessed in terms of the scores of the Functional Independence Measure (FIM) for daily living activities at the time of admission and after completion of the rehabilitation program. FIM is a standardized assessment tool, widely used for disability assessments. Its validity and reliability have been demonstrated in previous studies [18,19,20]. The Modified Barthel Index (MBI) was also used. The MBI is used to measure performance in activities of daily living (ADL) in patients with chronic disease; it evaluates the patient's independence through direct observation of the ADL of the patient along with an interviews. Furthermore, the PULSES profile was used to monitor and predict functional improvement after the rehabilitation program. The score is known to have a high correlation with the FIM and MBI scores [21,22]. At the times of admission and discharge, FIM, MBI, and PULSES scores of patients were determined so that improved functional performance was assessed by comparing the scores between admission and discharge after the rehabilitation program.

Risk factors between patients with and without atrial fibrillation were compared. If the variance according to Levene test was the same, Student t-test was used to conduct a univariate comparison between the two groups of continuous variables. The χ² test was used for univariate comparisons of nominal variables. A multiple linear regression model was used to analyze the effect of the program on the FIM, MBI, and PULSES scores. The multiple logistic regression model included factors that could be confounding variables, such as age, gender, hospitalization period, time taken from stroke onset to rehabilitation, valvular heart disease, congestive heart failure, and coronary artery disease, including nominal variables that were found to be statistically significantly different between patients with and without atrial fibrillation from the χ² test, to examine and quantify the relationship between the variables and atrial fibrillation independently. The significance level of the two-sided test was set to p<0.05. The SPSS ver. 17 for Windows (SPSS Inc., Chicago, IL, USA) was used for statistical analyses.

Among 87 patients with cerebral infarction, only 20 (22.9%) had atrial fibrillation at the time of admission. The frequency of atrial fibrillation increased as age increased. For example, in patients under 70 years, it was 6.1% while in patients over 70 years, it was 33.3%.

The average age of patients with atrial fibrillation was 75.1±5.1 years while the average age of patients without atrial fibrillation was 71.8±8.8 years. The difference was not statistically significant (p=0.116). No significant difference was seen in average morbidity period, diabetes, hypertension, dyslipidemia, smoking rate, or carotid atherosclerosis between the groups.

It took 11.5±10.7 days to transfer patients with atrial fibrillation from the Neurological Department that diagnosed cerebral infarction to the Department of Rehabilitation Medicine, while it took 10.5±8.3 days for patients without atrial fibrillation. Patients with atrial fibrillation had a longer hospitalization period but it was not statistically significant (p=0.067) (Table 1).

In the univariate comparison, patients with atrial fibrillation had significantly higher prevalence of valvular heart disease, coronary artery disease, and heart failure than patients without atrial fibrillation (χ²=10.818 and p=0.004, χ²=10.703 and p=0.002, and χ²=4.656 and p=0.046, respectively) (Table 1).

When factors, such as age, gender, hospitalization period, time taken from stroke onset to rehabilitation, diabetes, hypertension, dyslipidemia, smoking rate, and carotid atherosclerosis, were included in the multiple logistic regression analysis, atrial fibrillation showed a significant correlation with coronary artery disease (odds ratio [OR]=5.594, 95% confidence interval [CI]=1.592-32.843) and valvular heart disease (OR=4.104, 95% CI=1.122-15.020), while no correlation was seen with congestive heart failure (Table 2).

Patient motor function was evaluated with the MMT. Cognitive function was assessed using the MMSE-K. Neurological status before rehabilitation was assessed using the NIHSS. Regarding motor function, the non-Af group had better upper and lower extremity muscle power than the Af group before rehabilitation. However, there was no statistically significant difference between the groups (p>0.05).

In MMSE-K, mild cognitive impairment was detected in both groups. The total score of the non-Af group was better than that of Af group. However, there was no statistically significant difference (Af group=20.5±8.3 vs. non-Af group=23.2±7.6; p=0.214). In neurological status using NIHSS, the non-Af group showed a statistically significantly better result in the movement of lower extremities (Af group=2.4±1.3 vs. non-Af group=0.7±0.2; p=0.015). However, there was no statistically significant difference in the total score (Af group=17.5±6.8 vs. non-Af group=11.2±5.3; p=0.101).

In conclusion, there was no significant difference between the groups in terms of motor function, cognitive function, or neurological scale. In a univariate analysis including each item, there was no significant difference. In a multivariate analysis including each item, statistically significant differences remained, although the influence of atrial fibrillation on the FIM score at discharge decreased a little (MMT: B=-8.48, SE(B)=3.16, p=0.013; MMSE-K: B=-6.22, SE(B)=2.13, p=0.010; NIHSS: B=-7.39, SE(B)=2.54, p=0.009). In a multivariate analysis including all three items, the difference due to atrial fibrillation was still statistically significant (B=-5.87, SE(B)=2.01, p=0.027). In a multivariate analysis including each item, a statistically significant differences in MBI score at discharge due to atrial fibrillation remained (MMT: B=-6.92, SE(B)=3.27, p=0.028; MMSE-K: B=-7.96, SE(B)=4.05, p=0.021; NIHSS: B=-5.38, SE(B)=2.74, p=0.033). In a multivariate analysis including all three items, the difference of MBI score due to atrial fibrillation was still statistically significant (B=-5.19, SE(B)=2.52, p=0.031). In a multivariate analysis including each item, there was a statistically significant difference in PULSES score at discharge due to atrial fibrillation (MMT: B=-2.38, SE(B)=1.04, p=0.021; MMSEK: B=-2.07, SE(B)=0.74, p=0.032; NIHSS: B=-1.75, SE(B)=0.59, p=0.039). In a multivariate analysis including all three items, the difference in PULSES score due to atrial fibrillation remained statistically significant (B=1.01, SE(B)=0.49, p=0.043).

Because location, type, extent of infarcted lesions can affect the results of rehabilitation, we classified the infarcted lesions according to the Chaudhuri classification and by their locations. Regarding the relationship between functional recovery and the location and the extent of cerebral infarction, it is known that patients whose brain lesions are located in the brainstem rather than the cerebral hemispheres or in a cerebral subcortical area rather than a cerebral cortical area, or only in a restricted area, such as the putamen or thalamus, among the cerebral subcortical areas, show better results [23,24,25,26,27]. The results are shown in Tables 3 and 4. According to the Chaudhuri classification, group IV or V (deep infarct or combination of deep infarct and large superficial infarct) lesions were more prevalent in the Af group than the non-Af group. In the non-Af group, group II (small superficial infarct involving one cerebral lobe) lesions were detected more frequently. However, there was no statistically significant difference in the type or extent of brain lesions between the groups according to the Chaudhuri classification. Comparing the locations in both groups, there were more basal ganglia infarctions in the Af group, but the difference was not significant (p=0.087). In other areas of the brain, there was no statistically significant difference in the distribution of infarcted lesions between the groups. In the Af group, there were more brain lesions located in the brainstem and basal ganglia (including the putamen) that are known to show better results in rehabilitation compared with the non-Af group. However, the functional recovery results after rehabilitation were better in the non-Af group, seemingly contrary to the supposedly unfavorable brain lesions. Thus, location, type, and extent of infarcted lesions did not seem to affect rehabilitation outcomes in this study.

Patients with atrial fibrillation had significantly lower FIM, MBI, and PULSES scores at the time of admission to the Department of Rehabilitation Medicine (62.4±21.0 vs. 76.6±21.2, p=0.010; 53.4±19.7 vs. 66.8±23.8, p=0.024; and 15±1.3 vs. 11±2.2, p=0.035, respectively) (Table 5). Patients with atrial fibrillation also had significantly lower FIM, MBI, and PULSES scores after completion of the rehabilitation program (67.8±20.9 vs. 85.2±21.4, p=0.002; 57.9±9.2 vs. 72.9±24.0, p=0.012; and 13.8±1.4 vs. 8.5±2.8, p=0.017, respectively) (Table 5). Patients with atrial fibrillation also had significantly lower increases in FIM, MBI, and PULSES scores, obtained by subtracting the scores after the rehabilitation program from those before the program (5.4±2.4 vs. 8.7±4.1, p=0.001; 4.5±2.7 vs. 6.0±3.7, p=0.048; and 1.2±0.5 vs. 2.5±1.2, p=0.001, respectively) (Table 5). In the multiple linear regression analysis, atrial fibrillation correlated with a decrease of 11.34 in the FIM score at the time of discharge (B=-11.34, SE(B)=3.87, p=0.002), of 10.81 in the MBI score at the time of discharge (B=-10.81, SE(B)=4.02, p=0.012), and of 2.47 in the PULSES score at the time of discharge (B=-2.47, SE(B)=0.85, p=0.017). The effect of atrial fibrillation on the FIM score at the time of discharge was further reduced if the FIM score at the time of admission was included in the analysis. However, it still showed a statistically significant difference (B=-6.51, SE(B)=2.12, p=0.012). The MBI scores also showed that the difference between atrial fibrillation and the MBI scores at the time of discharge was reduced further if the MBI scores at the time of admission were included in the analysis, but it too still showed a significant correlation (B=-5.21, SE(B)=2.56, p=0.035). The PULSES scores also showed that the difference between atrial fibrillation and the PULSES scores at the time of discharge was reduced further if the PULSES scores at the time of admission was included in the analysis, but it too still showed a significant correlation (B=-1.05, SE(B)=0.56, p=0.041).