The present study was approved by the Ethics Committee of Fukui Prefectural Hospital (#11–36). Written informed consent was obtained from the patients or their relatives on enrollment into this study.
All inpatients with acute CVD introduced to the Department of Physical Medicine and Rehabilitation (PMR) of Fukui Prefectural Hospital between October 22, 2011 and March 31, 2013, were considered for inclusion. Patients who had suffered from ischemic and hemorrhagic strokes were included. Patients with intracranial tumors, traumatic brain injuries, subarachnoid hemorrhages due to ruptured aneurysms, or chronic subdural hematomas were excluded, as well as were pediatric patients. Standardized treatments were performed by board-certified neurologists or neurosurgeons at the time of hospitalization for all the patients.
The rehabilitation therapists in charge asked the participants to complete a questionnaire regarding their pre-stroke exercise routine and daily PA at the beginning of the rehabilitation therapy (Additional file 1). The questionnaire was prepared in accordance with the Japanese Kenko Nippon 21 Exercise Guide [22], in which the intensities of exercises and daily activities were estimated in metabolic equivalents (METs). For example, watching television in a sitting position, performing light household activities such as pressing clothes, walking at a speed of 67 m/min on a flat surface, and climbing stairs have intensities of 1, 2.3, 3, and 8 METs, respectively [22, 23]. The participants were asked to indicate all the activities they had performed for more than 1 h per week during the month prior to their admission. The therapists helped the participants complete the questionnaire if they needed assistance because of seeing or hearing difficulties; when the participants had cognitive impairment, their relatives were asked to fill the questionnaire. The participants’ maximal daily PAs were classified into integral METs, from 1 to 8 METs.
Basic information, such as sex, age, and type of disease, was gathered from hospital records, along with the following information: date on which the rehabilitation therapy was started; coexistence of aphasia, agnosia, or apraxia; ongoing or past treatment of diabetes mellitus, hypertension, hyperlipidemia, cardiac disease, malignancy, orthopedic complication, and psychiatric disease; smoking history, and CVD history. Stroke size was represented as the maximal diameter of the affected region measured on cross-sectional images of the initial computed tomography or magnetic resonance imaging findings. The measurement was performed under the supervision of a board-certified radiologist. Paresis was assessed according to the Brunnstrom Recovery Stage (BRS). Sensory disturbance was categorized as “no symptom/slight numbness” or “overt disturbance.”
When participants completed their inpatient rehabilitation program, gait independence was evaluated using the Functional Independence Measure (FIM), in which the activities of daily living of the patients were classified into seven levels according to their ability. Levels 1 and 2 indicated dependence on others, levels 3–5 indicated the need for varying degrees of assistance, and levels 6 and 7 indicated independence [24]. A physiatrist who was blinded to the premorbid PA level evaluated gait independence to avoid bias in the rehabilitation goal setting. The participants were then assigned to one of two categories according to their gait independence as follows: “walking alone” (WA) group, which included participants with FIM levels 6 and 7, and “not walking alone” (NWA) group, which included participants with FIM levels 5 and below.
Dementia was evaluated using the Japanese domestic scale presented by the Ministry of Health, Labor and Welfare, which is mainly used in the Long-term Care Insurance System in Japan. This scale focuses on how much service a patient/client needs, rather than on psychopathological severity [25]. The classes are as follows: class 0, no sign of dementia; class 1, signs of dementia, e.g. forgetfulness but an independent daily life; class 2, daily life involves surveillance because of signs of dementia, difficulty in communication, and differences in behavior; class 3, daily life is dependent on others because of the above symptoms; class 4, continuous help is needed from others because of the above symptoms; and class M, treatment is needed because of psychiatric, behavioral, or physical symptoms. In the present study, those in classes ≥2 were defined as having overt dementia.
Overlapping of coexisting diseases has been reported to be a factor of stroke outcome deterioration [11, 26]. The comorbidity index (CI) was defined as the total number of the following risk factors involved: hypertension, diabetes mellitus, hyperlipidemia, cardiac disease, malignancy, smoking history, and overt dementia.
Anonymous patient records were created in FileMaker Pro 8.5 ver. 2 (FileMaker Inc., Santa Clara, CA, USA), the data were calculated using Excel for Mac 2004 ver. 11.6.5 (Microsoft, Redmond, WA, USA) and transferred to StatView ver. 5 (SAS Institute, Cary, NC, USA) for statistical analysis.
Statistical analysis
Student’s t-test was used for participants’ ages, hospital stay length, and stroke size. Data are presented as mean ± SD.
Chi-squared test for independence was used to check for correlations between gait independence and the following factors: each daily activity, coexisting disease, and stroke type. As we found a possible tendency that participants with hemorrhagic stroke with ventricular perforation (HSVP) recovered poorly compared to those with other types of stroke, an inter-disease comparison was performed between HSVP and the other types of stroke.
Mann-Whitney U-test was used to determine the association between gait outcome and CI or maximal daily PA.
Logistic regression analysis was used to ascertain how increasing intensities of PA contributed to gait independence. The odds represented the contribution of premorbid PA to gait independence as it increased by 1 MET from a lower intensity.
Multiple logistic regression analysis was performed to identify the factors that contributed the most to gait independence. Possible explanatory variables; namely, age, CVD history, and premorbid PA were added to the analyses to check for confounding factors. Stroke size or the type of disease were not used as explanatory variables, because we were unsure whether premorbid PA has any effect on them, as vigorous PA might have prophylactic effects on arteriosclerotic disease [12, 17, 19]. CI and dementia were not used as explanatory variables either, because onset and treatment history of comorbidities were not confirmed.
A p - value of < 0.05 was considered statistically significant; Bonferroni correction was used for multiple analyses.