Participants and setting
The study was conducted at the Department of Neurology, Dongguan People’s Hospital, between January 1, 2018, and December 30, 2018. The inclusion criteria for the study were: (1) aged over 18 years; (2) first AIS occurred within 7 days before admission and was diagnosed according to the American Heart Association Stroke Council criteria [18]; (3) had complete brain magnetic resonance imaging (MRI) examinations; and (4) had an mRS score < 3 before admission. The exclusion criteria were: (1) transient ischemic attack, cerebral hemorrhage, subdural hematoma, or subarachnoid hemorrhage; (2) incomplete clinical data (e.g., incomplete brain MRI ); (3) died before discharge; (4) severe comorbidities (e.g., liver, kidney, heart, or respiratory failure, or malignant tumors); (5) unable to finish the assessment at follow-up because of severe hearing disabilities, visual disabilities, language disorders, or cognitive impairment; and (6) refused to sign a consent form.
The study protocol was approved by the Ethics Committee of Dongguan People’s Hospital. The consent of all subjects was obtained in accordance with the Declaration of Helsinki.
Data collection
Information about demographic variables (e.g., age, sex) and vascular risk factors (e.g., hypertension, diabetes mellitus, atrial fibrillation) was collected. We also recorded clinical data, including stroke severity, blood pressure, stroke subtype, lesion location, in-hospital treatment (antiplatelet, anticoagulation, and statins), and laboratory data. The severity of stroke was assessed using the National Institutes of Health Stroke Scale (NIHSS) [19]. The subtype of ischemic stroke was evaluated using the Trial of Org 10172 in Acute Stroke Treatment (TOAST) subtype system [20]. Laboratory data included fasting blood glucose, glucosylated hemoglobin A1c, total cholesterol, triglycerides, high-density lipoprotein, low-density lipoprotein, homocysteine, white blood cells, hemoglobin, fibrinogen, prothrombin time, and creatinine. Blood pressure was measured and recorded shortly after admission, and the laboratory data were tested within 24 hours after admission.
Definition of END
We evaluated NIHSS scores on the day of admission, and then daily over the
following 7 days after admission. We defined END as an increase in NIHSS score ≥ 2 points within 72 h after admission [21].
Patient follow-up
All follow-up assessments were completed by the patients. We assessed multiple outcomes at 12–15 months after the index stroke. Based on the WHO-ICF (2002) [15], we selected one measure to evaluate each level of functioning: body function impairment was assessed using the mRS, activity limitation was assessed using the Lawton ADL scale [22], and participation restriction was evaluated using the Chinese version of the reintegration to normal living index (RNLI) [23].
ADL is composed of basic ADL and instrumental ADL. The components of basic ADL include six different levels of abilities for toilet activity, feeding, dressing, grooming, physical ambulation, and bathing. The total basic ADL score is calculated by summing up the points for each item, with a maximum score of 24. The Lawton instrumental ADL score is composed of eight different domains: using a telephone, shopping, food preparation, housekeeping, laundry, mode of transportation, responsibility for own medications, and handling finances. The total instrumental ADL score is calculated by summing up the points for each item, with a maximum of 32. Therefore, the total ADL score ranges from 14 to 56. A higher ADL score reflects poorer ADL performance [22].
The RNLI is an 11-item questionnaire that targets nine domains affected by incapacitating events: mobility, self-care ability, daily activities, recreational activities, general coping skills, family roles, social activities, personal relationships, and presentation of self to others. Each item is scored from 1 (minimal reintegration) to 10 (complete reintegration), with a total scale range of 11–110. A lower RNLI score reflects poorer participation in normal living [23].
Poor outcome was defined as follows: body function, mRS score ≥ 3; activity limitation, ADL score > the 75% quartile; and participation restriction, RNLI score < the 25% quartile.
MRI assessments
Each participant underwent brain MRI, including T1-weighted imaging, T2-weighted imaging, and diffusion-weighted imaging (DWI), within 7 days of admission using a 3.0 T system (Sonata, Siemens Medical, Erlangen, Germany). DWI spin echo-planar imaging (EPI) (TR/TE/excitation = 2162/76/1, matrix = 128 × 128, field of view [FOV] = 230 mm, slice thickness/gap = 6 mm/1 mm, EPI factor = 47, acquisition time = 25.9 s) with three orthogonally applied gradients was used, with a b value of 0 and 1000. Axial SE T1 (TR/TE/excitation = 488/15/1, FOV = 230 mm, slice thickness/gap = 6 mm/1 mm, matrix = 256 × 256, time of acquisition = 1 min 24.8 s) and TSE T2 (TR/TE/excitation = 3992/110/2, turbo factor = 15, FOV = 230 mm, slice thickness/gap = 6 mm/1 mm, matrix = 512 × 512, time of acquisition = 1 min 55.8 s) images were also acquired.
A neurologist (JFQ) who was blinded to the patients’ clinical information measured the MRI variables. The MRI measurements included acute infarcts and preexisting abnormalities, as follows:
(1) Acute infarction. Both the site and volume of acute lesions in the DWI sequence were recorded. The sites of the acute infarcts were classified into cortical regions, subcortical regions, and infratentorial regions. Cortical regions included the frontal, temporal, parietal, and occipital lobes; subcortical regions included the white matter, basal ganglia, and thalamus; and infratentorial regions included the brainstem and cerebellum. Acute infarcts were defined as areas of restricted water diffusion identified on DWI with b values of 1000, together with hypointensity on the corresponding apparent diffusion coefficient map. The total area of acute infarcts on DWI was measured using manual outlines. The total volume was calculated by multiplying the total area by the sum of the slice thickness and the gap.
-
White matter lesions (WMLs). WML severity was graded using the four-point scale described by Fazekas et al [24]. WMLs included periventricular hyperintensities and deep white matter hyperintensities, which were scored separately on fluid-attenuated inversion recovery images.
-
Medial temporal lobe atrophy (MTLA). MTLA was evaluated using the Scheltens scale [25]. The MRI rater judged the severity of MTLA based on standard images of the coronary section of the MRI (range 0–4), with 0 representing no atrophy and 4 representing severe atrophy.
-
Microbleeds. These were defined as small (2–10 mm) hypointense lesions with a clear margin in susceptibility-weighted imaging [26]. Symmetric basal ganglia calcification and flow void artifacts of the pial blood vessels were excluded.
-
Silent lacunae. These were visible as focal lesions (diameter > 3 mm), with approximately the same intensity as cerebrospinal fluid on MRI [27].
Intra-rater reliability (kappa) tests were carried out on 20 stroke patients with the same MRI rater. The intra-rater agreements for the MRI measurements were good to excellent: volume of infarction-ICC: 0.82; WMLs: 0.86; MTLA: 0.86; microbleeds: 0.80; silent lacunae: 0.83.
Statistical analysis
Statistical analyses were performed using SPSS for Windows (v24.0, SPSS Inc., Chicago, IL, USA). Descriptive data are presented as proportions, means, or medians, as appropriate. To reduce the lack of randomization, we conducted a propensity score analysis using the SPSS custom dialog. Before matching, we performed a univariate analysis to compare variables between patients with and without END. The propensity score was then developed using a logistic regression model in which the variables with p < 0.05 in the univariate analysis were entered. After estimating the propensity score, we matched participants using a simple 1:1 matching. After matching was completed, the two matched groups with and without END were compared. Logistic regression models were performed to analyze the relationship between multiple outcomes and END, based on the matched data. In the multivariate logistic regression analyses, the poor outcomes served as dependent variables, while END and other confounding factors served as independent variables. The odds ratio (OR) of END was interpreted as the risk of poor outcomes. The significance level was set at p < 0.05 (two-sided).