Subjects
In this retrospective cohort study, data from 736 patients with acute ischemic stroke admitted to Suining Central Hospital, Sichuan Province, China, from January 1, 2015, to December 31, 2018, were recruited. The following were the criteria for inclusion: 1) patients were 18 years of age and older; 2) patients were hospitalized within 7 days of stroke start; 3) patients had a National Institutes of Health Stroke Scale (NIHSS) score of less than or equal to 3; and 4) patients had acute cerebral infarction confirmed by magnetic resonance imaging (MRI) during hospitalization.
The following were the criteria for exclusion: 1) previous history of stroke; 2) aphasia that made it impossible to assess cognitive function; 3) history of mental problems, neurological diseases, thyroid diseases, autoimmune diseases, or tumors; and 4) prestroke dementia or cognitive impairment. A total of 311 patients with acute mild ischemic stroke were eventually included in the final analysis (as shown in Fig. 1). The ethics committees at Suining Central Hospital approved this study in accordance with the Helsinki Declaration.
Data collection
Covariates
On admission, age, sex, body mass index (BMI), education, smoking status, alcohol use, hypertension, diabetes, atrial fibrillation, and other information were collected. Fasting venous blood was taken on the day after admission to assess fasting plasma glucose (FPG), HbA1c, blood lipid, and uric acid (UA) levels. Within 72 hours after hospitalization, MRI was performed. Within 24 hours of admission, the NIHSS was used to determine the severity of the stroke, with a score of less than or equal to 3 indicating mild ischemic stroke. At discharge, the modified Rankin scale (mRS) was used to assess functional outcomes.
Cognitive assessments
Cognitive function was assessed using the Chinese version of the Montreal Cognitive Assessment Scale (MoCA) 3-6 months after stroke onset14. The highest possible MoCA score was 30, and less than 26 was defined as cognitive impairment. A score of less than 25 was classified as cognitive impairment if the number of years of education was less than or equal to 12 years15, 16.
Statistical analysis
Data are reported as the mean ± standard deviation (SD) (Gaussian distribution) or median (range) (skewed distribution) for continuous variables and as numbers and percentages for categorical variables. χ2 (categorical variables), one-way ANOVA (normal distribution), or Kruskal–Wallis H test (skewed distribution) were used to detect differences in variables among different HbA1c groups (based on tertiles). To examine the effect of HbA1c on PSCI, we constructed three different models, namely, Model 1 (no covariates were adjusted for), Model 2 (only sociodemographic variables were adjusted for) and Model 3 (covariates are presented in Table 3). A 95% confidence interval was calculated for the effect sizes. We used smooth curve fitting (penalized spline method) to account for nonlinearity between HbA1c and PSCI as well as the generalized additive model (GAM). In addition, a two-piecewise binary logistic regression model was used to further explain the nonlinearity. Modeling was performed with the statistical software packages R (http://www.R-project.org, The R Foundation) and EmpowerStats (http://www. empowerstats.com, X&Y Solutions, Inc., Boston, MA). P values less than 0.05 (two-sided) were considered statistically significant.