Association of glycemic gap with stroke recurrence in patients with ischemic stroke

Abstract Background Glycemic gap, as a novel index of acute glycemic excursion, is associated with poor prognosis of different diseases. This study aimed to explore the association of the glycemic gap with long‐term stroke recurrence in patients with ischemic stroke. Methods This study included patients with ischemic stroke from the Nanjing Stroke Registry Program. The glycemic gap was calculated by subtracting the estimated average blood glucose from the blood glucose at admission. Multivariable Cox proportional hazards regression analysis was performed to explore the association between the glycemic gap and the risk of stroke recurrence. The Bayesian hierarchical logistic regression model was used to estimate the effects of the glycemic gap on stroke recurrence stratified by diabetes mellitus and atrial fibrillation. Results Among 2734 enrolled patients, 381 (13.9%) patients experienced stroke recurrence during a median follow‐up of 3.02 years. In multivariate analysis, glycemic gap (high group vs. median group) was associated with significantly increased risk for stroke recurrence (adjusted hazard ratio, 1.488; 95% confidence interval, 1.140–1.942; p = .003) and had varying effects on stroke recurrence depending on atrial fibrillation. The restricted cubic spline curve showed a U‐shaped relationship between the glycemic gap and stroke recurrence (p = .046 for nonlinearity). Conclusion Our study found that the glycemic gap was significantly associated with stroke recurrence in patients with ischemic stroke. The glycemic gap was consistently associated with stroke recurrence across subgroups and had varying effects depending on atrial fibrillation.


Supplementary Methods Fractional polynomial model
To account for the possibility of a non-linear relationship, we applied the fractional polynomial 1 terms of the glycemic gap into model 1.The best fitting fractional polynomial model was identified by the multivariable fractional polynomial method.The optimal glycemic gap level was when the first derive of the model was equal to 0. We used the delta method 2 based on standard errors to calculate the confidence intervals for the optimal glycemic gap level.

Bayesian hierarchical logistic regression model
According to the results of subgroup analyses, we hypothesized that the effects of glycemic gaps on stroke recurrence might differ in patients with or without atrial fibrillation.Initially we performed analyses in all patients, then we stratified the cohort by diabetes status to account for the effect of diabetes mellitus on stroke recurrence.To optimize the robustness and eliminate the influence of interaction, we selected the Bayesian hierarchical logistic regression model to explain the likelihood of stroke recurrence at the median follow-up time of 3 years separately in patients with or without atrial fibrillation.Analyses were implemented in Python (version, 3.7.10,"pymc3" package 3 ).
Since the coefficients of glycemic gap were estimated in the same model, we could directly quantify the difference of coefficients in patients with or without atrial fibrillation.In line with previous studies 4 , we employed the No U-Turn Sampler (NUTS) method, a type of Monte Carlo Markov Chain algorithm (setting: draws = 2000), to draw samples from the posterior predictive distributions and 90% highest probability density interval (HPDI).The differences could be substantial when the reference level of 0 were not included in the range of 90% HPDI.Model specification was presented as follows:

Table 4 . Hazard ratios for all-cause mortality according to glycemic gaps.
CI, confidence interval; HR, hazard ratio; NIHSS, National Institute of Health Stroke Scale.Model 1: unadjusted model.Model 2: adjusted by age, sex, hypertension, diabetes mellitus, atrial fibrillation, dyslipidemia, coronary heart disease, smoking status, alcohol consumption, stroke etiology, education years.Model 3: adjusted for covariates in model 2 and body mass index, NIHSS, hemoglobin, total cholesterol, triglyceride, high density lipoprotein, low density lipoprotein, prior antidiabetic agents, and the usage of antiplatelet drugs, anticoagulants, antihypertensive drugs and hypoglycemic agents at discharge.

Table 5 . Odds ratios for favorable outcome according to glycemic gaps.
CI, confidence interval; OR, odds ratio; NIHSS, National Institute of Health Stroke Scale.Model 1: unadjusted model.Model 2: adjusted by age, sex, hypertension, diabetes mellitus, atrial fibrillation, dyslipidemia, coronary heart disease, smoking status, alcohol consumption, stroke etiology, education years.Model 3: adjusted for covariates in model 2 and body mass index, NIHSS, hemoglobin, total cholesterol, triglyceride, high density lipoprotein, low density lipoprotein, prior antidiabetic agents, and the usage of antiplatelet drugs, anticoagulants, antihypertensive drugs and hypoglycemic agents at discharge.

Table 6 . Competing risk analysis of glycemic gaps for predicting stroke recurrence.
CI, confidence interval; HR, hazard ratio; NIHSS, National Institute of Health Stroke Scale.Model 1: unadjusted model.Model 2: adjusted by age, sex, hypertension, diabetes mellitus, atrial fibrillation, dyslipidemia, coronary heart disease, smoking status, alcohol consumption, stroke etiology, education years.Model 3: adjusted for covariates in model 2 and body mass index, NIHSS, hemoglobin, total cholesterol, triglyceride, high density lipoprotein, low density lipoprotein, prior antidiabetic agents, and the usage of antiplatelet drugs, anticoagulants, antihypertensive drugs and hypoglycemic agents at discharge.

Table 7 . Hazard ratios for stroke recurrence according to DM and AF.
AF, atrial fibrillation; CI, confidence interval; DM, diabetes mellitus; HR, hazard ratio; NIHSS, National Institute of Health Stroke Scale.Model 1: unadjusted model.Model 2: adjusted by age, sex, hypertension, dyslipidemia, coronary heart disease, smoking status, alcohol consumption, stroke etiology, education years.Model 3: adjusted for covariates in model 2 and body mass index, NIHSS, hemoglobin, total cholesterol, triglyceride, high density lipoprotein, low density lipoprotein, prior antidiabetic agents, and the usage of antiplatelet drugs, anticoagulants, antihypertensive drugs and hypoglycemic agents at discharge.