Commentary: Risk factors for mild cognitive impairment in type 2 diabetes: a systematic review and meta-analysis

Shanshan Wu^*

• The Affiliated Hospital of Liaoning University of Traditional Chinese Medicine, Shenyang, China

We read with great interest the article by Zhao and colleagues entitled " Risk factors for mild cognitive impairment in type 2 diabetes: a systematic review and meta-analysis " (1). We acknowledge Zhao et al.'s systematic review elucidating risk factors for mild cognitive impairment (MCI) in type 2 diabetes (T2DM) (1). Their identification of advanced age (≥60 years), prolonged diabetes duration (8-9 years), elevated HbA1c (>9%), low education (≤6 years), and cardiometabolic comorbidities as significant MCI predictors advances this field.Furthermore, smoking, hypertension, cardiovascular disease (CVD), insulin resistance, fasting plasma glucose(FPG), and high-sensitivity C-reactive protein(HS-CRP) were also significantly linked to higher MCI risk.However, methodological concerns regarding subgroup analyses warrant critical examination. A discrepancy exists between the statistical protocol (Section 2.5) and its implementation in Figure 8A. Per protocol specifications, a random-effects model is mandated when heterogeneity exceeds 50% (I² ≥50%). Given the ongoing controversy regarding sex differences in the risk of MCI among T2DM patients, Zhao et al. performed subgroup analyses by sex. The authors state in their original article (Page 8, Line 23) that a random-effects model was used due to significant between-group heterogeneity in effect sizes (I² = 81.5%; Figure 8A). However, Figure 8A reveals that a fixed-effects model was erroneously applied for this analysis. This discrepancy potentially biases the sex-specific risk estimates, leading to an erroneous conclusion that female sex is an independent risk factor for MCI in individuals with T2DM(1). This commentary aims to rectify the erroneously applied models mentioned above and to provide a correct re-analysis. We strictly adhered to the same dataset and inclusion criteria as those described by Zhao et al. Statistical analyses were conducted utilizing RevMan software (version 5.3). Dichotomous outcome variables were evaluated using odds ratios (ORs) accompanied by 95% confidence intervals (CIs). Heterogeneity among the included studies was quantified employing the I² statistic, with conventional thresholds designating low, moderate, and high heterogeneity as I² values exceeding 25%, 50%, and 75%, respectively (2). The fixed-effects model, which presumes homogeneity of effect sizes across studies, was applied in instances of low heterogeneity (I² ≤ 50%); this model derives the pooled effect estimate via inverse-variance weighting. Conversely, the random-effects model, accommodating inherent variability in effect sizes, was implemented where significant heterogeneity was present (I² > 批注 [姗吴1]: Thank You for Your Valuable Feedback Dear Professor, I hope this message finds you well. I am writing to express my sincere gratitude for the time and effort you dedicated to reviewing my manuscript. Your insightful comments and detailed suggestions have been immensely helpful in improving the quality of this work.I truly appreciate the thoroughness of your review and the constructive feedback you provided. Your expertise and attention to detail have not only strengthened the manuscript but also provided me with valuable learning opportunities.The highlighted sections have been revised in response to your Points 1 and 2. The highlighted sections have been revised in response to your Points 3. 50%). This approach incorporates between-study variance through inverse-variance weighting. 37Consequently, model selection was determined a priori based the observed I² value: the random-38 effects model was employed when exceeded otherwise the fixed-effects model was utilized. 39 A reanalysis of the sex-stratified data (from 8 studies) significant between-group 41 heterogeneity in effect sizes (I² = 55.2%; Figure 1), therefore, a random-effects model was employed. 42The