“…In this meta-analysis, subgroup analyses were further conducted to investigate the potential inter-study heterogeneity and examine the robustness of primary results 31 . We divided the subgroups as follows: geographical location (Asia 16–18 , North America 11, 12 , or Europe 9, 10, 13–15, 32 ), sex (male 10, 16, 17 or female 10, 16 ), duration of follow-up (<7 years 7, 9, 11, 14, 15, 17 or ≥7 years 10, 12, 13, 16, 18 ), sample size (<1000 13–15 or ≥1000 7, 9–12, 16–18 ), fasting status (yes 7, 11, 14, 15 or no 12, 13, 16 ), and NOS score (≤7 7, 11, 17 or >7 9, 10, 12–16, 18 ). We also performed analyses stratified by whether the studies had adjusted for potential confounding factors, including smoking, alcohol consumption, BMI, physical activity, history of CVD, history of diabetes, and serum vitamin B12/folate.…”