Associations among Body Mass Index, Waist-to-Hip Ratio, and Cognitive Impairment Tend to Follow an Opposite Trend and Are Sex Specific: A Population-Based Cross-Sectional Study in a Rural Area of Xi'an, China

Introduction: The relationship between obesity and cognitive impairment (CI) is highly heterogeneous in previous studies, which may be due to insufficient consideration of anthropometric indicators and sex. This study compared the cross-sectional relationships among body mass index (BMI), waist-to-hip ratio (WHR), and CI among people aged ≥40 years, and sex-specific relationships were also considered. Methods: This was a population-based cross-sectional study with a cluster sampling design. CI was defined as a Mini-Mental State Examination score lower than the cutoff value. Multivariate logistic regression was used. BMI and WHR were fitted as both restricted cubic splines and categorical data. Stratified analysis and interaction analysis were performed to explore the sex-specific relationship. Results: A total of 1,792 subjects (40.5% male) were analyzed, and 230 were confirmed to have CI. The relationships among BMI, WHR, and CI were significant (poverall = 0.023, pnonlinear = 0.097; poverall = 0.017, pnonlinear = 0.078, respectively) but exhibited an opposite trend in the total population in the analyses with BMI and WHR as restricted cubic splines. Further categorical analyses showed that subjects with a BMI <23 kg/m2 tended to have a higher risk of CI than those with BMI ≥23 kg/m2 (16.2% vs. 11.8%, p = 0.017; OR = 1.366 [0.969-1.926], p = 0.075), and subjects with a WHR >0.92 had a significantly higher risk of CI than those with a WHR ≤0.92 (11.7% vs. 16.2%, p = 0.011; OR = 1.619 [1.161-2.258], p = 0.005). In addition, the relationship between a low BMI and CI was more significant in males (p = 0.034), while the relationship between a high WHR and CI was more significant in females (p = 0.002). Further studies are needed to confirm the sex differences because of the marginal significance result in the interaction analysis (p = 0.051 for interaction term BMI × sex; p = 0.056 for interaction term WHR × sex). Conclusion: The relationships among BMI, WHR, and CI exhibit an opposite trend. A low BMI or high WHR was positively associated with CI, which was more prominent in males for a low BMI and females for a high WHR.