Frequency of Prediabetes According to Optimal Cut-Points for Body Mass Index in Saudi Population

Background and Objective: The prevalence and incidence of prediabetes are increasing worldwide. Body mass index (BMI) cut-off for prediabetes can vary. The objective of this study is to identify the optimal BMI cut-off that is associated with prediabetes. Methods: For the present study, we analyzed participants who are equal to or older than 18 years old. A total of 5498 were selected to be enrolled for the present study. All patients were from the population of the primary health and diabetic Centers at King Fahad Armed Forces Hospital. Participants were defined as having prediabetes according to selfreport, clinical reports, use of antidiabetic agents and HbA1c (5.7-6.4). All data were collected by personal interview and on the basis of a review of electronic medical data. Weight (kg) and height (cm) were measured by physician and nurse interviewers and recorded. The optimal sensitivity and specificity using different BMI cut-off values to predict the presence of prediabetes were examined by receiver operating characteristic curve analysis. Results: Of the 5498 participants analyzed, 2049 (37.3%) were male and 3449 (62.7%) were female with female to male ratio 1.7:1. Age was 42.7 ± 15.8 (minimum 18 years and maximum 105 years ) and males were significantly older than females (45.3 ± 16.6 vs. 41.1 ± 15.1 respectively, p<0.0001). BMI was 29.6 ± 6.9Vwhere females had higher BMI than males (30.1 ± 7.5 vs. 28.8 ± 5.5 respectively, p<0.0001). Prediabetes was present in 579 (10.5%), 89 (15.4%) were male and 490 (84.6%) were female with female to male ratio 5.5:1. Diagnostic performance of BMI in detecting prediabetes using optimal BMI cut-off values based on the shortest distance in ROC curve showed BMI values ranged from 28.50 to 29.50 in total population, 27.50 to 28.50 in male and from 28.50 to 29.50 in female. The AUC was 0.560 (95% CI, 0.503- 0.617) in male and 0.517 (95% CI, 0.490-0.544) in female. The predictive value of BMI in detecting prediabetes using BMI cut-off values based on the lowest significant association between BMI and the risk factors from the logistic regression analysis showed that the risk of prediabetes was significantly increased at BMI values as low as 21 kg/m2 and has not been increased as BMI increased. Applying this criterion to identify the cut-off values resulted in improvements in sensitivity, false negative rate and worsening in specificity and false positive rate. Using these lower BMI cut-offs resulted in a very small false negative rate ranging from 0.07 to 0.09. A significant positive association for total population was observed with BMI values starting at 21 kg/m2 and non significant association with BMI values for both genders. Conclusion: The diagnostic usefulness of BMI alone in defining obesity as risk for prediabetes is limited among men and women Saudi adults.