Talks
Presentation Type
Oral Presentation
Discipline Track
Community/Public Health
Abstract Type
Research/Clinical
Abstract
Background: Early diagnosis of type 2 diabetes (T2D) can delay its vascular complications, but the active search for T2D results in inefficient use of resources. Simple methods like anthropometry or clinical history could help prioritize laboratory testing in individuals with risk of T2D. We evaluated non-invasive clinical measurements to pre-screen for new T2D patients.
Methods: We used the 2017-2018 NHANES database as the discovery cohort, and identified age, sex, BMI, waist circumference, and family history of T2D (FHD) as predictor variables. Risk of new T2D was analyzed using odds ratios with logistic regression. The diagnostic accuracy was calculated using ROC curves. A binational cohort from both sides of the Rio Grande Valley border (RGV) was used for replication.
Results: Using NHANES, we identified 103 new cases of T2D among 3,037 participants. The odds of T2D were 3.83 for those over 40 years, between 1.49 and 6.07 for obesity classes, and 1.76 for FHD. The constructed score with age, BMI, and FHD had an area of 0.78 (sensitivity 0.88 and specificity 0.48). The model was twice as likely to detect new T2D among minority groups (Hispanics, Asian Americans, and African Americans vs. White Americans). The replication group showed better prediction and higher sensitivity for detection of new T2D for scores below 100 (e.g., sensitivity: 0.96 vs. NHAHES 0.88 at score of 70).
Conclusions: Our model detected 90% of the new T2D patients at a score of 70. We are now evaluating the gain in quality adjusted life-years with early T2D detection.
Academic/Professional Position
Undergraduate
Mentor/PI Department
Population Health and Biostatistics
Recommended Citation
Rios, Xavier; Restrepo, Blanca I.; and Lopez-Alvarenga, Juan Carlos, "Development and validation of a simple clinical construct for prediction of new type 2 diabetes mellitus" (2023). Research Symposium. 12.
https://scholarworks.utrgv.edu/somrs/theme1/track1/12
Included in
Development and validation of a simple clinical construct for prediction of new type 2 diabetes mellitus
Background: Early diagnosis of type 2 diabetes (T2D) can delay its vascular complications, but the active search for T2D results in inefficient use of resources. Simple methods like anthropometry or clinical history could help prioritize laboratory testing in individuals with risk of T2D. We evaluated non-invasive clinical measurements to pre-screen for new T2D patients.
Methods: We used the 2017-2018 NHANES database as the discovery cohort, and identified age, sex, BMI, waist circumference, and family history of T2D (FHD) as predictor variables. Risk of new T2D was analyzed using odds ratios with logistic regression. The diagnostic accuracy was calculated using ROC curves. A binational cohort from both sides of the Rio Grande Valley border (RGV) was used for replication.
Results: Using NHANES, we identified 103 new cases of T2D among 3,037 participants. The odds of T2D were 3.83 for those over 40 years, between 1.49 and 6.07 for obesity classes, and 1.76 for FHD. The constructed score with age, BMI, and FHD had an area of 0.78 (sensitivity 0.88 and specificity 0.48). The model was twice as likely to detect new T2D among minority groups (Hispanics, Asian Americans, and African Americans vs. White Americans). The replication group showed better prediction and higher sensitivity for detection of new T2D for scores below 100 (e.g., sensitivity: 0.96 vs. NHAHES 0.88 at score of 70).
Conclusions: Our model detected 90% of the new T2D patients at a score of 70. We are now evaluating the gain in quality adjusted life-years with early T2D detection.