|
Activity Number:
|
172
- Prediction and Misclassification in Biomedical Research
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
|
Abstract #313352
|
|
|
Title:
|
Practical Considerations When Building Prediction Equations for Blood Concentration Markers: Lessons Learned from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL)
|
|
Author(s):
|
Lillian Boe* and Pamela A Shaw and Daniela Sotres-Alvarez and Martha Daviglus and Robert Kaplan and Yasmin Mossavar-Rahmani and Ramon A Durazo-Arvizu
|
|
Companies:
|
University of Pennsylvania and University of Pennsylvania Perelman School of Medicine and University of North Carolina at Chapel Hill and University of Illinois College of Medicine and Albert Einstein College of Medicine; and Fred Hutchinson Cancer Research Center and Albert Einstein College of Medicine and University of Illinois College of Medicine
|
|
Keywords:
|
biomarker;
chronic disease;
diet;
measurement error;
prediction
|
|
Abstract:
|
Measurement error is a major issue in self-reported diet that can distort diet-disease relationships. Serum biomarkers avoid the bias in self-report. We developed prediction equations for serum biomarker levels of micronutrients assessed in the Study of Latinos: Nutrition and Physical Activity Assessment Study (SOLNAS) (n=445) and examined whether predictive accuracy was high enough to reliably detect underlying diet-disease associations in the parent Hispanic Community Health Study/Study of Latinos (HCHS/SOL) cohort (N=16,415) . Blood concentration biomarkers were collected for carotenoids, tocopherols, retinol, vitamin B12 and folate. Due to expense biomarkers were only available for the SOLNAS subset. We build regression-based prediction equations for 11 biomarkers based on participant characteristics, including two self-reported 24-hour dietary recalls. We used simulations to study the power of detecting the association between a true average concentration marker and a hypothetical incident diabetes survival outcome using the predicted biomarker level. While good power was observed for some nutrients, further work is needed to realize the full potential of dietary biomarkers.
|
Authors who are presenting talks have a * after their name.
