Modeling Longitudinal Observations with Excess Zeros and Measurement Error, with Application to Nutritional Epidemiology

• Activity Number: 487
• Type: Invited
• Date/Time: Wednesday, August 3, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Victor Kipnis*+ and Raymond James Carroll and Laurence S. Freedman and Douglas Midthune
• Companies: National Cancer Institute and Texas A & M University and Gertner Institute for Epidemiology and Health Policy Research and National Cancer Institute
• Keywords: regression calibration ; episodically consumed foods ; correction for measurement error

Abstract

Food frequency questionnaires have been the instrument of choice to assess long-term dietary intake in most cohort studies of diet and disease. It is now well appreciated that such questionnaires involve substantial measurement error, random and systematic, leading to distorted effects of diet. A popular method for correcting for this error, regression calibration, assumes a substudy with an unbiased short-term reference instrument. Application of this method to foods that are not consumed every day by everyone is problematic, since short-term reference instruments usually include a substantial proportion of subjects with zero intakes, leading to observations with excess zeros, in addition to within-person measurement error. Also, dietary intake is often analyzed relative to total energy intake to take into account dietary composition. We present a new bivariate model for short-term reference observations on food and total energy intakes with application to calibration of energy-adjusted associations between foods and disease. We exemplify the methodology by applying it to data from the US NIH-AARP Diet and Health Study and study its properties using simulations.