DESCRIPTION: (Adapted from the Applicant's Abstract): Measurement error in
risk factors is a commonly identified problem in cancer and nutritional
research. Many important examples can be found in epidemiological
investigations of the cancer prevention/ promotion properties of dietary
nutrients. Uncertainties in the reliability of questionnaire data, the
relevant time frame of observation, and the nutrient composition of food can
obscure and bias the effects observed in conventional studies. Measurement
error methods for epidemiological regression models attempt to deal with
this issue by incorporating information on the magnitude of uncertainties in
exposure estimates into procedures for designing and analyzing
epidemiological studies. New methods are proposed for a number of practical
and technical issues related to the measurement error problem and will be
applied to important ongoing studies in cancer epidemiology. The specific
aims fall into three broad areas: methods for exposure measurement error in
nutritional epidemiology, design issues for studies with measurement error,
and issues of special importance in longitudinal studies. In nutritional
epidemiology, a comprehensive approach will be developed for assessing the
relative impact of all major sources of errors in the assessment of nutrient
intakes. Ongoing research on special methods for ordinal categorizations of
nutrient intake will be extended to include simultaneous assessment of
several nutrients. In addition, further research on estimating the
reliability of noisy dietary measures for designs where: (a) an unequal
number of replicates per subject are available or (b) normality of the
dietary measures are not assumed will be performed. The design issues
feature further development and application of multi-stage adaptive designs
involving resampling for validation studies. Finally, new longitudinal
methods are proposed for dealing with measurement error in a risk factor
expressed as a proportion, such as the proportion of follow-up surveys
reporting aspirin use, and for estimation of the variance parameters of
random effects in a longitudinal validation study of dietary intakes and
nutrient concentrations in serum.
Error Notice
The database may currently be offline for maintenance and should be operational soon. If not, we have been notified of this error and will be reviewing it shortly.
We apologize for the inconvenience.
- The DCCPS Team.
