Application of a new statistical model for measurement error to the evaluation of dietary self-report instruments

Authors: FREEDMAN, LAURENCE - The Sheba Medical Center At Tel Hashomer; MIDTHUNE, DOUGLAS - National Cancer Institute (NCI, NIH); CARROLL, RAYMOND - Texas A&M University; COMMINS, JOHM - Information Management Services, Inc; ARAB, LENORE - University Of California; Baer, David; MOLER, JAMES - Information Management Services, Inc; Moshfegh, Alanna; NEUHOUSER, MARIAN - Fred Hutchinson Cancer Research Center; PRENTICE, ROSS - Fred Hutchinson Cancer Research Center; Rhodes, Donna; SPIEGELMAN, DONNA - Harvard School Of Public Health; SUBAR, AMY - National Cancer Institute (NCI, NIH); TINKER, LESLEY - Fred Hutchinson Cancer Research Center; WILLETT, WALTER - Harvard School Of Public Health; KIPNIS, VICTOR - National Cancer Institute (NCI, NIH)

Submitted to: American Journal of Epidemiology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/28/2015
Publication Date: 11/1/2015
Citation: Freedman, L.S., Midthune, D., Carroll, R.J., Commins, J.M., Arab, L., Baer, D.J., Moler, J.E., Moshfegh, A.J., Neuhouser, M.L., Prentice, R.L., Rhodes, D.G., Spiegelman, D., Subar, A.F., Tinker, L.F., Willett, W., Kipnis, V. 2015. Application of a new statistical model for measurement error to the evaluation of dietary self-report instruments. American Journal of Epidemiology. 26(6):925-933.

Interpretive Summary: Most statistical methods that adjust analyses for dietary measurement error treat an individual’s usual intake as a fixed quantity. However, in reality usual intake, if defined as the average intake over a few months, it will vary over the course of a year or more. The objective was to describe a new model that accounts for usual dietary intake variation over the course of a year or longer and for the proximity of biomarker measurements to self-reports within the framework of a meta-analysis. The new model was applied to the analysis of data on energy, protein, potassium and sodium and their densities using recovery biomarkers as reference instruments from five large validation studies conducted between 1999-2009 on U.S. adult population groups. Using this model, we estimated attenuation factors and correlations with true longer-term usual intake for single and multiple 24 hour dietary recalls (24HRs) and food frequency questionnaires (FFQs) and compared them with those obtained by the usual method. Compared to the usual method, the estimates using the time-varying model showed slightly larger values of the attenuation factor and correlation coefficient for FFQs and smaller values for 24HRs. With the new model, while four 24HRs had higher estimated correlations with truth than a single FFQ for absolute intakes of protein, potassium and sodium, for densities the correlations were approximately equal. Accounting for the time element in dietary validation is potentially important, and points towards the need for longer-term validation studies.

Technical Abstract: The objective was to describe a new model that accounts for usual dietary intake variation over the course of a year or longer and for the proximity of biomarker measurements to self-reports within the framework of a meta-analysis. The new model was applied to the analysis of data on energy, protein, potassium and sodium and their densities using recovery biomarkers as reference instruments from five large validation studies conducted between 1999-2009 on U.S. adult population groups. Using this model, we estimated attenuation factors and correlations with true longer-term usual intake for single and multiple 24 hour dietary recalls (24HRs) and food frequency questionnaires (FFQs) and compared them with those obtained by the usual method. Compared to the usual method, the estimates using the time-varying model showed slightly larger values of the attenuation factor and correlation coefficient for FFQs and smaller values for 24HRs. In some cases, the difference between the usual estimate and new estimate for multiple 24HRs was substantial. With the new model, while four 24HRs had higher estimated correlations with truth than a single FFQ for absolute intakes of protein, potassium and sodium, for densities the correlations were approximately equal. Accounting for the time element in dietary validation is potentially important, and points towards the need for longer-term validation studies.