Objective: The study was conducted to assess the relative validity of a 170-item semiquantitative food frequency questionnaire (SFFQ) adapted for use in the elderly.
Design and subjects: The study was carried out in a sample of 80 men and women aged 55-75 y participating in a community based prospective cohort study in Rotterdam, The Netherlands. The two-step dietary assessment comprised a simple self-administered questionnaire (20 min) followed by a structured interview with trained dietitians (20 min) based on the completed questionnaire. Multiple food records (FR) collected over a one year period served as reference method. 24 h urine urea was used as indirect marker for protein intake.
Results: Compared with FR, the SFFQ generally overestimated nutrient intake as reflected by difference in means and the ratio of SFFQ to FR. Energy adjustment reduced the observed overestimation. Pearson's correlation coefficients varied from close to 0.5 to about 0.9 for crude data, and after adjustment for age, sex, total energy intake, and for within-person variability in daily intake for 0.4-0.8. Cross-classification into quintiles resulted in correct classification into the same or adjacent quintile of 75.8% for crude and 76.8% for energy-adjusted data. Validation of protein intake estimated by SFFQ with protein excretion from 24h urine urea indicated overestimation of protein intake by SFFQ. Spearman correlation coefficient between protein intake estimated from urea excretion and SFFQ was 0.67.
Conclusions: Adaptation of a SFFQ for use in the elderly resulted in a valid and time-efficient dietary assessment instrument. Its ability to adequately rank study subjects according to their dietary intake support its application in epidemiological studies in the elderly.