Use of a Self-Administered Food Frequency Questionnaire in a Population 65 Years and Older

Little is known of the elderly\u27s ability to use a food frequency questionnaire to describe their dietary intake. This study examines the elderly\u27s ability to reliably describe their diets and how age, education, cognitive status, and gender may affect their ability to complete and return a food frequency questionnaire. The reproducibility of the questionnaire was tested by a repeated administration among 85 participants 65 years of age and older from the Preston, Idaho, area. Correlation coefficients between nutrient scores from the first and second administration ranged from 0.48-0.79 (total population), 0.44-0.88 (males), and 0.39-0.86 (females). Median values for the correlation coefficients were 0.60, 0.66, and 0.58 for total population, men, and women, respectively. Response rate and response quality were determined by distributing 4600 questionnaires to the residents of Cache County, Utah, who were 65 years or older. The overall response rate was 82.1%, 83.2% for men, and 81.3% for women. Little difference was found between the age, education level, and cognitive status of respondents compared to nonrespondents. Response quality was defined by the number of missing values per questionnaire. Age had a positive relationship with missing values. The linear regression model had a p-value significant at the p2=0.035), males (r2=0.020), and females (r2=0.044). The years of education had a negative relationship with the number of missing values. The p-value was significant at the p2=0.010), males (r2=0.004), and females (r2=0.018). The relationship between cognitive status and missing values was inconsistent. The p-values were significant at the