Abstract

The need for an inexpensive measure of dietary intake in intervention studies led to evaluation of the National Cancer Institute (NCI) Percentage Energy from Fat short instrument (PFat) in a subgroup of the Behavioral Change Consortium (BCC) intervention sites. The PFat's performance was evaluated using multiple nonconsecutive 24-h dietary recalls (24HR) as a reference instrument among participants at baseline in 4 demographically diverse intervention sites of the BCC. Mean estimates of percentage energy from fat for 24HR and PFat were within 2.1 percentage points of each other in all but 2 site/gender comparisons. 24HR and PFat estimates were not significantly different (P < 0.05) among men for 2 of 3 sites, and among women for 2 of 4 sites. Deattenuated Pearson correlation coefficients for the PFat and true intake (as estimated from the 24HR using a measurement error model) were significantly different from 0 (P < 0.05) for men and women in all sites, ranging from 0.52 to 0.77 among men and 0.36 to 0.59 among women. Besides gender and site, no other factors examined (age, education, smoking status, and BMI) consistently moderated validity estimates. If accurate assessment of diet at baseline (and presumably at follow-up) is essential, a more detailed instrument such as multiple 24HR may be warranted. The question of whether the PFat adequately measures change in diet is addressed in another article in this supplement.