How covert are covertly manipulated diets?

OBJECTIVE: To quantitatively assess subjects' ability to detect hedonic (pa latability), sensory and nutritional differences between covertly manipulat ed high-fat (HF) and low-fat (LF) diets. SUBJECTS AND DIETS: This study examined the response of 16 subjects (eight men, eight women) to 20 LF and 20 HF versions of manipulated foods. Average percentage protein:fat: carbohydrate (by energy) and energy density (ED) o f the two diets were 13 : 25 : 62, 371 kj/ 100 g and 13 : 50 : 37, 672 kJ/1 00 g, respectively. PROTOCOL: Subjects were asked to simultaneously assess the HF and LF versio ns of each food in terms of (i) subjective pleasantness of each food, (ii) perceived differences in appearance, smell, taste and texture of the foods, and (iii) for each sample to assess whether it was high or low in energy, protein, carbohydrate, fat, fibre, sugar and salt. ANALYSIS: Perceived pleasantness of HF and LF versions of the foods was com pared by analysis of variance. Comparisons used chi-squared tests of indepe ndence to assess departure from the null hypothesis of no perceived differe nce in remaining parameters (ii-iii). RESULTS: On average, subjects exhibited no significant preference for LF or HF versions of the foods (no difference 15 foods, three HF foods more plea sant, two LF foods more pleasant (P < 0.03)). On average there were no gene ral differences in comparison of sensory attributes that were consistently ascribable to the LF or HF foods, although there were numerous significant differences for individual foods. Subjects appeared unable to distinguish t he HF foods as being HF (66% of estimates) and guessed correctly 33% of the time. They were better able to categorize the LF foods correctly (53% corr ect). On aggregate 43% of HF and LF foods were correctly identified. Subjec ts appeared able to detect sensory differences between foods but not to rel ate them to energy or nutrient content of these foods. CONCLUSIONS: These data suggest that subjects are on average not able to pe rceive large differences in the fat content of diets manipulated in this ma nner. In general the assumption that the manipulation of such foods is cove rt appears to hold, but subjects were far better at correctly identifying c ertain food types (dairy-based) over others.