Malnutrition in hospitalised patients is often underestimated. The present study aimed to evaluate the validity of a questionnaire for the semi-quantitative evaluation of food intake compared to weighed records in patients who were hospitalised for the rehabilitation of neurological disorders.
Food intake at breakfast, lunch and dinner was evaluated in 319 in-patients, by weighing the meals and the residuals, and using a semi-quantitative questionnaire, during five consecutive days. The questionnaire represented, for each offered food, the pictures of the nonconsumed quantities. The consumption of each food was determined by weighing foods that were served and the residuals after the meal. As a measure of validity of the questionnaire, the agreement over chance (kappa statistic) between the questionnaire and the weight was calculated. Considering the weight as the gold standard, the sensitivity and specificity of the questionnaire in detecting patients who consumed <50% or 75% of the meals was calculated.
The agreement between the two measures was satisfactory (κ ≥ 0.70) or almost satisfactory (0.60 < κ < 0.70) for most of the foods, with the exception of fruit and the first course at dinner. The sensitivity and specificity of the questionnaire in detecting consumers of <50% or 75% of the offered foods were always >80%, except for bread and first course, as well as fruit at dinner.
The present study shows that this semi-quantitative questionnaire on food consumption reproduces with sufficient precision the measures obtained by weighing. The questionnaire appears also to be a valid and suitable instrument for the identification of patients with poor food intake in a neurorehabilitation hospital.