Validation of a self-administered form for recording food intake in hospital patients

Obejctive: Validation a self-administered form used by patients to record their food intake and compare the recorded data with the observed intake. Design: Data were obtained from an unselected cross-sectional group of hospitalized patients. Subejcts: Forty-five adult men and women volunteered to participate. Five of these dropped out. Methods: Observed intake at breakfast, lunch and dinner was obtained by recording the servings of food before they were served to the patients and subtracting weighed leftovers. At meal times the patients recorded food items eaten in fractions of amount served to the nearest 25%. Setting: Inpatients from five different wards at Rikshospitalet, Oslo. Results: There was a significant under-reporting of the number of foods served (P < 0.005) resulting in a significant underestimation of energy 231 kJ (P < 0.02). There was good agreement between the patients and the observers for the portions of most foods (Kappa 0.44-0.92, P < 0.00001). The differences in amount had little influence on the difference in total energy. The difference in number of foods correlated with the difference in energy (r = 0.68, P < 0.001) and with the difference in protein (r = 0.50, P < 0.01). Patients with an under estimation of energy above 20% had forgotten seven or more food items. Conclusions: For most patients, the self-administered form adapted to the hospital menu appears to have acceptable validity, but for some patients it was unacceptable, mainly owing to food items being omitted and not because of incorrect estimate of amounts of food.