An assessment of the use of simple methods to predict individual energy intakes for intervention studies

Objective: To investigate to what extent individual energy intakes can be predicted by rapid easily available low-cost estimation methods. Design: Data were obtained from a controlled dietary intervention study period of nine weeks in which the subjects should be weight stable. Subjects: Thirty-one male students in domestic and kitchen management aged 29 +/- 6 y. Methods: (Ij energy intake calculated from a quantitative food frequency questionnaire (FFQEI); (2) energy expenditure derived from estimates of basal metabolic rate (BMR) (fAO/WHO/UNU, 1985) based on weight, gender, age and low (1.55 x BMR), medium (1.78 x BMR) or high (2.10 x BMR) level of. activity. Level of activity was determined by questions concerning habitual activities lasting more than 20 min (WHOEE); (3) energy expenditure derived from individual recording in a specially prepared activity diary (ADEE). During the intervention, the subjects were to be fed test diets which should provide them with enough energy to keep them weight stable. The energy levels were established after taking both the FFQEIs, WHOEEs and ADEEs into consideration, and 10 MJ, 13 MJ, 15 MJ and 17 MJ per day were chosen because these levels were estimated to closely match the energy requirements of most of the subjects. The levels of energy were changed during the intervention period if the weight of the subjects fluctuated. The served level of energy at the last day of the intervention was denoted the weight maintenance energy intake (WMEI). WMEI was compared to FFQEI, WHOEE and ADEE in order to evaluate if one estimation method predicted WMEI better than the two others. Results: None of the three methods provided accurate estimates of WMEI of 13.3 +/- 1.8 MJ. However, WHOEE, gave the best estimate as demonstrated by the limits of agreement: -8.7 MJ to + 8.9 MJ for FFQEI, -5.4 MJ to +3.9 MJ for WHOEE and -7.2 MJ to +5.2 MJ for ADEE. The coefficients of correlation between the differences and the means of WMEI and FFQEI, WHOEE and ADEE were -0.8 (P less than or equal to 0.001), 0.1 (P = 0.6, NS) and -0.5 (P less than or equal to 0.01), respectively. The coefficients of variation were 34.6% for FFQEI, 11.3% for WHOEE, and 21.0% for ADEE. Conclusions: Although not precise, WHOEE showed the best agreement with the WMEI. These results demonstrate that a rapid and simple low-cost method predicted WMEI closely enough to avoid major weight fluctuations among these men during the intervention period.