1. ¹Departamento de Medicina Social, Universidade Federal de Pelotas, Pelotas, RS, Brazil
2. ²MRC Human Nutrition Research, Elsie Widdowson Laboratory, Cambridge, UK
3. ³Zoological Laboratory, Groningen University, Haren, the Netherlands
4. ⁴Centre for Isotope Research, Groningen, the Netherlands

Correspondence: H Haisma, Zoological Laboratory, Groningen University, Kerklaan 30, 9751 NN Haren Gn, Netherlands. E-mail: h.haisma@biol.rug.nl

Received 13 December 2003; Revised 2 June 2004; Accepted 17 June 2004; Published online 22 September 2004.

Abstract

Objective: Firstly, to compare food, and macronutrient intake as obtained from a single 24-h recall and a frequency questionnaire (FQ) covering a 14-day period in breast-fed infants aged 4 months of age. Secondly, nonbreast milk water intake (NB-WI, ml/day) was used as an estimation of energy and macronutrient intake, and NB-WI as calculated from FQ (NB-WI_FQ) was compared with NB-WI as measured using the dose-to-the-mother ²H₂O turnover method (NB-WI_DO) covering the same 14-day period.

Design: Cross-sectional.

Setting: Community-based study in urban Pelotas, Southern Brazil.

Subjects: In all, 67 breast-fed infants aged 4 months of age recruited at birth.

Main outcome measures: (1) Bias in estimations of food and macronutrient intake of the 24-h recall relative to FQ; (2) Bias in NB-WI_FQ relative to NB-WI_DO.

Results: In infants with an energy intake_FQ from complementary foods above the 50th percentile (1.03 kcal/day), estimations of water, tea, juice, and milk intake were not different between 24-h recall and FQ (n=34). Nor were estimations of energy and macronutrient intake (protein, fat, and carbohydrates) different between the two methods, and bias was nonsignificant. NB-WI_DO was divided into quintiles and compared with NB-WI_FQ. The first two quintiles included negative values for NB-WI_DO as a result of random errors of the ²H₂O turnover method. Subsequently, bias of NB-WI_FQ relative to NB-WI_DO was positive in the 1st (P=0.001) and 2nd quintile (P=0.638), respectively. Bias was negative for the three highest quintiles, and within this group, underestimation by FQ was significant for the 3rd and 4th quintile (-57.4%, P=0.019; -43.7%, P=0.019).

Conclusions: Firstly, at the age of 4 months FQ covering a 14-day period provides similar results on food and macronutrient intake as compared to a single 24-h recall for estimations of complementary liquid foods. Secondly, NB-WI_FQ appeared to be a good proxy for macronutrient and energy intake in breast-fed infants receiving other liquids. In infants with NB-WI_DO>0, the method provides a useful tool for the detection of bias from FQ, and results indicate an underestimation from FQ relative to the ²H₂O turnover method. This exercise could be applied wherever the ²H₂O turnover method is used in combination with conventional food consumption techniques for measuring intake of nonbreast milk liquids of breast-fed infants in whom solid foods have not yet been introduced. It would help interpreting estimations of macronutrient intake, and could be relevant to studies of dietary intake of infants and its relationship with growth and health.

Sponsorship: International Atomic Energy Agency through RC 10981/R1.