Assessment of body composition in lactating mothers in a rural African community using deuterium oxide

Abstract

Background and objective. The deuterium oxide (DO) dilution method for measuring body composition was validated against the widely used skinfold (SF) measurement-based equation of Durnin and Womersley. The study involved 10 lactating women living in a rural community in Nandi, Kenya and participating in a cross-sectional study aimed at determining their iron and vitamin A status.

Methods. The selection criteria were exclusive breast-feeding, infants between 2 and 4 months of age, maternal parity < 4, birth weight more than 2 500 g and no congenital abnormalities. Maternal and infant anthropometric measurements were taken. DO (approximately 0.1 g 2H2O/kg body water) was given orally to each mother accordingly to the Maastricht protocol (DO) for total body water (TBW) determination. The 2H enrichment of the urine was measured using gas-isotope-ratio mass spectrometry. Blood samples were collected. Serum retinol and ferritin were determined using high- performance liquid chromatography (HPLC) and enzyme-linked immunosorbent assay (ELISA) respectively. Body mass index (BMI) was determined as weight/height (m2). The Bland-Altman pair-wise comparison was used to compare maternal fat-free mass (FFM), body fat (BF) and percentage body fat (% BF) that were determined based on the DO and SF techniques.

Results. Maternal mean (± standard deviation (SD)) for parity, age, BMI, haemoglobin (Hb), serum ferritin and serum retinol were 3 (2), 26 (4) years, 23.4 (4), 12.1 (1.8) g/dl, 10.3 (4.0) µg/l and 0.696 (0.300) µmol/l respectively. The FFM, BF and % BF accordingly to the DO and SF methods respectively were 44.0 (4.7) kg v. 42.9 (5.9) kg, 16.7 (8.8) kg v. 17.8 (7.5) kg and 26.2 (8.1)% v. 28.4 (6.4)%. Limits of agreement for underestimation of FFM and % BF were 4.4 kg (SEE 3.4) and 11.6% (SEE 5.8) respectively. Bias in the measurement of FFM and % BF was 1.1 kg (SEE 1.9), and —2.2% (SEE 3.3) respectively.

Conclusion. The variability (5.8 - 17.5%) observed in the SF technique may result in lower prediction of % BF. This may be an important factor for community-based nutritional interventions that aim at improving the body composition of vulnerable groups such as pregnant and lactating women or subjectswith severe undernutrition.