A 24-h recall does not provide a valid estimate of absolute nutrient intakes for rural women in southern Ethiopia

ObjectiveTo investigate the relative validity of a multiple-pass interactive 24-h recall (24-HR) for assessing nutrient intakes of a group of rural Ethiopian women.MethodsDietary intakes from a 24-HR were compared with weighed record (WR) intakes collected on the same day for 58 women of child-bearing age from three subsistence farming communities in Sidama. The impact of memory lapses, use of average recipe data, and inaccurate portion size estimates on 24-HR intakes was also assessed.ResultsMedian daily intakes of energy and most nutrients (except fat and phytate) were lower (P < 0.05) by the 24-HR versus the WR. Controlling for energy densities decreased the discrepancies to lower than ±6%, except for vitamin C. No significant differences between the two methods existed for the contribution of six food groups to energy intakes (percentages). Fewer than 40% of 24-HR intakes were ±10% of the WR. After classifying intakes (per day) into quartiles, Cohen’s κ values were poor (<4.0) for protein, iron, retinol, and dietary fiber and fair (≥0.40 to ≤0.75) for energy and other nutrients; all values per megajoule were mostly fair. Bland-Altman plots confirmed a negative bias for daily energy and nutrient intakes with the 24-HR versus WRs. Discrepancies were attributed mainly to inaccurate portion size estimates.ConclusionIn this setting, the 24-HR cannot be substituted for the WR to assess absolute nutrient intakes for a group or the prevalence of inadequate intakes but could be used for energy-adjusted intakes. Prior training may improve the accuracy of the 24-HR.