The relative validity of the repeated 24 h recall for estimating energy and selected nutrient intakes of rural Ghanaian children.