Reliability and concurrent and construct validity of a food frequency questionnaire for pregnant women at high risk to develop fetal growth restriction

Abstract

Accuracy of dietary assessment instruments such as food frequency questionnaire (FFQ) is crucial in the evaluation of diet-disease relationships. Test-retest reliability and concurrent and construct validity of a FFQ were evaluated in 150 pregnant women at high risk to develop fetal growth restriction randomly selected from those included in the improving mothers for better prenatal care trial Barcelona (IMPACT BCN). The FFQ and dietary records were performed at baseline and 34-36 weeks of gestation. Test-retest reliability of the FFQ for 12 food groups and 17 nutrients was moderate (ICC = 0.55) and good (ICC = 0.60), respectively. Concurrent validity between food, nutrients and a composite Mediterranean diet score (MedDiet score) and food records was fair for foods and nutrients (ρ average = 0.38 and 0.32, respectively) and moderate (r = 0.46) for the MedDiet score. Validation with biological markers ranged from poor (r = 0.07) for olives to moderate (r = 0.41) for nuts. A fair concordance between methods were found for nutrients (weighted κ = 0.22) and foods (weighted κ = 0.27). The FFQ-derived MedDiet score correlated in anticipated directions with intakes of nutrients and foods derived by food records. The FFQ showed a moderate test-retest reliability and reasonable validity to rank women according to their food and nutrient consumption and adherence to the Mediterranean diet.