Reliability and Validity of a Short Dietary Intake Questionnaire for Retrospective Collection of Nutrients during Gestation in Autism Studies

Background:  

Nutrition needs increase during pregnancy and are critical for brain development. Evidence is accumulating for a role of gestational nutrition in autism etiology.  A new shorter tool for collection of maternal nutritional intake during pregnancy could facilitate research in this area.

Objectives:  

To develop a tool to collect dietary and supplement intake during pregnancy with a focus on nutritional factors and timing relevant to neurodevelopment and likely to influence autism risk, and assess the tool’s reliability and validity in a high-risk ASD population.

Methods:

Candidate nutritional factors were selected based on a thorough literature review.  Retrospective assessments of nutritional intakes during pregnancy were included in the ELEAT (Early Life Exposure Assessment Tool) during a pilot test conducted with participants from the MARBLES pregnancy cohort study of high-risk siblings of children with autism. Retrospective responses were compared with responses to supplement intake questions and/or the previously validated 2005 Block food frequency questionnaire (FFQ) prospectively collected by MARBLES during a pregnancy at least 2 years previously. Longer and shorter versions of the ELEAT dietary module were tested.  Nutrient values were calculated for the ELEAT dietary module using reported frequency of intake and nutrient values for foods from the USDA nutrient database obtained through NDSR.  Agreement between retrospectively assessed food and nutrient intakes and prospectively reported intakes based on supplement questions and the Block FFQ were evaluated using Kappa coefficients, Spearman Rank Correlation Coefficients (r_s) and Concordance Correlation Coefficients (ccc).  Asymptotic 95% confidence intervals (CIs) for Spearman correlations are based on Fisher’s Z transformation. 

Results:  

Supplement questions in both MARBLES and the ELEAT were completed by 114 women. MARBLES FFQ dietary intakes were compared among 54 women who completed the ELEAT long form and among 23 who completed the ELEAT short form.  Kappas were moderate for most supplements on whether or not they were taken, but modest for timing. Correlations across most individual food items and categories were fair to moderate on both the long and short ELEAT modules.  Most quantified nutrient values from the long form of the ELEAT were moderately correlated (r_s= 0.3 – 0.5) with those on the Block FFQ.  More nutrients, especially aggregate measures, based on the short ELEAT module had only modest, weak or even inverse correlations with the FFQ, however primary nutrients of interest displayed strong correlations: dietary folate equivalents, r_s=0.56 (CI: 0.19, 0.79); iron, r_s=0.59 (CI: 0.23, 0.80); fiber, r_s=0.52 (CI: 0.14, 0.77).

Conclusions:  

Responses on the ELEAT long form dietary and supplement modules were moderately reliable overall, even with recall after several years, and produced nutrient values that were moderately correlated to previously collected prospective measures. As with all FFQs, the ELEAT dietary module is not meant to assess exact nutrient intake for each participant, but rather can be used to rank participants on their responses in terms of food group intake, calcium, iron, folate, potassium, fiber, choline, vitamin K and vitamin C intake.  This short ELEAT dietary module can be added to autism studies to retrospectively assess maternal nutrient contributions to ASD etiology.