Reliability and Validity of Self-Reported Home Environment in Autism Studies

Friday, May 12, 2017: 12:00 PM-1:40 PM
Golden Gate Ballroom (Marriott Marquis Hotel)
P. Krakowiak1, D. Bennett2, D. J. Tancredi3, I. Hertz-Picciotto4, C. K. Walker5 and R. J. Schmidt6, (1)2825 50th Street, UC Davis, Sacramento, CA, (2)University of California, Davis, Davis, CA, (3)UC Davis School of Medicine, Sacramento, CA, (4)University of California at Davis, Davis, CA, (5)University of California, Sacramento, CA, (6)Public Health Sciences, University of California Davis, Davis, CA
Background:  Questionnaires can help advance research on environmental risk factors for autism when more detailed measurements are too expensive or not feasible, but only when these data are reliable and valid compared with gold standard measures.

Objectives:  To compare maternal retrospective report of home characteristics that can be used to assess environmental exposures during pregnancy and until child’s first birthday on the ELEAT (Early Life Exposure Assessment Tool) with prospectively collected responses, prenatal and postpartum home walkthrough visits, and public records in ASD-affected families.

Methods:  Participants (n=120) from the MARBLES (Markers of Autism Risk in Babies-Learning Early Signs) prospective cohort study of high-risk younger siblings of children with autism completed a structured telephone interview during the 1st half of pregnancy (Environmental Exposures Questionnaire [EQ1]) and then again with the ELEAT, a shorter instrument administered 2 or more years postpartum. MARBLES Study also conducted a home walkthrough visit at the homes of participating families during mid-pregnancy and again at 6 months postpartum. Additional information about the home was obtained from public records (Zillow.com). Home characteristics compared with the EEQ included type of residence, tap water source, and whether the residence was near an agricultural field or golf course. Home square footage reported on ELEAT was compared with public records. Planned analyses for comparisons with home walkthroughs will include type of residence, year/decade the home was built, heat and air conditioning sources, and type of flooring. Reliability was assessed with Cohen’s Kappa statistic (K); and validity with sensitivity (Se), specificity (Sp) and Youden’s index (Y=Se+Sp-1) for each home characteristic for matched addresses from the index time period (conception until child’s 1st birthday). Analyses were restricted to one address per family if more than one was reported during the index period.

Results:  Ninety-two families were included in analyses comparing ELEAT responses with EQ1. Over 80% of families lived in a single family home. Mothers recalled the type of residence (single family home vs. multiplex/apartment/condo) with 100% accuracy (K=1.00). Agreement coefficients for water supply (public supply, private well) and ¼ mile proximity to agricultural field or golf course were fair (K=0.30 and 0.38, respectively). Eighty-one families were included in the home square footage analysis, compared with public records. Home area was divided into 4 categories: <1000 sq. ft., 1000-1999, 2000-3000, and >3000. However, very few families lived in homes <1000 or >3000 sq. ft. and we used collapsed categories <2000 and ≥2000 instead. Agreement was substantial (K=0.77).

Conclusions:  Mothers of children with autism tended to recall information about the home they lived in during pregnancy and the child’s first year reliably, and their responses to home size were highly valid. However responses on proximity of their homes to agricultural fields and golf courses, which can be used to assess likelihood for drift pesticide exposure, were not as reliable, suggesting other methods of drift pesticide exposure are likely needed.

See more of: Epidemiology
See more of: Epidemiology