Abstracts: ISEE 20th Annual Conference, Pasadena, California, October 12–16, 2008: Contributed Abstracts
The goal of the SUPERB Study (Study of Use of Products and Exposure-Related Behaviors) is to determine optimal data collection methods for environmental exposures in longitudinal epidemiologic studies, taking into account acceptability and feasibility, as measured by retention, compliance, reliability, and error rates.
Three data collection modalities address exposure-related behaviors: annual computer assisted telephone interviews (Tier I); a series of eighteen monthly web-based surveys (Tier II); passive data collection involving home visits (Tier III). Domains of interest include food, household products and time-activity information, with a special focus on pesticides, metals, phthalates, acrylamide, PAHs, and benzene. The passive measures of exposure in Tier III involve weighing and scanning barcodes of food items, pesticides, and cleaning, other household and personal care products twice per season, separated by 1 week. During this week of monitoring, a video camera in the kitchen records food preparation methods, and subjects are asked to wear a GPS and activity monitor, as well as to collect food receipts. Participants are California residents and include young children and their parents, and older adults. When the study is concluded, de-identified data will be provided to the EPA to augment the Consolidated Human Activity Database (CHAD).
A total of 680 households have participated in Tier I; 1/3 are enrolled in at least one other tier. Lengthy interviews were sometimes difficult to complete, particularly in families with small children, and required numerous call-backs. Despite extensive efforts taken to facilitate the web-based survey and to make it user-friendly, retention in this Tier has been low, with well over 50% of participants not completing three consecutive months. In contrast, compliance and retention in the passive monitoring (Tier III) has been very high, with fewer than 10% of participants dropping out of this protocol (for most participants, this meant four seasons, each with two home visits). With regard to the quality of data obtained: 37% of consumer products found in at least one home contained a barcode that provided ample information from publicly available databases to identify the product. Forty five percent of the products had an uninformative barcode; these typically were store brand products and were identified by field staff and then added to our database such that these products became identifiable at other homes and on subsequent visits. Nine percent of the products had no barcode. At the end of the week, field staff were able to find and reweigh 91% of cleaning and 87% of personal care products. Passive measures to record food consumption and activity were also collected. On average, the first house visit took 2 hours, with the participant being involved for approximately 35 minutes, and the participant's time was reduced to 13 minutes on subsequent visits.
Our results suggest that longitudinal epidemiologic studies may benefit from the use of passive monitoring methodology without sacrificing data quality. The database of barcodes that the SUPERB study is assembling will be a valuable resource for other researchers.