Integrating Data from Multiple Time-Location Measurement Methods for Use in Exposure Assessment: The Multi-Ethnic Study of Atherosclerosis and Air Pollution (MESA Air)

Background: Environmental epidemiologic studies have often used time-location diaries (TLDs) and questionnaires to measure time spent in various microenvironments, but recently more studies have utilized GPS tracking.

Methods: Three tools for time-location monitoring were used simultaneously during two 2-week periods in a small subset (n=64) of the MESA Air cohort: GPS dataloggers, proximity monitors, and TLDs. GPS data were analyzed using a rule-based model based on time, speed, and spatial location. Proximity monitors developed specifically for this study to compensate for the limited resolution of the GPS units were placed in each participants home and vehicle, providing an aggregate count of the hours spent in each of those locations. Participants also completed comprehensive TLDs including information such as traffic conditions. These data were integrated to produce a single best estimate of time spent in each of five microenvironments: home indoors, home outdoors, other indoors, other outdoors, and in-vehicle.

Results: Initial analyses (n=26) suggest that the GPS data provided the most reliable estimates of time spent in-vehicle (mean 8.8±7.0 hrs/wk) as these units are not subject to recall bias (such as the TLDs) or human error in following the monitoring protocol (such as the proximity monitors). While GPS data were used to classify time at home versus other locations, these units were not precise enough to differentiate time spent indoors and outdoors at the same location. Proximity monitor data further divided time at home into indoors (mean 139.5±14.1 hrs/wk) and outdoors (mean 0.91±2.8 hrs/wk). Time in other locations was subdivided using TLDs: 15.8±9.1 hrs/wk indoors and 2.5±5.1 hrs/wk outdoors.

Discussion: This study combines multiple methods for characterizing time-location patterns among five microenvironments, each with distinct strengths and limitations. At this point no single method provides a gold standard for use in epidemiologic analyses.