Results and recommendations from an American Indian/Alaska Native observational seatbelt survey protocol development project

Background: American Indians/Alaska Native (AI/AN) mortality due to motor vehicle crashes is 3.3. times that of the overall U.S. rate. The Indian Health Service encourages implementation and evaluation of interventions to increase seatbelt use. Considerable variability exists in methods used to identify AI/AN seatbelt use rates. This project developed and reports a uniform means for AI/AN communities to conduct seatbelt surveillance. Methods: Retrospectively, we determined observational data collection methods and seatbelt use rates for 12 Tribes. SUDAAN analysis was used to estimate design effects from the clustered observational survey data. Prospectively, 17 tribes assessed traffic volume, identified a sampling frame, and conducted quarterly observational surveys at seven randomly selected locations. SAS Surveyfreq was used to calculate design effects (DEFF), intraclass correlations (ICC), and 95% confidence intervals of rates of seatbelt use. Results: Driver DEFF ranged from 2.2 to 31.9 (n=10,712 observations) in the retrospective phase and .7 to 13.7 in the prospective phase (n= 28,684 observations). Prospectively, driver ICC ranged from -.002 to .14. Seatbelt use varied considerably across tribes for drivers (22% to 72%) and passengers (11% to 75%). Average driver use was 54% (95% CI=52%-57%) and passenger use 48% (95% CI=45%-51%). Conclusions: Observation location selection procedures are important and the implications of ignoring design effects could be substantial. The final protocol recommends using a probability based sample (determined by average cluster size) that estimates seatbelt use rates with 95% CIs. Lower AI/AN seatbelt use rates highlight the disparity when compared to All-U.S. seatbelt use rates of approximately 80%.