Monday, October 20, 2014: 1:45 PM

Shahriar Shams, MA1, Lusine Abrahamyan, MD MPH PhD2, Gabrielle van der Velde, DC, PhD2 and Nicholas Mitsakakis, MSc PhD2, (1)Dalla Lana School of Public Health, University of Toronto, Toronto, ON, Canada, (2)Toronto Health Economics and Technology Assessment (THETA) Collaborative, University of Toronto, Toronto, ON, Canada
Purpose:   Whiplash injury following a traffic collision may lead to whiplash-associated disorder (WAD). The present study gives a detailed description of the health care utilization among patients with whiplash injury.  It evaluates the association between patients’ expected recovery at baseline and their health care costs one year after the accident. 

Method: We re-analyzed data from a previously conducted randomized clinical trial containing one year healthcare costs of 340 patients with whiplash injury following a motor vehicle accident. Detailed descriptive statistics were generated for health care utilization and cost. A two-part Zero Adjusted Gamma (ZAGA) regression model was used to investigate the association between cost and expected recovery, adjusting for important confounders. This model was selected primarily because it addressed issues with the cost data such as high skewness and presence of zeros. Model fitting was assessed with the use of diagnostics, while Bootstrap resampling was employed for model validation.

Result: The median cost of those who expected to “never get better” (N=74) was $7,354.  This cost was three times larger that those who expected to “get better slowly” (N=101) and more than five times larger than those who expected to “get better soon” (N=165). The association between expected recovery and cost was demonstrated by the covariate adjusted two-part model, where patients expecting to “never get better” had on average 2.66 times higher costs than those who expected to “get better soon” (p-value < 0.0001). Additional covariates with significant association with cost were Whiplash Disability Questionnaire (WDQ) score (p-value < 0.0001), Center for Epidemiologic Studies Depression score (CESD) (p-value = 0.0022) and Physical Component score (PCS) (p-value < 0.0001). The model was validated and seemed to fit the data well.

Conclusion: The ZAGA model was shown to perform well for the analysis of health care cost data. Expected recovery and other modifiable factors were found to have a significant effect on health care utilization and cost. These findings could have a significant impact on policy making by informing evidence-based guidelines for managing minor traffic injuries in Ontario.