The construction industry experience a high rate of injuries in general, and a high rate of fatal and disabling injuries in particular. The ability to predict injuries can help control and prevent them. However, because of the dynamic nature of the industry and the unique nature of each contractor and project traditional methods of prediction, such as historical averages, can provide inaccurate and misleading information. The purpose of this research is to develop knowledge that can be used in preventing traumatic injuries at commercial construction sites. This study will develop a model for predicting the risk of construction injuries as a function of the characteristics of contractor, project and construction worker. The main outcome of interest will be the rate of traumatic injuries occurring at commercial construction sites that require medical treatment by a licensed caregiver. The injuries studied will be the same as those that are required to be reported by OSHA. Data will be collected, for a minimum of 100 commercial construction contractors and 400 projects from commercial construction contractors and the State of Washington Department of Labor and Industries, and a data base created. Analysis will involve fitting Poisson regression models, with adjustments for over-dispersion, and for shared contractor and project effects. Model fitting will be accomplished using the GLIM Computer Software, developed by the Royal Statistical Society. The characteristics of commercial construction contractors, projects and workers will be investigated as explanatory variables in order to determine their significance in predicting injuries. Rates of injuries will be calculated as injuries/FTE. Relative risks and corresponding confidence intervals will be estimated for combined injuries, and for individual types of injury. Standard likelihood methods of generalized linear models will be used for testing the effect of each explanatory variable on the injury rate, both alone, and adjusted for the confounding effects of other explanatory variables.
