A composite risk score system for predicting agricultural injury among Colorado farmers

Abstract

Objective: This study aims to describe patterns of injury among Colorado farmers, identify potential risk factors, and develop a simple injury risk score system to predict the probability of agricultural work-related injuries. The ultimate goal is to develop a simple tool to identify Colorado farmers with high risk of agricultural work-related injuries. Methods: Data collected by the Colorado Farm Family and Hazard Surveillance project from January 1993 to December 1996 were used. This surveillance project interviewed 872 farmers (470 males and 402 females) on 485 Colorado farms in a statewide telephone survey in 1993 and 761 farmers (478 males and 283 females) on 478 farms by face-to-face interviews in an eight county area between 1993-1996. Seven hundred forty six out of 872 statewide participants (85.6%) were interviewed again in 1994, and six hundred twenty five out of those 746 participants (83.8%) were interviewed again in 1995. Participants were asked specifically about agricultural injuries which occurred in the previous 12 months, and the potential risk factors were evaluated. Injury risk scores were calculated using information from the logistic regression models estimated from first year statewide data. Then, the predictability of this injury risk score system was evaluated using data from the second and third year of the statewide surveys and data from the survey in the eight county area. Results: The agricultural injury rates among farmers interviewed by CFFHHS ranged from 7.4% to 11.5%. The leading causes of agricultural work-related injuries were overexertion (21% - 30%), animals (10% - 25%), falls (11% - 23%), and sharp objects (9% - 20%). Injuries mainly resulted in sprains and strains (30% - 38%), fractures (14% - 20%), and open wounds (5% - 20%). Farm machines caused only 6% - 9% of all agricultural work-related injuries in this study. It was found that injured farmers were significantly more likely to be: 1) males, 2) young farmers, 3) farmers with farming or ranching as main occupation, 4) farmers involved in animal products, 5) farmers worked 50-149 days per year in off- farm employment, 6) farmers lost sentimental things, 7) those who had legal problems, and 8) farmers with poor health status. Injury risk scores developed from logistic regression models showed good relationships with observed proportions of agricultural work-related injuries among those Colorado fanners. The final nomogram injury risk worksheet provided a simple tool to identify Colorado farmers with high risk of agricultural injuries. Conclusions: This study demonstrated that a simple injury risk score system, based on demographics of farmers, farming activities, and negative life events, is a reliable and valid tool for predicting risk of agricultural work-related injuries as a continuous phenomenon among Colorado farmers. The good correspondence of results in the cross-validation study also demonstrated that the injury risk score system can provide predictive evaluations of work-related injury risk in farmers even when used for different populations in other settings.