Within the realm of violence, work-related violence has recently been recognized as a major problem. During 2000 alone, 677 work-related homicides occurred, making homicide the third leading cause of occupational fatality, overall, and the second leading cause of occupational fatality for women (U.S. Department of Labor, Bureau of Labor Statistics, 2001). While there is an emerging literature pertinent to work-related homicides, there is a serious deficiency in the knowledge of non-fatal work-related violence and the associated risk factors. In a recent analysis of data extrapolated from the National Crime Victimization Survey for 1992-1996, based on a nationally representative sample of approximately 46,000 households, Warchol (1998) estimated that nearly two million acts of non-fatal work-related violence occurred annually. The true prevalence of occupational violence is unknown. This study was designed to identify the magnitude and consequences of the problem of work-related violence within a major occupational population and to identify specific risk factors, using a case-control design. In particular, this enabled determination of the relation between work-related violence in a cohort of registered and licensed practical nurses and: 1) personal exposures; 2) environmental situations/exposures in the work environment; and 3) characteristics of others in the environment (other workers, patients, visitors). The target population included all licensed registered (RN) and practical (LPN) nurses in Minnesota (79,128). This population involves one of the few professions at risk for violence for which a database of contact and some demographic information is available for selection of subjects. In this study, work-related violence was defined as the intentional use of physical force or emotional abuse, against an employee, that resulted in physical or emotional injury and consequences. This included physical assault (P A) and non-physical violence (NPV) (threat, sexual harassment, and verbal abuse). Work-related events included any activities associated with the nurse's job or events that occurred in his/her work environment; work-related travel was included. Pilot testing was conducted prior to each study phase that included a rigorous follow-up protocol. Prior to Phase I, 220 nurses were selected to pilot test the survey instrument and methods; one-half were sent surveys requesting a telephone number to be used, potentially, for clarification of missing information. Nurses were also assigned to one of two types of follow-up for non-response; approximately one-half were contacted by both mail and telephone to encourage response, while the remaining were contacted only by mail. Prior to Phase 2, pilot testing again was conducted to test the survey instruments. Initially, a specially-designed, comprehensive survey instrument was sent to a random selection of 6,300 RNs and LPNs who were licensed in the state of Minnesota. The purpose was to identify individuals who worked as nurses in Minnesota during a 12-month period, to identify nurses who did and did not experience work-related violence during that same study period, and to collect comprehensive data on their work-related PA and NPV experience occurring during the past year. Subsequently, a nested case-control design was used to examine the relation between potential risk factors and work-related PA. For each case (n=475), three controls (n=1425) were sampled from the population at risk during the study period. A questionnaire was then sent to the cases and selected controls to obtain data on work-related exposures, including the characteristics of nurses and significant others in the workplace and surrounding environmental factors. Cases were questioned about their exposures one month prior to and during the incident. Controls were questioned about their exposures on a randomly selected month from the study period to provide the person-time exposure information; key items were validated. An overall conceptual model was developed for the occurrence of work-related violence events, based on previous knowledge and the epidemiological model of human damage involving the dynamic interactions of a host, agent(s) and vehicles (or vectors) within the environment; this served as the basis for a more elaborate causal model that guided instrument development and study analyses. The ultimate goal of the data analyses was to estimate the impact of the above factors on work-related violence, controlling for important confounding factors. Analyses began with basic descriptive statistics on the sample and the consequences of reported events, and crude estimates of event rates. Selection of confounders for multiple logistic regression was based on a directed acyclic graph, derived from the causal model, following the methods described by Greenland et al. (1999). Confidence intervals for event rate estimates and regression coefficients were calculated by the bootstrap method (Efron and Tibshirani, 1993). Potential response bias was controlled by inversely weighting observed responses by probabilities of non-response estimated as a function of characteristics available from the licensing database (age, gender, license type, and home address: metropolitan versus non-metropolitan area) (Horvitz and Thompson, 1952). The probability of being eligible among the respondents across these same characteristics was used to estimate the unknown eligibility among non-respondents (Mongin, 2001). Based on pilot test results, it appeared that there were no differences in the response rates between the two methods of follow-up (mail only, and combination of mail and telephone). However, there were additional efforts and costs expended into completing the telephone followup process. These included extra charges for long distance calls, and a high level of personnel time; thus, the use of mail only appears be more efficacious for this population. For the comprehensive study, Phase 1, 78% responded; proportions of RNs and LPNs, were 80% and 74%, respectively. The PA adjusted rate was 13.2 per 100 persons per year. For NPV the adjusted rate was 38.8 per 100 persons per year. Patients/clients were reported most frequently as the source of PA (96%) and NPV (67%). For both PA and NPV, working in a nursing home/long-term care/rehabilitation facility increased risk the greatest in this study, based on multivariate modelling. The consequences of violence reported in this study deserve particular attention; those for NPV appeared even greater .than those for P A. For the case-control study, Phase 2, 76% responded. Full length surveys were returned for 324 cases and 946 controls. Weighted analyses of the environmental exposures identified important increased rates for: working primarily in a nursing home/long-term care or rehabilitation facility; working primarily in emergency and psychiatric/behavioral departments; and working in environments with a typical lighting level of less than "bright as daylight." Decreased rates were identified for personal protective devices carried by nurses, such as cell phones/personal portable alarms; if nurses provided their own cell phones or personal portable alarms (compared with someone else providing the device), this effect was even greater. This study is among the first such comprehensive efforts to identify the magnitude of the violence problem and specific risk factors in a major occupational population. From this effort, specific prevention and control strategies can be developed more realistically. In particular, attention to facilities, departments, patient populations, and activities that place these people at risk is of great importance and need to be addressed by employers and relevant safety committees. This violence affects not only the victim but, also, the employer, others in the work environment and significant others outside the work setting. Most importantly, this study serves as a basis for further in-depth research of specific risk factors, identified in the current effort, to examine additional opportunities for intervention efforts.