Graduation Year


Document Type




Degree Granting Department

Public Health

Major Professor

Jay Wolfson


Anger, De-escalation, Healthcare, Patient Care, Self-defense



In the summer of 2009, Tampa General care providers met with Hospital Administration to express concern that violence on care units was a growing problem and making it difficult to provide quality care. Nurses stated that such violence was one important reason many of their peers choose to retire. Administration took this situation seriously and formed a committee to gather information and submit suggestions to reduce the violence. The committee consisted of representatives from several nursing units, human resources, risk management, security, and administration. Duties assigned included investigation of the actual number of reports on all units and trends. The committee was also charged with the production of a report regarding reviewing other hospital data, literature review, and developing recommendations.

Internal reports indicated that the total prevalence of reported violence as well as the incidence per patient had increased annually since 2005. The hospital reports contradicted the national literature regarding the emergency department (ED) and psychiatric unit (Psych) being the two hospital units with the highest number of violent events. One possible reason for the difference is that these departments require all care providers to attend de-escalation and self-defense classes annually. Based on these findings, the researcher developed and adapted training similar to that of the ED for other units reporting aggressive, abusive, and violent patients. The committee approved a draft plan for implementation. Following presentation to Nursing Administration, some modifications were made, and the Internal Review Boards of the hospital and University of South Florida (USF) approved the project.

The hypothesis tested in this study was whether training in de-escalation and self-defense modifies providers' behaviors to prevent or reduce aggressive, abusive, or violent behavior by patients and visitors. The independent variable was training. The dependent variable was requests for assistance with unruly, angry, or violent patients or visitors. Event reports of the year prior were used for historical comparison. Event reports for the experimental period were assembled subsequent to the training for comparison.

Nursing Administration selected two units to receive the training intervention. The two units selected were neither the worst nor the best in numbers, but rather the middle. Nursing required that all training be scheduled in normal department meetings and that Nurse Managers of the units agree to participate. The research design presumed that at least 85% of care providers on a unit would attend the training. Schedules were developed to accommodate all care providers. The training was presented during June of 2010.

Experimental and comparison units were monitored each month for the number of reported violent events (Code Grays) on each unit. During the fourth months of monitoring, there was a data spike in the Cardiac Care unit. No action was taken until another spike occurred during the sixth month. It was determined that an error had occurred that partially invalidated the data from the Cardiac Care unit: the 85% participation rate among staff had not been reached. Monitoring continued for 12 months after the training. The Eldercare unit showed reduced requests for assistance. Overall, the Cardiac Care unit increased requests for assistance from the year before. Results were adjusted for patient census. Wilcoxon Signed Ranks Testing was performed and displayed using box plots to show how far the median changed during the research from one group to the next. The analysis compared prior year with the year following the interventions, and indicated that there was a movement toward a reduction of Code Grays. To determine if there was a difference between comparison units and experimental units 12 months after the training, Poisson Regression Analysis was utilized. When the comparison units were set as the reference, Poisson analysis indicated the events were decreasing on both units. The Cardiac Care unit did not have a statistically significant p value. The Eldercare unit had a p value of .019.

In conclusion, the results are mixed and statistically inconclusive. From the care providers' perspective, any reduction in violence is significant. The data regarding the training interventions indicates that there was an empirical, albeit not a statistically significant, change in Code Gray reports. Training may have reduced the violence on the Eldercare unit by nearly half.