CONNECT for quality: protocol of a cluster randomized controlled trial to improve fall prevention in nursing homes

Sixteen nursing homes will be randomized to one of two study arms, CONNECT + FALLS or FALLS alone. Subjects (staff and residents) are clustered within nursing homes because the intervention addresses social processes and thus must be delivered within the social context, rather than to individuals. Nursing homes randomized to CONNECT + FALLS will receive three months of CONNECT first, followed by three months of FALLS. Nursing homes randomized to FALLS alone receive three months of FALLs QI and are offered CONNECT after data collection is completed. Complexity science measures, which reflect staff perceptions of communication, safety climate, and care quality, will be collected from staff at baseline, three months after, and six months after baseline to evaluate immediate and sustained impacts. FALLS measures including quality indicators (process measures) and fall rates will be collected for the six months prior to baseline and the six months after the end of the intervention. Analysis will use a three-level mixed model.By focusing on improving local interactions, CONNECT is expected to maximize staff's ability to implement content learned in a falls QI program and integrate it into knowledge and action. Our previous pilot work shows that CONNECT is feasible, acceptable and appropriate.ClinicalTrials.gov: NCT00636675Efficacy trials [1] have shown that interventions that reduce multiple fall risk factors also lower fall rates, recurrent falls, and injurious falls in nursing home residents [2-5]. However, interventions in these trials were completed by specially-hired external study staff; prior attempts to move fall risk factor reduction into everyday practice by in-house staff have not been successful [6-10]. Quality Improvement (QI) interventions [11,12] are the current gold standard for introducing evidence-based care into nursing homes. These QI interventions provide the content for reducing falls but do not ensure that the processes needed to successfully implem