PROJECT SUMMARY/ABSTRACT This application responds to PA-16-188 "Mechanisms, Models, Measurement, & Management in Pain Research (R01)". Pain management in nursing homes is sub-optimal. Sixty-two percent of nursing home residents experience pain. Although cancer is an important reason for pain in nursing homes, arthritis and musculoskeletal problems affect over three quarters of persons over 65 years of age. Despite the high prevalence of painful conditions experienced by nursing home residents, non-malignant pain in nursing homes has not been thoroughly studied. Pain is seldom an isolated symptom and is often associated with/accompanied by other symptoms in common clusters. We do know that pain is under-reported in nursing homes and when documented, its treatment is often suboptimal. For nursing home residents, a “geriatric pharmacoparadox” exists – i.e., despite the growing population and the high burden of medication use, geriatric nursing home residents are systematically excluded from clinical trials. The resulting dearth of information requires rigorous non-experimental study designs to quantify the risks and benefits of medication use. This study extends the literature in two ways. First, it seeks to improve measurement of pain. Second, with improved measurement in hand, it seeks to expand knowledge on how best to relieve pain using pharmacologic approaches. This application builds on previous and ongoing work conducted by our team. We are currently working on characterizing non-malignant pain symptom clusters (cross-sectional and longitudinal) using latent variable modeling approaches. The work proposed in this application will further refine these models using a contemporaneous dataset. We will update our in-house data set to include the most recent data available (the Minimum Data Set 3.0 (MDS) and Medicare eligibility and claims data (Part A & D)). The specific aims are to: 1) Validate pain symptom clusters in current data; 2) Describe correlates of pain symptom clusters; 3) Describe analgesic use by pain symptom clusters; and 4) Quantify the extent to which medications are associated with outcomes such as improvements in pain and adverse events (e.g. falls/fractures, hospitalizations). Advanced statistical techniques including latent variable models, marginal structural models, and quantile regression will be used. Untreated, potentially undertreated and inappropriately treated non- malignant pain remains common in the nursing home setting. Non-malignant pain, and how best to relieve it, is understudied in the nursing home setting. This is a novel area of exploration that will provide needed information to form the basis of effective clinical strategies to improve non-malignant pain management in nursing homes.