Because pain is a subjective phenomenon, the documentation of which relies upon accuracy of self-report, little investigative work has been carried out in nursing homes, where dementia is prevalent. Assessment of non- verbal (i.e., behavioral) manifestations of pain may provide a useful alternative pain evaluation strategy for the cognitively impaired. In the three phases of this proposal, the investigator will carefully characterize the prevalence, patterns of presentation, and functional correlates of pain in residents of two long term care facilities. In addition, she will evaluate the pain assessment practices of nursing home personnel. Phase 1 investigations are designed to determine the prevalence and test characteristics (test-retest reliability and baseline variability) of self-reported pain as a function of cognition, after controlling for depression and components of the social network. The ability of chart review and physical examination to predict self-reported pain will also be determined. Phase 2 investigations are designed to develop a performance-based non-verbal (i.e., behavioral) pain observation protocol and then examine its interobserver reliability, relationship to cognitive performance, construct validity, and discriminant validity. Relationships between pain behavior and pain self-report, physical performance and pathology will also be examined. All protocols will be videotaped. Phase 3 investigations are designed to develop a naturalistic pain behavior observation protocol for use in the nursing home, and then determine the influence of such a protocol on nurses' assessments of pain. The relationship between nurses' familiarity with a particular patient, and their assessment of the patient's pain will also be explored. The results of the proposed investigations will facilitate future work toward developing appropriate pain management strategies for nursing home residents with a range of physical and cognitive impairment.