Lower respiratory infections account for up to 12% of all hospitalizations from nursing homes at a conservative incremental cost of over $200 million annually. Recommendations for management of LRIs in nursing home residents are based on almost no outcome data; nonetheless, reviews uniformly recommend that all nursing home patients with pneumonia be hospitalized or at least treated with parenteral antibiotics. The investigators hypothesize that there is a low-risk group of residents with LRIs in whom nursing home care can effectively produce favorable outcomes with fewer complications at substantially less cost than hospital care. The proposed prospective cohort study seeks to: 1) develop and validate an LRI risk index that distinguishes residents at low and high risk for mortality or functional decline; 2) examine the relationship between facility quality-of-care indicators, physician characteristics, and risk for mortality or functional decline; and 3) assess the relative cost of initial management in an acute care hospital versus initial nursing home care, particularly for low-risk residents. Community nursing homes in Central Missouri and St. Louis will be utilized in the study. Under specific protocol authorized by attending physicians, nursing home residents with specific or non-specific symptoms consistent with an LRI will be evaluated by a nurse practitioner. If an LRI is likely, supplemental assessment data will be obtained. Illnesses meeting the study definition of an LRI will also be followed by re-assessments at 30 and 90 days. All clinical information will be placed on structured forms in the resident's record, the data will subsequently be abstracted by research assistants. Data collected after initial and follow-up assessments will include: demographics; current and baseline functions; indicators of acute illness severity, such as respiratory rate; nutritional status; diagnoses and conditions; immunization history; LRI therapy; and outcomes. The study will also develop data on facility quality of care, physician characteristics, and the costs of care for illness. Using multivariable logistic regression or Cox Proportional Hazards modeling on two-thirds of the sample, a patient-level LRI risk index will be developed. The risk index will be validated on the remaining one-third of the sample. The study will further test whether facility and attending-physician characteristics show significant predictive ability after controlling for the other variables in the model. Finally, a simple cost analysis from a social perspective will be performed that compares nursing home initial care versus initial hospital care.
