Physician practice variation and its effects on patient outcomes have become a major concern during the 1980's. A great deal of research has attempted to identify and account for this variation. The prescribing of medications for institutionalized elderly and consequent adverse reactions have also been a focus for research. To date, there have been very few attempts to study variation in medication prescribing and to identify its sources. This is particularly true for drug prescribing in loosely regulated domiciliary care facilities. As a nation, we anticipate more aged people with more chronic disease and disability utilizing multiple drug regimens with potentially complex and costly adverse reactions. Also, many more of these people will be residing in very loosely regulated institutional environments. Therefore, we must concern ourselves with the quality of life, health care costs and outcome for these institutionalized elderly. Identifying the source of variation in physician prescribing for domiciliary care residents can potentially reduce the adverse consequences of medication use, reducing costs and improving patient outcomes in these settings. A framework is presented that incorporates characteristics from the prescriber, the resident, and the care setting into the prescribing process. This framework will be used to examine the effects of physician's individual characteristics on prescribing behavior. The effects of physician prescribing behavior on the outcomes of the domiciliary care population is also examined. The objective of this research is to identify the source of variation in physician drug prescribing in a domiciliary care setting and how prescribing behavior affects resident outcome. the hypotheses to be tested are: 1) Younger physicians with more geriatric training and experience; urban, group practices; hospital admitting privileges; and routine peer contract have more appropriate prescribing behavior in the domiciliary care population than older, less trained, physician's with more isolated practices. 2) Patients of physicians with more appropriate prescribing behavior have better outcomes than patients of physician's with poor prescribing behavior. The proposed one year study will supplement data gathered from two larger research projects that are currently in progress. Data from 35 Homes for the Aged and Disabled, and 96 Family Care Homes in North Carolina are being collected as part of two caregiver intervention studies. The proposed study will utilize resident, prescription, and facility 7,000 prescriptions, 1,200 residents and 150 physicians. Physician information will be collected to supplement the intervention studies' data in order to meet the stated objectives. An existing physician information database will be utilized, with additional information obtained through a mail and/or telephone survey to participating doctors. A Probit model will be used to control for resident and facility characteristics, in order to test for the probability of appropriate prescribing behavior. Additionally, controlling for resident and facility characteristics, a linear regression model will be used to test the effect of prescribing behavior on resident outcomes.
