A model predicting dentists' willingness to treat HIV-positive patients

Data for this study of dentists' willingness to treat HIV-positive (HIV+) individuals were derived from a survey of a probability sample of American general practitioner dentists (GPD). Data were received from 1,351 active GPD, which represented an 88% response rate. Because the outcome measure--willingness to treat HIV+ patients--is dichotomous, i.e., yes/no, logistic regression was selected as the statistical technique to be used for the creation of a predictive model. Seventeen independent variables were initially considered. The final and most parsimonious model contains six independent variables, of which perceived safety in treating HIV+ patients has the most predictive power. Fear of consequences for the practice, if HIV+ patients were seen, was also a powerful predictor, with a sense of ethical responsibility and a past history of treating HIV+ patients also being important predictive variables. Knowledge level about transmission of HIV and concern about risks associated with treating homosexuals were also significant.