The determinants of treatment choice in end-stage renal disease: can we generalize about decision making from...

Practitioner judgments about treatments for hypothetical end-stage renal disease patients were examined by two mailed surveys. It was hypothesized that treatment choice was a function of four hierarchically arranged sets of factors: disease- and treatment-specific, patient-specific, environment- and institution-specific, and practitioner-specific. The first survey identified six vignettes for which the case-specific factors alone did not yield a generally accepted treatment decision. These six cases were used in the second survey, whose results are reported here. Practitioner-specific characteristics were found to be only weakly related to treatment choice. Guttman scale analysis showed no significant practitioner propensity to use any given treatment. Instead, the number of colleagues in a respondent's renal unit picking a given therapy for a given patient was the strongest predictor of an individual's choice. Contextual factors--clinical details of the case and decision rules within the institution--appeared to overwhelm provider-specific tendencies; characteristics of the decision generally outweighed characteristics of the decision maker. This framework of factors may be useful for analyzing inter-provider variation. Consequences for certain approaches to the study of decision making (especially single-site studies and regression-based models) are noted.