Determinants of Physician Utilization: Approach to a Service-oriented Classification of Symptoms

The 495 residents in a probability sample of 1G0 households in a low-income urban section of Raleigh, N. C. were interviewed to identify the determinants of physician utilization in response to illness. A discriminant function analysis was used with data on 231 persons who experienced 396 symptom-complexes. Based on a random sample of one symptom-complex per person, 49 physician visits resulted from these symptoms. The significant variables in the stepwisc discriminant analysis were perceived seriousness of the symptoms, bed-loss days, perception of whether or not physicians could relieve the symptom, and race. These variables achieved a sensitivity of .776 and a specificity of .720 in correctly classifying individuals into the doctor visit or no doctor visit categories. These findings provide the rationale for developing a service-oriented classification system for symptoms.