Assessment of clinical features predicting streptococcal pharyngitis.

A total of 133 patients who consulted 4 general practitioners in Bergen 1988/89 for sore throat were examined. 8 clinical parameters with expected predictive value for identifying streptococcal tonsillopharyngitis were recorded. Clinical examination was validated against bacteriologic examination at a microbiological laboratory. The prevalence of streptococcal infection (group A, C and G) was 29%. An algorithm was constructed which identifies 3 groups with varying probabilities of streptococcal infection. A positive predictive value of 62% in the group with highest prevalence and a negative predictive value of 90% in the group with lowest prevalence was found. The consequences of performing a confirmative test only on patients in the group with uncertain prediction for streptococcal disease was elaborated. Although slightly reduced accuracy was demonstrated, due to diminished sensitivity, selective testing is recommended. Another algorithm was constructed for use in situations where no confirmative testing is available. The positive predictive value in the group with highest probability of streptococcal infection was 51%, and the negative predictive value in the group with lowest probability was 84%.