Validation and modification of streptococcal pharyngitis clinical prediction rules

Thomas G. McGinn, Joseph Deluca, Sushil K. Ahlawat, Ben Hur Mobo, Juan P. Wisnivesky

Research output: Contribution to journalArticlepeer-review

15 Scopus citations


Objective: To validate a simplified version of the Walsh clinical prediction rules (CPRs) for the presence of streptococcal pharyngitis in an inner-city, ethnically diverse population. Patients and Methods: This prospective study conducted in New York City, NY, from January 1, 1997, to May 31, 1997, consisted of 171 consecutive adult walk-in patients who presented with symptoms of upper respiratory tract infection and/or sore throat. The patients were assessed by using 5 clinical factors: cough, exposure to known streptococcal contact, temperature, tonsillar-pharyngeal exudates, and cervical lymphadenopathy. Throat cultures for group A β-hemolytic streptococcus were obtained from all patients. Clinicians assessing the patients were unaware of throat culture results, and those processing the throat cultures were blinded to the clinical predictors. Results: The prevalence of streptococcal pharyngitis was 24% (95% confidence interval, 18%-30%). The simplified version of the Walsh CPR for streptococcal pharyngitis predicted accurately the probability of a positive culture in our diverse population (area under the receiver operating characteristic curve, 0.71). The simplified CPR also showed clinically useful likelihood ratios and posterior probabilities. Conclusion: A simplified version of the Walsh CPR is accurate for diagnosing streptococcal pharyngitis in an inner-city population. This finding should provide clinicians more confidence in applying the CPR in similar clinical settings.

Original languageEnglish (US)
Pages (from-to)289-293
Number of pages5
JournalMayo Clinic Proceedings
Issue number3
StatePublished - Mar 1 2003

ASJC Scopus subject areas

  • Medicine(all)


Dive into the research topics of 'Validation and modification of streptococcal pharyngitis clinical prediction rules'. Together they form a unique fingerprint.

Cite this