[Multivariate model to predict Streptococcus pyogenes tonsillitis].

OBJECTIVE

to develop a multivariate model to predict the Streptococcus pyogenes isolation in patients with acute tonsillitis.

METHODS

cross-sectional analytic study on patients with acute tonsillitis without a recent history of antimicrobial consumption. We evaluated 14 signs and 18 symptoms. A pharyngeal culture was realized on 5% sheep blood agar. Group A streptococci was identified by standard methods.

STATISTICAL ANALYSIS

sensitivity, specificity, predictive values, chi2, Fisher's exact test, crude and adjusted odds ratio (OR) with 95% CI using dichotomical logistic regression with direct method and Hosmer and Lemeshow-goodness-fit test.

RESULTS

there were 213 participants, 37% were males; a mean age of 14.9 years. We isolated Streptococcus pyogenes in 15%, and 84 % of them had received antimicrobials. We identified signs and symptoms associated with Streptococcus pyogenes isolation: painful swallowing (OR=4.45, 95% CI = 1.13-17.53); tonsils with exudates (OR=3.20, 95% CI = 1.22-8.43); smelly breath (OR=2.78, 95% CI = 1.09-7.10); painful neck nodes (OR=2.70, 95% CI = 1.05-6.96). The presence of nasal symptoms was a protective factor (OR=0.25, 95% CI = 0.09-0.71).

CONCLUSIONS

the prevalence of Streptococcus pyogenes tonsillitis was similar to other reports. We found signs and symptoms associated to Streptococcus pyogenes isolation that allowed us to elaborate a decision algorithm.