The efficacy of tonsillectomy in the management of PFAPA syndrome: a systematic review and meta-analysis.

OBJECTIVE

To evaluate tonsillectomy's effectiveness in treating PFAPA syndrome through a systematic review and meta-analysis, providing evidence to guide clinical decisions and improve patient outcomes.

DATA SOURCE: The PubMed, Cochrane Library, and Google Scholar databases, covering the period from their inception up to October 2024.

METHODS

A meta-analysis of randomized controlled trials comparing surgical vs. conservative treatment was conducted to evaluate the postoperative complete remission rate, duration of fever, and frequency of fever episodes. Binary outcomes were analyzed using odds ratios (OR) and continuous outcomes using mean differences (MD) with 95% confidence intervals. The DerSimonian-Laird random-effects model was used, with the Hartung-Knapp-Sidik-Jonkman (HKSJ) adjustment, prediction intervals, and leave-one-out analysis as sensitivity analyses to address the small number of included studies.

RESULTS

Three randomized controlled trials (81 patients) met the inclusion criteria. Tonsillectomy significantly improved postoperative complete remission rates (pooled OR = 42.21, 95% CI: 9.34-190.74, P < 0.0001, I² = 0%), a finding confirmed by HKSJ sensitivity analysis (OR = 42.21, 95% CI: 2.93-609.09, P = 0.026) and robust in leave-one-out analysis. The pooled mean difference in fever duration was -3.15 days (95% CI: -6.78 to 0.49, P = 0.090), which was not statistically significant, with substantial heterogeneity (I² = 68.5%). Surgery significantly reduced fever episode frequency (MD = -0.41 episodes/person-month, 95% CI: -0.66 to -0.15, P = 0.002, I² = 0%). No postoperative complications were reported.

CONCLUSION

This meta-analysis suggests that tonsillectomy can provide significant benefits for children with PFAPA, particularly in achieving complete remission and reducing episode frequency. However, evidence on fever duration remains inconclusive, whereas the reduction in episode frequency was statistically robust. Given the small number of included studies, these results should be considered hypothesis-generating. Future large-scale, well-designed RCTs are needed to confirm these findings.