Network meta-analyses in clinical guideline development for cutaneous Lyme borreliosis

Article type
Authors
Torbahn G1, Hofmann H2, Dersch R3, Fingerle V4, Rücker G5, Meerpohl JJ6, Schmucker C6
1Cochrane Germany, Medical Center - University of Freiburg; Institute for Biomedicine of Aging, Faculty of Medicine, Friedrich-Alexander-University of Erlangen-Nuremberg
2Department of Dermatology and Allergology, Technical University of Munich
3Department of Neurology, Medical Center - University of Freiburg
4National Reference Center for Borrelia, Bavarian Health and Food Safety Authority, Oberschleissheim
5Institute for Medical Biometry and Statistics, Medical Center - University of Freiburg; Cochrane Germany, Medical Center - University of Freiburg
6Cochrane Germany, Medical Center - University of Freiburg
Abstract
Background: Clinical guidelines are developed through a 'multi-step process' that ensures that guidelines are feasible within the current clinical environment and that they are based on the best-available evidence. For example, for the treatment of erythema migrans (EM), the typical clinical sign of early skin infection in Lyme borreliosis (LB), the first line therapy is antibiotic treatment. However, controversies about the choice of the antibiotic agent, dose and length of treatment exist between different studies and recommendations.

Objectives: Due to competing interventions that have not been directly compared in studies for patient-relevant outcomes in the treatment of EM we made quantitative comparisons of interventions using network meta-analysis (NMA) to inform evidence-based treatment recommendations.

Methods: NMA’s were calculated with a frequentist approach using the R-package netmeta (Rücker et al.). Furthermore, we used the GRADE (grading of recommendations, assessment, development and evaluation) guidance on NMA to support decision making.

Results: Our evidence is based on 21 randomised-controlled trials. NMA for children and juveniles revealed no statistically significant differences between the different antibiotic agents and regimens for any outcome of interest. In adults, evidence from NMA suggested the lowest risk for retreatment was with the antibiotic agent azithromycin in comparison to doxycycline (OR 0.30, 95%-CI [0.10; 0.92], n=10 studies). Penicillin V regardless of dosage and treatment duration was the antibiotic with the lowest risk for any adverse event in comparison to doxycycline (OR 0.20, 95%-CI [0.06; 0.66], n=13 studies). Overall, quality of evidence was low or very low. The major uncertainty was the size of the effect estimate and a high risk of selection and detection bias in the included study pool.

Conclusions: NMA provides a useful tool when comparing competing interventions. However, some challenges in relation to the network connectivity, consistency and similarity of studies with respect to study design and populations remain in the use of NMA in guideline development.