Time-varying Bayesian Network Meta-Analysis
share
The presence of methicillin-resistant Staphylococus Aureus (MRSA) in complicated skin and soft structure infections (cSSSI) is associated with greater health risks and economic costs to patients. There is concern that MRSA is becoming resistant to other gold standard treatments such as vancomycin. While there are a number of review papers employing Bayesian Network Meta-Analyses (BNMAs) to investigate which treatments are best used to treat MRSA related cSSSIs, none have investigated whether the efficacy of treatments changes over time. This paper proposes two novel BNMA methods: Sig-BNMA, which allows treatments to follow a biologically-plausible sigmoidal time curve, and GP-BNMA, which models time effects non-parametrically. In a simulation environment, both proposed methods can detect time-varying trends which existing methods cannot. A dataset was agglomerated from nine existing review MRSA cSSSI review papers. It contains 58 studies comparing 19 treatments over 19 years. Sig-BNMA and GP-BNMA found all treatments to be approximately as effective at the end of the time-period as at the beginning. However, GP-BNMA found evidence of non-linear trends for linezolid, tedizolid, telavancin, and tigecycline; their efficacy relative to vancomycin increased until 2010, after which it declined. This is consistent with observations about vancomycin resistant MRSA in the literature.
READ FULL TEXT