Neoadjuvant chemotherapy regimens in treatment of breast cancer: a systematic review and network meta-analysis protocol

Abstract

The individual study effect size can be modeled as: $$ {y}_i=\mu +{\theta}_i+{\varepsilon}_i $$ (3) Further, θ i depend on between study variance (τ2) while ε i depends on the within study variance (\( {\sigma}_i^2 \)) alone and these entities distributed normally as θ i ~N(0, τ2) and \( {\varepsilon}_i\sim N\left(0,{\sigma}_i^2\right) \) respectively. Since random effect method considers two source of variation as between study variance as well as within study variance, the weights (w′ i ) associated with individual studies are inverse of this total variation: $$ w{\hbox{'}}_i=1/\left({\sigma}_i^2+{\tau}^2\right) $$ (4) The pooled effect estimate of y i may be given as below: $$ {\widehat{\mu}}_R=\sum {y}_iw{\hbox{'}}_i/\sum w{\hbox{'}}_i $$ (5) where y i is distributed as \( {y}_i\sim N\left(\mu, {\sigma}_i^2+{\tau}^2\right). \) Further, the variance of this estimated effect size may be represented as: $$ V\left({\widehat{\mu}}_R\right)=1/\sum w{\hbox{'}}_i $$ (6) Network meta-analysis The assumptions for network meta-analysis, i.e., ‘consistency’ for mixed treatment comparison and ‘similarity’ for indirect comparison, will be examined before analysis. [...]comparison adjusted funnel plots will be used in network meta-analysis. [...]it will facilitate evidence-based management of breast cancer patients. [...]this systematic review will be beneficial for wide audience including breast cancer patients, oncology professionals, insurers, policy makers, and researchers working in the field of oncology.