Desarrollo de una hoja Excel para metaanálisis de comparaciones indirectas y mixtas

Abstract

Financiación: ninguna.Las opiniones expresadas en este trabajo son responsabilidad de los autores, por lo que no reflejan necesaria mente el punto de vista de los organismos en los que trabajan. RESUMENEn investigación clínica, el metaanálisis ha servido generalmente para evaluar la eficacia y la seguridad de un tratamiento a través de la comparación con un único comparador. Las comparaciones indirectas, a través del método de Bucher, permiten combinar datos procedentes de estudios primarios de una forma sencilla cuando la información para las comparaciones directas es limi tada o inexistente. Las comparaciones mixtas permiten combinar las estima ciones procedentes de comparaciones directas e indirectas, aumentando el po der estadístico. En la actualidad, es necesario el desarrollo de aplicaciones informáticas para el cálculo de metaanálisis de comparaciones indirectas y mixtas. Por este motivo, se ha desarrollado una hoja de cálculo con Microsoft Office Excel para el cálculo de comparaciones indirectas y mixtas, de uso sencillo para investigadores clínicos interesados en revisiones sistemáticas con metaanálisis, pero no familiarizados con el uso de paquetes estadísticos más avanzados. La utilización de esta hoja de cálculo para las comparaciones indirectas y mixtas puede ser de gran utilidad en epidemiología clínica, con el fin de extender el conocimiento proporcionado por el metaanálisis tradicional cuando la evidencia de las comparaciones directas es limitada o inexistente. ABSTRACT Development of an Excel Spreadsheet for MetaAnalysis of Indirect and Mixed Treatment ComparisonsMetaanalyses in clinical research usually aimed to evaluate treatment efficacy and safety in direct comparison with a unique comparator. Indi rect comparisons, using the Bucher's method, can summarize primary data when information from direct comparisons is limited or nonexistent. Mixed comparisons allow combining estimates from direct and indirect comparisons, increasing statistical power. There is a need for simple appli cations for metaanalysis of indirect and mixed comparisons. These can easily be conducted using a Microsoft Office Excel spreadsheet. We deve loped a spreadsheet for indirect and mixed effects comparisons of friendly use for clinical researchers interested in systematic reviews, but nonfami liarized with the use of more advanced statistical packages. The use of the proposed Excel spreadsheet for indirect and mixed comparisons can be of great use in clinical epidemiology to extend the knowledge provided by traditional metaanalysis when evidence from direct comparisons is limi ted or nonexistent.

How to Perform and Interpret a Network Meta-Analysis for Indirect and Mixed Comparisons: Key Methodological Strategies

“…In addition, we cannot fail to mention that, despite being secondary studies, meta-analyzes may raise doubts regarding the compliance with ethical and legal principles, particularly in terms of the accuracy and adequacy of the findings (Li et al, 2011). The simplest application of a network meta-analysis for indirect and mixed comparisons consists of determining if treatment B is better than treatment A, and if treatment C is better than the same comparator A, taking into account the lack of studies directly comparing B and C (Tobías et al, 2014; Figure 1A). Thus, several statistical methods can be used to perform indirect comparisons that can combine data from preliminary studies when these are limited or simply non-existent ( Figure 1B), such as: the Bucher's method (Bucher, Guyatt, Griffith, & Walter,1997), also called the method for adjusted indirect comparisons; the meta-regression (Bayesian and frequentist inference approaches); and the hierarchical Bayesian models based on the Markov chain Monte Carlo (Catalá-López & Tobías, 2013;Tobías et al, 2014).…”

“…The simplest application of a network meta-analysis for indirect and mixed comparisons consists of determining if treatment B is better than treatment A, and if treatment C is better than the same comparator A, taking into account the lack of studies directly comparing B and C (Tobías et al, 2014; Figure 1A). Thus, several statistical methods can be used to perform indirect comparisons that can combine data from preliminary studies when these are limited or simply non-existent ( Figure 1B), such as: the Bucher's method (Bucher, Guyatt, Griffith, & Walter,1997), also called the method for adjusted indirect comparisons; the meta-regression (Bayesian and frequentist inference approaches); and the hierarchical Bayesian models based on the Markov chain Monte Carlo (Catalá-López & Tobías, 2013;Tobías et al, 2014). In this study, we only demonstrate the Bucher's method due to its relative simplicity and for being the most commonly used method in the scientific community (Catalá-López & Tobías, 2013;Tobías et al, 2014).…”

“…Thus, several statistical methods can be used to perform indirect comparisons that can combine data from preliminary studies when these are limited or simply non-existent ( Figure 1B), such as: the Bucher's method (Bucher, Guyatt, Griffith, & Walter,1997), also called the method for adjusted indirect comparisons; the meta-regression (Bayesian and frequentist inference approaches); and the hierarchical Bayesian models based on the Markov chain Monte Carlo (Catalá-López & Tobías, 2013;Tobías et al, 2014). In this study, we only demonstrate the Bucher's method due to its relative simplicity and for being the most commonly used method in the scientific community (Catalá-López & Tobías, 2013;Tobías et al, 2014). Mixed comparisons allow combining the estimates of direct and indirect comparisons, increasing the statistical power by reducing the confidence interval and increasing accuracy (Catalá-López & Tobías, 2013;Tobías et al, 2014).…”

“…In this study, we only demonstrate the Bucher's method due to its relative simplicity and for being the most commonly used method in the scientific community (Catalá-López & Tobías, 2013;Tobías et al, 2014). Mixed comparisons allow combining the estimates of direct and indirect comparisons, increasing the statistical power by reducing the confidence interval and increasing accuracy (Catalá-López & Tobías, 2013;Tobías et al, 2014). However, it should be noted that this methodology retains the identity of each intervention and can only be performed when the copyright) are currently available, such as STATA and SAS.…”

“…Thus, the network meta-analysis within a systematic review refers to a meta-analysis in which three or more treatments are compared using both direct comparisons of interventions within randomized controlled trials (RCT) and indirect comparisons across RCTs based on a common comparator. These meta-analyzes are usually referred to in the literature as network meta-analysis, multiple treatments meta-analysis (MTM) or mixed treatments comparison meta-analysis (MTC) (Higgins & Green, 2011;Li et al, 2011;Catalá-López & Tobías, 2013;Tobías, Catalá-López, & Roqué, 2014). In this article, we present some key concepts related to the performance and interpretation of a network meta-analysis for indirect and mixed comparisons.…”