Article type
Year
Abstract
Background: Network meta-analysis compares multiple treatments in terms of their efficacy and harm by including evidence from randomized controlled trials. Most clinical trials use parallel design, where patients are randomly allocated to different treatments and receive only one treatment. However, some trials use within person designs such as split-body, split-mouth and cross-over designs, where each patient may receive more than one treatment. Data from treatment arms within these trials are no longer independent, so the correlations between dependent arms need to be taken into consideration in statistical analysis. Ignoring these correlations may result in incorrect results.
Objectives: The main objective of this study is to develop statistical approaches to adjusting weights for dependent arms within special design trials. Consequently, data from those trials can then be analyzed alone with those from trials with parallel group design.
Methods: In this study, we demonstrated the following three approaches: the data augmentation approach, the adjusting variance approach, and the reducing weight approach. A network meta-analysis of periodontal regeneration was used to demonstrate how these approaches could be undertaken and implemented within statistical software packages, and to compare the results from different approaches. Results: The data augmentation approach requires meticulous calculations, but it can be extended trials with any number of treatment arms. The adjusting variance approach can be implemented within the network package in STATA, but it cannot be extended to trials with more than three treatment arms. In contrast, it is straightforward to use the reducing weight approach to set up the within-study variance-covariance matrix. However, the complex computations can only be undertaken by software packages.
Conclusions: In conclusion, the three approaches we proposed have their advantages and limitations. As they can achieve almost identical results, meta-analysts can therefore choose the one most suitable to their need.
Patient or healthcare consumer involvement: Our proposed approaches to adjusting the correlations within dependent arms can obtain more accurate and unbiased results for meta-analyses.
Objectives: The main objective of this study is to develop statistical approaches to adjusting weights for dependent arms within special design trials. Consequently, data from those trials can then be analyzed alone with those from trials with parallel group design.
Methods: In this study, we demonstrated the following three approaches: the data augmentation approach, the adjusting variance approach, and the reducing weight approach. A network meta-analysis of periodontal regeneration was used to demonstrate how these approaches could be undertaken and implemented within statistical software packages, and to compare the results from different approaches. Results: The data augmentation approach requires meticulous calculations, but it can be extended trials with any number of treatment arms. The adjusting variance approach can be implemented within the network package in STATA, but it cannot be extended to trials with more than three treatment arms. In contrast, it is straightforward to use the reducing weight approach to set up the within-study variance-covariance matrix. However, the complex computations can only be undertaken by software packages.
Conclusions: In conclusion, the three approaches we proposed have their advantages and limitations. As they can achieve almost identical results, meta-analysts can therefore choose the one most suitable to their need.
Patient or healthcare consumer involvement: Our proposed approaches to adjusting the correlations within dependent arms can obtain more accurate and unbiased results for meta-analyses.