Statistical approaches to adjusting weights for dependent arms in network meta‐analysis

Abstract

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 1 treatment. However, some trials use within person designs such as split-body, split-mouth, and crossover 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 accounted for within the statistical analyses. Ignoring these correlations may result in incorrect conclusions. The main objective of this study is to develop statistical approaches to adjusting weights for dependent arms within special design trials. In this study, we demonstrate the following 3 approaches: the data augmentation approach, the adjusting variance approach, and the reducing weight approach. These 3 methods could be perfectly applied in current statistical tools such as R and STATA. An example of periodontal regeneration was used to demonstrate how these approaches could be undertaken and implemented within statistical software packages and to compare results from different approaches. The adjusting variance approach can be implemented within the network package in STATA, while reducing weight approach requires computer software programming to set up the within-study variance-covariance matrix.

Evaluating network meta‐analysis and inconsistency using arm‐parameterized model in structural equation modeling

“…The result for fixed effects model, shown in Table , is identical to that of Su and Tu except a few second decimal digits, which is likely due to round‐off errors and the fact that the mvmeta package used by Su and Tu demands that users enter contrast data using treatment A as reference treatment, so artificial data must be imputed for studies that did not include treatment A . The estimated effect depended on r , especially for contrast AE . The result for random effects model (Table ) indicated that the value of r can also influence the estimate of heterogeneity estimate τ 2 .…”

“…Consequently, the outcomes of treatment arms are correlated. Recently, Su and Tu proposed three methods of data manipulation to account for within‐study correlations before routines such as mvmeta in R and STATA can be used to undertake NMA of the Lu‐Ades contrast‐based model . Under the arm‐parameterized SEM framework, the correlation can be easily specified and requires no prior data manipulation.…”

“…In light of the merit of arm‐parameterized models and various unexplored features that can be implemented in SEM, in this article, we attempt to implement the arm‐parameterized NMA model in SEM and explore how UWLS and inconsistency evaluation can be entertained within the SEM framework. In addition, Su and Tu recently proposed three methods that can incorporate studies of within‐person designs (for example, split‐mouth trials, cross‐over trials) into the Lu‐Ades model, so we also investigate how these studies can be included in our model. This article is organized as follows: we will first briefly review the Lu‐Ades model and demonstrate how it can be transformed into an arm‐parameterized model.…”

Comparison of the efficacy of supraglottic airway devices in low-risk adult patients: a network meta-analysis and systematic review

“…Risk ratios were used to measure the relative treatment effect on the risk of first-attempt insertion failure (failed or successful) and the occurrence of postoperative sore throat (yes/no), weighted mean differences were used to measure changes in OLP (cmH 2 O). For crossover trials, we used the adjusting variance approach to address correlations between different procedures within trials 22 . We evaluated the transitivity assumption by comparing the distribution of NMBAs used across studies.…”

Anesthetic efficacy in vital asymptomatic teeth using different local anesthetics: a systematic review with network meta-analysis

“…This renders it difficult, however, to synthesize the evidence using NMA because the independence assumption does not hold. Correlation between data can be adjusted by the standard approach, the reducing weight approach, and the adjusting variance approach [ 9 ].…”