that NMA can be greatly simplified by modelling treatment means rather than treatment contrasts using factorial analysis-of-variance (ANOVA) models, and that such analyses can produce identical or essentially the same results as analyses using baseline contrasts. Some methods for detecting inconsistency in meta-analysis networks based on baseline contrasts are relatively complex on account of the fact that baseline treatments may vary among trials and sources of inconsistency have to be traced through loops of the network. Most methods for analysis of NMA operate on pairwise contrasts of treatments with a baseline treatment or control, henceforth denoted as baseline contrasts.
Unity definition factor analysis trial#
Several methods have been proposed for detecting inconsistency in trial networks. In particular, consistency implies agreement between direct and indirect evidence on a treatment contrast. A key assumption of many methods for NMA is consistency of treatment effect estimates across designs, defined by the set of treatments tested. Such analyses combine different sources of pairwise treatment comparisons across trials, i.e., direct comparisons from trials that jointly test both treatments of interest and indirect comparisons from trials that only test one of the two treatments, but are connected through other treatments via the trial network.
When trials differ in design, i.e., in the sets of treatments tested, joint analysis may be done by what has come to be called network meta-analysis (NMA). In the simplest case, all trials comprise the same set of treatments, typically only two, i.e., a new treatment and a control or baseline treatment. Results from several randomized trials can be combined by meta-analysis methods. Conclusionįactorial analysis of variance provides a convenient framework for conducting network meta-analysis, including diagnostic checks for inconsistency. Moreover, a suitable definition of factors and effects allows devising significance tests for inconsistency.
We show that standard regression diagnostics available in common linear mixed model packages can be used to detect and locate inconsistency in trial networks.
This approach is in many ways simpler to implement than the more common approach of using treatment-versus-control contrasts. Inconsistency can be scrutinized by inspecting the design × treatment interaction. Network meta-analysis can be very conveniently performed using factorial analysis-of-variance methods. Potential inconsistency among different types of trial (designs) differing in the set of treatments tested is a major challenge, and application of procedures for detecting and locating inconsistency in trial networks is a key step in the conduct of such analyses. Network meta-analysis can be used to combine results from several randomized trials involving more than two treatments.
0 kommentar(er)
