Meta-analyses summarize data from head-to-head trials to evaluate pairs of treatments that have been directly compared.[1] However, in certain circumstances, multiple therapies are of interest, and no data is available on their direct comparison. In such cases, indirect treatment comparison (ITC) is performed for synthesizing evidence surrounding treatments of interest.[2] ITC assumes that the studies are similar and homogenous regarding the administered therapies, patient characteristics, and observed effects, and works best when the inconsistency between indirect and direct evidence is minimal or absent. 

Various methods of ITC have been developed depending on the availability of individual patient data (IPD) and summary level data (SLD). Some of these methods include naïve ITC, network meta-analysis (NMA), population-adjusted indirect comparisons (PAIC), simulated treatment comparisons (STCs), and matching-adjusted indirect comparisons (MAICs). The choice of these methods depends on the study design, the number of comparators available, and the degree to which the outcomes are measured. In addition, the extent of assumptions employed, methodological limitations, and inherent biases associated with each method also determine the choice of the ITC method.

Naïve ITC is based on SLD and is used when the treatments cannot be connected by a common comparator. It does not account for heterogeneity and excludes information from the placebo arms when comparing treatments, thereby introducing bias. Hence, this method is mainly avoided to preserve the randomization in trials during the analyses. 

Network Meta-Analysis (NMA) is perhaps the most popular of the ITC methods. It works with SLD, and compares treatments by combining indirect and direct evidence connected by a network of studies.[3] NMA is considered as the gold standard for ITCs. It offers more an exact estimate of the relative effects of treatments in the network than a single direct or indirect estimate. It also enables for the assessment of intervention ranking and hierarchy. To some extent, this bias can be reduced by using meta-regression, that addresses heterogeneity in treatment effects. It can assess how the effect of treatment changes with a covariate (a patient or methodological attribute). Unfortunately, the usage of meta-regression with NMAs becomes questionable in cases where the number of studies in a network is limited. Furthermore, this approach can only be used when there is a variance study or comparison with only minor variations in impact modifiers.[4,5] Moreover, covariate correction in aggregate-level data may result in ecological bias, which limits the interpretation of estimated results for subgroups. In such cases, Individual Patient Data (IPD) provides adjustments for covariates that cause inconsistencies (e.g., prognostic factors, effect modifiers, etc.). Hence, NMA that leverages IPD can be put to use for conducting analyses that can provide adjustments to reduce such inconsistencies.[6] 

The application of NMAs and their associated methods are often limited by insufficient evidence networks and heterogeneity across trials. This is resolved to certain extent through population-adjusted indirect comparison (PAIC), which is a targeted approach to enhance ITC.[7] It allows to overcome the challenges faced by NMAs by carrying out a targeted comparison between outcomes for specific treatments and factors. It includes two methods: simulated treatment comparisons (STCs) and matching-adjusted indirect comparisons (MAICs). These methods can help reduce the ambiguity in the comparisons with statistical adjustment. STCs do this by applying predictive equations, whereas MAIC relies on patient reweighting. 

STCs or MAICs can be used to conduct either “anchored” indirect comparison, in which each trial has a common comparator arm, or “unanchored” indirect comparison, in which the treatment network is disconnected (single-arm investigations). An anchored approach relies on “conditional constancy of relative effects”. In contrast, an unanchored approach works on a stringent assumption of “conditional constancy of absolute effects”. which is more demanding than the former and is not a widely accepted approach [8]  STCs are often appropriate in analyses where numerous comparators are available for a small set of outcomes, whereas MAICs are often suitable in cases with only one comparator but multiple outcomes. The precision of the equations in STC and the effective matching of populations in MAIC determine the dependability of the studies.

A task force report released in 2011 by the Professional Society for Health Economics and Outcomes Research (ISPOR) defines the fundamentals of conducting ITCs and assessing these studies for informed and efficient decision-making.[4,5] Though the methodological aspects of NMAs have received much attention from researchers, the other ITC methods are yet to be refined to a similar extent. The standardization of these methods is vital to increase their reliability and application.

Become A Certified HEOR Professional – Enrol yourself here!


[1] Ahn E, Kang H. Introduction to systematic review and meta-analysis. Korean J Anesthesiol. 2018 Apr;71(2):103-112. doi: 10.4097/kjae.2018.71.2.103. 

[2] Veroniki AA, Straus SE, Soobiah C, et al. A scoping review of indirect comparison methods and applications using individual patient data. BMC Med Res Methodol. 2016 Apr 27;16:47. doi: 10.1186/s12874-016-0146-y. 

[3] Tonin FS, Rotta I, Mendes AM, Pontarolo R. Network meta-analysis: a technique to gather evidence from direct and indirect comparisons. Pharm Pract (Granada). 2017 Jan-Mar;15(1):943. doi: 10.18549/PharmPract.2017.01.943

[4] Jansen JP, Fleurence R, Devine B, et al. Interpreting indirect treatment comparisons and network meta-analysis for health-care decision making: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 1. Value Health. 2011 Jun;14(4):417-28. doi: 10.1016/j.jval.2011.04.002. 

[5] Hoaglin DC, Hawkins N, Jansen JP, et al. Conducting indirect-treatment-comparison and network-meta-analysis studies: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 2. Value Health. 2011 Jun;14(4):429-37. doi: 10.1016/j.jval.2011.01.011. 

[6] Riley RD, Dias S, Donegan S, et al. Using individual participant data to improve network meta-analysis projects. BMJ Evid Based Med. 2022 Aug 10:bmjebm-2022-111931. doi: 10.1136/bmjebm-2022-111931. Epub ahead of print. 

[7] Phillippo DM, Dias S, Elsada A, et al. Population Adjustment Methods for Indirect Comparisons: A Review of National Institute for Health and Care Excellence Technology Appraisals. Int J Technol Assess Health Care. 2019 Jan;35(3):221-228. doi: 10.1017/S0266462319000333. Epub 2019 Jun 13. 

[8] Jiang Y, Ni W. Performance of unanchored matching-adjusted indirect comparison (MAIC) for the evidence synthesis of single-arm trials with time-to-event outcomes. BMC Med Res Methodol. 2020 Sep 29;20(1):241. doi: 10.1186/s12874-020-01124-6.

Related Posts