Occurrence and a short description of all comparison methods identified in the reviewed sample; the chosen categorisation into five larger categories is reflected in all figures describing the identified comparisons and was chosen to reflect the most important methodological differences between the comparisons
| Category | Method | n=418 | Short description |
| Quantitative, direct comparisons 74 (18%) | |||
| Head-to-head comparison | 60 | Direct comparison of two products as two parallel arms of one study, such as in a randomised controlled trial | |
| Baseline comparison | 14 | Comparing the outcome of one product measured at the baseline of a study and the outcome of the other product at the end of the study | |
| Quantitative, indirect comparisons 182 (44%) | |||
| Side-by-side comparisons (n=129, 31%) | Simple side-by-side comparison | 113 | Presentation of summary statistics for a variable (eg, objective response rate for ‘response’) by treatment arms. The treatment arms are from separate studies, and no statistical methods for cross-trial comparisons are applied (eg, the difference between objective response rates from different studies) |
| Pooled side-by-side comparison | 16 | Same as the simple side-by-side comparison, but the effect size from one or more of the comparators is derived from pooling results from several studies | |
| Inferential comparison with aggregate external data (n=40, 10%) | Matching-adjusted indirect comparison | 22 | Comparing individual patient data from the investigational product with aggregate data from one comparator from another study by means of re-weighting the individual patient data to match the baseline characteristics of the aggregate comparator data7 |
| Simulated treatment comparison | 8 | A regression-based approach estimating the effect of an investigational product based on individual patient data and adjusted for baseline characteristics compared with aggregate data for the comparator. The approach can have the additional element of simulation where samples are drawn from the joint covariate distribution of the aggregate data)18 | |
| Bucher method | 7 | Compares two or more products that have the same comparator (eg, placebo) via indirect adjustment19 | |
| Meta-analysis | 1 | Estimates the effects of two products using aggregate data from at least two independent studies. The combined (pooled) effect estimate is based on the weighted average of the independent studies20 | |
| Network meta-analysis | 2 | Compares more than two products with data from independent studies by combining direct and indirect evidence, here based on aggregate data21 | |
| Inferential comparison with patient-level external data (n=13, 3%) | Matched/weighted comparison | 4 | Indirect comparison based on matching patient-level data from each patient under the investigational treatment to data from the control group, or weighting data from the control group depending on their similarity to the treated patients (often weighted by the probability to receive the treatment based on several variables measured in treated and untreated patients) to create a comparable control group |
| Regression | 4 | Compares two products based on patient-level data in a regression model (eg, linear regression or Cox regression) | |
| Network meta-analysis | 5 | Compares more than two products with data from independent studies by combining direct and indirect evidence, here based on individual patient data21 | |
| Qualitative comparison | 162 (39%) | ||
| Partial overlap in the patient population | 50 | Instances where there was no complete overlap in indications for two products | |
| Non-preferred treatment | 44 | Any products marketed as non-preferred treatments, for example, second- or later-line products, therefore not needing to show improvement over earlier line/preferred products | |
| Adjunct treatment | 47 | Instances in which the investigational product is supposed to be used in combination with the comparator | |
| Unclear | 21 | All those instances, in which no quantitative comparison could be clearly identified | |