OBJECTIVES: Indirect treatment comparisons in HTA submissions are increasingly using uncontrolled or real world studies requiring population matching techniques. Prognostic factors (PFs) are fundamental to the conduct of meaningful population matching analyses (matching adjusted indirect comparisons [MAIC], simulated trial comparisons [STC]) yet there are no formal recommendations or consensus for their identification.

METHODS: We conducted a review of MAIC/STC analyses in ‘musculoskeletal diseases’ (excluding oncology) to determine how each analysis identified relevant PFs used for population matching.

RESULTS: Searches were run in Medline and Embase (Inception to 26-May-2023); 52 abstracts and 23 full-texts were reviewed. Eleven studies were included which consisted of 9 MAIC, 1 STC, and 1 both MAIC and STC. Indications considered included spinal muscular atrophy (SMA), rheumatoid arthritis, psoriatic arthritis, ankylosing spondylitis, haemophilia A and graft versus host disease.

PFs were predominantly selected based on commonly reported baseline characteristics between the trials being matched (3/11), clinical expert opinion (1/11) or a combination of both (2/11). Two studies failed to specify how PFs were selected. The three remaining studies used the literature to some extent to identify PFs. One referenced a previously conducted MAIC which utilised both trial and clinical opinion, another stated PFs were ‘validated through the literature’, although further details were not provided. Finally, one study demonstrated a methodical identification of PF in SMA which used a recently published systematic literature review of PF.

CONCLUSIONS: A key facet of evidence-based medicine is the comprehensive, unbiased identification of data typically by the conduct of systematic literature reviews. Clinical opinion, which featured in PF identification in several analyses, is placed on the lowest evidence level on the hierarchy of evidence. Whilst population matching techniques constitute an advancement in evidence-based medicine, greater consideration should be given to the unbiased identification of PFs in order to increase the validity of their results.