WORK PACKAGE 6

Methodological guidance on the analysis and interpretation of non-randomised studies to inform health economic evaluation

Objectives
  • Investigate the extent to which the findings of randomised and non-randomised studies differ when conducted for the same clinical question and explore potential reasons for observed discrepancies, in particular choice of analytical method

  • Assess whether different analytical methods used in non-randomised studies are likely to produce valid and unbiased estimates of relative effectiveness (a key input to economic evaluations)

  • Provide findings as methodological recommendations that help HTA agencies, regulators and the wider research community and other interested stakeholders to analyse and interpret non-randomised data in economic evaluations 

Methodology
  • Perform a large-scale meta-epidemiological review to obtain estimates of the discrepancy in treatment effects in randomised and non-randomised studies using various analytical methods

  • Develop empirically based recommendations that include input from international stakeholders gathered from a series of workshops

  • Pilot test recommendations using published NICE guidance

Outputs
  • Deliverable D6.1: Meta-epidemiological comparison of effect sizes in matched studies (available in 2022)
     

  • Deliverable D6.2: Recommendations on how to analyse and interpret non-randomised studies [PDF]
     

  • Dataset: Effect estimates obtained from randomised and nonrandomised studies (available December 2022)

Publications
  • Kent S, Salcher-Konrad M, Boccia S, Bouvy JC, De Waure C, Espin J, Facey K, Nguyen M, Rejon-Parrilla JC, Jonsson P (2021). The use of nonrandomized evidence to estimate treatment effects in health technology assessment. Journal of Comparative Effectiveness Research https://doi.org/10.2217/cer-2021-0108

Leads

National Institute for Health and Care Excellence

 

London School of Economics and Political Science

LSE Health

Research team

Pall Jonsson (PI)

Maximilian Salcher (PI)

Seamus Kent