An emulated target trial comparing I) the bevacizumab arm (n 65) of the ABC trial — a prospective, double-masked, multicentre randomised controlled trial undertaken in the United Kingdom during 2006–08 — with II) a synthetic arm from 31,151 eyes receiving aflibercept during routine care across 27 sites in England during 2012–18.
Workflow of emulated target trial estimating causal effects. L, confounding variables (age, sex, baseline read); A, independent variable (bevacizumab vs. aflibercept); Y, dependent variable (Δ in visual acuity from baseline to week 54); M, mediating variable (number of maintenance injections received during the study period). Abbreviations: EHRs, Electronic Health Records; SOC, Standard Of Care; NC, Negative Control; IPTW, Inverse Probability of Treatment Weighting; EM, Exact Matching; PSM, Propensity Score Matching.
Borrows the Bevacizumab arm from the ABC trial.
Assembled from Electronic Health Records:
src/amd_irf_srf_large_tables.sqlretrieves all treatments and readings for eyes with a indication of AMD.src/synthetic_eylea_arm_study_table.sqlretrieves vars and aligns all eyes in alignment with ABC and target-trial eligibility
In which numerous quasi-randomsiation methods are employed for conditioning on confounding variables (vars that are common causes of both x and y).
These cohorts are assembled in 0_cohorts_generation.Rmd using bespoke functions listed below:
| Method | Function |
|---|---|
| Negative control | — |
| Inverse Probability Treatment Weighting | fnc/inverse_probability_treatment.R |
| Exact matching | fnc/exact_matching.R |
| Propensity Score Matching | fnc/propensity_score_matching.R |
These cohorts are then saved to /data as .csv and read in for each analyses.
Causal assumption of exchangeability visualised in 1_exchangeability.
Baseline characteristics created in 1_table_one.
Noninferiority analysis contained in 1_noninferiority.
Superiority analysis contained in 1_superiority.
Time-to-event analysis contained in 1_time_to_event.