{Reference Type}: Journal Article
{Title}: Emulating randomised clinical trials in relapsing-remitting multiple sclerosis with non-randomised real-world evidence: an application using data from the MSBase Registry.
{Author}: Signori A;Ponzano M;Kalincik T;Ozakbas S;Horakova D;Kubala Havrdova E;Alroughani R;Patti F;Kuhle J;Izquierdo G;Eichau S;Yamout B;Khoury SJ;Karabudak R;Grammond P;Duquette P;Roos I;Butzkueven H;van der Walt A;Sormani MP;
{Journal}: J Neurol Neurosurg Psychiatry
{Volume}: 95
{Issue}: 7
{Year}: 2024 Jun 17
{Factor}: 13.654
{DOI}: 10.1136/jnnp-2023-332603
{Abstract}: BACKGROUND: To mimic as closely as possible a randomised controlled trial (RCT) and calibrate the real-world evidence (RWE) studies against a known treatment effect would be helpful to understand if RWE can support causal conclusions in selected circumstances. The aim was to emulate the TRANSFORMS trial comparing Fingolimod (FTY) versus intramuscular interferon β-1a (IFN) using observational data.
METHODS: We extracted from the MSBase registry all the patients with relapsing-remitting multiple sclerosis (RRMS) collected in the period 2011-2021 who received IFN or FTY (0.5 mg) and with the same inclusion and exclusion criteria of the TRANSFORMS RCT. The primary endpoint was the annualised relapse rate (ARR) over 12 months. Patients were 1:1 propensity-score (PS) matched. Relapse-rate ratio (RR) was calculated by mean of a negative binomial regression.
RESULTS: A total of 4376 patients with RRMS (1140 in IFN and 3236 in FTY) were selected. After PS, 856 patients in each group were matched. The ARR was 0.45 in IFN and 0.25 in FTY with a significant difference between the two groups (RR: 0.55, 95% CI: 0.45 to 0.68; p<0.001). The result of the emulation was very similar and fell within the 95% CI of that observed in the RCT (RR: 0.49, 95% CI: 0.37 to 0.64; p<0.001) with a standardised difference of 0.66 (p=0.51).
CONCLUSIONS: By applying the same inclusion and exclusion criteria used in the RCT and employing appropriate methodology, we successfully replicated the RCT results with only minor discrepancies. Also, even if the confounding bias cannot be fully eliminated, conducting a rigorous target trial emulation could still yield valuable insights for comparative effectiveness research.