Caution Using Propensity Score Adjustment for Multiple Sclerosis Studies

Propensity score adjustment has been used increasingly often in nonrandomized registries and real-world studies. These authors compared the placebo arm (A) from the AFFIRM phase 3 pivotal trial of natalizumab to the placebo arms (B) from the DEFINE and CONFIRM pivotal trials of dimethyl fumarate for treatment of multiple sclerosis.

Propensity score adjustments were successful in reducing baseline differences between the trials. Unadjusted analyses suggested superiority of placebo B for 6-month confirmed disability progression (0.64; 95% confidence interval, 0.47–0.88). After propensity score adjustments, placebo B continued to show consistent superiority over A. The authors acknowledge limitations of applying propensity score adjustments to …