“Unnatural” history: Modeling disease progression using observational data

Purpose: Cost-effectiveness analysis requires comparison of outcomes in treated and untreated populations. Data from randomized clinical trials (RCT) do not provide progression rates representative of the general population, while treatment effects in observational data may be biased due to non-randomization. We developed a novel approach for estimating untreated progression rates (controls) by using data from a nationally representative patient cohort, as well as RCT estimates.

Methods: We used data from the Sonya Slifka multiple sclerosis (MS) study. Disease progression was characterized by disability-based disease states and relapses. State transition probabilities were modeled using a first-order Markov model, adjusting for individual characteristics, disease status, and specific disease-modifying therapy (DMT). We developed an iterative multinomial logistic model, constraining the effects of DMT to match those reported by RCTs.

Results: After correcting for DMT treatment effects and risk factors, the probability of disease progression was greater for estimates based on all patients compared to estimates based on untreated individuals only. The 95% confidence intervals using the entire cohort were narrower than the intervals based on untreated patients only.

Conclusions: Untreated patients in our study had lower estimates of disease progression than the treated patients would have had if they remained untreated, suggesting patients forgoing treatment are likely to have milder forms of MS. Correcting for treatment effects in a more inclusive patient group likely provides a more realistic estimate of disease progression than characterizing progression in an untreated cohort. The use of a broader cohort also improves the precision of disease progression estimates.