Using short-term evidence to predict six-month outcomes in clinical trials of signs and symptoms in rheumatoid arthritis

Richard Nixon

Novartis

Objectives: A model is presented to generate a distribution for the probability of an ACR response at six months for a new treatment for rheumatoid arthritis given evidence from a one or three month clinical trial.

Methods: A model is presented to generate a distribution for the probability of an ACR response at six months for a new treatment for rheumatoid arthritis given evidence from a one or three month clinical trial. We denote j as the treatment arm within each clinical trial, and k the type of treatment used (k=1 for MTX, k=2 for biologic, k=3 for biologic plus MTX); t₀ as the earlier time point (either one month or three months depending on the model); and t₁ the six-month time point. We then denote n_j as the number of patients in treatment arm j for which there is data available at both time points, and r_jt as the number of patients achieving the ACR response criteria at time point t. We assume that r_jt0 and r_jt1 are binomially distributed with parameters φ_jt0 and φ_jt1 representing the probabilities of response, respectively. φ_jt1 is assumed to be dependent upon φ_jt0, in a logistic regression. To estimate the ACR response rate at time point t₁ from the ACR response rate at time point t₀, we fit the following statistical model:
r_jt0 ~ Binomial(n_j, φ_jt0)
r_jt1 ~ Binomial(n_j, φ_jt1)
logit(φ_jt1) = α_j+ βlogit(φ_jt0)
α_j~N(μ_k, σ²)

Results: The model is assessed by Bayesian predictive P-values that demonstrate that the model fits the data well.

Conclusions: The model can be used to predict the number of patients with an ACR response for proposed six-month clinical trials given data from clinical trials of one or three months duration

References:
[1] Nixon RM, Bansback N, Stevens JW, Brennan A and Madan J. Using short-term evidence as prior information in the design of six-month clinical trials of signs and symptoms in rheumatoid arthritis. Pharmaceutical Statistics. 2008 early view. DOI 10.1002/pst.351

Reference: PAGE 18 (2009) Abstr 1545 [www.page-meeting.org/?abstract=1545]

Poster: Methodology- Model evaluation

PDF poster/presentation (click to open)