ARTICLE SUPPLEMENTARY INFORMATION
(Supporting information is provided by the author to support the content
of the parent article)

An Evaluation of the Cost-Effectiveness of Rituximab in Combination with Chemotherapy for the First-Line Treatment of Follicular Non-Hodgkin's Lymphoma in the UK

Joshua A. Ray, Emma Carr, Gavin Lewis, Robert Marcus

– Value in Health Supplemental Information

Appendices

Appendix 1

Parametric functions

In order to further assess the impact of introducing additional uncertainty in the probabilistic sensitivity analysis, the asymptotic standard error was estimated for the probability of transitioning from the “Progressed” health state to death (an exponential distribution was assumed) and the probability of moving from “PFS” to the “Progressed” health state (where the exponential distribution was determined to be the best fit as determined by the R² statistic). The asymptotic standard error was calculation by taking the square root of the asymptotic variance, given by: where “r” is the number of events [20]. Estimation of the standard error for each comparison assumed that 20% of the patients were censored, based on available information from the randomized controlled trail comparing R-CVP and CVP [12].

In the one comparison where the Weibull function was determined to be the best fit (CHVP vs R-CHVP), the standard errors of the Lambda (λ) and Gamma (γ) were increased 3-fold.

Hazard Ratios

Hazard ratios were calculated from the KM curves reported in the clinical trial publications. Initially the cumulative hazard was calculated by taking the sum of the log survival for each treatment arm.

The hazard ratios were calculated by taking the ratio of the cumulative hazard for the R-chemotherapy treatment arm divided by the cumulative hazard for the chemotherapy-alone arm.

Given the difficulty of introducing uncertainty into the analysis due to the lack of access to patient level data, the standard errors of the respective hazard ratios were calculated using the method described by Tudur et al [37].

The standard error of the log hazard ratio is given by: where “R” is the number of patients randomized to the respective treatment arm and “E” is the number of events assuming 20% censoring of patients.

where i = “R-chemotherapy” or “chemotherapy”, respectively and N represents the sample size of the intent to treat population in each treatment arm.

Appendix 2

Sensitivity Analysis

Results of sensitivity analyses investigating the impact of including additional costs of treating adverse events and infusion site reactions to the R-chemotherapy treatment arms.

Acceptability Curves

Value in Health Article Supplementary Information