. 2019 Nov 11;38(2):205–216. doi: 10.1007/s40273-019-00855-9

Box 2.

Application of TRUST in a case study

NICE appraisal titled ‘Pembrolizumab for treating relapsed or refractory classical Hodgkin lymphoma’ (2018)
Pembrolizumab was compared with standard of care in patients who did and did not receive prior autologous SCT. This NICE appraisal suffered from a lot of uncertainty (Table 2). Most impactful uncertainty locations: model structure, transition probability and relative effectiveness estimates. Model structure: the main uncertainty was that the time at which patients would receive allogeneic SCT was modelled as a fixed time point (bias). Not included in PSA or scenario analysis (alternative time point submitted later upon request). Transition probability estimates: distributions to model OS were selected without appropriate justification (methods). Not included in the PSA but explored in scenario analysis. The company deemed their own trial to be immature for modelling OS (imprecision), and the estimation of post-allogeneic SCT OS was based on a study that included 13 patients (imprecision). Both included in PSA. Relative treatment effectiveness: using a naïve comparison for two single-arm studies (methods and bias). Not in PSA, but in scenario with matched-adjusted indirect comparison. Comparator data had a mixed population of patients who did and did not receive prior autologous SCT (bias). Not explored in PSA or scenarios (scenario with alternative data for one population submitted later upon request).
Unknown impact on cost effectiveness: context/scope and selection of evidence. Context/scope: comparators were omitted, best supportive care (unavailability of data) and nivolumab due to direction by NICE (unavailability and methods). No PSA or scenarios. Selection of evidence: a list of health economic publications on this topic was not provided (transparency). No PSA or scenarios.
Conclusions: 1. Example of the increasing number of applications for reimbursement based on single-arm evidence and immature survival data. 2. Most impactful uncertainties were issues with bias and indirectness, methods and imprecision and were not always included in the PSA. 3. Reaching a reimbursement decision under such circumstances is potentially associated with a significant risk.

NICE appraisal titled ‘Pembrolizumab for treating relapsed or refractory classical Hodgkin lymphoma’ (2018)

Pembrolizumab was compared with standard of care in patients who did and did not receive prior autologous SCT.

This NICE appraisal suffered from a lot of uncertainty (Table 2).

Most impactful uncertainty locations: model structure, transition probability and relative effectiveness estimates.

Model structure: the main uncertainty was that the time at which patients would receive allogeneic SCT was modelled as a fixed time point (bias). Not included in PSA or scenario analysis (alternative time point submitted later upon request).

Transition probability estimates: distributions to model OS were selected without appropriate justification (methods). Not included in the PSA but explored in scenario analysis.

The company deemed their own trial to be immature for modelling OS (imprecision), and the estimation of post-allogeneic SCT OS was based on a study that included 13 patients (imprecision). Both included in PSA.

Relative treatment effectiveness: using a naïve comparison for two single-arm studies (methods and bias). Not in PSA, but in scenario with matched-adjusted indirect comparison.

Comparator data had a mixed population of patients who did and did not receive prior autologous SCT (bias). Not explored in PSA or scenarios (scenario with alternative data for one population submitted later upon request).

Unknown impact on cost effectiveness: context/scope and selection of evidence.

Context/scope: comparators were omitted, best supportive care (unavailability of data) and nivolumab due to direction by NICE (unavailability and methods). No PSA or scenarios.

Selection of evidence: a list of health economic publications on this topic was not provided (transparency). No PSA or scenarios.

Conclusions:

1. Example of the increasing number of applications for reimbursement based on single-arm evidence and immature survival data.

2. Most impactful uncertainties were issues with bias and indirectness, methods and imprecision and were not always included in the PSA.

3. Reaching a reimbursement decision under such circumstances is potentially associated with a significant risk.

NICE National Institute for Care and Excellence, OS overall survival, PSA probabilistic sensitivity analysis, SCT stem cell transplant