Comparative Framework in VBAs and Impact of Adherence
The Ticagrelor-Clopidogrel Experiment
– Pierre Etienne MD, Chitra Lele, PhD
To dissipate uncertainty around real-world clinical outcomes for some newly approved drugs, health insurers have implemented various contracts and arrangements with drug manufacturers, which fall under the general category of value-based agreements (VBAs).
In an effort to translate clinical evidence into value-based payment methods, some manufacturers have compared the new treatment to standard of care or an adequate comparator while keeping some of the elements of the registration trials and respecting providers’ drug selection. The problem with this approach (which does without randomized treatment assignment) is that it opens the door to confounders which are difficult to adjust statistically. Here we look at a recently published article that exemplifies this difficulty (1).
To demonstrate the comparative advantage of ticagrelor in the real world, Astra Zeneca structured a Value Based Agreement (VBA) comparative framework against clopidogrel. Many elements of this experiment were based on the PLATO trial (RCT), which had demonstrated superiority for ticagrelor in reducing the incidence of recurrent myocardial infarction (MI) in patients with ACS. Between 2015 and 2018, 11 contract-years of VBAs were implemented totaling nearly 32,000 unique patients. The results show that ticagrelor consistently outperformed expectations. The authors conclude that comparative VBA frameworks of this nature may be utilized more broadly in future applications. They also concede that confounders are not always balanced in that type of comparative VBA framework and that adjustment for them may alter results.
One of these well recognized confounders is medication adherence (1,2). A high threshold of medication adherence (at least 80% of days covered [Proportion of Days Covered (PDC)] over 365 days or time from index ACS to recurrent ACS) was required to be included in VBA analyses (Step 4 of the I/E criteria). This high threshold was selected to align the VBAs with the RCT (PLATO) where investigator-reported medication adherence was above 80% for both ticagrelor and clopidogrel. An 80% PDC is desirable in a RCT but it’s unlikely in the real world for an extended, year-long period of treatment unless there are programs to improve it.
In the ticagrelor VBAs the proportion of adherent patients is different in the clopidogrel and ticagrelor groups. Of those patients with a prescription within 30 days of discharge with an ACS event diagnosis (step 2 of the I/E criteria), 58% and 43% were adherent in the clopidogrel and ticagrelor respectively. About 34% higher adherence in the clopidogrel group relative to the ticagrelor group makes the interpretation of efficacy and safety results difficult.
Why was there a smaller proportion of adherent patients in the ticagrelor group? Ticagrelor is a little more complicated to dose since it is twice a day whereas clopidogrel is once a day. Both are administered with low dose aspirin. Could this explain the difference in adherence?
Does this difference have anything to with the lower propensity of providers to prescribe ticagrelor in the population of interest 7% (15972/230113) versus clopidogrel 19% (43027/229992)?
Could aligning the VBA analysis adherence criteria to the RCT have had the effect of revealing providers’ bias?
Perhaps the authors could consider doing a post hoc confounder analysis to the extent possible, to understand the impact of adherence on comparative efficacy and safety, but we do not have a simple method to address another confounder, the providers’ drug prescription propensity.
We believe that a patient support program to improve adherence would be a good investment for this type of VBA, even if trying to remove bias outside of an RCT is not easy.
1. Bhandary D. et al. Translating Clinical Evidence Into Value-Based Payment Models: Pooled Analyses of Innovative Real-World Outcomes Agreements for Ticagrelor in the United States”, Am. J Manag.Care. 2020; 26:S275-S286)
2. Mohiuddin AK. Risks and Reasons Associated with Medication Non-Adherence. Am J Pharmacol. 2019; 2(2): 1017
3. Cutler RL, Fernandez-Llimos F, Frommer M,et al. Economic impact of medication non-adherence by disease groups: a systematic review. BMJ Open 2018;8: e016982
Chief Medical Officer and Co-founder
Pharma industry veteran with 30+ years in large Pharma & in leading small biotechs, spearheading large initiatives and securing funding, psychiatry practice and research for 10+ years.