The role of real-world evidence in health technology assessment: a case study of direct oral anticoagulants in the atrial fibrillation population

Ciminata, Giorgio (2019) The role of real-world evidence in health technology assessment: a case study of direct oral anticoagulants in the atrial fibrillation population. PhD thesis, University of Glasgow.

Abstract

Real-World Evidence (RWE) refers to any “data used for decision-making that are not collected in conventional Randomised Controlled Trials (RCTs), and is increasingly used in Health Technology Assessment (HTA) as an adjustment to the evidence coming from Randomised Controlled Trials (RCTs).

RWE can provide additional evidence concerning treatment safety and effectiveness, facilitate the identification of relevant subpopulations, and permit the inclusion and analysis of clinical endpoints not expected in RCTs but observed in real life. However, the use RWE in the context of HTA is still limited.

The aim of this thesis is to explore the role of RWE in economic evaluation by exploring methods to use with observational data and the role of RWE for a case study of direct oral anticoagulants (DOACs); a class of drugs, including apixaban, dabigatran and rivaroxaban, used for the prevention of stroke in the population affected by atrial fibrillation (AF). In addition to quantifying resource use and associated healthcare expenditure for the AF population in Scotland and evaluating propensity score methods for estimating the Average Treatment Effect (ATE), specific objectives are assessing cost as well as effectiveness and safety of DOACs using Scottish linked data.

Two cohorts, one consisting of patients with a diagnosis of AF or atrial flutter, and the other of patients on any oral anticoagulant (OAC) were identified from inpatient hospital records and prescribing data for the 1997 – 2015 study period. These data were complemented by outpatient attendances, the care home census and mortality records using individual patient data linkage.

As a first step, this thesis assessed the predictors of costs and estimated inpatient, outpatient, prescribing and care-home costs associated with AF, using population-based individual-level linked data. Inpatient admissions accounted for the majority of total costs and these were the main cost driver across all age groups.

Overall, inpatient cost contributions (~75 %) were constant across age groups.. This is offset by increasing care-home cost contributions. The inclusion of all available cost components is crucial for establishing overall costs, as these often extend beyond hospitalisation. Most importantly, the thesis found that patients’ age has a limited impact on overall AF-related cost, and therefore may not be the main driver of future growth of AF-related costs in an ageing Scottish population.

In order to identify an appropriate method for the comparative-effectiveness analysis, propensity score (PS) based method, such as PS matching, covariate adjustment including PS as covariate, and a series of Inverse Probability Weighting (IPW) methods were tested. A cohort of patients were followed from their first oral anticoagulant prescription to first clinical event (stroke and major bleeding) or death, and censoring was applied to treatment switching or discontinuation. In this methodological chapter, the approach that uses propensity scores (PS) as a covariate was identified as the most robust method to be used in the more comprehensive comparative-effectiveness analysis.

The comparative-effectiveness analysis, including additional clinical outcomes that were also used in the pivotal RCTs assessing the efficacy of DOACs versus warfarin in the AF population, found no statistically significant differences in risk of stroke for apixaban, dabigatran and rivaroxaban compared with warfarin. There were however, concerns over safety aspects of rivaroxaban, as it was associated with increased risk of all-cause mortality.

The hazard ratios estimated from the comparative-effectiveness analysis were used to populate a Markov model to evaluate the lifetime cost- effectiveness of DOACs compared to warfarin; one-way and probabilistic sensitivity analyses were carried out to assess the uncertainty around the findings and identify key drivers. At the £20,000 threshold, apixaban and dabigatran were found to be cost-effective in AF patients who are 50 years old when starting anticoagulation. Rivaroxaban, being the least effective intervention, was dominated by warfarin, being less costly but more effective than rivaroxaban.

This thesis shows the potential of RWE in general and within the Scottish healthcare setting. The findings highlight the importance of taking into account resource utilisation beyond hospital care, and assessing several comparative-effectiveness methods to understand strengths and limitation of each. Most importantly, the findings from this thesis have the potential to inform future research, prescribing patterns and provide real-world evidence for other healthcare settings, especially where rivaroxaban is the DOAC most widely prescribed.

Finally, this thesis shows that RWE generated from routinely collected linked data in Scotland, may well support the reassessment of prescription drugs accepted conditionally by the Scottish Medicine Consortium (SMC), an independent organisation that advises the NHS Health Boards about medicines, and would therefore support the SMC in making the final acceptance decision.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Keywords: real world evidence, health technology assessment, comparative effectiveness, atrial fibrillation, direct oral anticoagulant.
• Subjects: R Medicine > R Medicine (General)
• Colleges/Schools: College of Medical Veterinary and Life Sciences > School of Health & Wellbeing > Health Economics and Health Technology Assessment
• Supervisor's Name: Wu, Professor Olivia, Geue, Dr. Claudia and Langhorne, Professor Peter
• Date of Award: 2019
• Depositing User: Mr. Giorgio Ciminata
• Unique ID: glathesis:2019-76761
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 17 Dec 2019 08:19
• Last Modified: 12 Feb 2020 15:45
• Thesis DOI: 10.5525/gla.thesis.76761
• URI: https://theses.gla.ac.uk/id/eprint/76761