When new drugs are approved, often companies are requested to study remaining questions or uncertainties about effects or side-effects of the new drug. One way to study these is through so-called observational studies, in which the investigator observes what naturally occurs in daily care rather than influencing treatment decisions. Since the investigator only observes, he does not influence who is treated and who is not. Therefore characteristics of treated patients often differ from untreated (or other actively treated) patients and these may also change over time. These different patients may in turn also differ in their risk of suffering from a disease or experiencing a side-effect of a drug.
Other reasons for variation in the type of patients using a drug may be a government decision to allow the drug for treatment of another disease or to reimburse it so patients can actually use it, or a decision by doctors to more often use that drug. All these reasons may change the results of the study. Our study aims to show specific situations under which such changes do or do not occur.
We will perform this study using an example situation of patients with heart rhythm disorder that are treated with one of the blood thinners called rivaroxaban or warfarin to prevent stroke or clotting of the blood elsewhere in the body to occur. Ultimately, our study may help regulatory authorities to know how and when they should ask for studies, and how long these studies should be continued.
Background
For most medicines, post-approval data generation is important to further characterise or confirm their benefit-risk profile. Observational studies are especially suited to study e.g. effectiveness and safety outcomes but are also prone to various biases that may arise from variability in patient characteristics that may change over time due to deliberate decision-making during the drug life-cycle. Furthermore, increased sample size due to progress of the drug life-cycle will impact power to detect beneficial and adverse effects.
Objectives
This study aims to quantify the impact of variability in patient characteristics and increasing sample size during a drug life-cycle on relative effectiveness and safety estimates and precision for a case-study of rivaroxaban versus warfarin indicated for the prevention of stroke and systemic embolism in non-valvular atrial fibrillation. Furthermore, it aims to quantify the impact of significant regulatory, reimbursement and clinical decision-making on these estimates.
Study design
This is a retrospective study of rivaroxaban versus warfarin, studying multiple cohorts of patients at different points in time during the life-cycle within the UK Clinical Practice Research Datalink between 2011 and 2019. Cohorts will vary in sample size and duration of follow-up and may vary in terms of patient characteristics. Patients with an incident diagnosis of non-valvular atrial fibrillation will be followed until the occurrence of stroke or systemic embolism, major bleeding or angioedema, or until a censoring event.
Data analysis
Propensity scores will be used to identify patient variability between cohorts. To assess relative effectiveness and safety estimates for each cohort, we will model the data using Cox proportional hazards regression analyses and calculate hazard ratios and 95% confidence intervals, both unadjusted and adjusted for possible confounders. Individual estimates for each cohort and their evolvement over time will be described and visualised. The impact of decision-making will be assessed by adding interaction terms with these time points.
Composite endpoint of stroke (ischaemic or haemorrhagic) and systemic embolism, major bleeding, angioedema
Olaf Klungel - Chief Investigator - Utrecht University
Lourens Bloem - Corresponding Applicant - Utrecht University
Arno W Hoes - Collaborator - Maastricht University
Aukje Mantel-Teeuwisse - Collaborator - Utrecht Institute for Pharmaceutical Sciences
Helga Gardarsdottir - Collaborator - Utrecht University
Hubert Leufkens - Collaborator - Utrecht University
Menno E. van der Elst - Collaborator - Medicines Evaluation Board