Danish Medicines Agency
We are currently at a crossroads as regulators: New technologies enable us to incorporate new data sources in our regulatory evaluation of drugs and devices. A true paradigm shift with continuous benefit risk assessment based on not only Randomised trials but also real world evidence data.
University of Copenhagen
In spite of extensive testing of medicines before authorization, knowledge of their benefits and risks is inherently limited at the time of marketing authorization, even more so when products are authorized under the Conditional Marketing Authorisation pathway. Specific obligations, including commitments to perform post-authorisation studies (PAS) can be agreed upon between regulators and companies to resolve uncertainties about benefits and risks after authorisation. Whether these studies are actually conducted, whether this happens in an acceptable and agreed-upon timeframe, and, ultimately, whether uncertainties are indeed resolved is subject to debate. It was the topic of interest of regulatory science studies on the performance of the regulatory system in this respect.
University of Oslo
The lecture will give examples how the PRAC works using medication safety in pregnancy and lactation as the motivation example. The Pharmacovigilance Risk Assessment Committee (PRAC) is the European Medicines Agency's (EMA) committee responsible for assessing and monitoring the safety of human medicines. The PRAC is responsible for assessing all aspects of risk management of human medicines, including: the detection, assessment, minimisation and communication of the risk of adverse reactions, while taking the therapeutic effect of the medicine into account; design and evaluation of post-authorisation safety studies; pharmacovigilance audit.
Pragmatic randomized trials are designed to address real-world questions about options for care and to guide decisions. However, the characteristics which makes these trials pragmatic – including typical patients and care settings, clinically relevant comparators, unconcealed assignment to treatment, and longer follow-up – also make them vulnerable to post-randomization confounding from incomplete adherence and post-randomization selection bias. These sources of bias are common in observational epidemiology, and the use of analytic approaches pioneered for observational studies can improve inference from pragmatic trials. We propose causal inference guidelines tailored for the analysis of pragmatic randomized trials using methods from observational research.
Real-world evidence on drug safety and effectiveness is increasingly used to inform regulatory and policy decisions. Observational studies based on data from healthcare registers are often designed and analysed to mimic clinical trials. Cohorts of new users are followed, and confounding is dealt with using propensity score techniques. Studies focus mainly on differences between drugs and frequently do not address the variation among patients. Thus, they are not informative about important issues in real-world drug use, such as risk factors for adverse reactions or subgroups of patients who may experience little benefit from the drug.
Randomized clinical trials are commonly analyzed by the intention-to-treat (ITT) approach (participants analyzed in the group to which they were randomized, regardless whether they received or adhered to the allocated therapy) and per-protocol (PP) approach (only participants who fulfill the protocol requirements regarding inclusion, intervention, and outcome assessments). PP is a "best-case scenario" to reveal the actual effect of the investigated therapy, whereas ITT handles the bias associated with the non-random loss of participants. The approach with complementing ITT and PP/“on treatment” analyses is sometimes applied in observational, real-world studies. However, the implications for interpreting safety and effectiveness data based on this approach in a real-world setting might raise different challenges.
In register-based pharmacoepidemiological studies of medication effects and side-effects it is essential to determine treatment exposure at specific time points. I will present new methodologic developments based on the parametric Waiting Time Distribution with a focus on how it can be used in common study designs (cohort, case-control) and discuss how it provides a foundation for further developments in methods and applications in pharmacoepidemiology. The ultimate aim is to make the use of decision rules for prescription durations obsolete by establishing a statistically valid method for estimating medication effects and side-effects, and I will show how close (or far) we are from reaching that goal.
Large electronic health records (EHR) are a cost-effective resource for detection of intended and unintended treatment effects. Despite their advantages, these data have been criticized for their inability to include information on confounders related to the indication of use - a particular concern in epidemiological studies of drug effects. Traditionally, these studies rely on methods such as adjustment to minimize between-group differences and in the recent literature, the major focus has been on propensity scores methods that can make use of the growing number of variables from EHR data. The strength of restricting study populations, in the conduct of epidemiological studies of drug effects, is less well appreciated. The goal of restricting is ultimately to obtain less biased effect estimates by making patients more homogenous regarding potential confounding factors. The potential of restriction will be illustrated by the findings from a Danish cohort study mimicking the design criteria of a RCT.
University of Tampere and University of Helsinki
The statistics on primary health care database (AvoHILMO) are based on care notifications that are collected from health care units in the public sector on the basis of personal identity numbers. Care notifications contain data on service provider and the client's/patient's municipality of residence as well as information concerning admission, treatment, procedures and discharge. The National Institute for Health and Welfare (THL) can, on a case-by-case basis, grant permission to use registers for purposes of scientific research.
Norwegian Directorate of Health, Trondheim, and senior researcher, Norwegian Institute of Public Health
The Norwegian Directorate of Health is responsible for two nationwide registries. These registries, the Norwegian Patient Registry (NPR) and the Norwegian Registry for Primary Health Care (NRPHC), together cover all governmental-funded health care. The NPR (specialist health care) was established in 2008, while the NRPHC (primary health care) was established in 2017. Data from the NPR are extensively used in a large variety of epidemiological studies. Data from the NRPHC will increase its importance when the registry covers a longer time period. In this talk, the NRPHC is presented as a possible future research tool.
Drugs used in hospitals are often more expensive and more potent than those redeemed at community pharmacies. Still, databases containing in-hospital drug use are sparse and seldom validated. I will present our experiences with establishment and posterior validation of an in-hospital database: EPM-research. EPM-research comprises the Capital Region of Denmark with 1.8 million inhabitants. We hope that by sharing our experience, including considerable challenges when ensuring data quality, we can help colleagues wanting to establish similar databases. Key variables and examples of relevant studies will be presented.
Aarhus University Hospital
Biomarkers are used in everyday clinical practice in primary care and in the hospital setting for diagnosis, screening, monitoring, assessment of prognosis, and evaluation of treatment effects and safety. This talk provides an overview of the Danish healthcare research databases that contain information on routine individual-level biomarker data.