Traditionally the use of Real-World Data (RWD) in medicine development has focused on the following descriptive (non-comparative) and comparative studies:
Descriptive studies
- Description of the epidemiological burden of disease: incidence and prevalence, risk factors, rates of progression (and mortality)
- Description of symptoms and quality of life impacts for patients with the disease, as well as carer burden where applicable
- Description of the associated burden on healthcare (and other) services: treatment patterns, resource utilisation and cost
- Estimation of associations of possible study endpoints (efficacy) with long term outcomes: epidemiological ‘prediction’ or ‘disease’ models
- Epidemiological studies of safety post launch, often required to meet regulatory requirements of risk management (pharmacovigilance studies)
- Clinical validation studies for new endpoints, especially patient-reported and digital, to meet the standards set by regulatory agencies and health care systems for formal qualification and subsequent use
- Patient preference studies, utility estimation, discrete choice experiments and conjoint analyses (to understand the strength of relative preference for different health outcomes).
These methods were not the focus of the GetReal project. There is however a wide-ranging literature on best practice in performing these types of studies and analyses. Useful guidance can be found at:
- ISPE (International Society for Pharmacoepidemiology): Education and Learning Center
- ISPOR (international Society for Pharmacoeconomics and Outcomes Research): HEOR Resources
Comparative studies
- Prospective observational studies of the effectiveness and health care impact of existing and new therapies in usual practice (post-launch)
- Retrospective (database) studies of the effectiveness and health care impact of existing and new therapies in usual practice (post-launch)
These methods are described in more detail in Generating Real-World Evidence, and methods to adjust for potential bias are described in Adjusting for Bias in Non-Randomised Studies.