Background on Patient Registry Analysis & Data Management Working Group continued…

Patient registries are prospective, observational cohort studies of patients with or at risk for a particular disease and/or receiving a particular treatment / intervention. They can be used for understanding natural history, assessing or monitoring real-world safety and effectiveness, assessing quality of care and provider performance, and assessing cost-effectiveness.  Registries typically involve a more diverse group of patients than clinical trials and spotlight what happens in actual medical practice, which better reflect real-world management practices and outcomes than RCTs.

Moreover, patient registry information can build on clinical trial data.  For instance, if a preliminary cost-effectiveness model has already been developed, then registry data can be used to update the model to provide a more realistic picture of a product’s value. This will be an increasingly critical activity as payors, pricing authorities, guideline developers, and agencies involved in technology appraisal reassess a product’s cost-effectiveness in the years following its introduction.

Furthermore, registries allow collection of clinical effectiveness data, and they represent an opportunity to prospectively collect and report on important outcomes, such as health-related quality of life (HRQoL), health care resource utilization, and patient satisfaction. Data from registries may also be used to support other health economics/outcomes research (HE/OR) initiatives, such as cost-of-illness studies, economic modeling, cost effectiveness analyses, and capture endpoints such as long-term outcomes and patient utilities.

Patient registries are used extensively by the outcomes research community, government health care regulators, and the pharmaceutical, biotechnology, and medical device industries for risk assessment, evaluation of risk management interventions, and to address post-approval regulatory requirements (e.g., safety surveillance).

Registry evidence illustrates actual treatment results in the real world, and therefore, it is critical to improve analyses and interpretations relevant to this domain.  

Whereas for clinical trial data there are clear guidelines for the conduct and reporting of analyses (CONSORT), these often do not address the analyses required for patient registry data.  Although recent collaborative efforts (AHRQ, GRACE Principles) have developed guidelines for the design and conduct of patient registries, these initiatives do not address specific recommendations for data management and analysis of these data.  There are specific analysis challenges that are related to assessing 'real world' data without the structure of a clinical trial protocol.  For a clinical trial, the descriptive analyses are based on standard assessment tools and fixed assessment times, and rely on randomization to allow group comparisons free from bias due to patient selection and other factors.  Furthermore, full clinical trial data collection processes minimize missing information and misclassification that may be present in ‘real world’ data. 

The proposed manuscript is an effort to develop consensus on good research practices for patient registry data analysis.  It will also raise awareness of the statistical methodologies appropriate for the unique nature of registries, tackling issues such as missing data, varying and uncontrolled office visit schedules, participants with multiple diseases and other challenges that are much more prevalent in registries than randomized controlled clinical trials.