This meeting of the Digital Health Learning (DHLC) and Clinical Effectiveness Research Collaboratives (CERIC) focused on peer-reviewed journal publication frameworks that promote dissemination of evidence developed by learning health systems.
A premise of the learning health system is that data developed during the delivery of care will contribute evidence about the comparative effectiveness of preventive and therapeutic policies, treatments, and procedures. The expectation is that this information will be useful to delivery system leaders who make many decisions based on analyses of routinely collected health information in enterprise data warehouses or commercially available data, data from registries, or secondary analysis of clinical trial data. This approach to decision-making is sometimes at odds with some journals’ requirements for prior specification of testable hypotheses, and criteria for assessing the statistical significance of analyses of hypotheses that are specified after data are available to investigators who collected the data, and, increasingly, to the public.
Key issues raised at this meeting include:
- How should health system based investigations be designed, conducted, and reported to maximize their credibility? Ideally, this advice should recognize that these evaluations are often designed to provide actionable information in the shortest possible time.
- Does different guidance apply to prospective and retrospective analyses?
- When is it appropriate to adjust for multiple hypothesis testing?
- What role should prior registration play, for instance in ClinicalTrials.gov?
- What are the current barriers to publishing results of health system based assessments in peer-reviewed journals? What are appropriate approaches for addressing these barriers?
- How can researchers, journal editors, and decision-makers work together to develop standards to promote analysis and publication that maximizes the value of data from both randomized controlled trials and observational studies that are embedded in delivery systems?
Please view the meeting highlights for a detailed account of what was discussed at the meeting.
If you have any questions, please email LeadershipConsortium@nas.edu.