Advancing learning health care systems with best practice meetings: a mixed-methods study with insights from the Dutch Institute for Clinical Auditing.

backgroundNational clinical registries are widely used to benchmark quality indicators and improve care. Advancing registry-based auditing toward learning health care systems (LHCS) requires structural integration of continuous evaluation and improvement cycles. One way to operationalize the check phase of the Plan-Do-Check-Act cycle is through best practice meetings (BPMs), in which health care professionals transparently discuss inter-hospital variation in clinical outcomes and formulate targeted improvement actions. Since 2018, the Dutch Institute for Clinical Auditing (DICA) has organized BPMs across multiple registries. This study provides the first descriptive overview of the structure, participant experiences, and facilitators and barriers to the organization and conduct of BPMs.

methodsA descriptive mixed-methods study was conducted. Data on session characteristics (e.g. number, topics, format) and follow-up actions were obtained through a survey among coordinators of DICA registries. Participant satisfaction scores from evaluations of BPMs held between 2018 and 2025 were retrospectively extracted, where available. Semi-structured interviews were held with four registry coordinators and one project manager. Interviews were thematically analysed to identify facilitators and barriers to the organization and conduct of BPMs.

resultsIn total, 55 BPMs were held across 14 registries. Most were national and multidisciplinary. Frequently discussed topics included length of stay, treatment waiting times, Textbook Outcome, and mortality. Best practices and/or quality improvement actions were formulated for 22 of 55 BPMs (40%). The mean participant satisfaction score was 4.0 on a 5.0-point Likert scale. Five themes were identified through thematic analysis of interviews: preparation, setting and format, learning climate, implementation, and cyclical learning. Facilitators included sharing data in advance, skilled moderation, an open and safe atmosphere, and recurring themes; barriers included an online setting that hampered interaction, concerns regarding data quality, and discussion of too many topics during a single meeting.

conclusionBPMs appear to be a feasible and well-accepted instrument to support data-driven quality improvement across hospitals. Implementation of BPMs could support LHCS by operationalizing the learning cycle from benchmarking to shared learning, action planning, and follow-up. Future research should evaluate whether BPMs lead to implementation of improvement actions and measurable changes in clinical outcomes.