Registry-based randomized controlled trials merged the strength of randomized controlled trails and observational studies and give rise to more pragmatic trials

J Clin Epidemiol. 2018 Jan;93:120-127. doi: 10.1016/j.jclinepi.2017.09.017. Epub 2017 Sep 22.


Objectives: The objective of this study was to analyze the features of registry-based randomized trials (rRCTs).

Study design and setting: We systematically searched PubMed for rRCTs. Study selection was performed independently by two reviewers. We extracted all data in standardized tables and prepared descriptive summary statistics.

Results: The search resulted in 1,202 hits. We included 71 rRCTs. Most rRCTs were from Denmark and Sweden. Chronic conditions were considered in 82.2%. A preventive intervention was examined in 45.1%. The median of included patients was 2,000 (range: 69-246,079). Definition of the study population was mostly broad. Study procedures were regularly little standardized. The number of included and analyzed patients was the same in 82.1%. In half of the rRCTs, more than one registry was utilized. Various linkage techniques were used. In median, two outcomes were collected from the registry/ies. The median follow-up of the rRCTs was 5.3 years (range: 6 weeks to 27 years). Information on quality of registry data was reported in 11.3%.

Conclusion: rRCTs can provide valid (randomization, low lost-to-follow-up rates, generalizable) patient important long-term comparative-effectiveness data for relative little effort. Researchers planning an RCT should always check whether existing registries can be used for data collection. Reporting on data quality must be improved for use in evidence synthesis.

Keywords: External validity; Health registries; Internal validity; Patient registries; Pragmatic trials; Randomized controlled trials; Registry-based randomized trials; Study design.

MeSH terms

  • Denmark
  • Humans
  • Observational Studies as Topic
  • Pragmatic Clinical Trials as Topic*
  • Randomized Controlled Trials as Topic*
  • Registries*
  • Research Design
  • Sample Size
  • Sweden