Reproducible Evidence: Practices to Enhance and Achieve Transparency (REPEAT)

Study type

• Study type:
  Non-interventional study
• Scope of the study:
  Other

  Drug utilisation
• If ‘other’, further details on the scope of the study:
  Methodological

Non-interventional study

• Non-interventional study design:
  Cohort

Population studied

• Age groups:
  • Preterm newborn infants (0 – 27 days)
  • Term newborn infants (0 – 27 days)
  • Infants and toddlers (28 days – 23 months)
  • Children (2 to < 12 years)
  • Adolescents (12 to < 18 years)
  • Adults (18 to < 46 years)
  • Adults (46 to < 65 years)
  • Adults (65 to < 75 years)
  • Adults (75 to < 85 years)
  • Adults (85 years and over)
• Special population of interest:
  Renal impaired

  Hepatic impaired

  Pregnant women
• Estimated number of subjects:
  250

Study design details

• Main study objective:
  Our objective is to measure the current state of reproducibility for database research and empirically validate specific guidance/recommendations on what to report in order to achieve consistently reproducible and robust findings from healthcare database studies.
• Outcomes:
  The primary outcomes for each replicated study will be the primary outcomes reported in the original paper.
• Data analysis plan:
  •Descriptive frequencies for insufficient reporting on specific parameters•Standardized differences for baseline characteristics, incidence rates, and reported measures of association in original and replication•Calibration of measures of association in the original and replication•Relationship between lack of transparency in different areas (e.g. timing of cohort entry and follow up, algorithms to measure exposure, outcome, covariates etc.) and standardized differences in the original versus replication.•Proportion of studies with clear design or analysis flaws stratified by type. These may include immortal time bias, reverse causation, adjustment for intermediates,etc.•Plot measures of association with 95% CI for the original paper,replications using original methods, using plausible alternative choices, after external adjustment for residual confounding,after quantitative and probabilistic bias correction, after correction of clear design flaws,and using negative controls