Large randomised controlled trials (RCTs) are generally conducted across multiple centres. However, patient outcomes may be expected to differ between participating centres. Variability in patient outcomes between participating centres may impair the integrity and reliability of multicentre trial findings. Many clinicians believe that between-centre differences in multicentre RCTs represent a level of “background noise” that will reduce the chance of finding a statistically significant treatment effect.
The investigators will use data from several large multicentre RCTs conducted by the ANZCA Clinical Trials Network. Data collected in each trial will be used, to see if adjusting for trial centre ranking affects the primary findings of each trial. We will also combine key data (length of hospital stay and mortality rates) from all the trials and using this pooled data to look for any evidence of correlation between the ranking of trial centres by the incidence of perioperative complications overall, and their ranking by the trial treatment effect on that incidence. We will also conduct further analysis to see if these relationships are stronger for more complex or difficult trial interventions.
This extensive analysis will provide compelling evidence on the reliability of large multicentre trials to answer important clinical questions in anaesthesia and perioperative medicine, with equally important implications for research in other fields of medicine.
Professor Philip Peyton, Austin Health, Melbourne
Ms Vanessa Pac Soo, Centre for Epidemiology and Biostatistics, Melbourne
The project was awarded $A17,250 funding through the ANZCA research grants program for 2023.
In perioperative medicine, patient outcomes have been shown to vary substantially between hospitals due to potential differences in the characteristics of patients treated and/or quality of care.
Randomisation in multicentre trials is typically stratified by hospital at the design stage, and hospital is often adjusted for in the statistical analysis, However, it is unclear whether heterogeneity in care might also be associated with heterogeneity in trial treatment effects, which may confound trial results by weakening trial treament effect sizes.
Data from 6 large multicentre ANZCA CTN trials was used to seek evidence of this.
The main finding was that if trial randomisation is stratified by hospital and then accounted for in the statistical analysis, treatment effects for mortality and hospital length of stay appear to be robust to heterogeneity in patient outcomes across the participating hospitals.
This strengthens the assumption of the robustness of evidence on treatment outcomes from multicentre trials in anaesthesia and perioperative medicine.
The results have been presented at the 7th International Clinical Trials Methodology Conference 2024 in October in Edinburgh, Scotland.
Ref: Multicentre trials in anaesthesia: accounting for differences between hospitals.
Anurika De Silva1,2, Vanessa Pac Soo1,2, Philip Peyton3, Sabine Braat1,2
- Centre for Epidemiology and Biostatistics, Melbourne School of Population and Global Health, The University of Melbourne, Melbourne, Australia.
- Methods and Implementation Support for Clinical and Health (MISCH) research Hub, Faculty of Medicine, Dentistry and Health Sciences, The University of Melbourne, Melbourne, Australia.
- Austin Health, Melbourne, Australia.
Preparation for formal manuscript preparation and peer reviewed journal submission is underway (target journal BMC Trials).