Send to

Choose Destination
Ann Oncol. 2018 Oct 18. doi: 10.1093/annonc/mdy462. [Epub ahead of print]

Interpretation of Time-to-event Outcomes in Randomized Trials: an online randomized experiment.

Author information

Department of Biostatistics, Boston University School of Public Health, Boston, MA, USA.
Division of General Pediatrics, Department of Medicine, Boston Children's Hospital, Boston, Massachusetts.
Boston Medical Center, Department of Emergency Medicine, Boston, Massachusetts.
Section of General Internal Medicine, Department of Medicine, Boston University School of Medicine, Boston, Massachusetts.
Program On Regulation, Therapeutics, And Law (PORTAL), Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School.
National Heart, Lung, and Blood Institute's and Boston University's Framingham Heart Study, Framingham, MA.
Department of Epidemiology, Boston University School of Medicine, Boston, MA, USA.



Multiple features in the presentation of randomized controlled trial (RCT) results are known to influence comprehension and interpretation. We aimed to compare interpretation of cancer RCTs with time-to-event outcomes when the reported treatment effect measure is the hazard ratio (HR), difference in restricted mean survival times (RMSTD), or both (HR+RMSTD). We also assessed the prevalence of misinterpretation of the HR.

Patients and methods:

We performed a randomized experiment. We selected 15 cancer RCTs with statistically significant treatment effects for the primary outcome. We masked each abstract and created 3 versions reporting either the HR, RMSTD, or HR+RMSTD. We randomized corresponding authors of RCTs and medical residents and fellows to one of 15 abstracts and one of 3 versions. We asked how beneficial the experimental treatment was (0 to 10 Likert scale). All participants answered a multiple-choice question about interpretation of the HR. Participants were unaware of the study purpose.


We randomly allocated 160 participants to evaluate an abstract reporting the HR, 154 to the RMSTD, and 155 to both HR+RMSTD. The mean Likert score was statistically significantly lower in the RMSTD group as compared with the HR group (mean difference -0.8, 95% confidence interval, -1.3 to -0.4, pā€‰<ā€‰0.01) and as compared with the HR+RMSTD group (difference -0.6, -1.1 to -0.1, pā€‰=ā€‰0.05). In all, 47.2% (42.7% to 51.8%) of participants misinterpreted the HR, with 40% equating it with a reduction in absolute risk.


Misinterpretation of the HR is common. Participants judged experimental treatments to be less beneficial when presented with RMSTD as compared with HR. We recommend that authors present RMST-based measures alongside the HR in reports of RCT results.


Supplemental Content

Full text links

Icon for Silverchair Information Systems
Loading ...
Support Center