Objectives: To evaluate the empirical concordance between the hazard ratio (HR) and the median ratio (MR) in survival cancer studies.
Study Design and Setting: We selected all cancer survival articles from the New England Journal of Medicine published between 2000 and 2010. The qualitative concordance was estimated by the proportion of measured pairs in which the treatment effects for the MR and HR are in the same direction. The quantitative concordance was assessed through (1) the mean difference between the logarithms of the measures, (2) the Lin coefficient, and (3) the Bland–Altman plot.
Results: We retrieved 106 measured pairs (HR–MR) corresponding to 54 articles. Concordance was high, at both the qualitative (99 of 106, 93.4%) and quantitative levels (mean MR-to-HR ratio, 1.01; 95% confidence interval: 0.95, 1.06). However, the 95% Bland–Altman discordance limits indicate that the MR can be up to 50% higher or 50% lower than the HR.
Conclusion: The average concordance allows trialists to approximate HR from MR to determine sample size. However, the discordance limits are too great to consider that both measures are interchangeable. The actual policy to report HR only is not enough. Our results emphasize the need to attach descriptive survival measures to the HR.
To quantify the degree of disagreement between the two most popular methods for dealing with missing data: intention to treat (ITT) and per protocol (PP).
Study Design and Setting
We performed a systematic review of randomized two-armed clinical trials (CTs) published between 2001 and 2003, abstracted in PubMed and reporting both the ITT and PP analyses on a primary binary endpoint, out of which 74 papers were finally selected. The treatment effect of each CT was measured by the odds ratio, and the disagreement between them was quantified by the Bland–Altman method.
On average, the PP estimator provides greater values LogeORPP = 1.25·LogeORITT, (95% CI: 1.15, 1.35) than the corresponding ITT estimator, although the limits of concordance showed that the ratio between the two estimators varies greatly from 0.39 up to 2.53.
These results confirm that missing values may cause both systematic and unpredictable bias in CTs. Further efforts should be made to minimize protocol deviations and to use better statistical methods to highlight the drawbacks of missing information. In the presence of protocol deviations, the conclusion of a CT cannot rest on the single reporting of either the ITT or the PP approach alone.