Loading
1030 hazard ratios from 679 trials on one axis. A hazard ratio below 1 means the experimental arm did better on that endpoint; the line is the 95% confidence interval. Pick a cancer, a setting and an endpoint to see comparable trials side by side, then download the rows.
In plain words: every square left of the vertical line at 1 is a trial where the new treatment did better than the old one, and the further left, the bigger the difference; 0.5 means the risk of the event was halved. Each trial enrolled people who were already in the situation it describes (a particular stage, a particular prior treatment), so the ratios say how much better a treatment worked for that group, not how likely the event is for any one person. The trial page behind each row has the outcomes explained in plain words.
Each row is one endpoint from one trial, taken from the structured outcomes recorded in the corpus with the publication linked as its source. The square sits at the reported hazard ratio; its area grows with the number of participants in the arms (or the trial enrolment when arm sizes are not recorded), and its shade follows the trial's evidence score (phase, size, endpoint type, result, replication).
Trials are not pooled and hazard ratios are not adjusted: a ratio from a placebo-controlled adjuvant trial is not directly comparable with one against an active comparator in the metastatic setting. Use the setting and endpoint filters to keep like with like, and open the trial page for the population, comparator and follow-up behind each number.