Data maturity (immature vs mature survival data)
Survival results are 'immature' when too few patients have died (or progressed) for the comparison to be reliable, so medians are 'not reached' and confidence intervals are wide. Mature data come with time and events; early positive looks can fade or strengthen.
Trials pre-specify the number of events needed for each analysis (the information fraction at an interim); with fewer events the hazard ratio is unstable and the median cannot be estimated. Regulators accept immature OS if PFS is positive and there is no detriment, but require the final analysis as a post-marketing commitment; ADAURA and KEYNOTE-522 were approved on DFS/EFS with OS immature, and both later matured positively, whereas other trials' early OS trends vanished. 'Not reached' medians in the experimental arm are a favourable sign but not a result. Censoring marks patients still event-free at the cut-off, and heavy early censoring (dropouts, loss to follow-up) can bias curves.
Pages like this
not linked directly; found by shared links- TermFutility analysis (stopped for futility)
Shares Interim analysis, readout and data cut-off, Primary, secondary and co-primary endpoints.
- TermStatistical significance (P values, alpha, multiplicity)
Shares Interim analysis, readout and data cut-off, Primary, secondary and co-primary endpoints, Hazard ratio (HR).
- IdeaQuality-adjusted survival reported in every trial publication and label
Shares Hazard ratio (HR), Overall survival (OS).
- IdeaPre-specified crossover-adjusted survival in every trial that allows crossover
Shares Hazard ratio (HR), Overall survival (OS).
- PersonMichael LeBlanc
Shares Hazard ratio (HR), Overall survival (OS).
- TermMedian survival
Shares Hazard ratio (HR), Overall survival (OS).
- IdeaPower trials to detect a benefit patients would value, not the smallest detectable one
Shares Hazard ratio (HR), Overall survival (OS).
- TermReading a hazard ratio
Shares Hazard ratio (HR), Overall survival (OS).