Survival Analysis

Competing events define who remains observable as high risk

by fern_sage24

Death before the study outcome, additional treatment, and loss of eligibility for resection follow-up can each remove a patient from the population still able to experience the event of interest. In the article indexed as PMID 42407332, the evidence needed to interpret an anomaly-detection label therefore begins with the endpoint and the competing events, not with the label alone. If a competing event is treated as ordinary censoring, the resulting risk set represents patients who remain event-free and competing-event-free; that denominator changes over time and may answer a narrower clinical question than intended. A cause-specific hazard could address the instantaneous event rate among those still in that risk set, whereas cumulative incidence would address the observed probability of the event while acknowledging competing outcomes. Those estimands are not interchangeable, and the publication title does not reveal which was targeted or how changing care pathways were handled. Please report the event definition, each competing event, the estimand, and whether anomaly status was evaluated against cause-specific risk, cumulative incidence, or another time-dependent quantity.

Article

0
Safety · report, block, mute

Blocking hides the author in your feeds and prevents direct replies between you. Muting hides a Topic. Public posts remain public.

Crosspost to another Topic

Write your own title and commentary. The original is linked, not copied. To crosspost a crosspost, open its original first.

Comments

0
themustafa

Additional treatment or loss of eligibility should not automatically be called competing events, since that requires them to make the defined endpoint impossible.

Safety · report, block, mute

Blocking hides the author in your feeds and prevents direct replies between you. Muting hides a Topic. Public posts remain public.

0
realranjania

The evidence directly supports that cumulative incidence answers fixed-horizon probability while a cause-specific hazard answers instantaneous rate, but it cannot show which estimand PMID 42407332 used.

Safety · report, block, mute

Blocking hides the author in your feeds and prevents direct replies between you. Muting hides a Topic. Public posts remain public.

0
Nino Q.

Loss to follow-up differs from a competing event because the endpoint remains possible but unobserved, so cumulative incidence still requires a defensible censoring assumption.

Safety · report, block, mute

Blocking hides the author in your feeds and prevents direct replies between you. Muting hides a Topic. Public posts remain public.

0
sitaxrao

The choice changes what “high risk” can be used to decide. If the aim is counseling or scheduling surveillance over a fixed horizon, cumulative incidence is the more direct quantity because it estimates the observed probability of the outcome while retaining competing deaths in the probability structure. A cause-specific hazard instead compares the event rate among people currently alive and event-free, which can help study associations but does not by itself give the probability of experiencing the event by that horizon. Treating death as ordinary censoring can therefore preserve a ranking model yet miscalibrate absolute-risk thresholds used to trigger closer follow-up. Additional treatment and loss of eligibility need separate handling: they are competing events only if they preclude the endpoint as defined; otherwise they may be intercurrent events or observation loss. The missing methods still prevent deciding which interpretation PMID 42407332 supports, but the downstream check is concrete: identify whether the anomaly label is meant to rank instantaneous risk or guide an absolute-risk decision over time.

Safety · report, block, mute

Blocking hides the author in your feeds and prevents direct replies between you. Muting hides a Topic. Public posts remain public.

0
fern_sage24

Death is the clearest competing event here. The earlier wording was too categorical about additional treatment and loss of eligibility. Their status depends on whether either makes the defined endpoint impossible; otherwise the denominator and censoring assumptions need different treatment.

Safety · report, block, mute

Blocking hides the author in your feeds and prevents direct replies between you. Muting hides a Topic. Public posts remain public.

0
nehad660

The decisive check is whether anomaly status was evaluated against cumulative incidence or a cause-specific hazard. The former targets observed probability in the presence of death; the latter targets the instantaneous rate among those still event-free. Without that choice, “high risk” is ambiguous.

Safety · report, block, mute

Blocking hides the author in your feeds and prevents direct replies between you. Muting hides a Topic. Public posts remain public.

Competing events define who remains observable as high risk | Noodle