Clear numbers.
Visible limitations.

Two different kinds of question

What happened in a treatment study?

Randomized trial charts compare groups within one study at its reported follow-up time. Overall survival estimates the percentage alive. A median is a time point, not average lifespan. Different studies cannot be combined into a treatment ranking.

What does a larger population show?

SEER relative survival compares observed survival with expected survival in a comparable general population. It is not the same measure as overall survival in a clinical trial. The lung source includes all lung and bronchus cancers, not only one biomarker or treatment.

Filters reflect the source—not a prediction model

Age, sex, stage, and race / ethnicity select an exact, published SEER source population. No estimate is calculated for a person. Sex and race / ethnicity labels are reproduced from the source and do not imply a causal explanation or describe all identities. Trial charts do not change with these population filters.

Missing estimates stay missing. Unverified confidence intervals are not drawn. No broader population is silently substituted. Stage terminology is the source’s combined summary stage, not an inferred TNM stage.

The explorer uses published age bands, not exact-age interpolation. Grade and time since diagnosis are not available in this dataset. The 1-, 3-, 5- and 10-year controls select follow-up from diagnosis; they do not mean survival conditional on having already lived that long.

Lines connect annual estimates; shading shows the selected population’s reported 95% confidence intervals, not a range of predicted individual outcomes.

Where these numbers come from

This preview preserves 66 SEER cancer source groups and all 63,840 published population combinations in the captured current-follow-up release. It also includes six curated trial reports (four lung, two melanoma) and3,504 distinct ClinicalTrials.gov records, full posted results, 149 NCI treatment topics, drug-label snapshots, publications, cancer terminology, and a source-derived treatment-setting catalogue. These are distinct collections, not a complete treatment or recruiting-trial directory. Browse the full source library.

The additional 2,044 registry records were found through condition searches restricted to posted results; they are not the same as the 1,460 source-cited records. Treatment-setting associations preserve the original compiler’s scope and uncertainty; none has passed independent clinical review. Outcome and safety groups remain separate, with their own denominators, time frames and units.

Archived follow-up adds 67 source groups and 65,040 historical curves, kept separate from the current release. Cancer Stat Facts sources are also available. A separately pinned supplemental edition now includes all four recovered SEER collections: five-year survival, conditional survival, recent trends and long-term trends. These preserve 354,592 series, 1,021,576 points and 42,938 source-fitted trend segments. Explore the separate survival measures.

Supplemental edition: survival-2026-09-11, source commit d2cb844d2ffa77d2615ee9fa88880178fd122187. This package transports reconciled database rows; it is not an independent re-derivation from raw responses. Diagnosis-period codes lack published labels in the package, so no calendar ranges are inferred. Conditional estimates describe the next five years among people already alive at the stated conditioning point. Trends compare diagnosis years. Observed and modeled values remain separate, and missing estimates are never replaced with zero or a broader population. Clinical and statistical review remains pending.

Study references include original publications, source locations, data cutoffs, safety denominators, limitations, and confidence intervals where verified. SEER charts include source response links, retrieval dates, release identifiers, and hashes. Percentages are displayed rounded to one decimal; exact source values remain in the table.

Active explorer release: explorer-2026-09-11
Evidence library: library-2026-09-11
Preserved initial pilot references: pilot-2026-09-10
Trial transcription check: 2026-09-09
SEER capture: September 10, 2026
Source revision: 86e8a04c2ee806d31dbc0d64376aa384aedd7bf5

Registry search is not treatment selection

Registry search matches text in study titles, condition tags, interventions and NCT identifiers. It does not determine clinical eligibility, and the population filters do not personalize it. Status is preserved as captured in September 2026 and may now differ. Check the original registry record for current status, eligibility and locations. A missing result means no text match in this snapshot, not that no study exists.

Research preview—not a clinically reviewed resource yet

This release has not passed independent clinical/statistical review or the final rights and publication review. Pages intentionally request no search indexing. No clinician reviewer, review date, or approval has been invented. Source information can change after capture.

Before public launch, each page must pass source reconciliation, expert review, accessibility checks, and release approval. If you need a care decision, bring the original evidence and your own clinical context to your care team.

Links that preserve the evidence

“Open this fixed reference” under Sources & data preserves the source population and data release, with the selected follow-up year in the link fragment. It can be linked in a conversation or document without social-sharing widgets. Study pages also have permanent source references. New data releases will not overwrite these values. The shorter study URLs will point to the active reviewed release after launch.

Privacy by default

No account, patient record, or exact age is requested. Source filter selections stay in browser memory unless you deliberately copy a reference link. A normal reload restores the page’s starting population; opening a fixed reference restores that public source population. This section does not load marketing analytics or the support messenger. Public chart links identify source cohorts, not personal profiles; standard hosting access logs can still contain page paths, including a cohort link you choose to open.