The US cancer death rate falls below 70 per 100,000

Where things stand

• Latest official numbers. NCHS's final 2024 age-adjusted cancer death rate was 139.4 per 100,000, down from 141.8 in 2023 (Mortality in the United States, 2024, Data Brief 548, published Jan 29, 2026: https://www.cdc.gov/nchs/products/databriefs/db548.htm).
• Most recent quarterly figure. I downloaded the VSRR quarterly provisional dataset (data.cdc.gov 489q-934x; rows last updated 2026-06-23; 2025 data provisional as of Apr 19, 2026). It gives 137.6 for the twelve months ending 2025 Q3. The rolling twelve-month series drifts down slowly: 141.5 (2023 Q1), 139.4 (2024 Q4), 139.0, 138.4, then 137.6.
• 2025 provisional report. VSRR No. 44 (July 2026) counted 622,832 cancer deaths in 2025, up from 619,876 in 2024 (https://www.cdc.gov/nchs/data/vsrr/vsrr044.pdf). So the 2025 rate will be roughly 134–137.
• What YES needs. A rate below 70 means a ~49% fall from today's level.
• No reclassification route. A Congressional Research Service report (R49000, June 2026) says US work on moving to the new disease classification (ICD-11) is suspended because the US withdrew from the WHO. So a coding change won't move cancer deaths around.

Base rate and reference class

The reference class is NCHS's own long-run age-adjusted cancer mortality series: data.cdc.gov 6rkc-nb2q through 2018, plus NCHS data briefs for 2019–2024. I computed the following from it:

• The rate fell from 215.2 (1991) to 139.4 (2024).
• The largest one-year drop was 2.9% (2022, partly a rebound after the pandemic).
• The largest seven-year drop was 12.2% (2012–2019).

Publication lag matters. NCHS posts quarterly estimates about 8–9 months after the quarter ends, and final annual data about 13 months after the year ends. So the latest period that could be published by Dec 31, 2031 is roughly the twelve months ending Q1–Q2 2031.

• Getting from 137.6 to below 70 by then takes about 11% a year for ~5.75 years. That is roughly five times the fastest single-year drop ever recorded, kept up for six years.
• At the 1.5–2% a year trend, the rate would be about 122–126 by mid-2031.
• At 2% a year, reaching 70 takes about 33 years.

Timing within the window: only periods ending after Sep 16, 2026 count. The first one, the twelve months ending Sep 30, 2026, will probably not be published until mid-2027. So the chance by the 2026-12-31 and 2027-03-31 horizons is essentially nil.

Paths to YES

1. A medical revolution, perhaps driven by AI. Cures would have to be found, tested, approved, made at scale and given to most patients who would otherwise die, by about 2029–2030. Real 2026 advances are big for patients but small at the population level. Examples are the approval of daraxonrasib for pancreatic cancer, which nearly doubled survival, and Merck/Moderna's personalized melanoma vaccine. Even if some forecasters' fast-AI scenarios came true, clinical trials and adoption take years. My estimate for this path is at most about 0.1–0.15%.
2. A data error. NCHS would have to publish a mislabeled or miscomputed total-US age-adjusted cancer rate below 70 that the resolver accepts. I know of no such error in this series. I put this at about 0.05%.
3. A huge change in the population figures used as denominators. The population would have to be about double the current estimate. That is not plausible. The 2025 change in how the Census Bureau builds its population estimates moved rates by a few percent at most.

Strongest case for YES

The strongest case is an extreme AI takeoff that compresses drug development and rollout into a year or two before 2030, combined with my being wrong about how fast new treatments spread. Even then, the cure would have to reach most of the roughly 600,000 people who die of cancer each year, across many different cancer types, inside the window. I'd still bet heavily against it, so the case doesn't move me much.

Forecast

The chance of YES in any quarter starts near zero, around 0.002% to 0.005% per quarter through mid-2027, because no qualifying period can be published before then. It rises gradually to about 0.02% per quarter in 2031. That leaves a cumulative probability of about 0.22% by the end of 2031.

Other sources: the NCHS mortality dashboard (https://www.cdc.gov/nchs/nvss/vsrr/mortality-dashboard.htm); Targeted Oncology on the 2026 Annual Report to the Nation (Sept 2026); BBC and BioPharma Dive on the daraxonrasib approval (Sept 2026).

Question. By the last day of each quarter, has CDC's National Center for Health Statistics reported an age-adjusted US cancer death rate below 70 per 100,000 for any calendar year or any twelve months ending in a quarter?

Definitions.

• Cancer death rate: the national rate of deaths with malignant neoplasms as the underlying cause (ICD-10 C00–C97, or an NCHS-designated equivalent under a successor classification), for all cancer sites combined.
• US: the resident population of the 50 states and District of Columbia, all sexes, ages, races, and Hispanic origins combined; territories are excluded.
• Age-adjusted: directly standardized to the 2000 US standard population. A rate using a different standard counts only if NCHS also publishes the qualifying rate on the 2000 standard.
• Qualifying period: a complete January–December calendar year, or twelve consecutive months ending March 31, June 30, September 30, or December 31. The Observation period convention applies.
• Below 70: a published point estimate strictly less than 70 deaths per 100,000 standard population. Use the most precise numeric rate NCHS publishes in the release or its accompanying data; a value of 70.0 alone does not qualify, and a confidence interval extending below 70 does not suffice.

Resolves YES if NCHS publishes a qualifying rate for at least one qualifying period within the Resolution window and on or before the quarter-end. Final annual data, official provisional annual estimates, and official provisional twelve-month-ending-quarter estimates all count. A revision that first puts an eligible period below 70 counts on its publication date. Later revisions do not reverse a YES. No attribution to AI or any particular medical intervention is required.

Resolves NO if no qualifying publication exists by the quarter-end. Exclude crude rates, age-specific or demographic-subgroup rates, individual cancer-site rates, state or regional rates, incidence rates, survival statistics, projections, annualized three-month rates, year-to-date figures, and rates independently calculated from death counts or microdata. Publications that only repeat or revise observations ending on or before the opening date do not count.

Resolving source. NCHS publications and official NCHS data releases only: National Vital Statistics System annual reports (Deaths: Final Data and Mortality in the United States, including their tables); Vital Statistics Rapid Release annual provisional reports; and the VSRR Mortality Dashboard and associated downloadable data, selecting Cancer, Age-adjusted, Total US population, and 12 months ending with quarter. Official NCHS annual mortality rates disseminated through CDC WONDER also count. Successor NCHS products count if they publish the same defined statistic. The current annual-series example is Mortality in the United States, 2024, NCHS Data Brief 548, Figure/Table 4. Apply Resolution date (A3) to the publication or dated data release, not the observation period or query date; the named-source restriction supersedes the broader Public confirmation (A2) sources.