Armitage–Doll multistage model of carcinogenesis¶

A historical multistage cancer-incidence model in which r sequential changes imply an age-dependent rate proportional to t^(r−1) under its assumptions.

Version: v1 · 2026-09-28 · History
Domain-specific #: 8018
Domain group: Applied Sciences & Engineering
Origin domain: Medicine & Healthcare
Subdomains: Cancer Epidemiology, Multistage Incidence Modeling → Medicine & Healthcare
Aliases: Armitage–Doll model

Core Idea¶

The Armitage–Doll model is a mid-twentieth-century statistical account of age-related cancer incidence. It posits a series of r rate-limiting changes and gives, in the notation preserved by the frozen source, a rate proportional to the product of their transition factors times age t to the power r−1, divided by (r−1)!. Its distinctive reasoning move links a population age curve to a sequential-stage interpretation.

Armitage and Doll found a stage count often around five to seven in the common cancers they studied, but the frozen account itself notes fewer-stage cases such as retinoblastoma. The equation is a model of incidence patterns under assumptions, not an observed list of mutations in every tumor or a clinical prediction for an individual. The retained overview does not supply a full derivation or estimation protocol, so the entry stays at the conceptual/statistical level.

How would you explain it like I'm…

Many Steps to Sickness

Doctors noticed that some kinds of cancer happen much more to older people than to younger people. Two scientists made a math idea to explain that pattern: maybe several changes have to happen in the body one after another, like needing several doors to open in a row before you reach a room. The longer you live, the more time there is for all the doors to open. It is a guess that fits the pattern, not a list of what happened inside any one person.

Cancer as a Chain of Steps

The Armitage–Doll model is a math idea about why many cancers get much more common as people get older. It pictures cancer as needing several changes to happen in a row, each one being a step. If a cancer needs several steps, the chance of having it rises very steeply with age, and the steeper the rise, the more steps it suggests. For the common cancers they studied, the numbers often pointed to around five to seven steps, though some cancers seemed to need fewer. It describes patterns across whole populations, not what will happen to one particular person.

Multistage Model of Cancer Incidence

The Armitage–Doll multistage model is a mid-twentieth-century statistical model of how cancer incidence changes with age. It assumes that a cell must pass through a series of r rate-limiting changes, in order, before cancer appears. Under that assumption, the incidence rate rises in proportion to age raised to the power r−1, so a steep age curve suggests more stages. Fitting it to the common cancers they studied often gave around five to seven stages, while some cancers, such as retinoblastoma, fit fewer. The model is an interpretation of population incidence patterns under assumptions, not a direct observation of specific mutations and not a tool for predicting one patient's risk.

The Armitage–Doll model is a mid-twentieth-century statistical account of age-specific cancer incidence. It posits that malignancy requires a series of r rate-limiting changes, and in its standard form gives incidence at age t proportional to the product of the stage transition rates multiplied by t^(r-1) and divided by (r-1)!. Its characteristic reasoning move runs from the observed population age-incidence curve to a sequential-stage interpretation: the power of age fitted to the data suggests the number of stages. Armitage and Doll inferred stage counts often around five to seven for the common cancers they examined, while cases such as retinoblastoma fit fewer stages. The model is conditional on its assumptions and describes population incidence patterns; it does not claim to observe a particular sequence of mutations in each tumor, nor to predict an individual's risk. The account here stays at this conceptual and statistical level without a full derivation or estimation protocol.

Structural Signature¶

Sig role-phrases:

Population age t — Provides elapsed age on which incidence or mortality rate is indexed. It is constitutive. Counterfactual: Without age dependence the power-law signature of this model disappears.
Sequential stage count r — Sets the number of posited rate-limiting changes and the t^(r−1) exponent. It is constitutive. Counterfactual: An arbitrary fitted exponent without staged interpretation is only a generic age curve.
Stage-transition factors — Represent the p₁ through pᵣ terms whose product enters the rate coefficient. It is constitutive. Counterfactual: One unexplained global slope does not express the model's staged construction.
Incidence-rate readout — Connects the staged process to a population-level age pattern. It is constitutive. Counterfactual: A single patient's realized mutation history is not the rate curve itself.
Assumption and variation boundary — Limits interpretation to model conditions and cancer types, including fewer-stage counterexamples. It is boundary condition. Counterfactual: Treating five to seven stages as a universal individual count contradicts the frozen caveat.

What It Is Not¶

It is not a rule that every cancer always needs five to seven changes.
It is not a direct record of the ordered molecular mutations inside one person's tumor.
It is not a generic power-law fit stripped of its sequential-stage interpretation.
It is not a clinical risk calculator or procedure for predicting an individual's disease course.
Closest near-miss. Retinoblastoma's fewer-stage account in the source shows that stage count depends on disease and inherited context rather than a fixed constant for all cancers.

Scope of Application¶

Cancer-incidence history. Explains why age-dependent rate curves suggested multistage carcinogenesis.
Model comparison. Separates staged causal interpretation from a merely descriptive power law.
Population inference. Keeps cohort-level rate claims distinct from one person's outcome.
Assumption audit. Checks disease-specific stage counts and source limits before generalization.

Clarity¶

State age t, proposed stage count r, transition factors pᵢ, and population rate. The exponent r−1 follows the model's staged assumptions, not an observed count of one tumor's mutations. The frozen source's five-to-seven estimate is neither universal nor applicable to the retinoblastoma counterexample without adjustment. No individual clinical prediction follows from this short account.

Manages Complexity¶

One symbolic age-rate relation compresses a hypothesized sequence of changes into a tractable population pattern. That simplification made staged carcinogenesis thinkable from epidemiology, but it also hides heterogeneity in stage types, inherited context, and disease-specific dynamics.

Abstract Reasoning¶

Identify the population cancer rate and age variable under discussion.
State the assumed number r of sequential rate-limiting changes.
Read the coefficient as a product of model factors rather than measured individual mutation events.
Compare the implied t^(r−1) relation with the disease-specific observed pattern.
Keep mechanistic and personal conclusions outside what this population curve establishes.

Knowledge Transfer¶

The sequence-to-age-exponent idea can inform comparison of other multistage population models when transition assumptions and observed rate definitions are rebuilt. The specific r estimate and cancer interpretation do not transfer to every disease or person, and a superficially similar power curve in another domain is only an analogy without staged causal support.

Examples¶

Canonical¶

For a proposed r-stage process, the frozen expression gives rate at age t as N times p₁…pᵣ times t^(r−1) divided by (r−1)!. This illustrates how an extra hypothesized stage changes the age exponent; it does not infer the stage count of one patient's tumor.

Mapped back: Population age t → age in rate expression; Sequential stage count r → chosen hypothetical number of stages; Stage-transition factors → product p₁…pᵣ; Incidence-rate readout → population rate curve; Assumption and variation boundary → symbolic model, not personal diagnosis.

Applied / In Practice¶

The source's retinoblastoma discussion allows onset after fewer changes, sometimes in the presence of inherited predisposition. That case cautions against treating the five-to-seven estimate from many common cancers as a fixed biological law for every tumor.

Mapped back: Population age t → early-childhood occurrence; Sequential stage count r → fewer stages in this example; Stage-transition factors → not numerically estimated from packet; Incidence-rate readout → disease-specific pattern; Assumption and variation boundary → no universal stage count.

Structural Tensions¶

T1 — Compact Age Law versus Heterogeneous Disease Paths. One power relation organizes age-incidence data but cancer types and predispositions can alter plausible stage counts.

Diagnostic: Which disease and inherited context license the chosen r?

T2 — Population Pattern versus Individual Mechanism. An incidence curve can support staged causation without specifying the actual sequence of changes in one tumor.

Diagnostic: Is the claim about a population rate or a person's molecular history?

Structural–Framed Character¶

The skeleton is a representation mapping assumptions about unobserved sequential changes to an observable population age-rate pattern. The Armitage–Doll model connects r rate-limiting stages to an incidence rate proportional to t^(r−1) under its assumptions. Its approved parent is Representation, not an asserted strict statistical-model subtype.

Evaluative weight: Fit to age-incidence data and plausibility of assumptions matter; the exponent alone is not proof of the number of causal changes in a person.

Human-practice-bound: Epidemiologic definitions of population, incidence, age, and case influence how the relation is applied.

Institutional origin: This is a named mid-twentieth-century cancer-epidemiology model, not a timeless law of all sequential processes.

Vocabulary travels: “Multistage” and “power law” are used elsewhere but do not transfer the cancer-stage interpretation automatically.

Import versus recognize: The mapping strategy can guide other staged models only when their transition and observation assumptions are rebuilt.

Its character: A cancer-specific formal representation whose explanatory claim is conditional on its model assumptions.

Structural Core vs. Domain Accent¶

Skeletal core. A model translates a posited sequence of latent, rate-limiting transitions into a predicted age-dependent population rate.

Domain-bound accent. Armitage–Doll interprets those stages as carcinogenic changes and, under its assumptions, yields incidence proportional to age t raised to r−1. The observed curve concerns populations rather than a personalized mutation tally.

Why not prime. A similar power curve can arise without the same cancer process, and the stage interpretation does not apply universally. The named epidemiologic assumptions are constitutive, not optional decoration.

This entry is a kind of Representation.

Strict parent — Representation. The model maps a hypothesized multistage carcinogenic process into an age-rate equation for reasoning about incidence, preserving age dependence while suppressing detailed molecular histories. Statistical Model is related, but its live signature requires a declared family of probability laws on a sample space that the frozen age-rate overview does not specify.
Related — multistage carcinogenesis, retinoblastoma, and statistical model. They are the biological process, counterexample to a universal stage count, and formal modeling context rather than synonyms for this exact equation.

Relationships to Other Abstractions¶

Current abstraction Armitage–Doll multistage model of carcinogenesis Domain-specific

Parents (1) — more general patterns this builds on

Armitage–Doll multistage model of carcinogenesis is a kind of Representation Prime

Armitage–Doll represents a proposed sequence of carcinogenic changes with an age-dependent population incidence-rate equation.

Hierarchy path (1) — routes to 1 parentless root

Armitage–Doll multistage model of carcinogenesis → Representation → Abstraction

Neighborhood in Abstraction Space¶

Armitage–Doll multistage model of carcinogenesis sits in a sparse region of the domain-specific corpus (67^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Domain-Specific Indicators & Measurement Methods (26 abstractions)

Nearest neighbors

Orbital tuning — 0.85
Funnel Chart — 0.85
Lincoln Index — 0.85
Bartlett's theorem — 0.85
Disease Burden — 0.83

Computed from structural-signature embeddings · 2026-10-08

Not to Be Confused With¶

Any age-incidence power law. Tell: The Armitage–Doll account ties its exponent to posited sequential stages.
Universal mutation count. Tell: The source gives cancer-specific variation including fewer-stage retinoblastoma.
Individual risk forecast. Tell: The equation is a population rate model, not a personalized prediction.
Observed molecular sequence. Tell: The age curve alone does not identify each change in one tumor.

References¶

Frozen Wikipedia discovery revision: https://en.wikipedia.org/wiki/Armitage%E2%80%93Doll_multistage_model_of_carcinogenesis (revision 1155834898).
Preserved source candidate: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2007940/pdf/brjcancer00386-0010.pdf
Preserved source candidate: http://www.nature.com/bjc/journal/v91/n12/pdf/6602297a.pdf

The frozen Wikipedia revision is discovery provenance. The retained source set was reviewed for identity, formal or operational relation, and scope. The encyclopedia's structural synthesis is bounded to those claims; a thin authority surface is recorded as a nonblocking source-strengthening repair rather than concealed.