Background There are several intriguing features of amyotrophic lateral sclerosis (ALS). Some people with genetic susceptibility never develop ALS; ALS develops in late life and progresses rapidly; the same mutation can predispose to ALS or other diseases; ALS starts in one region and spreads; degeneration is specific to a subgroup of neurons; and ALS shows complex inheritance. Cancer shares many of these characteristics, and these have been incorporated into the Armitage-Doll multistage model. We therefore used this model to investigate the hypothesis that ALS is a multistage process.

Methods Incidence data by age and sex were generated from five ALS population registers, in Ireland, the Netherlands, Italy, Scotland and England; age and sex adjusted incidences were calculated for each register.

Findings 6274 cases were identified from 73 million people. There was a linear relationship between log incidence and log age in all five registers: Ireland R2=0·99, Netherlands R2=0·99, Italy R2=0·95, Scotland R2=0·97, England R2=0·95, overall R2=0·99. All five registers gave very similar estimates of the linear slope ranging from 4·5 to 5·1, with overlapping confidence intervals. Combining all five registers gave an overall slope of 4·8 (95% CI 4·5, 5·0) with similar estimates for males and females. In persons with known genetic mutations which predispose to ALS, the risk of ALS was higher, but the log-incidence/log-age slope was lower, as predicted by the multistage modelo.

Interpretation A linear relationship between the log incidence and log age of onset of ALS is consistent with a multistage model of disease. The slope estimate is about 5, suggesting that ALS is a 6-step process; the slope is 3–4 in persons with known genetic causes. These findings have implications for the search for causes of ALS, including occupational causes and gene-environment interactions.