REF-1186 · Foulkes et al., 2024 · British Journal of Sports Medicine · Paper · PubMed

Does elite-level running shorten life?

Not in the first 200 sub-4 milers: they lived 4.7 years longer than men of their age, era and country. No control for wealth, smoking, health or genes, so "not shorter", not "running added years".

Blimited evidenceMeasured on elite to world-class athletes

Foulkes et al. took the first 200 men in the Sub-4 Alphabetic Register (Roger Bannister in 1954 to the 200th runner in 1974; born 1928-1955; 28 countries) and found dates of birth and death by online searches of federation, Olympic and Wikipedia pages, checked by two further independent searches.

Runners with no evidence of death were counted alive at a censor date of 31 December 2023. Each runner's observed life years from the day of his first sub-4 mile were compared with the life expectancy of men of the same age, calendar year and nationality from United Nations life tables. Sixty (30%) had died, at a mean 73.6 ± 13.7 years; the 140 survivors averaged 77.6 years.

Overall the runners lived 4.74 years (95% CI 4.66 to 4.82) beyond expectation; by decade of the first sub-4 run the gain was 9.2 years (n = 22), 5.5 (n = 88) and 2.9 (n = 90), with shorter follow-up in each later decade. Olympians did no better than non-Olympians (4.13 vs 5.68 years).

Of seven men who died before 55 with a known cause, six died from trauma or suicide and one from pancreatic cancer. The authors read this as a challenge to the "U-shaped" idea that extreme exercise shortens life.

Our reading: it rules out a large life-shortening effect in this group; it cannot show that running added the years, because men who could run a sub-4 mile at 23 differ from the national average in health, genes and probably wealth, and none of that was measured.

Protocol

Design: retrospective cohort; primary outcome the mean difference between observed and population-expected life years.
Who: the first 200 men to break 4 minutes for the mile, from the Sub-4 Alphabetic Register (1759 runners listed as of 6 June 2022); age at the first sub-4 run 23.4 ± 2.8 years; Europe n = 88, North America 78, Oceania 22, Africa 12. No women have run a sub-4 mile, so the cohort is all male.
Exposure: having run a sub-4-minute mile; no information on lifelong training, health behaviours or cause of death for most.
Vital status: online searches (Olympic, international and national federation websites, Wikipedia) on 18 January 2024, repeated by two further investigators; no evidence of death = alive; censor 31 December 2023; for two runners only the year of birth was found and 31 December of that year was used.
Expected survival: United Nations life tables linked by country at the time of the sub-4 run, calendar year, sex and single year of age to 100 or to December 2023 (tables to 2021; 2022-2023 assumed equal to 2021); follow-up from the date of the first sub-4 run; Kaplan-Meier for the runners, Ederer II for expected survival; standard error by leave-one-out jackknife.
Subgroups: decade of the first sub-4 run (1950s, 1960s, 1970s); Olympians (n = 121) vs non-Olympians (n = 79).
Funding: "No funding sources were used in the direct preparation of this article"; personal fellowship and research-chair support to two authors; competing interests none declared. Foulkes, La Gerche and Haykowsky are also authors of REF-1185.

Key Results

GroupnYears beyond population-expected life (95% CI)Mean follow-up
All runners2004.74 (4.66 to 4.82); abstract "4.7 (4.7 to 4.8)"—
First sub-4 in the 1950s229.2 (8.3 to 10.1)67.0 years
1960s885.5 (5.3 to 5.7)58.2 years
1970s902.9 (2.7 to 3.1)51.3 years
Olympians1214.13 (3.97 to 4.3)—
Non-Olympians795.68 (5.6 to 5.76)—
Cohort detailValue
Dead / alive at 31 December 202360 (30%) / 140
Age at death73.6 ± 13.7 years (range 24.3-91.9)
Age of survivors77.6 ± 5.5 years (range 68.3-93.8)
Deaths before 55 with a confirmed cause7: six traumatic or suicide, one pancreatic cancer
Family clustering, as printed "among the first 200"20 sets of brothers (six sets of twins) and father-son pairs; the paper's own example (the Ingebrigtsen brothers) is from outside the first 200 and the count is cited to the full register, so it may describe all 1759 runners (our reading)
No life-shortening signal: on average the milers lived longer than men of their age, era and country; the authors call this a challenge to the upper end of the U-shaped hypothesis.
The gain shrinks by decade: 9.2, 5.5 and 2.9 years. The authors offer two reasons: improving population life expectancy, and 10-20 fewer years in which later cohorts could accrue the difference.
Olympians did not do better: a trend the other way (4.13 vs 5.68 years), which the authors take as evidence that socioeconomic advantages expected in Olympians do not mediate the effect.

Mechanisms

Measured: none. No physiology, training, behaviour or cause of death was measured; the paper compares survival only.
Authors' interpretation: the gain likely reflects endurance adaptations (a high VO2max, a strong predictor of cardiovascular and all-cause mortality), "favourable genetics" and "additional healthy lifestyle behaviours"; the 20 sets of brothers and six sets of twins among the 200 point to a genetic component of the performance itself.
Authors' interpretation: "It is possible that the extreme exercise required to run a sub-4 min mile is deleterious to the lifespan, but the effect is not large enough to overcome the positive effects of other factors seen in athletes" — a possibility this design cannot test.
Inference: the comparator is the whole national population, including men already ill in their early twenties; any healthy group selected at 23 (a "healthy entrant" effect) would look better on this comparison without any training effect.
Inference: the Olympian comparison is weak evidence about socioeconomic status: being an Olympian is a crude proxy for wealth, and nationality (Olympic selection differs by country) is mixed into it.

Practical Application

Does elite-level running shorten life? Not in the first 200 sub-4 milers: they lived 4.7 years longer than men of their age, era and country. No control for wealth, smoking, health or genes, so "not shorter", not "running added years".
For a masters athlete worried that years of hard training "wear out" the body: this cohort shows no average life-shortening after elite middle-distance running. It says nothing about lifelong training dose, and an individual cost such as the atrial fibrillation in REF-1185 can coexist with a long life.
Do not quote the 4.7 years as an effect of training. The design compares men selected for exceptional health and physiology at 23 with national averages; the authors themselves list genetics, diet, smoking and other factors as uncontrolled.
Women: no data; no woman has run a sub-4 mile and the authors had no comparable female database.

Limitations

No individual covariates: socioeconomic status, education, smoking, diet, body mass, disease and genetics were not measured; the comparison is with national life tables, not a comparable control group. The authors acknowledge this.
Vital status from web searches, with "no evidence of the athlete's death" counted as alive; deaths of less famous runners are more likely to be missed online, which would bias towards longer lives (inference).
Nationality at the time of the sub-4 run sets the life table; men who emigrated, or who came from countries with low national life expectancy (12 from Africa), are compared with tables that may not describe their circumstances (inference).
Follow-up is shorter for later decades, so the decade gradient mixes era with time at risk.
The confidence intervals are very narrow for 200 men with 60 deaths (overall 4.66 to 4.82 years); the leave-one-out jackknife is described in one sentence and individual-level spread is not shown (our reading).
Cause of death unknown for most; the authors excluded it from analysis.
Exposure is a single performance in early adulthood; no data on training before or after, so no dose-response.
Men only.
Funding: no direct funding; personal fellowship and research-chair support to two authors; no competing interests declared.
How this entry was checked and changed
Why this level

retrospective cohort, read from the full text (published PDF via the KU Leuven repository). The rubric's B+ row for a large observational cohort starts at 200; this one sits exactly at 200 with 60 deaths, no individual covariates at all, an external general-population comparator rather than a comparable control group, and vital status ascertained from online searches in which "no evidence of death" counted as alive, so it is graded one step down

Full text (checked 2026-10-22)
"Of the first 200 sub-4 min mile runners (100% male), 60 were dead (30%) and 140 were still alive" (full text).
"sub-­4 min mile runners showed an increase of 4.74 years (95% CI 4.66 to 4.82; n=200) beyond their predicted life expectancy" (full text).
1950s: 9.2 years (95% CI 8.3 to 10.1; n=22) during an average 67.0 years of follow-up; 1960s and 1970s: 5.5 years (95% CI 5.3 to 5.7; n=88) and 2.9 years (95% CI 2.7 to 3.1; n=90), follow-up 58.2 and 51.3 years (full text).
Non-Olympians (n=79): 5.68 years (95% CI 5.6 to 5.76); Olympians (n=121): 4.13 years (95% CI 3.97 to 4.3) (full text).
Average age at death 73.6±13.7 years (range 24.3–91.9); surviving runners 77.6±5.5 years (range 68.3–93.8); of the seven runners who died before age 55 with a confirmed cause, six were due to traumatic deaths or suicide and one was due to pancreatic cancer (full text).
Mean±SD age at completion 23.4±2.8 years; times between 3:52.86 and 3:59.9 min; Europe (n=88), North America (n=78), Oceania (n=22) and Africa (n=12); year of birth 1928 to 1955 (full text).
"In those runners who were still alive (no evidence of the athlete's death in a comprehensive search), their current age was calculated with a censor date of 31 December 2023" (full text).
"We did not account for specific socioeconomic status other than gender and nationality" (full text).
"It is possible that the extreme exercise required to run a sub-4 min mile is deleterious to the lifespan, but the effect is not large enough to overcome the positive effects of other factors" (full text).
"20 sets of brothers, including six sets of twins, and father and son combinations were among the first 200" (full text).