Fitness at 18 tracked with sepsis risk decades later in a Swedish cohort

Sweden's conscription records are one of epidemiology's more unusual assets: height, weight and cardiorespiratory fitness were measured at conscription to military service, at a mean age of 18.3 years, and those measurements can be linked to national health registers for decades afterwards [s1]. A study published this month in the Journal of Internal Medicine used that resource to ask a question the cohort has not previously been used for — whether adolescent fitness and body mass predict severe bacterial infection in middle age [s1].

The cohort

The register-based cohort followed almost one million Swedish conscripts, mean age 18.3 years, over three decades [s1]. Exposures were measured body mass index and cardiorespiratory fitness at conscription; outcomes were morbidity and mortality attributed to bacterial pneumonia, sepsis and infective endocarditis [s1].

The design's strength is that the exposures were measured rather than self-reported, at a uniform age, before any of the outcomes occurred. Its constraint is that it is a cohort of young men in one country, and nothing in it speaks to women or to other populations.

The results

High BMI and low cardiorespiratory fitness in late adolescence were both strongly associated with increased risk of all three infection outcomes in adulthood [s1].

The largest single association was for sepsis among those with obesity, with a hazard ratio of 3.1 (95% CI 2.7 to 3.5) [s1]. The risk did not begin at the obesity threshold: elevated hazards were observed already at high-normal BMI levels of 22.5 to 25 kg/m², compared with a reference of 18.5 to 19.9 [s1]. Risk of dying from bacterial infection increased gradually with higher BMI [s1].

Fitness moved in the other direction. High cardiorespiratory fitness, compared with low, was associated with a lower risk of contracting bacterial infections (HR 0.78, 95% CI 0.76 to 0.80) and of dying from them (HR 0.58, 95% CI 0.51 to 0.65) [s1].

The authors conclude that addressing these preventable risk factors in young people may be an effective measure to improve long-term health [s1].

What a hazard ratio from a cohort can and cannot say

This is an observational study, and the exposures are correlated with a great deal else. Adolescent BMI and fitness track socioeconomic position, chronic illness, smoking, and later-life BMI and activity, and a register study cannot fully separate a measurement taken at 18 from the four decades of behaviour that followed it. The paper's own framing — associated with, rather than causes — is the appropriate one.

Reverse causation is a live concern at the low end of the fitness distribution in particular: an 18-year-old with poor fitness may have had an undiagnosed condition that itself raised later infection risk. The finding that risk rises gradually across the BMI range, including within what is conventionally called normal weight, is harder to explain that way, but not immune to it.

What the study does add is a plausible mechanism-adjacent outcome that has been less studied than cardiovascular disease and cancer. Obesity and low fitness are established risk factors for those; severe bacterial infection has received less attention, and the effect sizes here are not small.

The other end of the life course

A second cohort published two weeks earlier looks at the same broad question — how activity relates to mortality — at the opposite end of adulthood, and is useful mainly as a reminder that the exposure is not fixed at 18 [s2].

The study drew on 58,168 multiethnic US postmenopausal women in the Women's Health Initiative Observational Study, with self-reported sedentary behaviour and recreational physical activity recorded at baseline (1993–1998, ages 50 to 79) and again after six years [s2]. Participants were grouped by transition — consistently non-sedentary, sedentary to non-sedentary, non-sedentary to sedentary, or consistently sedentary — and equivalently for activity level [s2]. Over a median 15.0 years of follow-up from the six-year assessment to March 2019, 17,354 deaths occurred, ranging from 1,336 respiratory to 5,111 cardiovascular deaths [s2].

Compared with the consistently non-sedentary group, both unfavourable sedentary transitions — becoming sedentary, or remaining so — carried higher risk of all-cause, cardiovascular, cancer and respiratory mortality [s2]. Compared with consistently low activity, both favourable activity transitions — taking up or maintaining high activity — carried lower mortality risk [s2].

The interaction is the interesting part. Significant interactions were observed between sedentary and activity transitions for all-cause and cardiovascular mortality (P-interaction < 0.01): unfavourable sedentary transitions were associated with elevated risk only among women who were also not achieving regular physical activity [s2].

Reading them together

The two cohorts are not comparable in design, population or exposure measurement — one uses objectively measured fitness in adolescent men, the other self-reported behaviour in older women — and no pooled inference is available from putting them side by side. What they share is direction and the shape of the finding: risk graded across the exposure rather than switching on at a threshold, and benefit associated with maintaining or improving rather than with a single measurement.

Neither is a trial. Neither can tell an individual what their own risk is or what to do about it. The Swedish study's contribution is a specific, previously under-examined outcome measured in a cohort large enough to see it; the Women's Health Initiative analysis's contribution is that a six-year change in behaviour in later life still showed up in fifteen years of subsequent mortality data [s1][s2].

Sources

• High BMI and low cardiorespiratory fitness in adolescence are associated with increased risk of severe bacterial infections in adulthood — Journal of Internal Medicine, 2025-11-23
• Longitudinal transitions in sedentary behavior and physical activity in relation to all-cause and cause-specific mortality among postmenopausal women — GeroScience, 2025-11-11