Pull up any chart of what people die from. Circulatory disease takes a large block. Cancer takes another. Infections, injuries, and respiratory disease each take a slice. Musculoskeletal disease is a sliver you can barely see, often a line so thin it needs its own color key.
Now look at what health advice spends its time on: muscle, strength, mobility, loading the skeleton.
That looks like a mismatch, and plenty of people treat it as one. If almost nobody dies of musculoskeletal disease, why does everyone talk about muscle?
The mismatch comes from how the chart is built, and it matters because the same misreading shapes what people decide to train.
What the chart’s own caption admits
I was recently reading a large Swedish registry study of young adults with type 1 diabetes and matched controls (Rawshani et al. 2018). One figure breaks deaths into color-coded causes by age group, the way most of these charts do.
In the caption, the authors note that only the primary cause of death was assessed, so cardiovascular disease may be a significant contributor to deaths filed under “Endocrine causes.”
So one of the biggest blocks in their chart is partly heart disease that got filed somewhere else. That’s a limit of the data, and the authors were upfront about it.
A death certificate requires a single underlying cause. The person completing it must pick one, and the picks must sum to 100%. In math terms, that makes the chart a partition: a whole split into pieces that don’t overlap. That’s great for bookkeeping and poor for explaining why someone died.
Real causes overlap
Epidemiologists think about cause differently. An outcome happens when a complete set of contributing causes is present at once. Any one piece is usually not enough on its own, yet remove it and that particular route to the outcome closes (Rothman 1976; VanderWeele 2017). The philosopher J. L. Mackie worked out the same logic in 1965.
Because each death can be credited to every piece of the set that produced it, the shares attributed to different causes routinely add up to more than 100% (Rowe et al. 2004). A death certificate can’t do that. Its categories have to add up to exactly 100%.
So when you read “musculoskeletal disease: 1% of deaths,” you have not learned that muscle does 1% of the causal work. You have learned that in 1% of cases, someone judged a musculoskeletal condition to be the single best label for the final event. Those are different claims, and only one of them is in the data.
The labeling is also noisier than most people assume. When pathologists compared 272 Connecticut autopsies with the death certificates, 29% of deaths moved to a different major disease category, and in another 26% the category held but the specific disease changed (Kircher et al. 1985). In 450 autopsies at a Japanese geriatric hospital, the certificate matched the autopsy 48% of the time overall, and just 9% of the time for pneumonia (Mieno et al. 2016), the label that absorbs many end-of-cascade deaths.
How a fall becomes “Infections”
Here’s one path that ends up in the “Infections” block.
Someone loses muscle mass and power over two decades, slowly, without noticing. Their margin for error narrows. One day they trip, and where a stronger version of them would have caught themselves, they fall and break a hip. That means surgery and weeks of barely moving. Lying in bed sets up pneumonia, and the pneumonia is what kills them.
Every step in that chain happened, but only the last one gets recorded.
Only the last step gets recorded
A common path from lost strength to a death filed under infections.
- Muscle and power fadeover decades
- A tripmargin too thin to recover
- A fall
- Hip fracture
- Surgery and bed restreserve drains faster
- Pneumonia
- Death certificate“Infections”
Happened, never recordedThe only link the chart counts
The numbers on this path are large. About one in five older adults dies within a year of a hip fracture (Downey et al. 2019). In the first three months, the risk of death runs five to eight times that of peers the same age and sex, and it never fully returns to their level (Haentjens et al. 2010). In a Korean national cohort of nearly 15,000 hip-fracture patients over 65, those who developed pneumonia in the hospital had two to three times the mortality of those who didn’t (Jang et al. 2020). This is one of the most common ways a long decline in physical capacity turns into a death certificate that never mentions physical capacity.
The capacity itself drains faster than people expect. When healthy, moderately active adults around 67 years old spent 10 days in bed, a shorter stay than many hospitalizations, they lost about a kilogram of leg muscle and their rate of muscle protein synthesis fell by about 30% (Kortebein et al. 2007). Knee strength dropped 13%, stair-climbing power 14%, and aerobic capacity 12% (Kortebein et al. 2008). And none of them were sick going in.
Three ways to name the mistake
Scientists and philosophers have described this mistake in a few different ways, and each one adds something.
Proximate versus distal causation. Death certificates are built to capture the final event. Everything upstream is invisible, because the form has no box for it.
The seen and the unseen. Prevention leaves no trace. The fall that didn’t happen, because someone could catch themselves, generates no code, no bar segment, no data point. Anything whose job is to keep you out of a category will look unimportant when you only count what’s inside the categories.
Mistaking the label for the thing. The ICD code is a filing decision someone makes under time pressure, and reading “Infections, 12%” as a description of biology confuses the index card with the patient.
Death isn’t the only outcome that matters
There’s a simpler problem, too: death may be the wrong outcome to look at.
Musculoskeletal conditions sit near the bottom of mortality tables and near the top of disability tables. Low back pain alone is the leading cause of years lived with disability worldwide: about 619 million people affected and 69 million years of healthy life lost to disability in 2020 (GBD 2021 Low Back Pain Collaborators 2023). It kills almost no one, but it takes away a huge number of good years.
If your question is “how long until I’m dead,” mortality share is the right thing to look at. If your question is “how many years can I live in a body that does what I ask it to,” it isn’t.
What the chart can’t show: reserve
The idea tying all of this together is physiological reserve, the gap between what your systems can do at maximum and what daily life demands of them. Reserve absorbs shocks. It’s why two people with the same illness have different outcomes, and why the same fall is an embarrassment at 40 and a terminal event at 80.
Reserve never appears on a death certificate, because it decides whether you enter a diagnostic category, not which one you land in. It shows up clearly in data that follow people forward in time, though.
In the PURE study, which followed nearly 140,000 adults in 17 countries, each 5 kg lower grip strength was associated with about a 16% higher risk of death from any cause and a 17% higher risk of cardiovascular death. Grip strength predicted death better than systolic blood pressure did (Leong et al. 2015). Grip strength doesn’t cause those deaths; it’s a quick way to read how much reserve someone has.
Among 122,007 adults who completed a treadmill test, higher fitness meant lower long-term mortality with no observed upper limit. The least fit group carried about five times the risk of the fittest, a gap larger than the one attached to smoking, diabetes, or coronary artery disease in the same patients (Mandsager et al. 2018).
Low fitness carried more risk than smoking or diabetes
Adjusted hazard ratio for death from any cause among 122,007 adults after a treadmill test. 1× means no extra risk.
Least fit group compared with…
Having the condition vs not having it
Show as a table (with 95% confidence intervals)
| Comparison | Hazard ratio | 95% CI |
|---|---|---|
| Least fit vs elite | 5.04 | 4.10–6.20 |
| Least fit vs high | 3.90 | 3.67–4.14 |
| Least fit vs above average | 2.75 | 2.61–2.89 |
| Least fit vs below average | 1.95 | 1.86–2.04 |
| End-stage kidney disease | 2.78 | 2.53–3.05 |
| Smoking | 1.41 | 1.36–1.46 |
| Diabetes | 1.40 | 1.34–1.46 |
| Coronary artery disease | 1.29 | 1.24–1.35 |
| High blood pressure | 1.21 | 1.16–1.25 |
On evidence like this, the American Heart Association argued in 2016 that fitness should be treated as a clinical vital sign, because adding it to traditional risk factors improves how well doctors sort patients by risk (Ross et al. 2016).
None of these measurements show up on a cause-of-death chart, yet all of them predict what ends up on one.
Training the reserve, without overselling it
I want to be careful here, because this is where arguments like this one usually get oversold.
Aerobic training of almost any kind raises VO2max. In a meta-analysis of 28 controlled trials in healthy 18- to 45-year-olds, continuous endurance training improved it by about 4.9 mL/kg/min and high-intensity intervals by about 5.5; head to head, intervals came out ahead by a small margin, about 1.2 mL/kg/min (Milanović et al. 2015). Because VO2max is among the strongest single predictors of all-cause mortality we have, intervals earn a place in your week, especially when time is short.
That doesn’t mean intervals beat everything else at protecting the heart. SAINTEX-CAD randomized 200 people with stable coronary artery disease to interval or continuous training and found equal improvements in exercise capacity and blood-vessel function (Conraads et al. 2015). SMARTEX-HF randomized 261 people with heart failure across three arms and found intervals no better than continuous training at reversing heart enlargement, its primary endpoint (Ellingsen et al. 2017).
No exercise modality has hard-outcome trial evidence approaching what statins, blood pressure control, or quitting smoking have. The evidence is strong enough without stretching it: aerobic training, including intervals, raises a trait that independently predicts how long and how well you live. Strength training works on the other reserve, the one that decides whether a stumble stays a stumble. If you want to see where you stand on each, the sitting-rising test checks the strength-and-balance side and the one-mile walk test estimates the aerobic one.
What to do with this
So instead of asking what kills people, which a cause-of-death chart can’t really answer, ask how big your margin is and whether it’s growing or shrinking. That question has answers you can measure: strength, aerobic capacity, balance, the ability to get off the floor unassisted. Those numbers tell you where you stand today and where you’re headed. If you want the risk-calculator view of the same idea, your heart’s 10- and 30-year odds runs the American Heart Association’s PREVENT equations. A cause-of-death chart can only tell you how the story ended for someone else.
Most people have never measured their own margin, and it only takes a few simple tests.
This article is for general education. It discusses mortality risk in populations, not your personal risk, and creates no medical or other professional relationship. Talk with your physician before starting a new exercise program, and never delay seeking medical advice because of something you read here.
Sources
Every number in this article traces to one of these 19 peer-reviewed sources. In-text citations use the author and year.