How much walking actually reaches your hip

Somewhere after forty, bone starts leaving. Between 1% and 2% of it a year, quietly, with no symptom attached, until something breaks that should not have broken ^[4].

Guidelines agree on the answer: exercise. Which kind, and how much, is less settled. The World Health Organization says 150 to 300 minutes a week of moderate activity, plus strength work — but that is a recommendation for general health, not a bone prescription ^[3].

In September 2026 a group in China pooled 124 randomised trials, 18,429 adults over forty, and tried to answer the question properly: which exercise, and what dose ^[1].

First, whose bones these were

Not Indian ones. The largest contributors were the United States (22 trials), China (14), Australia (13) and England (9). No Indian trial is named in the paper's main text; the full country list sits in a supplement we have not read. Everything below is pooled international evidence, and it should be read that way until somebody runs the Indian version.

Two other things about who was studied, because they change what the numbers mean:

• 90 of the 124 trials enrolled only women. Ten enrolled only men. The authors call sex-balanced trials in this field an urgent need, and they are right.
• 18,429 is the total across the whole paper, not the denominator of any single result. Bone density at the spine rests on 6,553 people, the femoral neck on 6,462, the total hip on 3,670, and fractures on 11,132. Any headline that attaches 18,429 to one finding is wrong ^[1].

Walking won

At all three measured sites — lumbar spine, femoral neck, total hip — brisk walking or jogging came first. At the spine it shares that place, as the next paragraph shows.

That is the headline, and the trial count is why we are printing it next to the number. Combined aerobic-and-resistance training produced exactly the same point value, 0.013 — but from 39 trials and 1,456 people, with a much tighter interval of 0.010 to 0.016 ^[1].

Same number. Wildly different amounts of evidence behind it. If you only read the ranking, walking and combined training look like first and second place. If you read the intervals, they are the same finding, and the combined one is better supported.

The total hip result is the one to be most careful with. It is the biggest effect in the paper — +0.021 g/cm² — and it comes from three trials and 68 people ^[1]. No other exercise type reached significance at the total hip.

There is exactly one finding in the entire paper that the authors rate as moderate certainty, and it is not about walking. It is mind-body exercise — tai chi, yoga — at the femoral neck: +0.007 g/cm², interval 0.002 to 0.013 ^[1]. Everything else in the paper is low, or low-to-moderate, or very low.

How much

The paper converts every exercise in every trial into one shared unit: METs-minutes per week. A MET is a multiple of your resting metabolism — sitting still is 1, brisk walking is roughly 4 ^[5].

Its answer, pooled across all six exercise types, is about 600 METs-min a week, which the authors translate as two to three hours of brisk walking ^[1].

But the paper also publishes the thresholds for brisk walking specifically, and they are higher: 427 for the spine, 753 for the femoral neck, 789 for the total hip ^[1].

Do that arithmetic and the guideline number runs into trouble.

At the fastest brisk pace, 150 minutes a week reaches 750. The femoral neck threshold is 753. That is a miss of 0.4% — a hairline, not a gap, and we are not going to draw it as one. The total hip miss is 4.9%. At any slower pace, both misses widen.

Forty minutes a day — 280 minutes a week — produces 933 to 1,400 METs-min depending on how fast you walk, and clears all three thresholds at every pace in that range.

150 minutes a week is the number everyone quotes. On this paper's own figures, it lands at the spine and stops there.

One honest caveat, which the authors state themselves: these thresholds are "exercise modality derived approximations… rather than definitive mechanical or clinical cut-offs", and a MET measures metabolic cost, not the mechanical strain that actually builds bone ^[1]. So this is "below the modelled figure", not "not enough for your hip".

And more is not indefinitely better. The dose-response curve is an inverted U for every exercise type except mind-body, with gains peaking somewhere around 900 to 1,100 METs-min. The authors do not claim over-exercising damages bone — they suggest the downturn more likely reflects people dropping out of the most intensive trial arms, and the very small number of trials at those doses ^[1].

Now the part the headline will not carry

Bone density is a surrogate. It is a number on a scan. What a person actually cares about is whether a bone breaks, and treatment-related changes in density only partly predict that ^[2].

The same paper measured fractures, in a smaller analysis: 26 trials, 11,132 people.

Read the third row. Brisk walking's interval runs from 0.44 to 1.06 — it crosses 1.0, which means this analysis did not find a statistically significant fracture reduction for it ^[1].

The exercise that built the most bone did not significantly reduce fractures. And the two that did show a fracture reduction — mixed aerobic, and tai chi and yoga — showed no significant bone-density gain at the spine. All of it is low certainty, and a 71% reduction estimated from 119 people is not something anyone should build a decision on.

The best-supported safety finding in the paper is this one: no exercise type increased fracture risk. If you have been avoiding walking because you are frightened of breaking something, that is the number for you.

What "clinically meaningful" actually means here

This is the part most coverage will get wrong, so it is worth being exact.

The authors did not pick an arbitrary threshold for "big enough to matter". They anchored it to biology: since adults over forty lose 1–2% of bone density a year, they defined a meaningful gain as 1% to 2% of baseline ^[1][4].

Which means hitting it does not make your bones stronger than they were. It cancels roughly one year of loss. The authors note that individual exercise types rarely reached even the 2% upper end. Nothing in this paper supports "reverses osteoporosis" or "rebuilds bone".

How much to trust any of it

• 30% of the 124 trials were rated high risk of bias, 48% had some concerns, and only 23% were low risk ^[1].
• Three quarters had concerns about performance bias — unavoidable, because you cannot blind someone to whether they are jogging.
• Tests for small-study effects were significant at the lumbar spine (p=0.003) and total hip (p=0.033), meaning small trials reported unusually large effects ^[1].
• Overall confidence across the network ranged from moderate to very low.

And then the authors say the quiet part themselves, about their own headline finding:

"several small trials evaluating brisk walking or jogging reported outsized effect estimates… these specific estimates may be inflated and should be interpreted judiciously." ^[1]

It is the most useful line in the paper.

One sentence we could not verify. In the section describing the network, the paper reports a statistical test for the femoral neck as χ²=2303.9 with 97 degrees of freedom and a p-value of 1.00. Those two numbers are not compatible — a statistic that large on that many degrees of freedom gives a p-value close to zero. The paper's other three sites are internally consistent. We assume a typographical error, we do not know which of the two numbers is wrong, and so we have not used that sentence for anything.

What this does not license

• It does not show that walking prevents fractures. In this analysis, it did not significantly reduce them.
• It does not show that exercise reverses bone loss. The benchmark is cancelling one year of decline.
• It does not show that too much exercise harms bone. The authors explicitly offer other explanations for the downturn.
• It does not show that yoga beats walking. Mind-body has higher certainty, not a higher effect. Those are different claims.
• It is not Indian evidence.

We have deliberately left out the physiology — why a step loads bone, why light resistance work often fails to, why bone stops responding to the same load repeated forever. All of that sits in the paper's discussion, cited to other work, and we would rather source it properly and publish it separately than repeat it on this paper's authority.

This is not medical advice. If you have had a fracture, have been told you have osteoporosis or osteopenia, or are on treatment for either, the exercise that is right for you is a conversation with your doctor, not a number from a pooled analysis of somebody else's trials.

Updated 23 September 2026, before publication. Rechecked line by line against the paper. The summary and the fracture lines now say "no significant reduction" instead of stronger wording, the country list names only what the paper names, and the dose and bias details match the paper's own figures.