Deep divefrom Episode 10Fat loss7 November 202615 min

Can Training Make Every Step Cheaper?

A deep dive into one paper: what was known before, what the researchers did, what the numbers say and what they don't.

The paper: Effects of Exercise Training on the Metabolic Cost of Walking in Health and Disease (Sports Medicine)

Chapters

Transcript

The script the two AI voices read from. The times are worked out from the script, not heard, so a line can start a few seconds early or late.

Read the whole episode

Cold open

Host 1You heard the trailer — this is the whole story. Welcome to the Hastrup Review of Science, where artificial intelligence reads real science, so you don't have to pretend you understood the paper.

Host 2Today is one paper, a systematic review and meta-analysis in the journal Sports Medicine. It asks whether regular exercise training makes walking cheaper for the body, in healthy adults and in people with stroke, Parkinson's disease or multiple sclerosis.

Host 1When you say walking gets cheaper, cheaper in what currency, exactly?

Host 2In energy. The metabolic cost of walking is how much energy your body burns to move a given distance, and a lower cost means you cover the same ground for less fuel.

Why walking has a price

Host 1Why would anyone spend a career studying how much fuel a person burns walking to the shops?

Host 2Because walking is the one movement almost everyone does every day, and its price decides how far and how long people can go. In the wider field, researchers have measured this for over a century, first with bags that collected a walker's breath, and today with lightweight masks that read oxygen use breath by breath.

Host 1So they measure how much oxygen you breathe in, and turn that into energy?

Host 2Yes, that's the core of it, a method called indirect calorimetry. You walk at a steady speed, the mask tracks oxygen used and carbon dioxide produced, and from that you get energy per metre, often scaled to body weight.

Host 1Is the price the same for everyone, or does it change over a lifetime?

Host 2It changes. The paper opens by noting that the cost rises with healthy ageing and in neurological conditions, so an older adult or someone after a stroke pays more for every step.

Host 1How much more are we talking, roughly, for someone who has had a stroke?

Host 2This paper doesn't give that figure, but in the wider literature people after a stroke often spend well over half again as much energy per metre as healthy peers, sometimes close to double. For older healthy adults the extra is usually smaller, more in the range of a tenth to a fifth.

Host 1And why does that extra cost matter in daily life, beyond feeling a bit more tired?

Host 2Because if each step is expensive, walking uses up a bigger share of what your heart and lungs can deliver. Then people walk less, get less fit, and the cost climbs further, a loop that researchers in this area worry about a lot.

What was known before

Host 1Before this paper, what did people already believe about exercise and walking cost?

Host 2There were many separate trials, mostly small. Some found that aerobic training, resistance training or walking practice lowered the cost, and others found almost nothing, so the picture was patchy.

Host 1Why would small trials disagree so much if they're all measuring the same thing?

Host 2Partly because a trial with twenty people can easily miss a modest effect or exaggerate one by chance. They also used very different programmes, groups of people and walking speeds, so it was hard to know whether the disagreement was noise or real differences.

The open question

Host 1So what was the question nobody had answered yet?

Host 2Whether chronic exercise training, meaning a programme over weeks or months rather than one session, reliably lowers the cost of walking. And if it does, whether the type of training, the dose or the intensity changes how much, and whether patients benefit as much as healthy people.

Host 1By dose, do you mean how many weeks, how many sessions a week, and how long each one lasts?

Host 2Exactly those three, plus intensity, how hard each session is. The authors describe this as the first synthesis to look at all of that across different populations together.

How they did it

Host 1How do you even pull together twenty-odd studies that each did something slightly different?

Host 2They ran a systematic review, which means a predefined search, here of three large databases called Medline, Web of Science and CINAHL, following the standard reporting rules for such reviews, known by the acronym PRISMA. Every study that met their criteria went in, not just the ones they liked.

Host 1What did they end up with once the search was done?

Host 2Twenty-seven studies, with thirty-six intervention arms and fifty-four outcomes, in one thousand one hundred and seventy-seven people. Seven hundred and three were healthy adults and four hundred and seventy-four were patients.

Host 1What's the difference between a study, an arm and an outcome in that count?

Host 2A study is one paper, an arm is one training group within it, so a study comparing aerobic and strength training against a control has two arms. An outcome is one measurement of walking cost, and some studies measured it at several speeds.

Host 1Which patients were included, and were the groups evenly balanced?

Host 2Not at all. Stroke had nine studies, while Parkinson's disease and multiple sclerosis had just one study each, so the disease side is really mostly a stroke story.

Host 1If one study can contribute several outcomes, doesn't that risk counting the same people twice?

Host 2Yes, and that's the problem their statistical method is built for. They used robust variance estimation, which lets several linked results from one study enter the analysis while correcting for the fact that they aren't independent.

Host 1So it's a way of using all the measurements without letting one study shout louder than it should?

Host 2That's a fair way to put it. Each study's results are allowed in, but their combined weight is kept honest.

Host 1How do they compare a walking cost measured in oxygen in one study with one measured in calories in another?

Host 2They convert each result into a standard effect size, here one called Hedges' g. It expresses the improvement in units of how much people normally vary, so different measures land on one common scale.

Host 1What counts as a big or a small number on that scale?

Host 2By the usual rule of thumb, about zero point two is small, about zero point five is medium, and about zero point eight is large. It's a convention, not a law, but it helps you place the results.

Host 1And then how did they test whether training type or dose made a difference?

Host 2With meta-regression, which asks whether studies with a certain feature, say longer programmes or higher intensity, tend to show bigger effects. They put training type, duration, intensity, frequency and session length in, along with population and body mass index.

Host 1Body mass index, the weight-for-height number your doctor quotes, why was that in the model?

Host 2Because carrying more mass changes the mechanics and the energy of walking, so it might change how much training can help. They used the average for each study, not each person's own value, and that detail matters later.

A word from our sponsor

Host 1This episode is brought to you by Steeper, the hill you carry in your coat pocket.

Host 2Unfold Steeper onto any flat pavement and your walk instantly costs more. Finally, value for your oxygen.

Host 1I take mine to the supermarket. The dairy aisle is now a north face, and the cashier has started carrying a flag.

Host 2In an unblinded trial of one host, Steeper beat no hill with an effect size of nine. The control group simply walked home.

Host 1Side effects include altitude, a light dusting of snow, and goats.

Host 2Steeper. Fold it away and your street remembers it was once a mountain.

What they found

Host 1So, after all that pooling, does exercise training actually make walking cheaper?

Host 2Yes, on average. Across all types of training, the pooled effect was a Hedges' g of zero point four, with a p value of zero point zero zero one.

Host 1On the scale you described, that sits between small and medium, is that right?

Host 2Right, a small to moderate improvement. It's a real average shift, not a dramatic one, and it says nothing about how many calories any one person saves.

Host 1And what does the p value of zero point zero zero one tell me, in plain words?

Host 2That a result this strong would be very unlikely if training truly did nothing. It tells you the effect is probably not zero, not that it's large.

Host 1Did healthy people and stroke patients improve by the same amount?

Host 2No. Healthy adults showed a g of zero point four nine, with a p value of zero point zero zero one, while people after stroke showed zero point two four, with a p value of zero point zero six five.

Host 1So in stroke the effect was about half the size, and it didn't quite clear the usual bar for being sure?

Host 2Correct, it just missed the conventional cut-off of zero point zero five. That isn't proof of no effect in stroke, it's a smaller, less certain one, from only nine studies.

Host 1Was that difference between the groups itself solid, or could it be a fluke?

Host 2The meta-regression found that population did moderate the effect, with a p value of zero point zero four three, after accounting for the training characteristics. So the gap held up statistically, though not by a wide margin.

Host 1What about the body mass index result you mentioned earlier?

Host 2Studies whose participants had a higher average body mass index showed smaller training effects, with a coefficient of minus zero point two and a p value of zero point zero four. Again, a modest association, close to the line.

Host 1Does that mean training simply works less well for heavier people?

Host 2Not necessarily, and that's the catch I flagged. It compares studies by their average, so it could reflect differences between those study groups, like age or illness, rather than what happens inside any heavier individual.

Host 1And the big practical question: which kind of training worked best, aerobic or strength?

Host 2None stood out. Training type, programme length, intensity, how often people trained and how long each session lasted did not significantly change the effect.

Host 1So I could pick whichever kind of exercise I like and expect the same benefit?

Host 2That goes further than the data. The authors say plainly that the meta-regression may have been underpowered, meaning there were too few studies to detect real differences, so 'no difference found' is not the same as 'no difference exists'.

Host 1How consistent were the studies with each other overall?

Host 2Moderately. A measure of inconsistency called I squared stayed below about sixty-two percent, and the risk of bias in the studies was also rated moderate.

Host 1What does an I squared of around sixty percent mean for how much I should trust the average?

Host 2It means a fair share of the spread between studies reflects real differences rather than chance alone. The average is meaningful, but individual programmes may do noticeably better or worse than it.

Why it matters

Host 1Stepping back, what's the main idea I should take from this?

Host 2That the energy price of walking isn't fixed. Across many programmes, regular training nudged it down, most clearly in healthy adults.

Host 1Why would training make walking cheaper in the first place, what changes in the body?

Host 2This paper doesn't test mechanisms. In the wider field, the usual suspects are stronger muscles doing each step with less effort, better coordination with less wasted co-contraction, and fitter muscle cells that burn fuel more efficiently.

Host 1For someone in their sixties who feels walks are getting harder, what does this mean for them?

Host 2It's supporting evidence that a training programme can make walking itself less costly, not just make you fitter for other things. It doesn't say which programme, or how much any one person will gain.

Host 1And for someone recovering from a stroke, should they read this as bad news?

Host 2No. The effect was smaller and less certain, which fits a harder problem, since a stroke often changes how the body moves, not just how fit it is. It suggests training alone may not be enough to close the gap, and that is a question for stroke rehabilitation research to sharpen.

Host 1What can we definitely not conclude from this paper?

Host 2We can't say how much exercise is enough, or which type is best, or that heavier individuals benefit less. And with one study each, we can say essentially nothing specific about Parkinson's disease or multiple sclerosis.

Honest caveats

Host 1Are there weaknesses in how the evidence was put together that worry you?

Host 2A few. Twenty-seven studies is a modest base for testing six or seven possible moderators at once, and many of the underlying trials were small.

Host 1Does it matter that the studies measured walking at different speeds?

Host 2It can. Walking cost depends strongly on speed, so if people walk faster after training and are tested at their own chosen pace, comparisons get tricky, and the abstract doesn't say how they handled that.

Host 1Were the trials all comparing training against people who did nothing?

Host 2The abstract doesn't specify the comparison groups, and that's something to check in the full paper. Moderate risk of bias usually means issues like unblinded assessors or missing data in some trials.

Host 1Is there anything about the healthy group that might make it look better than it is?

Host 2Possibly. Healthy volunteers who sign up for exercise trials tend to be motivated and well, and seven hundred people across many small studies can still lean towards particular ages or settings.

Host 1So the honest summary is a modest, real average benefit, with the details still blurry?

Host 2Yes, that's a fair summary. The direction is reasonably clear, the size is small to moderate, and the recipe is unknown.

What to watch next

Host 1What would you want to see next to make this clearer?

Host 2Larger trials that directly compare training types and doses head to head, in the same people, at the same fixed walking speed. That answers the recipe question far better than comparing across studies.

Host 1And for the neurological conditions, what's missing?

Host 2Simply more studies, especially in Parkinson's disease and multiple sclerosis, where this review had just one each. In stroke, it would help to know whether pairing fitness training with gait retraining shrinks the cost more than either alone.

Host 1Would it help to know whether the benefit lasts after people stop training?

Host 2Very much. The abstract reports changes after programmes, not long-term follow-up, so whether walking stays cheaper months later is still open.

Sign-off

Host 1That's the whole story. Training makes the walk a little cheaper, nobody yet knows the exact recipe, and our sponsor would like it more expensive again.

Host 2Every weekend this same feed carries the roundup, the week's new papers in one short episode. Thanks for listening to the Hastrup Review of Science.