Low-carbohydrate diets

Most dietary trials in osteoarthritis compare a diet to nothing, which means they compare a diet to being paid attention. This one did not. It ran a low-carbohydrate arm, a low-fat arm and a control arm side by side, and only one of the two diets did anything — which is a more informative result than either arm alone could have produced.

The trial

Adults aged 65 to 75 with knee osteoarthritis followed a low-carbohydrate diet, a low-fat diet, or their usual eating, for twelve weeks. Pain was assessed every three weeks in two ways: self-reported, and evoked — functional tasks designed to provoke the symptom rather than ask about it. Serum oxidative stress was measured before and after.

The low-carbohydrate arm reduced pain intensity and unpleasantness on some of the functional tasks and on self-report, relative both to the low-fat diet and to the control. It also reduced serum oxidative stress and leptin against both comparators. And the size of a person's fall in oxidative stress was related to the size of their fall in functional pain.

The low-fat arm did not produce those changes. That is what makes the design worth the space: if dietary attention, food-diary keeping and clinic visits were producing the effect, both diet arms should have moved.

What the mechanism is proposed to be

Not inflammation, and not — on the authors' reading — weight. The measured mediator is oxidative stress, and the second signal is leptin, an adipokine with direct effects on chondrocytes that fell specifically on the low-carbohydrate arm. Neither is established as causal by a correlation inside a pilot, but they are at least measured mediators rather than assumed ones, which is unusual in this literature.

The weight problem

Carbohydrate restriction takes weight off. The strongest randomized evidence anywhere in this section — the IDEA trial and the meta-analysis behind it — is that diet-induced weight loss improves knee pain, function and inflammatory markers, with a clear dose-response. No trial in this topic has held weight change constant between arms, so the honest position is that some or all of the low-carbohydrate effect may be the weight, arriving by a route that happens to be effective.

The low-fat arm complicates that objection usefully rather than settling it. Low-fat diets also produce weight loss, and that arm did not reduce pain — but the trial does not report the two arms as weight-matched, so this is a reason to run the comparison, not a reason to consider it made.

The follow-up

The 2020 trial suggested the benefit was larger in women and in non-Hispanic Black participants, in a sample too small to say so confidently. A subsequent study recruited fourteen non-Hispanic Black women with knee osteoarthritis for six weeks of the same diet, and reported significant changes in daily pain, pain interference, stiffness, physical functioning, depressive symptoms, food knowledge and food security, with evoked pain falling on timed-walk and chair-stand tasks.

It has no control group. Fourteen people improving over six weeks in an uncontrolled study is what regression to the mean, attention and expectation look like, and the study is best read as extending the observation to a population the parent trial under-sampled rather than as confirming it.

Practical notes

The trials specified a low-carbohydrate pattern rather than ketosis, and nothing in this literature tests a ketogenic diet against a joint endpoint. Twelve weeks is the longest anyone has run it. Carbohydrate restriction interacts with glucose-lowering medication, which is a clinical matter these trials did not address.

What would change this entry

A weight-matched comparison of low-carbohydrate against low-fat, long enough to image. If the pain difference survives with weight held equal, this becomes one of the more interesting findings in the section — a dietary composition effect independent of body mass. If it does not, the entry folds into body weight, and that is a clean answer too.