Alcohol

Alcohol appears in this subcategory not because the data say to limit it for a joint, but because the data have been asked and have answered. It is the entry where the machinery of nutritional epidemiology is most visible, because the finding and its collapse are in the same paper.

What the pooled observational data showed

A systematic review and meta-analysis gathered 29 observational studies covering 25,192 people with osteoarthritis. Any alcohol consumption was associated with less osteoarthritis — odds ratio 0.79 (95% CI 0.68–0.93) — and so was weekly or more frequent drinking (0.79, 0.65–0.97). The association held when the studies were grouped by outcome: radiographic osteoarthritis 0.83 (0.70–0.98), hand 0.80 (0.66–0.95), knee 0.85 (0.72–0.99).

Read on its own, that is a consistent protective signal across joints and study designs, and it is the source of a claim that circulates widely.

And what happened when it was adjusted

The same authors ran the analysis a second way, splitting studies by whether they had adjusted for any covariate. In unadjusted data the odds ratio was 0.70 (0.55–0.89). In studies adjusted for anything at all, it was 0.93 (0.78–1.10) — a confidence interval comfortably crossing 1.

Their own conclusion is that the meta-analysis provides evidence to dispel the notion that alcohol protects against osteoarthritis. The apparent effect lives entirely in the studies that did not control for anything.

Why the signal was there

Three structures, all of them familiar and all of them present here. The comparison group of non-drinkers contains people who stopped drinking because they became unwell — the sick-quitter problem, which makes abstainers look worse than they are. Drinking tracks with income, social participation and mobility; people who go out drink more, and people whose knees let them go out have better knees. And most of these studies are cross-sectional, so the exposure is measured after the disease has arrived.

The genetic evidence agrees

Mendelian randomization uses genetic variants as proxies for lifetime exposure, which sidesteps reverse causation and much of the confounding. A meta-analysis pooling eleven such studies, plus the authors' own analyses across four lifestyle exposures and five arthritis outcomes, found no causal association between alcohol and arthritis: odds ratio 1.02 (95% CI 0.94–1.11).

The same analysis is not blind to lifestyle effects. Smoking came out at 1.44 (1.27–1.64), holding for both osteoarthritis (1.44) and rheumatoid arthritis (1.37). Coffee came out marginally positive. Alcohol did not move. A separate Mendelian randomization of eighteen dietary factors against knee osteoarthritis included both frequency of alcohol intake and alcoholic beverages, and neither was among the exposures that reached significance.

One distinction worth keeping

Alcohol has a different and more persistent inverse literature in rheumatoid arthritis. Rheumatoid arthritis is autoimmune and osteoarthritis is not, and findings do not cross between them — the same caution this section applies to the marine-oil meta-analysis, whose pain effect was carried almost entirely by rheumatoid trials.

Practical notes

Nothing here is a statement about whether to drink. Alcohol's health effects are large, well documented, and outside a cartilage site's scope entirely. What this entry can say is narrow and worth saying plainly: the specific claim that alcohol protects joints does not survive contact with adjusted data or with genetic instruments.

What would still be worth knowing

Every estimate above is an average across a range that under-represents the heaviest drinkers, and no cohort has looked at alcohol against post-traumatic osteoarthritis or against recovery after cartilage surgery. If alcohol affects joints at all, tissue repair after an injury is where the rest of the literature would predict it — and that is the one place nobody has looked.