"Drink more water" looked like a 20% cut in kidney stone risk in a 61-study review. In the stronger studies it shrank to 4% and could be chance. Coffee's link barely moved

Ask anyone who has passed a kidney stone what they were told afterwards and the answer usually starts with water. A review published in Nutrients on 3 November 2026 tried to do something broader: pool the observational research on nearly everything people eat and drink, across 61 studies published between 1993 and 2026. Its summary reads like a familiar list. More fluid, fruit, dairy and coffee on one side; meat, sugary drinks, salt and ultra-processed food on the other.

The more interesting part is a second pass most readers will never see. Some of the 61 studies measured diet and stones at the same moment or looked backwards, and a diagnosis can change what people eat and drink. So the authors reran the numbers using only prospective cohorts rated at low or moderate risk of bias, where diet is recorded first and stones are counted later. That is where water slipped. Across 11 reports, people with the highest fluid intake had about 20% lower risk than those with the lowest (HR 0.797). In the eight stronger reports the gap was about 4% (HR 0.961, 95% CI 0.874 to 1.057), which can't be told apart from no difference. Tea went the same way in its high-against-low comparison, from HR 0.925 to 0.985.

Coffee didn't budge: HR 0.845 overall, 0.841 in the stronger cohorts, and about 9% lower risk per daily serving in the wider analysis. Fruit held too (0.849, then 0.824), as did dairy (0.926, then 0.918). Sugar-sweetened drinks stayed on the wrong side (1.340, then 1.302, the latter from a single report), and salt looked slightly worse (1.129, then 1.194). Dairy is the finding that runs against old instinct, since stones are often made of calcium. The usual explanation is that calcium in a meal binds oxalate in the gut, so less of it reaches the urine. That favours food; the authors note that calcium supplements taken between meals may not behave the same way.

Before anyone pours the water away: a faded estimate is not evidence that water doesn't work. The stronger water analysis had extremely high inconsistency between studies (I² 93%), and the authors suggest lost statistical power may explain it. They kept hydration as the cornerstone of their advice, though their own conclusion calling fluid protection robust goes further than that sensitivity analysis does. The review also left out randomised trials by design, and those are where the water question has been tested directly. In a five-year Italian trial of 199 people after a first calcium stone, 12 of 99 told to drink plenty had a recurrence, against 27 of 100 who weren't. A much larger US trial, reported in The Lancet in 2026, gave 1,658 people with a stone history coaching and financial incentives to drink more. Urine volume rose modestly, and stone events came out at 19% against 20% (HR 0.96, 95% CI 0.77 to 1.20). That tested a programme for changing a habit rather than water itself, so it can't settle the question, but it didn't back the easy version of the advice either.

Some of the other numbers invite more confidence than they've earned. Beer and wine both came out protective (HR 0.817 and 0.839), which is a reminder of what observational data can and can't do rather than a reason to drink; alcohol's other harms don't appear in this table, and spirits showed no significant link. Ultra-processed food looks dramatic at HR 1.532, but rests on two reports that disagree sharply (I² 86%). The steepest figure of all, 1.891 for a diet with a high acid load, comes from one report. And while the authors advise limiting animal protein, their own pooled estimate for it was flat (1.006, and 0.99 in the stronger cohorts). Meat was the signal (1.153), from just two reports.

Read the whole table with those limits in mind. The studies define high and low intake differently, and many relationships seen in high-against-low comparisons vanished when intake was treated as a continuous amount: water, salt and sugary drinks among them. The review pooled first, repeat and existing stones together, relied heavily on questionnaires, and counted several comparisons from the same cohorts, which the authors concede may overstate precision. It doesn't report results by stone type either, and advice for uric acid stones isn't the same as for calcium ones. What it offers is a ranking of where the evidence is sturdy. On that ranking, fruit, dairy and fewer sugary drinks are among the sturdier bets, and a stone-prone coffee drinker finds no reason here to give up the cup.