More than half of 1,027 older Chinese adults with diabetes called their diet light. Urine tests put 94% over the salt limit, and the saltiest quarter more often showed an early kidney warning sign

Kidney trouble in diabetes often shows in the urine before it shows in the usual blood test. A small leak of the protein albumin can appear while filtering still looks normal, which makes it a useful place to look for early links with diet. Researchers at the Third Xiangya Hospital in Changsha did that for salt, in a paper published in BMJ Open on 23 October.

They started with 270,900 people who had routine check-ups at the hospital's health centre between January 2017 and October 2024. Most were dropped for being under 60, and others for missing results, known kidney disease or a filtering rate (eGFR) below 60. That left 1,027 people with type 2 diabetes, 66% of them men, with an average age of about 67. The authors describe them as health-seeking, community-dwelling people. Most were retired, with check-ups arranged by former employers.

Salt intake was estimated from a morning urine sample collected between 8 and 10 am, using the Kawasaki formula. The average came to 8.12 g a day, and 969 people (94.35%) were over the WHO's 5 g limit. Yet 545 had described their diet as light. Only 58 people in the whole study weren't over the limit, so at least 487 of those light eaters were, on this estimate.

An albumin-to-creatinine ratio (UACR) of 30 to 300 mg/g counted as early kidney impairment, and 459 people (44.69%) were classed as positive. Sorted into four equal groups by salt, 38% of the lowest group were positive (97 of 255) and 56% of the highest (145 of 257). Those percentages are our arithmetic from the paper's table, before any adjustment.

After adjusting for age, weight, blood pressure, fasting glucose, blood lipids, eGFR, alcohol, food preferences and medicines, the top group (9.46 g or more) had 1.62 times the odds of the lowest group (6.63 g or less). The steps between were less clear: 0.99 for the second group and 1.31 for the third, and neither interval excluded no difference. A curve fitted across the range looked roughly straight, and the authors saw no obvious safe threshold. With so few people under 5 g, though, the study says little about what happens at the recommended limit.

The link also wasn't uniform. Among the 678 people with an eGFR of 90 or more, the top group's odds were 2.20 times the lowest group's. Among the 349 with an eGFR of 60 to 89 the figure was 0.91, with a wide interval. The authors suggest fewer people and more established damage may explain this, but they didn't formally test the difference and call these findings hypothesis-generating.

The salt number deserves some caution. A validation study of 1,424 adults aged 18 to 69 in Zhejiang compared the Kawasaki formula with real 24-hour collections. It overestimated by 18.1 mmol of sodium a day on average, which is about 1 g of salt, and put 65.6% of people in the wrong band when intake was split into four. That's a different group of people, and even 1 g less would leave the average here above the limit. But it does mean any one person's figure is rough, and so is their place in a salt group.

There's another wrinkle the paper doesn't discuss. The Kawasaki formula and the UACR both divide by creatinine measured in urine, and in this study both came from morning samples at the check-up. Someone with low urine creatinine that morning would be pushed up on both scores at once, which could create some of the link without any effect of salt. That's our reading and we can't say how much it matters here, but it's a good reason to want results from full 24-hour collections.

The design leaves other gaps. It's one snapshot, so it can't show which came first, and the paper describes no repeat urine test to confirm a raised reading. HbA1c, the usual measure of long-term blood sugar, wasn't available, and neither were specific drug classes such as renin-angiotensin blockers and SGLT2 inhibitors, or education and income. The authors say longitudinal studies are needed. They also point to randomised trial evidence that modest salt reduction lowers both blood pressure and albumin excretion in people with type 2 diabetes or impaired glucose tolerance, which says more about cause than this survey can.