Measured salt intake tracks diabetes risk, and self-report never showed it

Design

pooled prospective cohort analysis with Cox models, exposure by two to four 24-hour urine collections per participant

Population

3,173 adults without major chronic disease from three cohorts, mean age 62.2 years, 69% women

Primary outcome

incident type 2 diabetes over a median 13.6 years

Effect

161 cases; highest versus lowest sodium quartile HR 2.66 (95% CI 1.57-4.51); per 1,000 mg/day HR 1.32 (1.16-1.52); potassium HR 0.87 (0.69-1.11); per unit sodium-to-potassium ratio HR 1.26 (1.10-1.45)

Dietary sodium has been studied against diabetes many times using food questionnaires, with inconclusive results — largely because people cannot report how much salt they eat. This analysis used two to four 24-hour urine collections per person, the reference method, in 3,173 adults from three prospective cohorts without major chronic disease, and followed them for a median 13.6 years, during which 161 developed type 2 diabetes.

The highest quartile of sodium excretion carried 2.66 times the risk of the lowest (95% CI 1.57-4.51). Each additional 1,000 mg per day of sodium was associated with 32% higher risk (HR 1.32, 1.16-1.52). Potassium excretion showed nothing (0.87, 0.69-1.11), but each unit rise in the sodium-to-potassium ratio carried 26% higher risk. Adjustment included total energy intake and body mass index.

The authors themselves note that residual confounding may explain part of this: high salt intake tracks with processed food, and processed food brings much else with it. Even so, the exposure measurement here is far better than in the literature it supersedes, and adjusting for BMI and energy makes the crude 'they just ate more' explanation harder to sustain. The practical consequence for an Indian clinic is that salt advice, usually given for blood pressure and often given vaguely, has a second reason behind it — and that the sodium-to-potassium ratio points at the same intervention from both ends: less packaged food, more fruit and vegetables.

• Give specific salt advice: pickles, papad, packaged snacks and namkeen, not just 'less salt'
• Raise potassium through fruit and vegetables rather than substitutes, and avoid salt substitutes in chronic kidney disease
• Do not order a 24-hour urine to guide advice — this is a research method, not a clinical test
• The message overlaps entirely with blood pressure advice, so it costs no extra consultation time
• Residual confounding is likely; treat this as strengthening existing advice rather than as a new mechanism

The statistics, in plain English

A hazard ratio of 2.66 with an interval from 1.57 to 4.51 rests on only 161 diabetes cases, which is why the interval is so wide — the effect is real in direction but its size is uncertain. The per-1,000 mg estimate of 1.32 (1.16-1.52) is more stable because it uses the full range of exposure rather than comparing extremes. Potassium's interval (0.69-1.11) crosses 1.0, so there is no detectable association, not a demonstrated absence of one. Repeated 24-hour collections substantially reduce the measurement error that made earlier questionnaire-based studies unreliable, which is the main reason to take this analysis more seriously than its predecessors.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.