Back

Accelerated biological age linked to high normal serum sodium in general healthcare electronic medical records and NHANES

Rabinowitz, J.; Green, O.; Kwon, D.; Burak, N.; Darawshi, M.; Belsky, D.

2026-08-26 public and global health
10.64898/2026.08.23.26361167 medRxiv
Show abstract

Recent epidemiological studies suggest poor hydration is a modifiable risk factor for aging-related chronic disease. We tested whether serum sodium was associated with accelerated biological aging. We analyzed data from 363,286 adults (18-80 years) from 20 years of electronic medical records from a large healthcare system, as well as 24,611 adults (18-80 years) from National Health and Nutrition Examination Survey (NHANES) continuous (1999-2018). Seven key biomarkers were used to calculate biological age (BA) using the Klemera and Doubal method. We then reran the calculation using only the four variables with highest correlation with age as a robustness check. In both models, there was a significant linear association between age adjusted serum sodium and advanced biological aging, especially in the young cohorts. In the 7-variable model, in the Leumit dataset, the males in the highest sodium level versus the lowest, had a biological age that was 0.88 (95% CI 0.68-1.08) years accelerated and for females 2.32 (2.14-2.51) years. In NHANES dataset biological age of males at the highest sodium level was 1.92 (0.98-2.87) years accelerated as compared to those in the lowest sodium group. For females, the largest difference was for those 41-50 (1 year, .30-1.79). Increased serum sodium in the normal range is associated with accelerated biological aging in the general population, especially among people aged 18-50. Intervention studies are needed to confirm the link between hydration and biological aging.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.