Sweat sodium composition and sweat loss estimation through wearable sensors and predictive equations in dry and humid hot conditions

David Bandiera^1,2Jean de Bardonnèche²Delphine Margout-Jantac³Léa Dubois^2,3Nisrine El Allaoui³Jonathan Rubio²Jean-Christophe Aubin²Sebastien Racinais^2,3Antonio Tessitore¹Yannis Pitsiladis^1,4*

• ¹Universita degli Studi di Roma Foro Italico, Rome, Italy
• ²CREPS Montpellier-Font Romeu, Montpellier, France
• ³Universite de Montpellier, Montpellier, France
• ⁴Hong Kong Baptist University, Hong Kong, Hong Kong, SAR China

Introduction: Individualized sweat testing is essential for tailoring hydration and nutrition strategies, as water and sodium losses during exercise vary greatly across athletes. The validity of a wearable sweat sensor (S1, Flowbio) and a handheld analyzer (LAQUAtwin, Horiba Advanced Techno) for measuring sweat sodium concentration ([Na+]) was tested against flame photometry (FP). Additionally, whole body sweat loss (WBSL) estimated by the S1 and by a sweat rate calculator (SRC) was compared to the scale-based method. Methods: Twenty-three recreationally active participants (11 males, 12 females) completed two sessions in hot and dry (40°C, 36% rh) and hot and humid (30°C, 81% rh) controlled environmental conditions on a cycling ergometer (74 ± 12 minutes, 1.9 ± 0.4 W/kg). Participants were instrumented with two S1 sensors and absorbent patches placed on each upper arm. Sweat was extracted from patches to measure [Na+] with LAQUAtwin and FP. Nude body mass was measured to the nearest 0.005kg before and after exercise, with fluid intake monitored to determine WBSL. The influence of the method and the condition on the measure of sweat [Na+] and WBSL was investigated with linear mixed-effects models. Results: The estimated marginal means of sweat [Na+] in dry conditions for S1 and LAQUAtwin were equivalent (both 53 mmol/L, p=0.952) and significantly lower than FP (63 mmol, both -10 mmol, p<0.001). No significant interaction effects were observed between methods and conditions. For WBSL, the S1 estimation (1.479 kg) was not different than the scale measure (1.432 kg, 0.047, p=0.624) while the SRC estimation (1.202 kg) was significantly lower than the scale and S1 (both p<0.001), without interactions effects. Conclusion: S1 offers equivalent and more practical collection of sweat [Na+] compared to the LAQUAtwin during indoor cycling ergometer exercise. However, measurements from both devices should currently be interpreted with caution and not considered equivalent to laboratory-grade analyses. Furthermore, S1 is an adequate tool during indoor cycling ergometer exercise to estimate WBSL when scale measurements are impractical, while SRC was found to underestimate fluid loss.