Supplementary Material for: Pilot study of a wearable hydration monitor in haemodialysis patients

Abstract
Introduction
We aimed to assess the validity and reproducibility of a wearable hydration device in a cohort of maintenance dialysis patients.
Methods
We conducted a prospective, single-arm observational study on 20 haemodialysis patients between January-June 2021 in a single centre. A prototype wearable infrared spectroscopy device, termed the Sixty device, was worn on the forearm during dialysis sessions and nocturnally. Bioimpedance measurements were performed 4 times using the body composition monitor (BCM) over 3 weeks. Measurements from the Sixty device were compared with the BCM overhydration index (litres) pre- and post-dialysis and with standard haemodialysis parameters.
Results
12 out of 20 patients had usable data. Mean age was 52 ± 12.4 years. The overall accuracy for predicting pre-dialysis categories of fluid status using Sixty was 0.55 [ᴋ= 0.00; 95% CI -0.39 – 0.42]. The accuracy for the prediction of post-dialysis categories of volume status was low [accuracy = 0.33, ᴋ = 0.08; 95% CI = -0.13 to 0.3]. Sixty outputs at the start and end of dialysis were weakly correlated with pre- and post-dialysis weights (r = 0.27 and r = 0.27, respectively), as well as weight loss during dialysis (r= 0.31), but not ultrafiltration volume (r = 0.12). There was no difference between the change in Sixty readings overnight, and the change in Sixty readings during dialysis (mean difference 0.09 ± 1.5 kg), [t(39) = 0.38, p = 0.71].

Conclusion
A prototype wearable infrared spectroscopy device was unable to accurately assess changes in fluid status during or between dialysis sessions. In the future, hardware development and advances in photonics may enable the tracking of interdialytic fluid status.