Continuous measurement of pulse arrival time for identification of blood pressure changes in patients undergoing hemodialysis-a feasibility study

Background and hypothesis. Intradialytic hypotension (IDH) is a common complication during hemodialysis, associated with immediate and long-term risks. Current monitoring methods lack timely warning signals for preemptive intervention. Continuous non-invasive measurement of blood pressure providing real-time insights into hemodynamic changes might enable the early detection of impending IDH. Pulse arrival time (PAT), which can be measured non-invasively and continuously by electrocardiogram (ECG) and photoplethysmography (PPG), is inversely correlated with blood pressure and might serve as such. Thus, this study explores the feasibility of using PAT to track blood pressure changes in end-stage renal disease patients undergoing hemodialysis. Methods. A feasibility study was conducted with 44 patients [mean age 64.9 (18.1 SD) years] undergoing hemodialysis, collecting data at the Klinikum rechts der Isar of the Technical University of Munich, School of Medicine. The PAT was determined by an algorithm for detecting the R-wave in the ECG and the beginning of the pulse curve in the PPG. The study focused on changes in PAT and blood pressure during three predefined time periods of 1 hour during dialysis. Statistical analysis included equivalence testing using the TOST (two one-sided t-tests) approach, correlation analysis, and classification analysis. Results. Patients' mean systolic blood pressure (BP) was 137 (30.1 SD) mmHg, diastolic BP 68.4 (18.4 SD) mmHg and PAT 233 (42.3 SD) ms. The study revealed an inverse association between PAT and BP changes, demonstrating significant correlations [systolic/diastolic blood pressures (SBP/DBP) r = -0.45, P < .001; DBP r = -0.25, P = .003] and equivalence (inverted changes in PAT and SBP (P1 = .002 and P2 = .013) and DBP (P1 < .001 and P2 = .037), respectively). Classification analysis using relative PAT changes showed moderate results for categorizing patients as SBP decliners, risers, or stable, with an F1-score of 0.67 (sensitivity 62.7%; specificity 70.9%) and an area under the curve of 0.70 from receiver operating characteristic analysis. Conclusions. The study demonstrates that tracking BP changes using PAT as a non-invasive method in hemodialysis patients is feasible. However, the reliability of PAT measurements appears to vary among patients, indicating that this method may not be universally applicable. The observed inverse association between PAT and BP changes, especially during the start to middle phase of dialysis, highlights the potential of this technique to complement traditional monitoring methods. These exploratory findings open avenues for further research to validate the utility of PAT, particularly in predicting IDH episodes and assessing its behavior in scenarios of low BP or minimal variations.