Comparative Diagnostic Performance of Remote Dielectric Sensing and Lung Ultrasound for Detecting Pulmonary Congestion in Stable Ambulatory Heart Failure Patients

Abstract

Aims: To compare the diagnostic accuracy of remote dielectric sensing (ReDS) and lung ultrasound (LUS) for detecting pulmonary congestion in stable ambulatory patients with chronic heart failure using a blinded composite clinical–biomarker reference standard. Study Design: Prospective single-center diagnostic accuracy study. Place and Duration of Study: Single center, conducted between March and November 2024. Methodology: A total of 112 stable outpatients with chronic heart failure underwent sequential clinical assessment, ReDS measurement, and standardized eight-zone LUS during the same visit. Congestion status was adjudicated by a blinded senior cardiologist using a pragmatic composite clinical score combined with NT-proBNP ≥1500 ng/L as the reference standard. The prespecified ReDS threshold was >35%. The LUS threshold of ≥24 B-lines was derived using the Youden index and considered exploratory. The primary outcome was area under the receiver operating characteristic curve (AUC). Secondary outcomes included sensitivity, specificity, likelihood ratios, agreement, and correlation. Results: Fifty-nine patients (52.7%) met criteria for pulmonary congestion. ReDS demonstrated an AUC of 0.831 (95% CI 0.748–0.899), and LUS demonstrated an AUC of 0.845 (95% CI 0.764–0.909), with no significant difference (P = .80). At >35%, ReDS sensitivity was 71.2% and specificity 81.1%. At ≥24 B-lines (exploratory), LUS sensitivity was 69.5% and specificity 83.0%. Overall accuracy was 75.9% for both methods. Inter-method concordance was 83.0% (κ = 0.65). Continuous measures correlated modestly (r = 0.36, P < .001). Inter-observer agreement for thresholded LUS classification was substantial (κ = 0.75). Conclusion: ReDS and LUS demonstrated comparable discrimination for detecting pulmonary congestion in stable ambulatory heart failure patients. Modest correlation and clinically relevant discordance suggest complementary physiological information. The pragmatic composite reference standard and absence of invasive validation represent limitations. Further multicenter validation is warranted.