Objectives Controversial remains in regard to the usefulness of OptiVol alerts in HF prediction. We aimed to investigate the reliability and predictive potency of OptiVol FI alone as well as multiple device diagnostics in HF prediction in real world practice.

Methods Patients who underwent ICD or CRT-D implantation with feature of intrathoracic impedance monitoring in our single centre from Sep 2010 to Jan 2013 were included in this observational study. In-office follow-up was required in month 3, 6, 12 and every 12 month after that. Unscheduled follow-up was arranged upon any alert reported. Worsening HF was defined as hospitalisation or presentation of symptoms or signs of HF.

Results 50 patients with 26 in CRT-D and 24 in ICD enrolled in this study. 121 visits collected during average 273 ± 161 days (90-730) follow-up. Overall 62 alerts and 17 adjudicated HF events were observed. 14 alarm events were associated with worsening HF. Sensitivity and positive predictive value by nominal threshold 60Ω-d was 82.4% and 22.6% respectively. The unexplained alerts rate was 77.4% and 1.34 per person year. Chi-square test indicated no significant correlation between OptiVol alerts and cardiac worsening events. Logistic regression analysis showed that only HRV was a significant predictor to cardiac exacerbation (p = 0.001). LVEF and daily activity contributed little to the prediction. HRV 80 ms in combination with OptiVol alert could improve the PPV to 82.4% and negative predictive value to 89.1%.

Conclusions OptiVol FI predicted HF with high unexplained events rate and low PPV. HRV was an important independent predictor to worsening heart failure. OptiVol alert in combination with HRV in analysis could benefit cardiac worsening early identification.