Abstract We have tested the accuracy of detection of cardiac arrest in a cardiac event recorder that has the 4x4cm sensor implanted, the recorder external, and the two connected by wireless. Storage and transmission is activated by the patient when symptoms are present or automatically when cardiac arrest is detected, in which case the recorder sounds a loud alarm to alert bystanders. It is the automatic mode that requires database testing of accuracy. Materials We obtained an appropriate database by mapping the chest in 60 ICD patients in sinus rhythm (SR) and induced ventricular fibrillation (VF), and deriving bipolar ECGs from adjacent electrodes located where the electrodes on the subQ sensor would be. (We then implanted the device in 2 dogs and demonstrated the equivalence of ECGs recorded on the skin to subcutaneous ECGs). A blinded selection of 10 of these 60 patients, together with 14 records from the MIT/BIH database, formed a testing set; the others were available for development and training of the detection algorithm. Results In all 60 patients, ECGs from closely-spaced electrodes properly located on the chest appeared suitable for automatic VF detection. Formal testing with the 24-patient database yielded sensitivity and specificity of 100% in detecting VF. Conclusion It appears a 4cm square subcutaneous sensor with electrodes at its corners can reliably monitor cardiac activity, accurately detect the occurrence of cardiac arrest, transmit the ECG to an external receiver, and produce an alarm when a life-threatening arrhythmia is detected.
Read full abstract