Opportunistic detection of atrial fibrillation during ambulatory blood pressure measurement

Abstract

Atrial FibrillaƟon (AF) is a potenƟally devastaƟng cardiac arrhythmia. Early detecƟon is crucial but difficult due to the someƟmes asymptomaƟc and/or paroxysmal nature of the condiƟon. Non-invasive blood pressure (BP) monitors are one way to opportunisƟcally screen for AF. This thesis aims to invesƟgate the possibility of detecƟng AF during ambulatory blood pressure monitoring (ABPM), a test that involves mulƟple BP measurements over 24 hours. The analyƟcal validity of this technique was assessed using mixed effects models that compared RR intervals from ECG with oscillometric pulse intervals obtained from a brachial cuff. RR intervals could be esƟmated from oscillometric intervals with a random measurement error of < 10 ms. ExisƟng AF detecƟon algorithms were evaluated to determine the impact of this error. Several algorithms had high diagnosƟc accuracy and were robust to small amounts of measurement error. The technique was clinically assessed on 104 paƟents referred for ABPM. Alongside their rouƟne ABPM test, cuff pressure and single-lead ECG were recorded for each parƟcipant. AddiƟonally, 10 paƟents with known AF were also recruited. Oscillometric pulse intervals, idenƟfied during BP measurements, were analysed for signs of AF. When single BP measurements were considered, sensiƟviƟes and specificiƟes > 90% were achievable. When all BP measurements from an ambulatory study were considered even higher sensiƟviƟes and specificiƟes were reached. This suggests AF detecƟon during ABPM is possible, however the results are limited by the study populaƟon. Pulse interval series that gave rise to false posiƟve results were examined and three potenƟal causes were highlighted: presence of non-AF arrhythmias and ectopic beats, bleed valve artefact and movement artefact. These findings should guide further research in this area with parƟcular focus on refining the technical aspects of the technique and carrying out further clinical assessments to address the limitaƟons of this work.