Accurate detection of atrial fibrillation using a smartphone remains uncertain: a systematic review and meta-analysis

Abstract

Abstract Introduction Early diagnosis of atrial fibrillation (AF) is essential to reduce complications such as stroke, and improve patient quality of life. Novel screening techniques using smartphone camera photoplethysmography (PPG) can be used for AF detection, but their clinical applicability remains unclear. Purpose To assess the diagnostic accuracy of smartphone PPG compared to conventional ECG for AF detection. Methods We performed a systematic review of MEDLINE, EMBASE, Cochrane library, and other databases (1980-October 2019), including any study or abstract where smartphone finger-tip PPG was compared with a reference ECG (1, 3 or 12-lead). Outcomes were sensitivity (SE), specificity (SP), positive and negative predictive value (PPV; NPV) and overall accuracy. Bivariate hierarchical random effects meta-analysis was performed for studies with confidence intervals for SE and SP, and funnel plots were used to identify publication bias. Study quality was assessed using the established QUADAS-2 tool by two independent graders. Results 1350 publications were screened, of which 17 studies were included in the systematic review (7 full text publications and 10 abstracts), providing 21 comparisons of accuracy for AF detection. Most studies were based in secondary care and small (range n=33 to 1095), with a total of 5469 participants including 1384 with AF. Only 4 studies were multicentre. Smartphone applications used were Cardiio Rhythm, Fibricheck, Preventicus and Heartbeats, with 7 studies not specifying the tool used. Overall SE and SP for AF detection were high, ranging from 76 to 100%, and 85 to 100% respectively. PPV ranged from 54 to 100% and NPV from 77 to 100%, with overall accuracy between 61 and 99%. The meta-analysis included 12 comparisons from 10 studies (n=2714; 936 with AF). The pooled SE was 93% (95% CI 90–96%) and SP 97% (95% CI 95–99%); Figure 1A. QUADAS-2 assessment demonstrated poor quality of studies overall, with a high or unclear risk of bias in at least one domain for all studies. There was clear evidence of publication bias; Figure 1B. Conclusions PPG offers the potential for large scale, non-invasive, patient-led screening of AF. However, current evidence is limited to biased, low quality studies often with unrealistic results for AF detection. These are insufficient to advise clinicians on the true value of current smartphone PPG technology. Figure 1. Meta-analysis & publication bias Funding Acknowledgement Type of funding source: Public grant(s) – EU funding. Main funding source(s): BigData@Heart EU/EFPIA IMI 116074.