The ABC‐Stroke Risk Score and Effects of Atrial Fibrillation Screening on Stroke Prevention: Results From the Randomized LOOP Study

Background The ABC‐stroke score is a risk scheme for prediction of stroke or systemic embolism (SE) in atrial fibrillation (AF). This study sought to examine whether the score could be useful in predicting stroke in AF‐naïve individuals and risk stratifying for AF screening. Methods and Results The LOOP (Atrial Fibrillation Detected by Continuous ECG Monitoring Using Implantable Loop Recorder to Prevent Stroke in High‐Risk Individuals) study randomized 6004 AF‐naïve individuals aged 70 to 90 years with stroke risk factors to either screening with an implantable loop recorder and anticoagulation upon detection of new‐onset AF episodes ≥6 minutes, or usual care. A total of 5781 participants had available ABC‐stroke score at baseline and were included in this secondary analysis: 4170 (72.1%) with an estimated stroke/SE risk ≤1%/year versus 1611 (27.9%) with an estimated stroke/SE risk >1%/year. Having an annual ABC‐stroke risk >1% was associated with stroke/SE, stroke/SE/cardiovascular death, and all‐cause death (hazard ratio, 1.82 [95% CI, 1.44–2.21], 2.17 [95% CI, 1.80–2.62], and 2.19 [95% CI, 1.87–2.56], respectively). For screening with implantable loop recorder versus usual care, no significant reduction in these study outcomes was obtained in any ABC‐stroke risk groups (P>0.0500 for all), with no signal toward interaction (P interaction>0.2500 for all). Similar findings were yielded when assessing the ABC‐stroke score as a continuous variable. Conclusions In an elderly, AF‐naïve population with additional stroke risk factors, a higher ABC‐stroke score could identify individuals with increased stroke risk. However, this risk score may not be useful in pinpointing those more likely to benefit from AF screening and subsequent preventive treatment. These findings should be considered as hypothesis generating and warrant further study. Registration URL: https://www.clinicaltrials.gov; unique identifier: NCT02036450.


Figure S1. Cumulative incidences of primary and secondary outcomes according to TnT subgroups
The figure shows the absolute risks of stroke/SE, stroke/SE/cardiovascular death, and all-cause death in the entire study cohort according to high-sensitivity TnT subgroups.Cumulative incidences were plotted using the Kaplan-Meier estimator for all-cause death and the Aalen-Johansen estimator for other outcomes with death as competing risk.HRs and p-values were determined in cause-specific Cox models adjusted for sex, age, body mass index, weekly alcohol consumption, smoking pack years, hypertension, diabetes, prior stroke, heart failure, valvular heart disease, ischemic heart disease, and peripheral artery disease.

Figure S2. Cumulative incidences of primary and secondary outcomes according to TnI subgroups.
The figure shows the absolute risks of stroke/SE, stroke/SE/cardiovascular death, and all-cause death in the entire study cohort according to high-sensitivity TnI subgroups.Cumulative incidences were plotted using the Kaplan-Meier estimator for all-cause death and the Aalen-Johansen estimator for other outcomes with death as competing risk.HRs and p-values were determined in cause-specific Cox models adjusted for sex, age, body mass index, weekly alcohol consumption, smoking pack years, hypertension, diabetes, prior stroke, heart failure, valvular heart disease, ischemic heart disease, and peripheral artery disease.

Figure S3. The associations of primary and secondary outcomes with continuous TnT
The figure shows the risks of stroke/SE, stroke/SE/cardiovascular death, and all-cause death in the entire study cohort according to high-sensitivity TnT as a continuous variable.Hazard ratios were determined with the median TnT value (13 ng/L) as reference, in cause-specific Cox models adjusted for sex, age, body mass index, weekly alcohol consumption, smoking pack years, hypertension, diabetes, prior stroke, heart failure, valvular heart disease, ischemic heart disease, and peripheral artery disease.The colored areas represent 95% confidence intervals.SE, systemic embolism; TnT, troponin T.

Figure S4. The associations of primary and secondary outcomes with continuous TnI
The figure shows the risks of stroke/SE, stroke/SE/cardiovascular death, and all-cause death in the entire study cohort according to high-sensitivity TnI as a continuous variable.Hazard ratios were determined with the median TnI value (15 ng/L) as reference, in cause-specific Cox models adjusted for sex, age, body mass index, weekly alcohol consumption, smoking pack years, hypertension, diabetes, prior stroke, heart failure, valvular heart disease, ischemic heart disease, and peripheral artery disease.The colored areas represent 95% confidence intervals.SE, systemic embolism; TnI, troponin I.

Figure S5 .
Figure S5.The associations of primary and secondary outcomes with troponin percentile ranks.

Figure S6 .
Figure S6.The associations of ILR screening effects on primary and secondary outcomes with the estimated ABC-stroke risk at baseline.

Figure S7 .
Figure S7.The associations of ILR screening effects on primary and secondary outcomes with troponin percentile ranks.

Figure S8 .
Figure S8.Cumulative incidences of AF diagnosis according to TnT subgroups.

Figure S9 .
Figure S9.Cumulative incidences of AF diagnosis according to TnI subgroups.

Table S2 . Screening effects for ILR vs control according to ischemic stroke subtypes.
Crude event rates are presented as event number per 100 person-years.Hazard ratios were estimated in a univariable cause-specific Cox regression model stratified for ABC-stroke risk group.