not-yet-known not-yet-known not-yet-known unknown 3.6 Predictors of arrhythmogenic diagnosis Only the presence of a first degree AV-block (p=0.01) and the presence of a longer HV interval during EP study (p=0.041) were significant in predicting an arrhythmogenic diagnosis in univariate analysis, as seen in table 3. The presence of any type of QRS conduction abnormality borderline missed significance (p = 0.051), as well as LBBB (p=0.09). A trend towards age at time of implant as predictor of arrhythmogenic event was observed as well (p=0.107). Traffic accident, trauma and number of syncope were not significant. Multivariate analysis by binary logistic regression including age, 1st degree AV-block, any QRS conduction abnormality and HV interval explained merely 6% of variation in the dependent variable, being arrhythmogenic diagnosis. None of the above noted variables contributed in a statistically significant manner in multivariate analysis. Discussion Main findings This prospective study monitored real-world ILR battery longevity in patients with unexplained syncope. The main findings are: With a median time to EOL of 42 months, 99.5% of ILRs reached predefined battery longevity according to manufacturers’ product performance specifications. Time to EOL among patients with same ILR models varied up to 33 months A relevant arrhythmogenic diagnosis for syncope was detected in 27% of the patients (73% sick sinus syndrome, 20% AV-block and 7% VT). Median time to relevant arrhythmogenic diagnosis measured 10 months. Half of the arrhythmogenic diagnoses were detected within the first year, 72% at two years and 100% after 3.5 years. Further monitoring up to 6 years did not yield new arrhythmogenic diagnosis explaining syncope. In multivariate analysis no predictors for arrhythmogenic syncope could be withhold. Real-world battery longevity To the best of our knowledge this is the first study to assess real-world battery longevity of ILR in patients with unexplained syncope. Our study shows that the predefined battery longevity, as specified in the manufacturers’ product performance specifications is met in 99.5% of the patients. However, the time to EOL among patients with same ILR models varied up to 33 months indicating a wide range of ILR monitoring time. The explanation for this variety remains unclear, although different programming settings, number of transmissions via remote monitoring and varying use of symptom triggered ECG storage might have an impact on battery longevity. Such analysis was outside the scope of the current study. However, it is reassuring that despite patient tailored programming, prespecified battery longevity is met in the vast majority of patients. Therefore, ILR battery longevity, as provided by the manufacturer, can be considered as the minimum guaranteed time of ILR monitoring irrespective of ILR programming. Time relationship between arrhythmogenic diagnosis and ILR battery longevity Optimal battery longevity to obtain the maximum yield of ILR monitoring in patients with unexplained syncope is not known. In our study median time to diagnosis approximated 10 months of ILR monitoring, although with a broad interquartile range of 2 to 25 months. The systematic review by Solbiati in 2017 (23), found a mean time to diagnosis of 134 days (=4.5 months), which clearly is shorter than our results. This can be explained by differences in the definition of diagnostic yield, which was defined in our study as a clinically relevant arrhythmia related to the index syncope, while the review by Solbiati included studies describing any arrhythmia. More recent registries show relatively large differences regarding median time to diagnosis: 9 months by Kreimer et al (24), 3 months by Huemer et al (25), 7 months by Smith et al (26), 4 months by Mueller-Leise (27), further underlining the great variability and differences in diagnostic work-up. In the first month, 16% of arrhythmogenic diagnoses were detected. If all patients had undergone a 30 days external Holter monitoring, an arrhythmogenic diagnosis could have been made, obviating the need for ILR implantation. From cost-effectiveness it would be interesting to balance the cost of 30 day continuous external Holter in all patients against avoiding 16% ILR implants. At one and two years of monitoring respectively 54% and 73% of arrhythmogenic diagnoses were made. The Kaplan-Meier curve did not show any flattening in the cumulative diagnostic yield until 43 months, reflecting the unpredictable nature of arrhythmogenic syncope among these patients. Whether a late event is associated with the index syncope, is hard to prove and open to debate, but nevertheless remains clinically relevant as it warranted in most cases appropriate therapy as shown by our study. After 43 months of monitoring, no relevant arrhythmias were detected. This would indicate that a minimum ILR battery longevity of 3.5 years could capture most relevant arrhythmias in patients with unexplained syncope and that extended battery longevity beyond 3.5 years will not further increase the diagnostic yield of ILR monitoring. Diagnostic yield of ILR Diagnostic yield, defined as a relevant arrhythmogenic diagnosis potentially explaining the index syncope but irrespective of recurrent syncope, was 27% in this study. This diagnostic yield is in line with the systematic review of Solbiati (23) in which an arrhythmogenic diagnosis was found in 26.5% of patients. A recent study by Smith et al. (26), executed in a large academic hospital, established a diagnostic yield of 27.8% in syncope, thereby resembling our findings among a comparable population. Other retrospective and observational trials (22, 25, 27-31) have reported a diagnostic yield varying between (5.4 to 55.6%). This difference in diagnostic yield might be explained by the differences in duration of ILR monitoring (1-4 years), patient selection by pre-implant diagnostic work-up, definition of arrhythmogenic diagnosis, heterogeneity of population and shifting practice towards the implantation of ILRs earlier in the diagnostic process. Additional diagnoses with no relation to syncope, were present in 58 patients (18.8%), mostly being atrial fibrillation (AF) and flutter. Therefore automated ILR algorithms to detect these arrhythmias should not be deactivated among these patients, despite the possible higher load of alerts (32), which may be a burden for EP staff members (33). In our population 89% of patients with arrhythmogenic diagnosis (24% of the total population) received a pacemaker or ICD. Recent real-world data show pacemaker implantation in 87% of diagnosed patients, resembling our results (34). In 3% of patients with relevant arrhythmia, a conservative approach was selected by the treating physician, which differs from the prospective ILR registry PICTURE in 2011(15), where just over 60% of events led to implantation of a pacemaker or ICD. This underscores the importance of careful patient selection for ILR implantation, as detection of a significant pause in patients with, for example, vasovagal syncope should not directly lead to pacemaker implantation, reflecting the double-edged sword of ILR monitoring (35). The ideal ILR candidate, is a patient in whom clinically significant arrhythmias have not yet been confirmed, but for whom an arrhythmogenic diagnosis would result in therapeutic measures. Predictors of arrhythmogenic diagnosis Regarding predictors of arrhythmogenic diagnosis, presence of a first degree AV-block was a significant feature (p=0.010) in univariate analysis, as was seen in the study by Ahmed et al (36). Longer duration of His-to-Ventricle (HV) interval in electrophysiology study was found to be a significant predictor as well (p=0.041). These findings were not significant in multivariate analysis. Whether EP studies are routinely performed or not, has major impact on the population receiving ILRs, since patients with syncope and bifascicular block who have significantly prolonged HV interval might directly be guided towards pacemaker (37, 38). The presence of any QRS duration abnormality (>120ms) borderline missed significance which is in line with one of the first large loop recorder registries (Spanish Reveal Registry) (39), but is in contrast to a Korean study by Lee et al (40), where presence of a bundle branch block was significant in multivariate analysis. Presence of RBBB was found to be predictive by Huemer et al (25). Several studies (31, 36, 40-42) also found age to be a predictor of pacemaker implantation, while in our study merely a trend was present (p=0.107). Finally, trauma secondary to syncope, was not statistically significant (p=0.601), which is in contrast to Ahmed et al (36) (p<0.0001), Palmisano (41) (p=0.039), as well as the tendency (p=0.179) seen by Magnusson (31). A possible explanation for this is the lack of a stringent definition of minor/major trauma, as well as the length of in-hospital monitoring after the event, which might be longer in severe trauma. Number of syncope was not found to be significant, in contrast to the Spanish Reveal Registry (39) were recurrent syncope led to more diagnoses than pre-syncope and palpitation. Our study, however, did not differentiate in this matter, as only patients with true syncope were included. The absence of robust predictors proves that the nature of syncope in our patients was truly unexplained and therefore were entitled to an ILR. On the contrary, Kreimer et (24) proposed a risk score (based on LBBB, RBBB, coronary artery disease, previous syncope and AF) thereby identifying patients with a higher probability of an arrhythmogenic diagnosis. Safety of ILR implantation Finally, ILRs result in a low number of serious adverse events such as device extrusion due to ineffective wound sutures (1 in 100 patients) as shown by Deneke et al (43) and Maines et al (29), and is confirmed by our analysis, as only one patient experienced skin protrusion. Four patients opted for early extraction due to local discomfort. No predictors of adverse events were found in our analysis, but previous research found that younger age at implantation lead to more complications (44). Limitations Several limitations are presents in this study, including an unavoidable selection bias as whether or not a patient was to be implanted with an ILR, was decided by the treating physician and therefore may vary accordingly. However, this reflects the real-world clinical setting. In this study, influence of ILR programming on ILR battery longevity was not investigated. The study and the conclusions only pertain to patients with unexplained syncope and not patients with cryptogenic stroke (another population likely to benefit from ILR monitoring). Conclusion This prospective, observational study showed that real-world ILR battery longevity matched industry projected longevity for the vast majority of patients with unexplained syncope. Relevant arrhythmogenic diagnosis was detected in 27% of patients, with a median time to diagnosis of 10 months and reaching a plateau in diagnostic yield after 43 months. As such an ILR battery longevity of minimum 3.5 years is recommended to maximize the diagnostic yield in this population.