ABC | Volume 111, Nº2, August 2018

Original Article Sampaio et al Monitoring of rhythm, atrial fibrillation and stroke Arq Bras Cardiol. 2018; 111(2):122-131 Table 2 – Monitoring period (hours) by study groups Variables Sample (n = 52) Stroke/TIA (n = 26) Controls (n = 26) P-value Holter Recording time 23.5 ± 0.6 23.4 ± 0.8 23.5 ± 0.4 0.948 Loss (artifacts) 0.6 ± 1.4 0.6 ± 1.7 0.6 ± 1 0.162 PoIP Connection period 156.5 ± 22.5 148.8 ± 25.6 164.3 ± 15.8 0.024 Recording time on the first day 19.2 ± 3.4 19.1 ± 2.5 19.2 ± 4.2 0.514 Recording period 148.8 ± 20.8 143.9 ± 23.3 153.7 ± 16.9 0.080 Loss (artifacts) 50.9 ± 26.2 45.6 ± 26.3 56.1 ± 25.5 0.081 Wilcoxon Mann-Whitney test for independent samples; monitoring period had been planned to be up to 24 hours by Holter andu p to 168 hours (7 days) by PoIP. Comparison of recording periods between Holter and PoIP on the first day: p < 0.001 W difference between the groups, despite higher transmission loss for artifacts among PoIP control subjects. PoIP signal losses were caused by loss of connection (6.8%) and recording signal loss in the server (Table 2). In the first 24 hours, longer period was required for Holter recording (23.5 ± 0.6 hours) as compared with PoIP (19.2 ± 3.4 hours) (p < 0.001). In the stroke/TIA group, PoIP monitoring was started after 5.4 ± 2.7 days of stroke/TIA during hospitalization, and a shorter connection (p = 0.02) and recording period was observed with PoIP (Table 3). Arrhythmias AF was detected in one patient by Holter monitoring and in 6 patients by PoIP in the stroke/TIA group, and in only one control by PoIP. Regarding other supraventricular arrhythmias, further cases of nonsustained AT and frequent AT or SVES were identified by Holter monitoring in patients aged 65 years or older in the stroke/TIA group (p = 0.04 and 0.04, respectively). In two cases, differential diagnosis of AT and nonsustained AF required revision by the two observers (RFS and EBS). It is worth mentioning, however, that patients who had AF also had AT, and therefore, a misinterpretation of electrocardiographic tracings would not affect the results, due to the occurrence of both conditions in the same patient. PoIP monitoring revealed that there were no significant differences between the groups regarding tachycardia (Table 4), and all patients with AF also had AT. Comparisons between Holter and PoIP results showed a higher proportion of AT identified by PoIP in both stroke/TIA (p=0.004) and control (p=0.02) groups. Also, PoIPmonitoring revealed a higher proportion of patients with frequent AT or SVES in the stroke/TIA (p = 0.01) and control (p = 0.02) groups considering total monitoring period, but no difference was found between the groups in the first 24 hours. Discussion In the present study that included 52 patients older than 59 years, prolonged rhythm monitoring was performed in 26 patients with acute cerebrovascular events, and initiated only 5 days (mean) after the event. The main findings were high prevalence of arterial hypertension and diabetes mellitus, some connectivity problems and problems related to PoIP signals’ recording, and similar profile of cardiac arrhythmias between the study groups. The most frequent comorbidities were arterial hypertension (84.6%) and diabetes mellitus (51.9%), with similar distribution between the groups studied. This result was expected, since these variables were used in the PSM model, and both comorbidities are also included in the CHADS2 and CHA2DS2-VASc scores. Although these scores provide simple methods for predicting an individual risk of ischemic stroke, the risk estimated by these instruments represent only part of the overall risk (statistical agreement of 0.5). In other words, not all patients with a CHADS2 score equal to 0 or 1 have a low risk, and hence the clinical decision not to anticoagulate patients based only on this score may be erroneous. Despite the higher specificity of a CHA 2 DS 2 -VASc score ≥ 2, this still underestimates the risk. 15 For this reason, we analyzed with particular interest the higher prevalence of smoking in stroke/TIA patients (p=0.038), especially among patients older than 65 years (p = 0.045). A recent meta-analysis showed that smoking is associated with a modest increase in AF, and that quitting smoking reduces but not eliminates the associated risk of the disease. 16-18 Nevertheless, the addition of smoking to the score does not improve the risk prediction of stroke or TIA. 19 Monitoring by mobile phone Although PoIP and Holter monitoring systems had similar performance in the first 24 hours, there were problems with signal connection and transmission during PoIP monitoring. Loss of connection with the cell phone provider accounted for 6.8% of total monitoring time, shorter recording time in the server and lower data losses due to artifacts (Table 2). Loss of connectivity was greater in hospitalized (stroke) patients (p = 0.024). For better interpretation of this result, we measured the strength of the provider signal using the Network Monitor® 126