ROLE OF PERIPHERAL PULSE MEASUREMENT IN DIAGNOSIS OF PAROXYSMAL ATRIAL FIBRILLATION AFTER ISCHEMIC STROKE

ROLE OF PERIPHERAL PULSE MEASUREMENT IN DIAGNOSIS OF PAROXYSMAL ATRIAL FIBRILLATION AFTER ISCHEMIC STROKE

By

Fathy M Afifi, Khaled M Sobh, Mohammad A Morsy*,

Hussein Awad El-Gharieb, Mahrous I Seddeek

and Ahmed A Abd Al-Raheem

Departments of Neurology and Cardiology*, Al-Azhar Faculty of Medicine

ABSTRACT

Background: Ischemic stroke comprises 85% of all strokes. Atrial fibrillation (AF) is the most common cause of cardioembolism and a leading cause of ischemic stroke.  The diagnosis of paroxysmal AF (PAF) after cerebral ischemia is often difficult to establish because paroxysmal episode may be short and may elude stroke unit monitoring in the acute phase.

Objectives: Detection of PAF after ischemic stroke and investigating feasibility and diagnostic accuracy of measurement of the peripheral pulse at the radial artery as a simple non-invasive screening tool for PAF in patients after acute ischemic stroke.

Patients and methods: This study was carried on 52 patients with acute ischemic stroke admitted at Neurology Department, Al-Azhar University Hospitals from 1^st November 2014 till the 1^st of June 2015. All patients were submitted to complete medical history, complete physical examination, full neurological examination, routine laboratory investigations, CT scan of the brain, conventional ECG registration and echocardiography. All patients and their relatives were undergoing on training program for peripheral pulse measurement and standardized introduction on the pathophysiology of cardioembolic stroke.

Results: Patients with PAF were 11 (21.2 %), patients with AF were 22 (42.3 %), mean age ± SD of patients with AF was 60.273 ± 10.743 years and that of patients with PAF was 59.682 ± 8.283 years. Patients had PAF were 18% males, 82% females, 90.9% had a history of hypertension, 91% had diabetes mellitus, 72.7% had hyperlipidemia, 9% were smokers, 72.7% had IHD (ischemic heart disease) and 81.8% had a history of previous stroke. Patients had AF were 22.7% males, 77.3% females, 90.9% had a history of hypertension, 72.7% had diabetes mellitus, 50% had hyperlipidemia, 13.7% were smokers, 86.4% had IHD  and 90.9% had a history of previous stroke. AF or PAF related ischemic stroke was associated with older age, female gender, hypertension, diabetes mellitus, no smoking and history of previous stroke. Hypertension and diabetes mellitus were significant predictors of newly diagnosed (NDAF) AF or PAF in acute ischemic stroke. Measurements by health care professionals had sensitivity 81.82, specificity 95.12, positive predictive value 81.82, and negative predictive value 95.12 with accuracy 92.31. Measurements by patients themselves had sensitivity 42.86, specificity 53.57, positive predictive value 18.75, and negative predictive value 78.95 with accuracy 51.43. Measurements by the relatives of the patients had sensitivity 55.56, specificity 72.22, positive predictive value 33.33, and negative predictive value 86.67 with accuracy 68.89

Conclusion: Early diagnosis of AF facilitated the detection of patients who should receive oral anticoagulant treatment to decrease risk of stroke. Primary prevention of cardioembolic stroke can be achieved by early discovering AF patients. Patients with diabetes, hypertension, IHD or history of previous stroke are at increased risk of AF or PAF. We suggest that all patients with AF were PAF at first and then converted to established AF. Early detection and management of PAF reduce the occurrence of established AF.

Key Words: stroke, atrial fibrillation.

INTRODUCTION

     Stroke is defined as acute focal loss of perfusion to a vascular territory of the brain, resulting in ischemia with corres-ponding loss of neurologic function. Ischemic stroke comprises 85% of all strokes (Hill, 2005).  Acute stroke is one of the leading factors of morbidity and mortality worldwide. Stroke ranks as third most common cause of death in industrialized countries (Hacke et al., 2003).

      Atrial fibrillation is the most common cause of cardioembolism and a leading cause of ischemic stroke.  The diagnosis of PAF after cerebral ischemia is often difficult to establish because paroxysmal episode may be short and may elude  stroke unit monitoring in the acute phase. However, the diagnosis of AF is of particular clinical relevance because antithrombotic treatment significantly reduces the risk of recurrent embolism and death (Kallmünzer et al., 2014).

    The detection rate of AF after stroke has progressively increased by extending the duration and intensity of cardiac monitoring for this purpose, innovative medical devices and implantable event recorder have been suggested as long term diagnostic tools after cryptogenic stroke.  However, high  socioeconomic  expense , malcompliance and the invasiveness of some of these approaches  currently limit their use to a minority of affected patient , while the growing number of stroke survival is lacking access to diagnostic screening tools (Cotter  et al ., 2013). For primary prevention  , the measurement of peripheral pulse is currently the evidence based method of choice for screening among individuals aged 65 years  or older and the only diagnostic tool recommended  by international guidelines ( Kamel et al ., 2013 ).

     The cause of ischemic stroke remains unknown in approximately one third of patients. Some proportion of these cryptogenic strokes may be caused by undetected AF. Because asymptomatic and paroxysmal AF is common and increases the risk of stroke as much as chronic AF.  As a result, clinical guide-lines recommend at least 24 hours of cardiac monitoring after stroke as an opportunity to make a delayed diagnosis of PAF and in practice, patients are commonly admitted to telemetry beds for 24 to 48 hours of continuous ECG. However, there is no agreement on the optimum type and duration of monitoring. Despite this uncertainty, detection of AF is important, because treatment with anticoagulation decreases the annual risk of recurrent stroke by two thirds (Kamel et al., 2009). The present study was a trial to detect PAF after ischemic stroke using peripheral radial pulse.

Patients AND METHODS

     This study was carried on 52 patients with acute ischemic stroke admitted at the Neurology Department in Al-Azhar University Hospitals from 1^st November 2014 till the 1^st of June 2015. All patients were submitted to the following:

     Complete medical history, complete physical examination, full neurological examination, routine laboratory investiga-tions including; complete blood count, blood urea, creatine,  lipid profile, random blood sugar, AST, ALT, albumin, bilirubin (direct, indirect ), PT, PTT and INR , ESR, CRP, uric acid, CT scan of the brain,  connection to multimodal monitor-ing system including 6 leads continuous ECG registration and echocardiography.

     All patients or their legal represen-tatives gave written informed consent before inclusion in our study. All patients and their relatives were undergoing on training program for peripheral pulse measurement and standardized introduc-tion on the pathophysiology of cardioem-bolic stroke. We guided them to standardized pulse measurement at the left radial artery 3cm proximal to the wrist. In addition to personal training, the patients and relatives received schematic instruction materials and stopwatch.

     The training had two educational objectives:

● Learning to distinguish between rhy-thmic pulse sensations (normal or absolute arrhythmic sensation, suspected AF).

● Performing a measurement of the pulse rate using stopwatch.

- The training was considered successful if 2 subsequent measurements were accomplished reliably. In case of tiring or insufficient learning success after 30 minutes, the tutorial was interrupted and continued by the next day. Tutorial without success on the second day were considered as a failure and led to cessation of further measurements.

 - Three different modalities of measure-ments of peripheral pulse were investi-gated: 1-measurement by health care professional with positive experience on cardiovascular assessment of patients. 2- Measurement by a relative of the patient after successful completion of the tutorial. 3- Self measurement by the patient after successful completion of the tutorial. Measurements of participants as well as our measurements were then compared with simultaneous blinded ECG to evaluate diagnostic accuracy parameters.

Inclusion criteria: All patients with acute ischemic stroke.

Exclusion criteria: Patients with cardiac pacemaker were excluded from our study.

Statistical Analysis: The collected data were tabulated and analyzed using SPSS version 21 software (Spss Inc, Chicago, ILL Company).  Categorical data were presented as number and percentages while quantitative data were expressed as mean and standard deviation. Chi square test (X²), “Z” test and student “t” tests were used as tests of significance, Odds ratio (OR) and the corresponding 95% CI were calculated when applicable. Logisitic regression analysis was done to determine the significant predictors (independents) of stroke with AF. The accepted level of significance in this work was stated at 0.05. Sensitivity, Specificity, PPV, NPV and accuracy were used.

RESULTS

     Fifty two patients with acute ischemic stroke were included in this study. 50 % of the sample were males and 50 % were females, with mean age of 57.3 years ranging from 34 – 85 years. Prevalence of PAF among the study group was 21.2% while 78.8% of patients were negative for PAF. Mean age of patients with PAF wass 59.68 with SD ± 8.28, and the mean age of patients without PAF was 55.47 with SD ± 7.5. 81.82 % of the females were stroke patients with PAF, while 18.18 % of the males were stroke  patient with PAF, and this difference was statistically significant. Percentage of occurrence of PAF in nonsmoker patient was 90.91%, while in smoker patients was 9.09 %, and this difference was statistically significant. Percentage of occurrence of PAF in diabetic patient was 90.91%, while in non-diabetic patients was 9.09 % and this difference was statistically significant.  Percentage of occurrence of PAF in patient with old stroke was 81.82%, while in patients with no history of old stroke was 18.18% and this difference was statistically significant. Percentage of occurrence of PAF in patient with hypertension was 90.91%, while in patients with no hypertension was 9.09% and this difference was statistically significant. Percentage of occurrence of PAF in patient with history of IHD was 72.73 %, while in patients with no history of IHD was 27.27%, and this difference was statistically significant (table 1).

     Prevalence of AF among the study group was 42.3%, while 57.7 % of patients were negative for AF. Mean age of patients with AF was 60.273 with SD ± 10.743, and the mean age of patients without  AF was 56.439 with SD ± 7.068 and this was statistically insignicant. 77.27% of the females were stroke patients with AF while 22.73% of the males were stroke patient with AF, and this difference was statistically significant. Percentage of occurrence of AF in non-smoking patient was 86.36 %, while in smoking patients was13.64% and this difference was statistically significant. Percentage of occurrence of AF in diabetic patient was 72.73 %, while in non-diabetic patients was 27.27 % and this difference was statistically significant (table 2).

      Measurement of PAF by health care professional was with sensitivity 81.82, specificity 95.12, positive predictive value 81.82, and negative predictive value 95.12 with accuracy 92.31. Measurement of PAF by the patients themselves was with sensitivity 42.86, specificity 53.57, positive predictive value 18.75, and negative predictive value 78.95 with accuracy 51.43. Measurement of PAF by the patients themselves was with sensitivity 55.56, specificity 72.22, positive predictive value 33.33, and negative predictive value 86.67 with accuracy 68.89 (table 3).

Table (1): Demographic data of the studied sample and other variables:

Table (2): Relationship between AF and other parameters.

Table (3): Measurements of PAF by different subjects.

DISCUSSION

     In our study, the percentage of paroxysmal AF detected by health care professional measurements using peri-pheral pulse measurement and continuous ECG monitoring was 21.2%.This was in agreement with Tayal et al. (2008) with 23%, Elijovich et al. (2011) with 20% and Miller et al. (2013) with 17.3%. However, our results disagreed with Guanalp et al. (2006) with 34.8 %, Higgins et al. (2013) with 42 %, and Merce et al. (2013) with 35.7%. This disagreement came due to the difference in the advanced techniques for detection of paroxysmal AF as ambulatory holter monitoring, mobile cardiac out-patient telemetry, external loop recording and implantable loop recording were used.

In our study, the mean age of stroke patients with PAF and AF was the higher than patients without PAF or AF. These results were in agreement with Alberts and Eikelboom (2012) who stated that patients with atrial fibrillation-related stroke tend to be older than other patients with stroke and Amin (2013) who found that the risk of AF or PAF related stroke increase with aging.

     The present study showed that AF related stroke tends to be more in females than males, and it showed significant difference between the two groups. These results were in agreement with Kamel et al. (2009) who said that there are increased risks of PAF and AF related strokes in females.

       In our study concerning the history of hypertension, it was found in significant high percent in stroke patients with and without AF. These results were in agreement with Di Legge et al. (2012) who stated that hypertension (HTN) is the single most important modifiable risk factor for stroke. HTN contributes to 60% of all strokes. Also in agreement with Alberts and Eikelboom (2012) who stated that the major risk factors for stroke in individuals with atrial fibrillation is history of hypertension. On the other hand, Kamel et al. (2009) found that Detection of AF was more associated with the absence of hypertension in stroke patients as hypertension is strongly associated with lacunar and atherothrom-botic strokes. This difference may be explained by difference in genetic and ethnic state between studied subjects.         

      In our study, diabetes mellitus was detected in 73% of stroke patients with AF compared to 40% of stroke patients without AF, and this difference was statistically significant. Also, diabetes mellitus was detected in 91% of stroke patients with PAF compared to 44% of stroke patients without PAF, and this difference was statistically significant. Our results were in agreement with Alberts and Eikelboom (2012) who found that the diabetes is one of the major risk factors for stroke in individuals with atrial fibrillation and in concordance with findings of Kallmünzer et al.( 2014) who reported that diabetes was found more frequently among stroke patients with cardiac arrhythmias mainly AF. These results also agreed with Sposato et al.(2012) who stated that diabetes mellitus is independent predictors of newly diagnosed AF (NDAF) in stroke patients as diabetes mellitus might contribute to the impaired cardiovascular autonomic function seen in stroke survivors and  diabetic dysautonomy could explain this finding by the parasympathetic and sympathetic phenomena are possible mechanisms of paroxysmal AF.

     We found in our study that percentage of occurrence of PAF and AF in non-smoking patients was higher than smoking patients, and this difference was statistically significant. This was supported by the study of Bansil and Karim (2004) who found that stroke patients with AF were less likely to be smokers and may be related to smokers having diffuse atherosclerotic disease and developing strokes earlier because of other causes. Also, Ruigómez et al. (2009) stated that there is a relative  risk in non-smokers. Bugnicourt et al. (2013) also reported that stroke patients with NDAF were less likely to be active smokers.

     In our study, concerning the history of previous stroke, it was found in 90.91% and 53.33% of stroke patients with and without AF respectively, and the difference was statistically significant. Also, it was found in 81.82% and 0% of stroke patients with and without PAF respectively, and the difference was statistically significant. This was supported by the study of Alberts and Eikelboom (2012) who found that the major risk factors for stroke in individuals with atrial fibrillation are previous stroke. In concordance with our study, Bugnicourt et al. (2013) reported that history of previous stroke was associated with NDAF in ischemic stroke patients.

     Furthermore, there was a significant relation between the IHD in stroke patients with and without AF. In our study, IHD was detected in 86.36 % and 43.33% of stroke patients with and without AF respectively. Also, we found that there is a significant relation between IHD in stroke patients with and without PAF. IHD was detected in 72.73 % and 14 34.15% of stroke patients with and without PAF respectively.

     These results were in agreement with Bugnicourt et al. (2013) who mentioned that previous coronary artery disease  was associated with NDAF in stroke patients. Also, Kallmünzer et al. (2014) revealed that structural heart disease (IHD) were occurred more frequently among stroke patients with cardiac arrhythmias mainly AF.

     Throughout our study, we got three measurements:

● Measurements by health care profes-sionals were with sensitivity 81.82, specificity 95.12, positive predictive value 81.82, and negative predictive value 95.12 with accuracy 92.31. This agreed with Kallmünzer et al. (2014) who found that sensitivity was 96.5, specificity 94, positive predictive value 82.1 and negative predictive value 98.9 with accuracy 88.7.

● Measurements by patients themselves were with sensitivity 42.86, specificity 53.57, positive predictive value 18.75, and negative predictive value 78.95 with accuracy 51.43. Number of participating patients was only 35 because the rest of patients failed to continue in our tutorial. This agreed with Kallmünzer et al. (2014) who found that sensitivity was 54.1, specificity 96.2, positive predictive value 76.9 and negative predictive value 90 with accuracy 63.5.

● Measurements by the relatives of the patients were with sensitivity 55.56, specificity 72.22, positive predictive value 33.33, and negative predictive value 86.67 with accuracy 68.89. Number of patients was 45 because the rest of realtives failed to continue in our tutorial. This agreed with Kallmünzer et al. (2014) who stated that sensitivity was 76.5, specificity 92.9, positive predictive value 78.8 and negative predictive value 91.9 with accuracy 77.6.

CONCLUSION

● Early diagnosis of AF facilitates the detection of patients who should receive oral anticoagulant treatment so decrease risk of stroke.

● The low educational level of the patients and their relatives made their training difficult and their peripheral pulse measurements were less accurate than that of health care professionals.

● Patients with diabetes, hypertension, IHD or history of previous stroke are at increased risk of AF or PAF.

● All patients with AF were PAF at first and then converted to established AF. So, early detection and management of PAF reduces the occurrence of established AF.

RECOMMENDATIONS

● Prolonged continuous cardiac monitor-ing after acute ischemic stroke are recommended at least for 7-10 days for diagnosis of silent AF or PAF.

● Good training of patients or their relatives to measure peripheral pulse make their measurements more accurate and can be used as a powerful tool to select the patients in need for further cardiological monitoring.

● We recommend the use of ambulatory holter monitoring, mobile cardiac outpatient telemetry, external loop recording or implantable loop recording to get results that are more accurate.

●  More studies should be done with large numbers of patients.

REFERENCES

1. Alberts M and Eikelboom J (2012): Anti-thrombotic therapy for stroke prevention in non-valvular atrial fi brillation. Lancet Neurol . 11 (12): 1066–1081.

2. Amin  A (2013): Oral anticoagulation to reduce risk of stroke in patients with atrial fibrillation: current and future therapies. A+current+and+future+therapies++++Alpesh+Amin+2013" Aging."Clin HYPERLINK "A+current+and+future+therapies++++Alpesh+Amin+2013%22ClinHYPERLINK%20%22http://www.ncbi.nlm.nih.gov/pubmed/?term=Oral+anticoagulation+to+reduce+risk+of+stroke+in+patients+with+atrial+fibrillation%3A+current+and+future+therapies++++Alpesh+Amin+2013" HYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/?term=Oral+anticoagulation+to+reduce+risk+of+stroke+in+patients+with+atrial+fibrillation%3A+current+and+future+therapies++++Alpesh+Amin+2013"IntervHYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/?term=Oral+anticoagulation+to+reduce+risk+of+stroke+in+patients+with+atrial+fibrillation%3A+current+and+future+therapies++++Alpesh+Amin+2013" Aging."HYPERLINK "A+current+and+future+therapies++++Alpesh+Amin+2013%22ClinHYPERLINK%20%22http://www.ncbi.nlm.nih.gov/pubmed/?term=Oral+anticoagulation+to+reduce+risk+of+stroke+in+patients+with+atrial+fibrillation%3A+current+and+future+therapies++++Alpesh+Amin+2013" HYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/?term=Oral+anticoagulation+to+reduce+risk+of+stroke+in+patients+with+atrial+fibrillation%3A+current+and+future+therapies++++Alpesh+Amin+2013"IntervHYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/?term=Oral+anticoagulation+to+reduce+risk+of+stroke+in+patients+with+atrial+fibrillation%3A+current+and+future+therapies++++Alpesh+Amin+2013" Aging."IntervHYPERLINK "A+current+and+future+therapies++++Alpesh+Amin+2013%22ClinHYPERLINK%20%22http://www.ncbi.nlm.nih.gov/pubmed/?term=Oral+anticoagulation+to+reduce+risk+of+stroke+in+patients+with+atrial+fibrillation%3A+current+and+future+therapies++++Alpesh+Amin+2013" HYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/?term=Oral+anticoagulation+to+reduce+risk+of+stroke+in+patients+with+atrial+fibrillation%3A+current+and+future+therapies++++Alpesh+Amin+2013"IntervHYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/?term=Oral+anticoagulation+to+reduce+risk+of+stroke+in+patients+with+atrial+fibrillation%3A+current+and+future+therapies++++Alpesh+Amin+2013" Aging." Aging;8:75-84.

3. Bansil, S and Karim H (2004): Detection of atrial fibrillation in patients with acute stroke. Journal of Stroke and Cerebrovascular Diseases. 13(1): 12-15.

4. Bugnicourt  J, Flament M and Guillaumont  M (2013): Predictors of newly diagnosed atrial fibrillation in cryptogenic stroke: a cohort study. Euro J Neurol. 20 (10):1352-9.

5. Cotter P, Martin P, Ring L, Warburton E, Bilham M and Pugh P (2013): Incidence of atrial fibrillation detected by implantable loop recoredesr in unexplained stroke. Neurology. 80:1546-50.

6. Di-Legge S, 22Koch,%20G"Koch G, 22Diomedi,%20M"Diomedi M, 22Stanzione,%20P"Stanzione P and 22Sallustio,%20F"Sallustio F (2012): Stroke prevention: managing modifiable risk factors. StrokeHYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/23213626/" Res Treat., 2012: 391538.

7. ElijovichHYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/?term=Elijovich%20L%5BAuthor%5D&cauthor=true&cauthor_uid=19426887" HYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/?term=Elijovich%20L%5BAuthor%5D&cauthor=true&cauthor_uid=19426887"L, 22Josephson%20SA"Josephson S, Fung G and Smith W (2011): Intermittent atrial fibrillation may account for a large proportion of otherwise cryptogenic stroke: a study of 30-day cardiac event monitors. J Stroke HYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/19426887"CerebrovascHYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/19426887" Dis. 18(3):185-9

8. Gunalp M, Atalar E and Coskun F (2006): Holter monitoring for 24 hours in patients with thromboembolic stroke and sinus rhythm diagnosed in the emergency department. Adv Ther.,23:854-60.

9. Hacke W, Kaste  M,  Bogousslavsky J and  Brainin M (2003): European Stroke Initiative Recommendations for Stroke Management - Update. Cerebrovasc. Dis. 16:311–37

10. Higgins P, Macfarlane B, Dawson J, and Mcinnes G (2013):  Noninvasive cardiac event monitring to detect atrial fibrillation after ischemic stroke: randomized,controlled trial. Stroke, 44:2525-31.

11. Hill M. (2005): Diagnostic Biomarkers for stroke: A stroke neurologist’s prespective. Clinical Chemistry. 51: 2001-2002.

12. Kallmünzer B, Bobinger T, Kahl N, and  Kopp M (2014): Peripheral pulse measurement after ischemic stroke: A feasibility study. Neurology.12;83 (7):598-603.

13. Kamel H., Elkind M, Bhave P, and Navi B (2013): Paroxysmal supraventricular tachy-cardia and the risk of ischemic stroke. Stroke. 44(6):1550-4.

14. Kamel H,  Lees K, Lyden P and Teal P (2009): Delayed Detection of Atrial Fibrillation after Ischemic Stroke. Journal of Stroke and Cerebrovascular Diseases. 18 (6): 453-457.

15. Merce J, Garcia M Mundel M and Freudenberger T (2013): Insertable cardiac event recorded in detection of atrial fibrillation after cryptogenic stroke:an audit report. Stroke. 44:2007-09.

16. Miller D, Khan M, Schuntz L, Simpson J and Katramados A (2013): Outpatient cardiac telemetry detect a high rate of atrial fibrillation in cryptogenic stroke. J neurol Sci. 324:57-61.

17. Ruigómez  A, Johansson  S, Wallander M and  Edvardsson N (2009): Risk of cardio-vascular and cerebrovascular events after atrial fibrillation diagnosis. International Journal of Cardiology, 136(2): 186-192.

18. SposatoHYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/?term=Sposato%20LA%5BAuthor%5D&cauthor=true&cauthor_uid=20727789" HYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/?term=Sposato%20LA%5BAuthor%5D&cauthor=true&cauthor_uid=20727789"L, Klein F, Jáuregui A,  Ferrúa M and  Klin P (2012): Newly diagnosed atrial fibrillation after acute ischemic stroke and transient ischemic attack: importance of immediate and prolonged continuous cardiac monitoring. J Stroke HYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/20727789"CerebrovascHYPERLINK "http://www.ncbi.nlm.nih.gov/pubmed/20727789" Dis.21(3):210-6.

19. Tayal  A, Tian M and Kelly K (2008): Atrial fibrillation detected by mobile cardiac outpatient telemetry in cryptogenic TIA or stroke. Neurology .71:1696-701.