UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Autonomic nervous system parameters to predict the occurrence of ischemic events after transient ischemic… Guan, Ling 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2017_november_guan_ling.pdf [ 7.16MB ]
Metadata
JSON: 24-1.0357024.json
JSON-LD: 24-1.0357024-ld.json
RDF/XML (Pretty): 24-1.0357024-rdf.xml
RDF/JSON: 24-1.0357024-rdf.json
Turtle: 24-1.0357024-turtle.txt
N-Triples: 24-1.0357024-rdf-ntriples.txt
Original Record: 24-1.0357024-source.json
Full Text
24-1.0357024-fulltext.txt
Citation
24-1.0357024.ris

Full Text

  Autonomic Nervous System Parameters to Predict the Occurrence of Ischemic Events after Transient Ischemic Attack or Minor Stroke by  Ling Guan  MBBS, Sichuan University-West China School of Medicine, 2008 M.Sc., The University of British Columbia, 2012  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY  in  THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Experimental Medicine)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  October 2017  © Ling Guan, 2017ii  Abstract  Background  Stress response is tightly regulated by the autonomic nervous system (ANS) which can be measured by heart rate variability (HRV). Traditional risk factors and acute triggers for ischemic stroke and transient ischemic attack (TIA) are considered as chronic and acute stressors, respectively. These risk factors all contribute to the recurrent ischemic events and are related to ANS dysfunction. The moderate predictive value of ABCD2 score may be due to assessing a limited number of stressors.  Aims, Objectives and Hypotheses We proposed to 1) assess whether HRV parameters, as markers of ANS function and stress, can predict secondary ischemic events after TIA or minor stroke, and 2) compare the HRV-based predictive tools with ABCD2 score. We expected that using HRV indicators can enhance the prediction of ischemic events.   Methods This is a prospective cohort study. Patients developed TIA or minor stroke within 48 hours were eligible. The main study variables included: ABCD2 score, HRV assessment from 24-hour Holter recording, and psychological stress. HRV measurement included calculations of both absolute values and changes of HRV frequency-domain parameters: high frequency (HF), normalized HF, HF+low frequency, and total power. Patients were followed for 90 days to assess iii  the development of outcome events. Logistic regression was employed for data analyses. Area under the curve (AUC) and diagnostic tests were used to assess models’ predictive power.  Results Final analyses include data collected from 201 patients. The most useful HRV predictors were Daytime HF changes (AUC=0.70) and Morning HF value (AUC=0.61). AUCs for the Best Stress Model and the Most Practical Model were 0.82 and 0.76, respectively, which were significantly higher than AUC of ABCD2 score (AUC=0.63), p <0.05. The optimal cut-off points for Daytime HF changes and Morning HF might be increase of 12.5% and 50 ms2, respectively. The exploratory models that involved both values and changes of HF had AUCs ≥0.82.   Conclusions Assessing the effects of stress on ANS may be an innovative way to stratify the risk of TIA or minor stroke. Models using HRV parameters, especially HF, provide superior predictive values to ABCD2 score. Future research is needed to validate these results.    iv  Lay Summary  ABCD2 score, a classic tool to identify patients’ risk of ischemic stroke after transient ischemic attack (TIA), has only moderate predictive value. Most ischemic stroke risk factors are body stressors, which affect the autonomic nervous system’s (ANS) function. Our study aims to assess the predictive value of ANS-related parameters (the heart rate variability, HRV) to identify patients who are at a higher risk of developing ischemic events after TIA or minor stroke. We assessed personal information, ABCD2 score, HRV data (by 24-hour Holter recording) and psychological stress in 201 patients with TIA or minor stroke. Patients were followed for 90 days to assess the occurrence of secondary ischemic events and death. We found that HRV-based stress models significantly improved the predictive value of ABCD2 score. Assessing ANS function may represent an innovative way to identify individual’s risk of developing secondary ischemic events after initial TIA or minor stroke.   v  Preface  The overall study was designed and implemented under the leadership of Dr. Jean-Paul Collet. The study was also supported by Dr. Garey Mazowita, Dr. Devin Harris, Dr. Victoria Claydon, Dr. Rollin Brant, Dr. Yongjun Wang, Dr. Yilong Wang and Dr. Hui Lin. This study was approved by University of British Columbia Children’s & Women’s Clinical Research Ethics Board (CW14-0196 / H14-00435, July 8, 2014) and Beijing Tiantan Hospital affiliated to Capital Medical University Clinical Research Ethics Board (KY2014-015-02, May 6, 2014).   Ling Guan’s contribution included:  1. Developing the whole study protocol;  2. Designing the “Case report form (CRF)”, “Consent form”, “Screening form”, “Recruitment record”, and “Follow up record”; 3. Applying the two ethics approvals (mentioned above) for this study; 4. Contacting patients, obtaining informed consent, recruiting and monitoring study patients;  5. Organizing all data collection for every patient, including personal and clinical information (clinical symptoms, ABCD2 score, imaging results, preliminary and final diagnoses, physiological indices and treatment, historical and present diseases, and family history), 24-hour Holter and HRV data recording, psychological stress assessment, as well as 3 months’ follow up; 6. As the study manager, responsible for organization and daily management;  7. Designing the data analyses strategy and conducting all data analyses and interpretation, including HRV analyses, PSS score analyses and clinical assessment;  vi  8. Performing all statistical analyses;  9. Writing up of all this thesis  Dr. Collet guided and supported the whole process. Supervisory committee Dr. Garey Mazowita, Dr. Devin Harris, Dr. Victoria Claydon and Dr. Rollin Brant gave me great assistance in understanding the important theoretical concepts, evaluating clinical assessment, analysing HRV, and performing statistical analyses. Dr. Yongjun Wang, Dr. Yilong Wang and Dr. Xingquan Zhao provided suggestions to improve of study design and performance. All physicians, especially Dr. Weiqi Chen and Dr. Xin Liu, in Department of Neurology at Beijing Tiantan Hospital provided supports on study recruitment and follow up. Dr. Hui Lin, Dr. Zheng Liu and all specialists in the Cardiac Electrophysiology Laboratory helped with scheduling and conducting the Holter testing, as well as solving practical problems. Dr. Xia Meng helped with the ethics application in Beijing Tiantan Hospital and study management.         vii  Table of Contents  Abstract .......................................................................................................................................... ii Lay Summary ............................................................................................................................... iv Preface .............................................................................................................................................v Table of Contents ........................................................................................................................ vii List of Tables .............................................................................................................................. xiii List of Figures ...............................................................................................................................xv List of Abbreviations ................................................................................................................ xvii Acknowledgements ......................................................................................................................xx Dedication .................................................................................................................................. xxii Chapter 1: Introduction ................................................................................................................1 1.1 Executive Literature Review ............................................................................................... 3 1.2 Research Questions and Study Aims .................................................................................. 5 Chapter 2: Background .................................................................................................................7 2.1 Minor Ischemic Stroke and Transient Ischemic Attack ...................................................... 9 2.1.1 Definition .................................................................................................................... 9 2.1.2 Diagnosis ................................................................................................................... 11 2.1.3 Etiological Classification .......................................................................................... 12 2.1.4 Prognosis ................................................................................................................... 13 2.1.5 Risk Factors .............................................................................................................. 17 2.1.6 Risk Stratification ..................................................................................................... 19 2.1.7 Problems Needed to be Solved ................................................................................. 20 viii  2.2 Stress, Autonomic Nervous System, and Health .............................................................. 23 2.2.1 Stress ......................................................................................................................... 23 2.2.2 Physiological Function of the Autonomic Nervous System ..................................... 24 2.2.3 Assessment of ANS Function ................................................................................... 27 2.2.3.1 HRV assessment ............................................................................................... 29 2.2.3.2 Using HRV to assess dynamic ANS activity .................................................... 35 2.2.3.3 Use of HRV in clinical research ....................................................................... 38 2.2.4 ANS in Stress Responses: from Adaptation to Disease ............................................ 39 2.2.4.1 Adaptation ......................................................................................................... 39 2.2.4.2 Disease progression and triggering ................................................................... 43 2.3 Stress-Related Risk Factors of Ischemic Stroke and Autonomic Nervous System .......... 48 2.3.1 Hypertension and ANS Dysfunction......................................................................... 48 2.3.2 Diabetes and ANS Dysfunction ................................................................................ 50 2.3.3 Dyslipidemia and ANS Dysfunction......................................................................... 52 2.3.4 Atherosclerosis and ANS Dysfunction ..................................................................... 52 2.3.5 Cardiovascular diseases and ANS Dysfunction ........................................................ 54 2.3.6 Atrial Fibrillation and ANS Dysfunction .................................................................. 55 2.3.7 Aging and ANS Dysfunction .................................................................................... 56 2.3.8 Genetic Factors and ANS Dysfunction ..................................................................... 57 2.3.9 Psychological Stress and ANS Dysfunction ............................................................. 57 2.3.10 Sedentary Life and ANS Dysfunction ...................................................................... 58 2.3.11 Smoking and ANS Dysfunction ................................................................................ 59 2.3.12 Alcohol Consumption and ANS Dysfunction ........................................................... 59 ix  2.3.13 Recent Infections and ANS Dysfunction .................................................................. 60 2.3.14 Prior TIA and ANS Dysfunction .............................................................................. 61 Chapter 3: Rationale....................................................................................................................63 Chapter 4: Prior Hypotheses on Identification of HRV Predictors ........................................66 Chapter 5: Objectives and Hypotheses ......................................................................................73 5.1 Primary Objective and Hypothesis ................................................................................... 73 5.2 Secondary Objectives and Hypotheses ............................................................................. 74 5.3 Exploratory Objectives and Hypotheses ........................................................................... 74 Chapter 6: Methods .....................................................................................................................76 6.1 Clinical Care and Patients Monitoring in Study Setting ................................................... 76 6.2 Study Design ..................................................................................................................... 78 6.3 Study Population ............................................................................................................... 78 6.4 Recruitment ....................................................................................................................... 80 6.5 Study Variables and Assessment ...................................................................................... 80 6.5.1 Demographic and Clinical Information .................................................................... 81 6.5.2 Assessment of ABCD2 Score ................................................................................... 82 6.5.3 Assessment of Autonomic Activity (Heart Rate Variability) ................................... 83 6.5.3.1 24-hour Holter monitoring ................................................................................ 83 6.5.3.2 HRV assessment ............................................................................................... 84 6.5.3.3 Preliminary selection of HRV parameters ........................................................ 87 6.5.4 Assessment of Perceived Psychological Stress ......................................................... 89 6.6 Follow-Up ......................................................................................................................... 89 6.7 Outcome Events Identification ......................................................................................... 90 x  6.7.1 Definition of Ischemic Events................................................................................... 90 6.7.2 Assessment and Validation of Outcome Events ....................................................... 93 6.8 Study Outcomes ................................................................................................................ 94 6.9 Sample Size and Power Calculation ................................................................................. 95 6.10 Statistical Analyses ........................................................................................................... 96 Chapter 7: Results........................................................................................................................98 7.1 Recruitment ....................................................................................................................... 98 7.2 Identification and Validation of Outcome Events .......................................................... 101 7.2.1 Identification of Possible Outcome Events ............................................................. 101 7.2.2 Validation of Suspect Outcome Events .................................................................. 101 7.2.3 Description of Cases ............................................................................................... 102 7.3 Demographic and Clinical Information .......................................................................... 104 7.4 HRV Parameters, PSS Score and ABCD2 Score............................................................ 109 7.4.1 Univariate Predictive Models.................................................................................. 117 7.4.2 Diagnostic Performance of ABCD2 Score, PSS Score and the Selected HRV Parameters ........................................................................................................................... 120 7.4.2.1 ABCD2 score .................................................................................................. 120 7.4.2.2 PSS score ........................................................................................................ 121 7.4.2.3 Daytime HF changes ....................................................................................... 121 7.4.2.4 Morning HF value ........................................................................................... 125 7.4.3 Kaplan-Meier Curves for ABCD2 Score, PSS Score and the Selected HRV Predictors ............................................................................................................................ 128 7.5 HRV-based Stress Predictive Models ............................................................................. 134 xi  7.5.1 Bivariate Predictive Models .................................................................................... 134 7.5.2 Full Predictive Models ............................................................................................ 134 7.6 The Best Stress Model and the Most Practical Model .................................................... 141 7.6.1 The Best Stress Model ............................................................................................ 141 7.6.2 The Most Practical Model ....................................................................................... 145 7.7 Exploratory Stress Model with Combined HRV Variables ............................................ 149 7.7.1 Exploratory Stress Model-1 .................................................................................... 149 7.7.2 Exploratory Stress Model-2 .................................................................................... 153 Chapter 8: Discussion and Conclusion ....................................................................................157 8.1 Predictive Values of HRV Parameters ............................................................................ 158 8.2 HRV Predictors versus ABCD2 Score ........................................................................... 166 8.3 The Use of PSS Score in the Model................................................................................ 167 8.4 The Use of Multiple Testing ........................................................................................... 168 8.5 HRV-based Stress Predictive Models ............................................................................. 172 8.6 Clinical Relevance, Unique Aspects and Strengths ........................................................ 176 8.7 Important Issues, Challenges and Limitations ................................................................ 181 8.8 Future Directions ............................................................................................................ 193 Bibliography ...............................................................................................................................196 Appendices ..................................................................................................................................254 Appendix A NIHSS Score ...................................................................................................... 254 Appendix B TOAST Classification for Ischemic Stroke ........................................................ 259 Appendix C ABCD, ABCD2, ABCD3 and ABCD3-I Scores ............................................... 262 Appendix D The General Picture of Autonomic Nervous System ......................................... 264 xii  Appendix E Heart Rate Variability Measurement .................................................................. 266 Appendix F Acqknowledge Software ..................................................................................... 270 Appendix G Perceived Stress Scale ........................................................................................ 271 Appendix H Definition of the Outcome Events ...................................................................... 273 Appendix I Analyses of LF and VLF ..................................................................................... 277    xiii  List of Tables  Table 2.1 Identified risk factors for ischemic stroke .................................................................... 18 Table 2.2 Comparison of sympathetic and parasympathetic function .......................................... 26 Table 2.3 Some tests of ANS function.......................................................................................... 28 Table 2.4 Frequency-domain measures in short-term and long-term spectral HRV analyses ...... 34 Table 2.5 Behavioral and physical adaptation during mild level of acute stress .......................... 41 Table 7.1 Demographics and clinical information ...................................................................... 105 Table 7.2 Comparisons of demographic and clinical information between event group and event-free group .................................................................................................................................... 108 Table 7.3 HRV, PSS score and ABCD2 score in study patients ................................................ 110 Table 7.4 Comparisons of HRV, PSS score and ABCD2 score between event group and event-free group .................................................................................................................................... 113 Table 7.5 Univariate predictive models ...................................................................................... 118 Table 7.6 Diagnostic performance of ABCD2 score .................................................................. 120 Table 7.7 Diagnostic performance of PSS score ........................................................................ 121 Table 7.8 Diagnostic performance of Daytime HF changes ....................................................... 124 Table 7.9 Diagnostic performance of Morning HF value ........................................................... 127 Table 7.10 Multiple stress models .............................................................................................. 136 Table 7.11 AUC test metrics for the selected models ................................................................. 140 Table 7.12 Performance of the Best Stress Model ...................................................................... 144 Table 7.13 Performance of the Most Practical Model ................................................................ 148 Table 7.14 Performance of Exploratory Stress Model-1 ............................................................ 152 xiv  Table 7.15 Association between risk exposure and development of secondary ischemic events..................................................................................................................................................... 153 Table 7.16 Performance of the Exploratory Stress Model-2 ...................................................... 156 Table 8.1 Testing performance of ABCD2 score, Morning HF value and PSS score, and combined tests ............................................................................................................................. 171 Table 8.2 Comparison of diagnostic performance between ABCD2 score, BSM and MPM .... 177   xv  List of Figures  Figure 2-1 Kaplan-Meier survival curves for stroke and all adverse events ................................ 16 Figure 2-2 Power spectral density graph ...................................................................................... 33 Figure 2-3 Spectral analysis of HRV in a young subject at rest and during 90º tilt ..................... 37 Figure 2-4 Anti-inflammatory reflexes by Autonomic Nervous System ...................................... 42 Figure 2-5 Chronic stress, the nervous system and the development of stress-related disorders . 45 Figure 2-6 Possible link between stress, ANS and progression of ischemic stroke ..................... 47 Figure 3-1 The logic of the HRV-based stress predictive model .................................................. 65 Figure 4-1 24-hour rhythm of HF in healthy population .............................................................. 67 Figure 4-2 24-hour rhythm of HF, LF and TP in healthy people, patients with diabetes and with chronic angina ............................................................................................................................... 68 Figure 6-1 Selected HRV parameters ........................................................................................... 88 Figure 7-1 Study recruitment flow chart ..................................................................................... 100 Figure 7-2 The 90-day event-free curve ..................................................................................... 103 Figure 7-3 Boxplots for HRV parameters between event group and event-free group .............. 115 Figure 7-4 ROC curve for Daytime HF changes ........................................................................ 123 Figure 7-5 ROC curve for Morning HF value ............................................................................ 126 Figure 7-6 Kaplan-Meier curves for ABCD2 score .................................................................... 130 Figure 7-7 Kaplan-Meier curves for PSS score .......................................................................... 131 Figure 7-8 Kaplan-Meier curves for Daytime HF changes ........................................................ 132 Figure 7-9 Kaplan-Meier curves for Morning HF value ............................................................ 133 Figure 7-10 ROC curves for the Best Stress Model ................................................................... 143 xvi  Figure 7-11 ROC curves for the Most Practical Model .............................................................. 147 Figure 7-12 ROC curves for Exploratory Stress Model-1 .......................................................... 151 Figure 7-13 ROC curves for Exploratory Stress Model-2 .......................................................... 155 Figure 8-1 Multiple testing by ABCD2 score, HRV and PSS score .......................................... 170   xvii  List of Abbreviations  AF Atrial fibrillation  ANS Autonomic nervous system AUC Area under the curve BP Blood pressure CI Confident interval  CNS Central nervous system CT Computed tomography CTA Computed tomography angiography CUS Carotid ultrasound CV Coefficient of variance DBP Diastolic blood pressure DSA Digital subtraction angiography DWI or DW-MRI Diffusion-weighted magnetic resonance imaging FFT Fast Fourier transformation HF High frequency  HPA axis Hypothalamus-pituitary-adrenal axis HRV Heart rate variability ICC Intraclass correlation coefficient  LAA Large-artery atherosclerosis  LDL-C Low-density lipoprotein cholesterol xviii  LF      Low frequency MI Myocardial infarction MRA Magnetic resonance angiography MRI Magnetic resonance imaging NIHSS National Institutes of Health Stroke Scale NN Normal-to-normal  NPV Negative predictive value OR Odds ratio PAF Paroxysmal atrial fibrillation  PNS Parasympathetic nervous system PPV Positive predictive value PSD Power spectral density PSS Perceived stress scale RCT Randomized control trial ROC curve Receiver operating characteristic curve RR Relative risk RSA Respiratory sinus arrhythmia RR Relative risk SBP Systolic blood pressure  SDNN Standard deviation of normal-to-normal intervals SNS Sympathetic nervous system T2D Type-2 diabetes xix  TCD Transcranial Doppler sonography TEE Transesophageal echocardiography TIA Transient ischemic attack TOAST Trial of Org 10172 in Acute Stroke Treatment TP Total power  TTE Transthoracic echocardiography ULF   Ultra-low frequency VLF Very low frequency     xx  Acknowledgements  I offer my most sincere gratitude to my dear supervisor, Dr. Jean-Paul Collet, for his GREAT patience, tolerance and care on me. He supervised me for both my Master and PhD. His profound knowledge guided me keep improving my research thought and work. He always took time out from a busy schedule to be there for me. I have learned kindness, integrity, professionalism, rigor, tolerance, generosity and modesty from him, which benefits me beyond academia. I have no words to fully describe how I appreciate all your witty guidance, tireless assistance and hearty care on me during these years.  I am grateful to my committee members Dr. Garey Mazowita, for all his vast clinical knowledge, constant support and encouragement, as well as careful revision on my doctoral dissertation. Your advices on both clinical and research perspectives extended my research horizon. I thank Dr. Devin Harris for his deep insight in neurology which provides solid clinical basis for my research. I appreciate Dr. Victoria Claydon. Your deep insight in HRV research gave me great support in understanding my study and conducting HRV analysis. Your advices in every meeting guided me to further explore my study. I extend special thanks to Dr. Rollin Brant who is an amazing professor and brilliant statistician. He guided me from my Master study. He provided me wise ideas and methods on data analysis that helped my research too much. Thank you very much for your great patience and support throughout all my graduate study.   It is my honour to be a member of 2012 PhD cohort (and 2009 Master cohort) in Experimental Medicine which is a supportive and open program. I always get timely assistance from my program. I am extremely fortunate to be a member of our research team. Thank all my colleague students: Mojgan Gitimoghaddam, Michelle Chakraborti, Sravan Jaggumantri, xxi  MirKaber Mosavian Pour, MirSohail Fazeli, Tammie Dewan, and Nataliya Yuskiv. They are always there to listen, to trust, and to offer selfless support.  Finally, I owe special thanks to my grandparents (Hepu, Xiuyun, Ying and Shifen), my parents (Nan and Baohong) and my husband (Guopeng), as well as my sweet friends (Ding, Xiaoyu, Ka, Hao, Lijia, Huan, Dongning, Mailan, Juelu, Yun, Qian, Xinxin and many others) who are always by my side, share my ups and downs in this journey, and support my choices and dreams.      xxii  Dedication  To the love in my heart and all life experiences 1  Chapter 1: Introduction  This study on testing the values of autonomic nervous system (ANS) function and stress in predicting the development of a secondary ischemic event after acute transient ischemic attack (TIA) and minor stroke was proposed at the end of 2011. I conducted the systematic literature review, proposed the research questions, prepared the study protocol and other documents, organized the study setting, as well as applied for two ethics approval from 2012 to July 2014. After the ethics approvals were obtained, I conducted a pilot study to assess the study feasibility from July to Sep 2014 with a sample size of 10 patients. Then, the recruitment of 201 patients for the main study was started in October 2014 and ended in Feb 2016. Follow-up on all study patients ended in May 2016. This dissertation presents the entire study, starting with this general introduction.  In Chapter 2, I present the study background on TIA and minor stroke based on a systematic literature review. I also introduce important concepts related to stress, the Autonomic Nervous System (ANS) and heart rate variability (HRV) (as a measurement of ANS). At last, I show the correlation between ANS dysfunction and the risk factors of ischemic stroke.  Based on the associations among the risk factors of ischemic stroke, stress, and ANS, in Chapter 3, I explain the logic/rationale that constructed our ANS/HRV-based stress model for ischemic events prediction within 90 days after TIA or minor stroke. 2  As a pioneer study to evaluate the values of ANS/HRV/stress assessment in predicting ischemic events, Chapter 4 describes the selection of HRV parameters (the prior hypotheses) that includes the type of HRV parameters, the time periods for HRV absolute values, and the time intervals for HRV changes. Then Chapter 5 is based on a priori hypotheses explained in Chapter 4 to propose the aims and overall objective of the study.  In Chapter 6, I illustrate the study methods including the study design, the study population, recruitment, the study variables and assessment, follow-up, the study outcomes and statistical analyses.  The study results are shown in Chapter 7. Basically, I evaluated all candidate HRV parameters, and identified the best and the most practical ones. I showed the diagnostic performance and identified the potential cut-off points of the selected HRV predictors. Furthermore, I established different multivariate models and found the best and the most practical ones; both significantly improved the predictive value of ABCD2 score. At last, I developed two exploratory stress models with a combination of HRV values and HRV changes.  In Chapter 8, I discuss the study results, implications, clinical relevance, strengths, challenges and future directions.  In the following text of this introduction chapter, I will provide a general picture of study background, research questions and overarching study aims.    3  1.1 Executive Literature Review  Cerebrovascular diseases with mild or transient symptoms mainly comprise the transient ischemic attack (TIA) and minor stroke (Easton et al., 2009; Sacco et al., 2013). However, up to 23% of strokes are preceded by a TIA or minor stroke (Rothwell & Warlow, 2005), which makes the two events to be considered as warning signals for severe ischemic stroke rather than benign events (Easton et al., 2009). Therefore, improving prevention during the short window between TIAs and severe ischemic stroke is essential (Jauch et al., 2013; Powers et al., 2015). It is well demonstrated that early effective treatment for ischemic stroke (including drugs and operations) provides benefits to patients (Furie et al., 2011; Goldstein et al., 2006; Jauch et al., 2013; Kernan et al., 2014; Powers et al., 2015; Sacco et al., 2006); however, it may also cause adverse events such as minor bleedings (around 3% for dual antiplatelet treatment) (Diener, Bogousslavsky, & Brass, 2004) and intracranial hemorrhage (around 2% for dual antiplatelet treatment and 6% for intravenous thrombolysis) (Diener et al., 2004; Geeganage et al., 2012; Miller, Simpson, & Silver, 2011). Therefore, to precisely evaluate the patients’ risk and identify the ones who need further treatment and hospitalization is critical.   The general concept of “stress” describes a threatened state caused by any forms of internal or external disturbing forces, which are named “stressors” (Chrousos & Gold, 1992; Chrousos, 2009; Johnson, Kamilaris, Kamilaris, Chrousos, & Gold, 1992; Lusk & Lash, 2005). “Stress response” is a counteracting force to neutralize the effect of stressors and re-establish homeostasis 4  (Chrousos & Gold, 1992; Johnson et al., 1992; Lusk & Lash, 2005). The stress response is tightly regulated by the ANS which is a prompt reaction system and plays a critical role in modulating stress response (Tracey, 2002; Ulrich-Lai & Herman, 2009). ANS activity can be measured by heart rate variability (HRV), which is defined as the fluctuations in the intervals between normal heartbeats (Task Force of ESC, 1996). Given that HRV measurement is a reliable, non-invasive, and convenient measurement of overall cardiac autonomic activity (Kleiger, Stein, & Bigger, 2005; Montano et al., 2009; Task Force of ESC, 1996; Thayer & Sternberg, 2006), the values of HRV may be indicators of the overall stress of an individual, which is a combination of chronic life stress and acute stress, at the time of assessment. The change of HRV during a period may reflect the changes in stress, which shows dynamic feature of ANS activity (Malliani, Pagani, Lombardi, & Cerutti, 1991; Montano et al., 1994, 2009; Pumprla, Howorka, Groves, Chester, & Nolan, 2002).  Many identified risk factors of ischemic stroke, including age, hypertension, diabetes, smoking, alcohol consumption, and sedentary lifestyle are considered as chronic stressors (Chrousos, 2009). Acute risk factors/triggers of ischemic stroke such as recent infections, psychological shock and recent TIAs act as acute stressors (Goldstein et al., 2006, 2011; Meschia et al., 2014). On the other hand, some risk factors are consequences of stressors. For example, hypertension is a stressor for ischemic stroke; however, the development of hypertension is related to many other stressors such as aging, diet and smoking, which makes hypertension as a stress-5  caused condition (Ford & Cooper, 1991; Kornitzer, Dramaix, & De Backer, 1999). Diabetes, as a metabolic stressor for the body, may be a consequence of obesity, chronic inflammation and insulin resistance (Mokdad et al., 2015; Petit et al., 2001). Exposure to chronic or excessive acute stress may cause dysfunctional ANS adaptation to stress; the dysfunctional ANS will further reduce the control of the stress responses (Chrousos & Gold, 1992; Johnson et al., 1992; Lusk & Lash, 2005). Such vicious circle ultimately makes ischemic stroke a stress-related condition, which might be seen as the “final endpoint” of the effects of multiple stressors.  All potential risk factors may contribute to the occurrence of secondary ischemic events after TIA or minor stroke. However, only a limited number of risk factors/stressors are traditionally assessed in clinical practice, which may be one reason for the moderate predictive value of traditional tool ABCD2 score in stroke development after TIA (Chandratheva, Geraghty, & Rothwell, 2011; Kiyohara et al., 2014; Perry et al., 2011). Given the association between risk factors of ischemic stroke, stress and ANS, it would be possible to use HRV data to identify the personalized risk of a second ischemic event after initial TIA or minor stroke.   1.2 Research Questions and Study Aims   This study was developed to answer the following research questions:  i. Can HRV parameters (assessed within 48 hours after initial TIA or minor stroke) predict the occurrence of a secondary ischemic event? 6  ii. What is the best HRV indicator to predict a secondary ischemic event after the initial TIA or minor stroke? iii. What is the best HRV-based model to predict a secondary ischemic event after the initial TIA or minor stroke? iv. How does this predictive model compare to ABCD2 score? Can it improve the predictive value of ABCD2 score?  As a pioneer study to examine this topic in stroke research area, the overarching aim of our study is to identify the best HRV indicator and to develop the HRV-based models for predicting secondary ischemic events after TIA or minors stroke. We expect that HRV-based predictive models would be superior to the traditional ABCD2 tool.  7  Chapter 2: Background   American Heart Association (AHA) and American Stroke Association (ASA) published the latest Heart Disease and Stroke Statistics 2017 At-a-Glance (AHA/ASA, 2017; Benjamin et al., 2017), which shows that: • Incidence and prevalence: - Each year, about 795,000 people experience a new or recurrent stroke. Approximately 610,000 of these are first attacks, and 185,000 are recurrent attacks. - In 2013, worldwide prevalence of stroke was 25.7 million, with 10.3 million people having a first stroke. • Mortality: - Stroke was the second-leading global cause of death behind heart disease in 2013, accounting for 11.8% of total deaths worldwide. - When considered separately from other cardiovascular diseases, stroke ranks the fifth among all cause of death in the US, killing nearly 133,000 people a year.  During the past several decades stroke has been responsible for high morbidity, mortality, and social burden (Go et al., 2014; Mozaffarian et al., 2016). A stroke is commonly understood by the non-medical population, and portrayed by the lay press as a “brain attack” (Sacco et al., 2013). 8  Strokes are categorized into ischemic and haemorrhagic sources, with ischemic stroke accounting for 87% of all stroke events (Sacco et al., 2013). The minor (ischemic) stroke and transient ischemic attack (TIA) are two types of cerebrovascular diseases with mild or transient symptoms and non-disabling consequences (Easton et al., 2009; Sacco et al., 2013). TIA and minor stroke are markers of reduced cerebral blood flow, and are considered as “warning signals” for possible occurrence of a more severe ischemic stroke (Easton et al., 2009; Johnston et al., 2006). TIA and minor stroke however, offer a unique opportunity to initiate treatment that can forestall the onset of permanently disabling injury (Lavallée et al., 2007; Rothwell et al., 2007). In guidelines TIA and minor stroke are recommended to receive urgent treatment with careful medical care and management (Adams et al., 2007; Furie et al., 2011; Goldstein et al., 2006; Jauch et al., 2013; Kernan et al., 2014; Powers et al., 2015; Sacco et al., 2006).  In general, TIA and minor stroke form part of the ischemic stroke continuum, with common risk factors and pathophysiological processes (Easton et al., 2009; Rothwell, Buchan, & Johnston, 2006; Sacco et al., 2013). In clinical practice, the evaluation criteria, treatment strategy and early prognosis for TIA and minor stroke are on the same track (Adams et al., 2007; Furie et al., 2011; Goldstein et al., 2006; Jauch et al., 2013; Kernan et al., 2014; Powers et al., 2015; Sacco et al., 2006). Studies in the population with TIA and minor stroke focus on many aspects, including identification of the risk factors (Dennis, Bamford, Sandercock, & Warlow, 1989; Easton et al., 2009; Grau et al., 2001), assessment of prognosis (Johnston et al., 2007a; Merwick et al., 2010; 9  Rothwell et al., 2005), primary and secondary prevention (Giles & Rothwell, 2007; Johnston et al., 2016; Kennedy et al., 2007; Wang et al., 2013), as well as clinical management (Giles & Rothwell, 2007; Lavallée et al., 2007). In this thesis, I describe a study examining possible predictors of the occurrence of secondary ischemic events in the following 3 months after TIA or minor stroke. Potential predictors of recurrent ischemic events included traditional risk factors as well as a new approach based on the assessment of patient’s stress within 48 hours after the occurrence of initial TIA or minor stroke.  In this chapter, I will first describe ischemic stroke and TIA in terms of definition, diagnosis, classification, prognosis, risk factors and risk stratification in section 2.1. Following that, I will introduce the concepts related to stress, describe the stress regulatory system in particular the Autonomic Nervous System (ANS), and explain the progressive shift from “stress adaptation” to “stress-related diseases” in section 2.2. At last in section 2.3, I will show the association between the risk factors of ischemic stroke and TIAs, stress and ANS dysfunction.   2.1 Minor Ischemic Stroke and Transient Ischemic Attack 2.1.1 Definition Ischemic stroke is defined as an acute episode of focal or global neurological dysfunction caused by brain, spinal cord, or retinal vascular injury as a result of infarction (Hicks et al., 2015; Sacco et al., 2013). Ischemic stroke with mild and non-disabling symptoms is considered as minor 10  ischemic stroke. Patients with minor stroke should have a total score of the National Institute Health Stroke scale (NIHSS) (Brott et al., 1989) (Appendix A) less than 4, with each item ≤ 1 and normal consciousness (Furie et al., 2011; Jauch et al., 2013).  TIA is defined as a transient episode of neurological dysfunction caused by focal cerebral, spinal cord or retinal ischemia without permanent cerebral infarction (Easton et al., 2009). The definition of TIA has changed several times over the recent decades. In the past, TIA was defined as time-based, with the presence of focal neurological symptoms or signs lasting <24 hours (Easton et al., 2009). However, multiple studies worldwide have proved that this arbitrary time threshold is too broad, because 30% to 50% of classically defined TIAs show brain injury on diffusion-weighted magnetic resonance imaging (DWI or DW-MRI) (Easton et al., 2009). With these observations in mind, scholars proposed a tissue-based definition in 2002: “a brief episode of neurological dysfunction caused by focal brain or retinal ischemia, with clinical symptoms typically lasting less than one hour, and without evidence of acute infarction” (Albers, Caplan, & Easton, 2002; Johnston, 2002). With advances in neuroimaging use, many cases with an evidence of infarction despite resolved or very subtle symptoms should be diagnosed as minor stroke rather than mischaracterized as TIAs.  11  2.1.2 Diagnosis  At patients’ initial presentation, a comprehensive assessment should include identification of their signs and symptoms consistent with a focal neurologic deficit, and the timing of symptoms onset and resolution (Caplan, 2009; Easton et al., 2009; Sacco et al., 2013). This is crucial, especially for TIA, because TIA symptoms often resolve by the time of presentation (Caplan, 2009; Easton et al., 2009; Sacco et al., 2013). The medical history and family history should elicit risk factors associated with ischemic diseases, such as hypertension, diabetes mellitus, obesity, dyslipidemia and cigarette smoking, as well as personal and family history of cerebrovascular diseases and hypercoagulability disorders (Caplan, 2009; Easton et al., 2009). The physical examinations should include measurement of vital signs and symptoms, and comprehensive neurologic and cardiovascular examinations (Caplan, 2009; Easton et al., 2009; Sacco et al., 2013). More importantly, the diagnostic evaluation of ischemic stroke and TIA should be initiated as soon as possible to stratify the risk of recurrent events (Lavallée et al., 2007; Rothwell et al., 2007). According to AHA/ASA guidelines, the goals of the diagnostic evaluation are to explore the mechanism and the origin of patient’s symptoms and to exclude non-ischemic etiologies (Easton et al., 2009; Sacco et al., 2013). Normal neuroimaging examinations include Computed Tomography (CT) and MRI (Caplan, 2009; Easton et al., 2009; Sacco et al., 2013). Vessel imaging includes Magnetic Resonance Angiography (MRA), Computed Tomography Angiography (CTA) and carotid ultrasound/transcranial Doppler (CUS/TCD) (Caplan, 2009; Easton et al., 2009; Sacco 12  et al., 2013). Cardiac assessment mainly contains electrocardiography (ECG), transthoracic or transesophageal echocardiography (TTE or TEE) and Holter monitoring  (Caplan, 2009; Easton et al., 2009). Routine laboratory testing is also needed to help with the diagnosis (Caplan, 2009; Easton et al., 2009; Sacco et al., 2013).  Patients may have transient focal neurological symptoms that are not attributable to a focal cerebral ischemia; such conditions may imitate a TIA and therefore be labeled TIA mimics, or imitate an ischemic stroke and are labeled stroke mimics (Barrett, Levine, & Johnston, 2008; Hand, Kwan, Lindley, Dennis, & Wardlaw, 2006). The rate of TIA mimics ranges from 10% to 48.5% (Nadarajan, Perry, Johnson, & Werring, 2014; Prabhakaran, Silver, Warrior, McClenathan, & Lee, 2008), depending on the clinical settings. Frequent TIA mimics include seizures, fainting/syncope, migraine headaches, mood disorders and others (Caplan, 2009; Nadarajan et al., 2014). Such conditions need to be attended to and the diagnosis of TIA ruled out.   2.1.3 Etiological Classification Ischemic stroke or TIA may not be a final diagnosis. There are many possible causes for TIA/stroke symptoms. For a practical etiological classification, the Trial of ORG 10172 in the Acute Stroke Treatment (TOAST) criteria (Adams et al., 1993) (Appendix B) is used in “AHA/ASA Guidelines for Prevention of Stroke in Patient with Stroke and TIA” (Furie et al., 2011; Kernan et al., 2014; Sacco et al., 2006), as well as in both clinical practice and research. The 13  etiological classification has implications for risk factor identification, prognosis and urgency of assessment, and is a prerequisite for prescribing the correct therapies for secondary prevention (Easton et al., 2009; Sacco et al., 2013). According to TOAST, ischemic stroke is classified into five subtypes (Adams et al., 1993): (i) large-artery atherosclerosis (LAA) related stroke, (ii) cardioembolism related stroke, (iii) lacunar stroke, (iv) stroke of other determined etiology and (v) stroke of undetermined etiology (cryptogenic). Other classification systems used in clinical practice and research include OCSP (Oxford Community Stroke Project) (Bamford, Sandercock, Dennis, Burn, & Warlow, 1990), NINDS (National Institute of Neurological Disorder and Stroke) (NINDS Committee, 1990) and ASCO (Phenotypic) (Amarenco, Bogousslavsky, Caplan, Donnan, & Hennerici, 2009).   2.1.4 Prognosis TIA and minor stroke should not be considered as benign conditions; they are warning signs for the occurrence of ischemic stroke (Easton et al., 2009). Beyond all risk factors, prior TIA and minor stroke episodes provide additional risk information for a secondary ischemic event. The risk of stroke in 90 days after TIA was reported up to 20%, as noted both in primary literature (Chandratheva, Metha, Geraghty, Marquardt, & Rothwell, 2009; Coull, Lovett, & Rothwell, 2004; Johnston, Gress, Browner, & Sidney, 2000; Kleindorfer et al., 2005; Rothwell & Warlow, 2005; Wu et al., 2007) and a systematic review (Giles & Rothwell, 2007). Johnston et al. determined the 14  prognosis in the 90 days after emergency diagnosis of TIA and found that among 1707 patients, 428 (25.1%) developed adverse events including stroke, cardiovascular hospitalization, death and recurrent TIA; in these cases, more than 50% of adverse events occurred within the first 4 days (Figure 2.1) (Johnston et al., 2000). Similarly, Coull and colleagues reported in 2004 that the risk of developing ischemic stroke after a prior TIA or minor stroke was up to 5% within the first 48 hours, 8% at 7 days, 11.5% at 30 days and 18.5% in 90 days (Coull et al., 2004). Rothwell et al. showed in 2005 that around 17% of ischemic strokes were preceded by a warning TIA; in 43% of cases this warning TIA occurred within one week of the subsequent stroke (Rothwell & Warlow, 2005). Additionally, Chandratheva and others reported in 2009 that the stroke risks in the early few hours, 6-, 12-, and 24-hour after the first TIAs were 1.2%, 2.1%, and 5.1%, respectively (Chandratheva et al., 2009). Of note, these studies were conducted in either United States or United Kingdom, in patients receiving standard management and usual treatment in clinical practice. In recent years, the recurrent stroke rate in the 90 days after an initial TIA or minor stroke has decreased to around 7% to 13% with standard treatment in randomized control trials (RCT) (Johnston et al., 2016; Wang et al., 2013, 2017). One recent multicenter RCT conducted in China (CHANCE trial) showed a recurrent ischemic stroke rate in 212 TIA/minor stroke patients of 8.2% in those receiving clopidogrel–aspirin, compared to 11.7% in 303 patients receiving aspirin alone (Wang et al., 2013). Nevertheless, the overwhelming evidence of the high risk of recurrent stroke 15  after an initial event still underscores the need for urgent evaluation and treatment of patients with TIA or minor stroke. 16  Figure 2-1 Kaplan-Meier survival curves for stroke and all adverse events  The figure is adapted from the published article by Johnston et al. (Johnston et al., 2000)  A: Kaplan-Meier survival curves for all adverse events that include stroke, TIA, hospitalization for a cardiovascular event, and death.  B: Kaplan-Meier survival curves for patients with different numbers of risk factors. Risk factors included age >60 years, duration of symptoms >10 minutes, diabetes mellitus, weakness associated with TIA, and speech impairment associated with TIA.   17  2.1.5 Risk Factors  Ischemic stroke and TIA are associated with the same spectrum of risk factors (Furie et al., 2011; Goldstein et al., 2006, 2011; Kernan et al., 2014; Meschia et al., 2014; Sacco et al., 2006). AHA/ASA guidelines list established chronic risk factors for stroke and TIA: age, gender and heredity, hypertension, hyperglycemia, dyslipidemia, obesity, cardiovascular diseases, atrial fibrillation, sleep apnea, asymptomatic carotid stenosis, sickle cell disease, hyperhomocysteinemia, hypercoagulability, physical inactivity, nutrition, cigarette smoking, drug abuse and alcohol consumption (Goldstein et al., 2006, 2011; Kernan et al., 2014; Meschia et al., 2014). Besides the traditional chronic risk factors that predispose to the occurrence of TIA and ischemic stroke, some “acute risk factors/triggers” may precipitate this process. Identified acute triggers of ischemic stroke include recent TIAs, recent infections, surgery (general and cardiac), cervical trauma and manipulation, pregnancy and the postpartum state, addicted drugs, medications, diurnal fluctuations, winter season, air pollution, sudden psychological stress, anger and negative emotions, and sudden changes in posture (Elkind, 2007; Guiraud, Amor, Mas, & Touze, 2010; Koton, Tanne, Bornstein, & Green, 2004; Sharma et al., 2015; Tofler & Muller, 2006). These risk factors are summarized in Table 2.1. The statistics of the risk factors and their associations with ANS dysfunction will be further described in section 2.3.   18  Table 2.1 Identified risk factors for ischemic stroke  Chronic risk factors Acute risk factors/triggers  Modifiable Non-modifiable   Hypertension Diabetes Dyslipidemia Obesity Atrial fibrillation Cardiovascular diseases Other cardiac events Asymptomatic carotid stenosis Sickle cell disease Metabolic syndrome Sleep apnea Migraine Hyperhomocysteinemia Hypercoagulability Elevated lipoprotein Postmenopausal hormone therapy Cigarette smoking Alcohol abuse Drug abuse Diet and nutrition Physical inactivity Age Gender Low birth weight Race-ethnicity Genetic factors Infections Psychological/mental stress Anger Negative emotions Sudden changes in posture Winter season Diurnal fluctuations Air pollution Surgery Cervical accident and manipulation Pregnancy and postpartum Medications   Information is summarized from AHA/ASA guidelines and published articles  (Adams et al., 2007; Elkind, 2007; Furie et al., 2011; Goldstein et al., 2006; Johnston et al., 2006; Kernan et al., 2014; Sacco et al., 2006; Sharma et al., 2015)19  2.1.6 Risk Stratification  To stratify or evaluate the risk of early occurrence of an ischemic stroke after TIA or minor stroke, several scoring systems (ABCD, ABCD2, ABCD3 and ABCD3-I, Appendix C) have been developed and widely used in clinical practice. The ABCD score was developed in 2005 by Rothwell and colleagues (Rothwell et al., 2005) to predict the risk of ischemic stroke after TIA. The ABCD score consists of 4 components, age, blood pressure, clinical features of TIA, and duration of symptoms. In 2007, Johnston et al (Johnston et al., 2007a) proposed a new risk score, the ABCD2, in which the presence of diabetes mellitus was added as another component to the original ABCD score. In 2010, Merwick et al (Merwick et al., 2010) proposed two novel scores to predict early risk of stroke after TIA, namely, ABCD3 and ABCD3-I scores. In the ABCD3 score, dual TIA (the presence of ≥2 TIA symptoms within 7 days) is added to the ABCD2 score. In the ABCD3-I score, the presence of abnormal findings on neuroimaging (i.e., carotid stenosis and/or abnormal acute DWI on brain MRI) is further added to the ABCD3 score (Merwick et al., 2010).  The predictive ability of ABCD and ABCD2 scores has been investigated by many studies (Cancelli et al., 2011; Chandratheva et al., 2011; Giles et al., 2011; Kiyohara et al., 2014; Merwick et al., 2010; Perry et al., 2011; Purroy et al., 2012; Sanders et al., 2012; Sheehan et al., 2009, 2010; Wardlaw et al., 2015). Most results showed that ABCD and ABCD2 scores have moderate predictive power (area under the curve (AUC) between 0.55 to 0.7) (Chandratheva et al., 2011; Kiyohara et al., 2014; Merwick et al., 2010; Perry et al., 2011; Purroy et al., 2012; Sanders et al., 20  2012; Sheehan et al., 2010; Wardlaw et al., 2015). The moderate predictive ability may be partially attributed to the intrapersonal heterogeneity, as it is difficult to qualify the effect of risk factors on an individual basis. For instance, “diabetes” has different degrees of severity; and individuals have different ways coping with the consequences of chronic metabolic stress. Furthermore, other important risk factors such as psychological stress, recent infections, smoking and physical inactivity, as well as factors that have not been identified, are excluded in the classic clinical assessment. All of them however, may contribute to the development of ischemic outcomes. The imaging assessment for TIA and minor stroke (ABCD3-I) improves the predictive power of secondary ischemic events, with AUC >0.8 (Ay et al., 2009; Calvet et al., 2009; Merwick et al., 2010). However, the use of imaging is costly and makes scoring technology dependent, therefore affecting universal use (not available in every medical setting and certainly not at home) and delaying scoring process. The limitation of current tools for personal risk prediction calls for exploration of a new tool that is valid, precise and convenient for use in the emergency environment and other settings, to determine the risk of a secondary ischemic event in patients with TIA or minor stroke and to help provide orient medical care.  2.1.7 Problems Needed to be Solved  People who are living with chronic risk factors and suffer an acute TIA or minor stroke, are considered to be at high risk of developing a secondary ischemic event (Johnston et al., 2000; 21  Prencipe et al., 1998). Effective early treatment for ischemic stroke mainly includes drugs for antiplatelet (single or dual) and anticoagulation, and operations such as intravenous thrombolysis and endovascular treatment (Furie et al., 2011; Jauch et al., 2013; Kernan et al., 2014; Powers et al., 2015). It is well known that early treatment and management may help to reduce the volume of brain damaged by ischemia, promote recanalization of blocking vessels, and reduce the risk of early recurrent ischemic stroke (Furie et al., 2011; Jauch et al., 2013; Kernan et al., 2014; Powers et al., 2015). However, these treatment may also create safety issues including minor bleedings (around 3% for dual antiplatelet treatment) (Diener et al., 2004) and fatal intracranial hemorrhage (around 2% for dual antiplatelet and 6% for intravenous thrombolysis) (Diener et al., 2004; Geeganage et al., 2012; Miller et al., 2011). Therefore, an urgent assessment of the risk of second ischemic events after acute TIAs or minor stroke is of the utmost importance for patients and families to be aware, and for medical care-givers to provide necessary treatment and to develop a personalized management plan for preventing the onset of serious ischemic outcomes (Kruyt, Biessels, Devries, & Roos, 2010; Luitse, Biessels, Rutten, & Kappelle, 2012; Rothwell et al., 2007).  However, it remains difficult to precisely identify the risk of secondary ischemic events after TIA or minor stroke for several reasons. Firstly, not all chronic risk factors for ischemic stroke have been well understood, including those poorly measured and those unidentified. Furthermore, acute triggers of ischemic stroke are even more difficult to measure precisely than chronic risk 22  factors, because most acute triggers happen without being noticed or have different expressions due to interindividual variation or time variance (Elkind, 2007; Guiraud et al., 2010; Koton et al., 2004; Sharma et al., 2015; Tofler & Muller, 2006). Even if determining the long-term risk of stroke is available based on the use of several identified chronic risk factors (Furie et al., 2011; Kernan et al., 2014), it still remains a challenge to predict “when a stroke will occur” in a short future (Elkind, 2007). Stroke research seeks to move from a single focus on chronic risk factors to additionally include acute triggers, as it is important to predict the short-term risk of stroke and, to identify the “stroke-prone state” (Elkind, 2007; Guiraud et al., 2010; Koton et al., 2004; Sharma et al., 2015; Tofler & Muller, 2006).   It brings about the significance of our study that aims to find an efficient and simple approach to include the effects of both chronic risk factors and acute triggers, to identify the risk of secondary ischemic events after TIA and minor stroke. My approach in this study was to consider that each identified risk factor is a source of stress for the body, which makes possible to assess the comprehensive effect of all these factors by assessing the ANS function. In the next section, I will describe the relationship between stress (acute and chronic) and ANS, and the health consequences.    23  2.2 Stress, Autonomic Nervous System, and Health  In this section, I will describe the concepts related to stress, stress responses, stress regulatory system ANS, and possible shift from physiologic regulation to stress-related disorders.   2.2.1 Stress  In the central construct of Hans Selye’s stress theory, the notion of “homeostasis” refers to the stability of physiological systems that maintain life (Selye, 1955; Chrousos & Gold, 1992). The generalized definition of “stress” describes a state of threatened homeostasis caused by any form of internal or external disturbing forces, which are called “stressors” ( Selye, 1955; Chrousos & Gold, 1992; Johnson et al., 1992; Lusk & Lash, 2005). All life experiences of a person (i.e. the accumulation of stress experiences) contributes to building an idiosyncratic “stress profile” specific to the individual. This “stress profile”, measured at one specific time, includes both previous and current stress experiences: physical, physiological, psychological and environmental (Chrousos & Gold, 1992; Johnson et al., 1992; Lusk & Lash, 2005).  The “stress response” or “adaptive response” is a counteracting force that puts forth to neutralize the effects of stressors and re-establish homeostasis (Chrousos & Gold, 1992; Johnson et al., 1992; Lusk & Lash, 2005). The stress response is a succession of processes that follows the perception of stress by the brain (Chrousos, 1998; McEwen, 1993). The two main peripheral pathways of the stress regulatory system include the hypothalamus-pituitary-adrenal (HPA) axis 24  and the ANS (Chrousos & Gold, 1992; Ulrich-Lai & Herman, 2009). The ANS generates prompt reaction compared to the HPA axis, and thus plays a critical role in modulating the stress response at early stage (Tracey, 2002; Ulrich-Lai & Herman, 2009).  2.2.2 Physiological Function of the Autonomic Nervous System The autonomic nervous system is one part of the peripheral nervous system that regulates physiological processes without conscious control (Robertson, 2004). The ANS comprises two major divisions: the sympathetic nervous system (SNS) and the parasympathetic nervous system (PNS) (Robertson, 2004) (Appendix D). ANS innervates most organs and controls important physiological processes, such as blood pressure, heart rate, body temperature, weight, digestion, metabolism, fluid and electrolyte balance, sweating, urination, defecation, sexual response, and others (Table 2.2) (Robertson, 2004).  It is generally considered that sympathetic and parasympathetic divisions typically function in opposition to each other (Curtis & O’Keefe, 2002; Robertson, 2004). SNS prepares “fight or flight” responses when faced with stress; while the PNS promotes “rest and digest” responses when relaxed (Curtis & O’Keefe, 2002; Robertson, 2004). However, in many normal physiological conditions, sympathetic and parasympathetic activity cannot be simply ascribed as "fight" or "rest". The neural regulation of physiological function is mainly affected through the interaction of SNS and PNS activity. The activation of either SNS or PNS outflow is accompanied by the 25  relative inhibition of the other (Montano et al., 2009). Therefore, it is helpful to consider the complementary nature of SNS and PNS as they work synergistically, for instance in the anti-inflammatory responses (Robertson, 2004; Tracey, 2002) shown in section 2.2.4.26  Table 2.2 Comparison of sympathetic and parasympathetic function    Sympathetic Nervous System  Parasympathetic Nervous System Heart 1. Positive chronotropic effect: increases heart rate 2. Positive dromotropic effect: increases conduction velocity in atrial-ventricular junction 3. Positive inotropic effect: increases cardiac contraction 1. Negative chronotropic effect: decreases heart rate 2. Negative dromotropic effect: decreases conduction velocity in atrial-ventricular junction 3. Negative inotropic effect: decreases cardiac contraction  Blood vessel  Constricts the coronary artery though α-receptor, while dilates the vessels through β-receptor Has little effect on constriction (in most organs) and dilation (in several organs)  Immune system Anti- or pro-inflammation Anti-inflammation  Respiratory system Dilates bronchi Contracts bronchi  Digestive system  Inhibits peristalsis  Improves peristalsis  Urinary system  Inhibits urinate  Promotes urinate  Reproductive system Inhibit reproductive function  Enhances reproductive function  Eyes Dilates pupil Contracts pupil  Skin Contracts arrector muscle; Increases sweating  NA Metabolism  Promotes glycogenolysis;  Promotes adrenal hormone secretion Promotes glycogen synthesis;  Promotes insulin secretion Information is summarized from the book Primer of Autonomic Nervous System, Roberson 2004 (Robertson, 2004).27  2.2.3 Assessment of ANS Function  The ANS dynamically controls the response of the body to a range of external and internal stimuli/stressors, providing physiological stability in an individual (Chrousos, 2009). Because most of the ANS is inaccessible to direct physiological testing, in clinical settings the most widely used techniques entail the assessment of an end-organ response to an ANS physiological provocation (Freeman, 2006; Freeman & Chapleau, 2013). Some clinical tests for measuring the ANS function (Freeman, 2006; Low, 2003; Vallbo, 2004; Zygmunt & Stanczyk, 2010) are summarized in Table 2.3. Among these tests, HRV assessment is a non-invasive, convenient, valid and reliable method in measuring ANS function, which will be described in this section. 28  Table 2.3 Some tests of ANS function  Type of testing  Strength Limitation Testing of cardiovascular modulation   Heart rate variability (HRV) Non-invasive, convenient, practical, valid and reliable (described in the following text)   Limiting analysis of ectopic waves (such as permanent AF)  Heart rate and blood pressure assessment in normal condition or in Valsalva maneuver test, deep breathing, isometric handgrip test, cold pressure test, orthostatic test, head-up tilt test and baroreflex sensitivity test   Short time-consuming  Assessing both SNS and PNS on cardiovascular modulation Only assessing ANS response to a rapid change of stress   Testing of neurotransmitter levels  Catecholamines and acetylcholine assessment  More direct   Invasive  Not precise   Testing of sudomotor function Quantitative sudomotor axon reflex test (QSART), thermoregulatory sweat test  Precisely assessing ANS modulation on sweat gland    Not assessing cardiovascular modulation Requiring precautions for electrical safety Microneurography Muscle or skin sympathetic nerve activity (MSNA or SSNA)    Precisely assessing SNS  Invasive  Not assessing PNS  Not assessing cardiovascular modulation Information is summarized from published articles (Freeman, 2006; Low, 2003; Vallbo, 2004; Zygmunt & Stanczyk, 2010)29  2.2.3.1 HRV assessment HRV is defined as the fluctuations in the intervals between normal heartbeats (Task Force of ESC, 1996). The measurement of HRV is mainly based on three methods: frequency-domain method, time-domain method, and non-linear method; within each, different parameters reflect different aspects of the ANS function (Appendix E) (Task Force of ESC, 1996). The time-domain method is the simplest one based on the statistics of Normal-to-Normal (NN) intervals (Task Force of ESC, 1996). The frequency-domain method is based on the generation of power spectral density (PSD) which provides the basic information of the distribution of the power (i.e. variance) in different frequencies (Task Force of ESC, 1996). The description of “variability” should include not only a function of time, but also elementary oscillatory components such as frequency and amplitude (Montano et al., 2009). Frequency-domain method provides such information of the data (Task Force, 1996; Montano et al., 2009). The standards of non-linear method are still needed (Task Force of ESC, 1996). The spectral analysis of HRV includes short-term analysis and 24-hour long-term analysis (Task Force of ESC, 1996). Short-term recording of 2-5 minutes has three main power distributions across frequencies: very low frequency (VLF: 0.003-0.04 Hz), low frequency (LF: 0.04-0.15 Hz), and high frequency (HF: 0.15-0.4 Hz); the 24-hour long-term recording can also assess the ultra-low frequency (ULF <0.003 Hz) (Task Force of ESC, 1996). Table 2.4 lists selected frequency–domain measures. Methods for the calculation of PSD are generally classified 30  as nonparametric with Fast-Fournier transformation (FFT) (one of the most common algorithms) and parametric with autoregression; in most instances, both methods provide comparable results (Task Force of ESC, 1996). The frequency-domain analyses of HRV has been demonstrated as a valid procedure for assessing autonomic function in both clinical and experimental settings (Kamath & Fallen, 1993; Malliani, Lombardi, & Pagani, 1994; Montano et al., 1994, 2009; Myers, Martin, & Magid, 1986). HF power is considered to be solely regulated by the PNS, with high HF power representing increased PNS activity (Berntson et al., 1997; Kleiger et al., 2005; Montano et al., 2009; Task Force of ESC, 1996). LF is mediated by a complex mixture of SNS and PNS modulation (Berntson et al., 1997; Kleiger et al., 2005; Montano et al., 2009; Task Force of ESC, 1996). The physiological explanation of VLF component is less defined because it represents the influence of the peripheral vasomotor, renin-angiotensin systems, thermoregulation, hormonal changes, as well as other uncontrolled factors (Berntson et al., 1997; Kleiger et al., 2005; Sollers, Sanford, Nabors-Oberg, Anderson, & Thayer, 2002; Task Force of ESC, 1996; Taylor, Carr, Myers, & Eckberg, 1998), although it shows utilities in cardiovascular research (Bigger et al., 1992). LF/HF ratio (between 0 and 1) reflects the balance of the sympathetic and parasympathetic functions, with a higher ratio representing SNS dominance and a lower ratio representing PNS dominance (Berntson et al., 1997; Kleiger et al., 2005; Montano et al., 2009; Task Force of ESC, 1996). HF or LF in normalized unit (between 0 to 1) represents the relative value of each component in proportion to 31  the sum of HF and LF (Kleiger et al., 2005; Task Force of ESC, 1996). Similar to LF/HF ratio, HF or LF proportions emphasize the controlled and balanced behavior of the two branches of the ANS while minimizing the effect of the changes in total power on the values of HF and LF components (Kleiger et al., 2005; Montano et al., 2009; Task Force of ESC, 1996). In short-term analysis, VLF, LF and HF compose the total power (TP) (Task Force of ESC, 1996) which may reflect the overall ANS activity but crudely because of the dubious measure VLF (Task Force of ESC, 1996). It is generally considered that a higher TP represents increased overall ANS activity, while a lower TP indicates decreased ANS activity (Task Force of ESC, 1996). The sum of HF and LF (HF+LF), although not a traditional parameter, may be considered as the fraction of HRV that can be totally explained by ANS modulation based on the physiology of HF and LF, and therefore may represent a more precise indicator of the overall ANS activity. We consider that a higher HF+LF value represents increased overall ANS activity, while a lower HF+LF value indicates decreased ANS activity. We have published these data in several articles (Guan et al., 2014; Guan, Collet, Ruitenbeek, & Eibensteiner, Samantha Kissoon, 2014; Fazeli, Pourrahmat, Liu, Guan, & Collet, 2016).  HRV spectral analyses show decent levels of reliability with coefficients of variation (CV) ranging from 6% to 15%, and intraclass correlation coefficient (ICC) from 0.73 to 0.96 in healthy population (Guijt, Sluiter, & Frings-Dresen, 2007; E. B. Schroeder et al., 2004). In patients with cardiac diseases, the CV of HRV frequency-domain measures vary from 4% to 25%, lower than 32  the CV of time-domain measures (up to around 30%) (Piepoli et al., 1996). This indicates that frequency-domain measures of HRV may be more reproducible than time-domain measures. The ICC of HRV frequency-domain measures range from 0.75 to 0.90 in patients with chronic heart failure (Piepoli et al., 1996). In patients with diabetes, ICC for time and frequency domain parameters are from 0.82 to 0.97 (Nussinovitch, Cohen, Kaminer, Ilani, & Nussinovitch, 2012). Logarithmic transformation on the absolute values of HRV could improve the reproducibility of HRV measures (Ponikowski et al., 1996). There are some technical requirements for HRV spectral analyses to ensure its quality. The sampling rate has to be properly chosen. The optimal range is 250–500 Hz or higher (Pinna, Maestri, Cesare, Colombo, & Minuco, 1994; Task Force of ESC, 1996). Ectopic beats, arrhythmic events, missing data and noise effects may alter the estimation of the PSD of HRV. Deletion of ectopic waves may cause selection bias and missing of important data if the occurrence of ectopy is just related to ANS status (Lippman, Stein, & Lerman, 1994). Studies showed remarkable differences of both time and frequency domain parameters between deletion and interpolation methods (Lippman et al., 1994; Salo, Huikuri, & Seppänen, 2001). Therefore, proper interpolation on successive beats on the HRV signals is suggested to be used for editing ectopic waves in HRV analyses (Kamath & Fallen, 1995; Lippman et al., 1994; Salo et al., 2001).    33  Figure 2-2 Power spectral density graph  Red area: ULF and VLF components, 0-0.04 Hz;  Blue area: LF component, 0.04-0.15 Hz;  Yellow area: HF component, 0.15-0.4 Hz;  White area: very high and ultra-high frequency which are highly influenced by respiration and other physiological factors, > 0.4 Hz 34  Table 2.4 Frequency-domain measures in short-term and long-term spectral HRV analyses  Variable  Units  Description  Frequency range    Analysis of short-term recordings (5 min)   5-min total power ms2 The variance of NN intervals over the temporal segment Approximately ≤0.4 Hz VLF ms2 Power in very low frequency range ≤0.04 Hz LF ms2 Power in low frequency range 0.04-0.15 Hz LF norm n.u. LF power in normalized unites LF/(LF+HF) * 100%  HF ms2 Power in high frequency range 0.15-0.4 Hz HF norm n.u. HF power in normalized unites HF/(LF+HF) * 100%  LF/HF  Ratio LF [ms2]/HF [ms2]        Analysis of entire 24 hours   Total power ms2 Variance of all NN intervals  Approximately ≤0.4 Hz ULF ms2 Power in the ultra low frequency range ≤0.003 Hz VLF ms2 Power in the very low frequency range 0.003-0.04 Hz LF ms2 Power in the low frequency range 0.04-0.15 Hz HF ms2 Power in the high frequency range 0.15-0.4 Hz α  Slope of the linear interpolation of the spectrum in a log-log scale Approximately ≤0.4 Hz Information is summarized from the published article by the Task Force of The European Society of Cardiology and the North American Society for Pacing and Electrophysiology (Task Force of ESC, 1996) VLF: very low frequency; LF: low frequency; LF norm: normalized LF; nu: normalized units; HF: high frequency; HF norm: normalized HF; LF/HF: ratio of LF to HF; ULV: ultra low frequency35  2.2.3.2 Using HRV to assess dynamic ANS activity As the activity of ANS is dynamic, HRV is constantly changing over the entire 24 hours. Single HRV values reflect the instantaneous ANS activity or ANS at a specific time, while the changes of HRV between periods reflect dynamic ANS activity in regulating the cardiovascular response to changes in stress (Malliani et al., 1991; Montano et al., 1994, 2009; Pumprla et al., 2002). Therefore, HRV may serve as a proxy for the neurological mechanisms that guide flexible control on physiology and behavior in the context of stress. This provides an important window into understanding stress and health (Thayer, Åhs, Fredrikson, Sollers, & Wager, 2012). Two examples of the changes of HRV corresponding to stress changes are described below. One example of dynamic stress is a change in posture. Montano et al. showed the changes of HRV spectral paramters from at rest to during 90° tilt in young subject (Figure 2.3) (Montano et al., 2009). As the total power is markedly reduced during tilt, the absolute values of both LF and HF are decreased. However, the proportion of HF (i.e. HF n.u.) largely increased at rest, compared to during tilt (Montano et al., 2009).  Another example is the “uncoupling” and “recoupling” theories proposed by Godin et al ( Godin & Buchman, 1996). The “uncoupling theory” suggests that during stress response, the organ responsiveness to autonomic signaling is diminished indexed by decreased HRV and further; this process is enhanced with disease severity (Ellenby et al., 2001; Godin & Buchman, 1996). The uncoupling theory indicates that when stress (either acute or chronic) is not well controlled, the 36  autonomic balance is altered with decreased PNS activity and relatively increased SNS activity, which leads to reduced HRV (Montano et al., 1994, 2009; Thayer et al., 2012). Conversely, recovery from diseases (or other stress context) is characterized by the recuperation of variability in HR or “recoupling” of neuro-organ communication (Ellenby et al., 2001; Godin & Buchman, 1996). “Recoupling” theory signifies the return of ANS modulation, more specifically, increases in parasympathetic activity and suppression of sympathetic activity (Ellenby et al., 2001; Godin & Buchman, 1996).   37  Figure 2-3 Spectral analysis of HRV in a young subject at rest and during 90º tilt    The figure is adapted from the published article by Montano (Montano et al., 2009) The top panels show RR intervals. The middle panels are PSD graphs. Pie charts show relative distribution of HF and LF. VAR: variance; VLF: very low frequency; LF: low frequency; HF: high frequency; au: absolute units; nu: normalized units  38  2.2.3.3 Use of HRV in clinical research The clinical relevance of HRV was first appreciated in 1965 when Hon and Lee noted that fetal distress was preceded by alterations in beat-to-beat intervals (Horn & Lee, 1965). During the 1970s, Ewing et a1. designed a number of simple bedside tests of short-term R-R differences to detect autonomic neuropathy in diabetic patients (Ewing, Martyn, Young, & Clarke, 1985). The association of a higher risk of post-infarction mortality with reduced HRV was first shown by Wolf et al. in 1977 (Wolf, Varigos, Hunt, & Sloman, 1978). In 1981, Akselrod et al. introduced power spectral analysis of HRV to quantitatively evaluate beat-to-beat cardiovascular control (Akselrod et al., 1981). These frequency domain analyses contributed to the understanding of autonomic background of R-R interval fluctuations in heart rate recording (Pagani et al., 1986; Pomeranz et al., 1985). The clinical importance of HRV became appreciated in the late 1980s when it was confirmed that HRV was a strong and independent predictor of mortality following acute myocardial infarction (Bigger et al., 1992; Kleiger, Miller, Bigger, & Moss, 1987; Malik, Farrell, Cripps, & Camm, 1989).  Today, HRV method is an established tool in cardiology research, having well-documented association with cardiac ischemia, cardiovascular and cerebrovascular diseases, and cardiac death (Al-Qudah, Yacoub, & Souayah, 2015; Fyfe-Johnson et al., 2016; Micieli & Cavallini, 2008; Thayer et al., 2012; Thayer, Yamamoto, & Brosschot, 2010; Thayer & Lane, 2007; Yperzeele et al., 2015). It is also increasingly being used for a range of clinical and psychophysiological 39  applications (Beauchaine, 2001; Kleiger et al., 2005; Stein & Kleiger, 1999), including informing risk stratification (Fyfe-Johnson et al., 2016; Huikuri & Stein, 2013).  2.2.4 ANS in Stress Responses: from Adaptation to Disease One of the important roles of ANS is to regulate the stress response (Chrousos, 1998, 2009; Robertson, 2004). Typically, ANS, via sympathetic and parasympathetic branches, provides an instantaneous physiological response to stress that provokes immediate physiological state alterations through neural innervation of the target organs (Chrousos, 1998, 2009; Robertson, 2004). ANS has different responses to different stress situations: normal acute stress, chronic stress, and additional acute stress in the context of chronic stress, which provides a neuromodulation basis to understand the progression from “adaptation” to “stress-related diseases” (Kloet, Joëls, & Holsboer, 2005).   2.2.4.1 Adaptation  Acute stress is defined as stress that lasts for a period of minutes to hours (Johnson et al., 1992). Normally, the stress response to acute stress is physiological and adaptive, with the preservation of homeostasis in the short run (Chrousos & Gold, 1992; Johnson et al., 1992; Lusk & Lash, 2005; Selye, 1955) (Table 2.5). The neuroregulation system is activated during this reaction process in order to control the stress response (Charmandari, Tsigos, & Chrousos, 2005).  40  A typical example of stress adaptation is the physiological inflammatory response (Black, 2003). The inflammatory response is a local protective response which relates to the ANS reaction to stressors (for instance an infectious agent) (Black, 2003). The inflammatory response is beneficial to the body in terms of defending against infection, repairing tissues, adapting to stress and ultimately restoring homeostatic state (Medzhitov, 2008). Part of the pro-inflammatory action is produced by SNS, while the anti-inflammatory response is affected by both SNS and PNS. As described by Tracey (Tracey, 2002), there are three anti-inflammatory reflexes: “the cholinergic anti-inflammatory pathway”, “the sympathetic anti-inflammatory pathway” and “the humoral anti-inflammatory pathway” (Figure 2.4). The cholinergic anti-inflammatory PNS pathway is a rapid neuronal reflex to control the inflammatory response local (Tracey, 2007). The efferent vagus nerves produce acetylcholine to effectively reduce the production of pro-inflammatory cytokines. In addition, the systemic humoral anti-inflammatory pathway (Tsigos & Chrousos, 2002) and the sympathetic anti-inflammatory pathway (Tracey, 2002) are triggered, with the release of stress hormones, including cortisol, aldosterone, epinephrine and norepinephrine to elicit anti-inflammatory effect (Tracey, 2002). In this situation PNS and SNS act synergistically to control the inflammation/stress response. 41  Table 2.5 Behavioral and physical adaptation during mild level of acute stress    Behavioral adaptation  Physical adaptation   Increased arousal and alertness Increased cognition, vigilance and focused attention Euphoria or dysphoria Suppression of appetite and feeding behavior Suppression of reproductive behavior Contrain of stress response    Oxygen and nutrients directed to the CNS and stressed body sites Altered cardiovascular tone, increased blood pressure and heart rate Increased respiratory rate Increased gluconeogenesis and lipolysis Detoxification from endogenous or exogenous toxic products Inhibition of growth and reproductive systems Inhibition of digestion-stimulation of colonic motility Contrain of stress response   Information is summarized from the published articles by Selye and Chrousos ( Selye, 1955; Chrousos & Gold, 1992)  42  Figure 2-4 Anti-inflammatory reflexes by Autonomic Nervous System   The figure is adapted from the published article by Tracey (Tracey, 2002) Inflammatory products produced in damaged tissues activate afferent signals that are relayed to the nucleus tractus solitarius; subsequent activation of vagus efferent activity inhibits cytokine synthesis through the cholinergic anti-inflammatory pathway (‘the inflammatory reflex’). Information can also be relayed to the hypothalamus and the dorsal vagal complex to stimulate the release of ACTH, thereby activating the humoral anti-inflammatory pathway. Activation of the SNS by flight-or-fight responses, or through direct signaling, can increase local concentrations of adrenaline and noradrenaline, which can suppress inflammation further. 43  2.2.4.2 Disease progression and triggering  When the source of stress persists for days to months (chronic), the physiological stress response is affected, which may cause a variety of stress-related disorders (Black, 2003; Schneiderman, Ironson, & Siegel, 2005).  As shown in Figure 2.5 (Chrousos, 2009), when the chronic stressors, such as aging, diet, sedentary life, smoking and psychological stress, continuously and cumulatively affect the stress systems (ANS and HPA axis), excessive catecholamine (by ANS) and cortisol (by HPA axis) are produced. These stress hormones affect the target tissues and cause various metabolic disorders, such as insulin resistance, high blood glucose, visceral obesity, dyslipidemia and high blood pressure. These metabolic disorders and endothelial dysfunction, as “secondary” stressors, may progressively destroy the function of ANS and ultimately lead to cardiovascular and cerebrovascular diseases (Chrousos, 2009).  In other words, inappropriate response to initial stress could be a source of new stress, leading to a sustained vicious circle of mutual reinforcement toward the development of chronic or even severe conditions (Chrousos, 2009; Kloet et al., 2005; Schneiderman et al., 2005). Therefore, it becomes difficult to distinguish between causes and consequences in such dysregulated cascade. The stress system is to a large extent “nonspecific” and meant to interact with internal or external perturbations in a quite similar manner (Kloet et al., 2005).  44  The established traditional risk factors of ischemic stroke, such as age, smoking and obesity, are typical chronic stressors (Chrousos, 2009; Schneiderman et al., 2005). Long term exposure to these chronic stressors leads to a progressive dysfunctional response to stress (Chrousos, 2009; Schneiderman et al., 2005), and in particular, to a limited PNS function to control inflammation (Licht et al., 2010). New conditions are created by those chronic stressors, such as hypertension, hyperglycemia and dyslipidemia. These stress-related disorders act as secondary stressors that aggravate the situation, create new pathological cascades and finally contribute to the development of cardio- and cerebro-vascular diseases (Chrousos, 2009).  On top of that, an additional acute event is a source of excessive (or unopposed) stress in the context of original chronic ones. This overloaded stress may further impair the ANS function (Schneiderman et al., 2005). The acute triggers for ischemic stroke can be seen as sources of acute stress to the body that superimposes on the original chronic stress to increase the overall level of stress; and then directly precipitates the ischemic events (Elkind, 2007; Guiraud et al., 2010; Koton et al., 2004; Sharma et al., 2015; Tofler & Muller, 2006). The dynamic nature of stroke development follows such the pattern in which the acute triggers (such as recent infections and TIA episodes) superimposed on chronic risk factors, serve to precipitate vessel blockage and the consequent cerebrovascular ischemia (Elkind, 2007; Guiraud et al., 2010; Koton et al., 2004; Sharma et al., 2015; Tofler & Muller, 2006).  45  Figure 2-5 Chronic stress, the nervous system and the development of stress-related disorders   The figure is adapted from the published article by Chrousos (Chrousos, 2009) Chronic stress leads to the development of a series of disorders. ABP, arterial blood pressure; ACTH, adrenocorticotropic hormone; APR, acute-phase reactants; AVP, arginine vasopressin; CRH, corticotropin-releasing hormone; iCRH, immune CRH; E, epinephrine; E2, estradiol; GH, growth hormone; HPA, hypothalamic–pituitary–adrenal; IGF-I, insulin-like growth factor I; LC, locus ceruleus; LH, luteinizing hormone; Ne, norepinephrine; T, testosterone; TG, triglyceride. 46  In summary, the adaptive stress system activates a complex repertoire of physiological and behavioral compensatory responses in an attempt to mitigate the adverse consequences relevant to stress and keep homeostasis. However, this compensatory response may fail in the context of chronicity, especially when a new acute stressor is added on. The improper stress responses (excessive magnitude) make the adaptive responses turning prolonged and maladaptive, which relentlessly disrupts normal physiological pathways, damages body biomechanisms, and progressively contributes to the development of stress-related diseases.  A similar logic then may be applied in probing the association between stress, ANS and development of ischemic stroke, as portrayed in Figure 2.6. The stress response plays a role in each phase of stroke progression, as does the ANS modulation on stress response. The vicious cycle may be described as one wherein the initial stress (as risk factors) affects ANS function and causes ANS dysfunctional response to stress, which will cause the development of stress-related disorders (acting as secondary stress) that further impairs ANS function and causes occurrence of TIA or minor stroke. Finally, the initial and secondary stressors, along with dysfunctional ANS will contribute to the development of secondary ischemic events. This procedure leads to an accumulation of stress that affects the entire body, which potentially promotes the development of initial TIAs and the secondary ischemic events.   47  Figure 2-6 Possible link between stress, ANS and progression of ischemic stroke    This process illustrates how initial stress (as risk factors) affects ANS function and causes ANS dysfunctional response to stress, which will cause the development of stress-related disorders (as secondary stress) that further impairs ANS function and causes occurrence of TIA or minor stroke. Finally, the initial and secondary stressors, along with dysfunctional ANS will contribute to the development of secondary ischemic events.48  2.3 Stress-related Risk Factors of Ischemic Stroke and Autonomic Nervous System Dysfunctional ANS response to stress accounts for the development of many stress-related disorders, such as hypertension, hyperglycemia, hyperlipidemia, cardiovascular and cerebrovascular disorders (Kernan et al., 2014; McEwen, 1993, 2007; Chrousos, 2009). These stress-related disorders combined with other stressors/risk factors including aging, smoking, alcohol assumption, sedentary life, psychological stress, may ultimately lead to (i) the development of TIA or minor stroke, and (ii) the recurrence of ischemic stroke or other ischemic events (Furie et al., 2011; Kernan et al., 2014). From this perspective, the traditional risk factors for ischemic stroke are under neuromodulation, which supports the neurogenic hypothesis of ischemic stroke development (Marwah, Doux, Lee, & Yun, 2007). In this section, I will explain the relationship between autonomic dysfunction and the main risk factors of ischemic stroke that include hypertension, diabetes, dyslipidemia, atherosclerosis, cardiovascular disease, atrial fibrillation, psychological stress, aging, physical inactivity, smoking, alcohol consumption, recent infections and recent TIAs.    2.3.1 Hypertension and ANS Dysfunction Hypertension is defined as a systolic blood pressure (SBP) ≥140 mmHg or a diastolic blood pressure (DBP) ≥90 mmHg (Mozaffarian et al., 2016). As a long-term medical condition, hypertension is a consequence of multiple stressors, such as excess salt, obesity, smoking, alcohol, and genetic factors (Ford & Cooper, 1991; Kornitzer et al., 1999); on the other hand, hypertension itself is a source of chronic stress to the body and a typical risk factor for ischemic stroke (Goldstein et al., 2006, 2011; Meschia et al., 2014). 49  The prevalence of hypertension among patients with a recent ischemic stroke is around 70% (Lawes, Bennett, Feigin, & Rodgers, 2004; Sacco et al., 2006). The relative risk (RR) of stroke is around 4 for people aged 50, and around 3 for those aged higher than 60 (Goldstein et al., 2006; Whisnant, Wiebers, Fallon, Sicks, & Frye, 1996). The risk of the first ischemic stroke is directly related to BP starting with an SBP of 115 mm Hg (Lawes et al., 2004), while the relationship between BP and recurrent stroke is relatively under studied but is presumably similar (Furie et al., 2011; Goldstein et al., 2006; Kernan et al., 2014; Sacco et al., 2006). Meta-analyses of RCTs have shown that BP lowering is associated with a stroke risk reduction of 30% to 40% (BPLT Trialists’ Collaboration, 2003; Lawes et al., 2004). The risk reduction is greater with larger reductions in BP (BPLT Trialists’ Collaboration, 2003).  It has been confirmed for several decades that SNS hyperactivity and PNS underactivity are central in the etiology of early and borderline hypertension, as well as sustained essential hypertension (Ferrara et al., 1988; Grassi et al., 2000; Julius, Pascual, & London, 1971; Mark, 1996; Palatini & Julius, 2009). Relevant to the pathological progression of hypertension, total peripheral resistance and cardiac output are the two determinants of arterial pressure (Carthy, 2014; Mancia & Grassi, 2014). Cardiac output is determined by heart rate and stroke volume which are regulated by both PNS and SNS among other regulatory systems (Carthy, 2014; Mancia & Grassi, 2014). The peripheral resistance is determined by functional and structural narrowing in small arteries and arterioles which can be explained by two main systems: an overactive renin-angiotensin system and an overactive SNS, leading to increased stress responses (Esler, Lambert, & Schlaich, 2010; Steptoe & Kivimäki, 2012). In established hypertension, a “neuro-adrenergic” overdrive is evident. Hyperactivity of SNS has been found in both hypertensive males and females (Grassi, 2009); in young, middle-aged, and elderly people with hypertension (Grassi et al., 2000); 50  in pregnancy-induced hypertension (Schobel, Fischer, Heuszer, Geiger, & Schmieder, 1996); and in systo-diastolic hypertension or an isolated elevation of systolic blood pressure (Esler, 2014). The magnitude of the elevation in SNS may be related to the magnitude of hypertension among other factors (Grassi, Cattaneo, Seravalle, Lanfranchi, & Mancia, 1998). Beta-blocker, as a competitive antagonist of beta-adrenergic receptor, is therefore widely used to control hypertension (Wiysonge, Bradley, Volmink, Mayosi, & Opie, 2012). Numerous studies have demonstrated the association between autonomic dysfunction measured by impaired HRV (HF, LF, TP, RMSSD) and hypertension (Liao, Cai, Barnes, & Al., 1996; J Maver, Struci, & Accetto, 2004; E. Schroeder et al., 2003; Singh et al., 1998). HF as an estimate of PNS is significantly lower in hypertensives than in normotensives, after adjusting for age, race, smoking, diabetes and education (Liao et al., 1996). Moreover, it has been reported that people with the lowest quartile of HF had 2.44 (95% CI: 1.15, 5.20) fold risk of hypertension than those with the highest quartile of HF (Liao et al., 1996), suggesting that reduced vagal function and the imbalance of sympatho-vagal function are associated with the risk of developing hypertension.   2.3.2 Diabetes and ANS Dysfunction Similar to hypertension, Type-2 diabetes (T2D) is also a consequence of long-term cumulative chronic stress (obesity, insulin resistance, unhealthy lifestyle) (Chan et al., 1994; Mokdad et al., 2015; Petit et al., 2001); and acts as a source of stress to the body (Goldstein et al., 2006, 2011; Meschia et al., 2014). T2D is associated with a substantially increased risk of the first ischemic stroke (Furie et al., 2011; Goldstein et al., 2006; Kernan et al., 2014; Sacco et al., 2006). The adjusted RR is in the range of 1.5 to 6 (Folson et al., 1999; O’Donnell et al., 2010; Sarwar et al., 2010). Pre-diabetes 51  encompasses impaired fasting glucose, impaired glucose tolerance, and intermediate elevations in HbA1c (5.7%– 6.4%), all of which increase the risk of the first stroke (Lee et al., 2012; Selvin et al., 2010). In the absence of diabetes, insulin resistance is associated with double risk for ischemic stroke (Rundek et al., 2010). Around 60% to 70% of ischemic stroke patients have pre-diabetes or diabetes (Ivey et al., 2006; Matz et al., 2006). Additionally, diabetes is reported to associate with a 60% increased risk for a secondary stroke (RR, 1.59; 95% CI, 1.07–2.37) (Callahan et al., 2011; Kaplan et al., 2005). Abundant evidence has demonstrated that an altered balance of PNS and SNS, mainly explained by an attenuated parasympathetic activity and an elevated sympathetic activity, are causative factors that can trigger a cascade of inflammatory/stress responses in the development and progression of diabetes (Cryer, 2013; Malpas & Maling, 1990; Wellen & Hotamisligil, 2005). Increased circulating insulin may drive a regional increase in sympathetic activity that cause insulin resistance (SNS mediated vasoconstriction antagonizes the insulin’s effect on glucose uptake) (Schlaich, Straznicky, Lambert, & Lambert, 2015; Surwit & Feinglos, 1988). Studies focused on the effect of stress and catecholamines in impairing glycemic control support SNS involvement in the pathophysiology of T2D (Carnethon, Golden, Folsom, Haskell, & Liao, 2003; Franklin et al., 2008; Tarvainen, Laitinen, Lipponen, Cornforth, & Jelinek, 2014). Data from a study in dogs has shown that the hepatic sympathetic nerves inhibit liver glucose uptake while hepatic sympathetic denervation leads to an increase in hepatic glucose uptake (Yi, Fleur, La Fliers, & Kalsbeek, 2010). An attenuated PNS activity is shown to contribute to the development of insulin resistance and T2D with showing significantly lower values in all HRV parameters (HF, LF, TP, SDNN, RMSSN, pNN50) in diabetic patients, compared to healthy controls (Kudat et al., 2006; Liao, Cai, Brancati, & Al., 1995; Tarvainen et al., 2014). The treatment strategies 52  recommended for patients with metabolic disorders, such as diet, exercise and control weight, are usually associated with sympathetic inhibition and parasympathetic improvement, indicated by increases in HF and HF norm, and decreases in LF norm (Pagkalos et al., 2007; Schlaich et al., 2015; Straznicky et al., 2005).   2.3.3 Dyslipidemia and ANS Dysfunction The development of dyslipidemia is a consequence of overall effect of genetics, obesity, lifestyle and other factors (Genest et al., 2009; Reiner et al., 2011). On the other hand, there is a modest association between elevated total cholesterol or low-density lipoprotein cholesterol (LDL-C) and increased risk of ischemic stroke (Ebrahim et al., 2006; Iso, Jacobs, Wentworth, Neaton, & Cohen, 1989; Kurth et al., 2007; Leppälä et al., 1999). For every 1 mmol/L (38.7 mg/dL) increase in total cholesterol, there is an approximately 25% increase in the risk of ischemic stroke (Bots et al., 2002; Horenstein, Smith, & Mosca, 2002). Dyslipidemia has been demonstrated to be associated with SNS activation and PNS suppression (Lambert, Straznicky, & Sari, 2013). Evidence has shown that low HRV (HF, LF SDNN, RMSSD) is associated with high LDL (Christensen, Toft, Christensen, & Schmidt, 1999; Kupari, Virolainen, Koskinen, & Tikkanen, 1993; Liao et al., 1998), total cholesterol (Christensen et al., 1999; Kupari et al., 1993; Liao et al., 1998) and fatty acid concentrations (Marfella et al., 2000), suggesting an impaired PNS function in dyslipidemia individuals.   2.3.4 Atherosclerosis and ANS Dysfunction Multiple stressors including hypertension, dyslipidemia, cigarette smoking and infections contribute to the chronic process of atherosclerosis (Fruchart, 2004; Kullo, Gau, & Tajik, 2000), 53  which includes the progressive development of the plaques over years, and acute rupture of the plaque at a late stage (Elkind, 2006; Withers & Toggweiler, 1993). Meanwhile, atherosclerosis is a source of chronic stress to the body and one of the most important and independent risk factors for ischemic stroke. It is responsible for the thrombosis and occlusion of large brain arteries (large-artery atherosclerosis subtype), associated with an increased risk of small-vessel stroke (lacunar subtype), and also partially contributing to the embolism (cardioembolic subtype) (Elkind, 2006).  According to the prevailing theory proposed by Russell Ross and colleagues (Ross, 1999), atherosclerosis development is predominantly a cascade of inflammation/stress response mediated events, from initiation through progression, rupture and ultimately, to the thrombotic and embolic complications (Hansson, 2005; Hansson, Robertson, & Söderberg-Nauclér, 2006). During the process, ANS dysfunction is characterized by the stimulation of SNS and down regulation of PNS, with subsequent impairment of the tight control of inflammation reaction (Libby, 2002; Peter Libby, 2012; Marwah et al., 2007). The relationship between ANS dysfunction and atherosclerosis development has been well documented (Gidron, Kupper, Kwaijtaal, Winter, & Denollet, 2007; Harris, 2004; Peter Libby, 2012; Marwah et al., 2007). Decreased parasympathetic function (decreased HF and HF norm and increased LF/HF) and increased sympathetic function (elevated heart rate) have been reported to be correlated with the progression of coronary artery atherosclerosis (Huikuri et al., 1999; Manfrini, Pizzi, Viecca, & Bugiardini, 2008) including coronary artery occlusion and plaque rupture (Heidland & Strauer, 2001). ANS also plays a crucial role in thrombogenesis by regulating the chronic stress responses (Shebuski & Kilgore, 2002; Watson, Shantsila, & Lip, 2009).   54  2.3.5 Cardiovascular Diseases and ANS Dysfunction  Cardiovascular diseases share the same risk factors with cerebrovascular diseases such as hypertension, diabetes, dyslipidemia, aging, smoking, physical inactivity, family history, etc., (Goff et al., 2014; Pearson et al., 2002), and thus they can be seen as a consequence of multiple stressors. At the same time, cardiovascular disease is a well-documented independent risk factor for ischemic stroke (Goldstein et al., 2006, 2011; Meschia et al., 2014). Acute myocardial infarction (MI) is a common cardiovascular disorder which refers to a condition of blocked coronaries by thrombosis, impairing blood flow to the heart, leading to a process of “ischemic cascade” and then causing myocardial cells to function improperly or die (Reed, Rossi, & Cannon, 2017). Studies and review articles have shown that during the first 30 days post MI, patients had a 30-fold increase in the rate ratio of ischemic stroke, compared to sex and age matched general population without MI (Sundboll et al., 2016; Witt et al., 2006; Witt, Jacobsen, Weston, & Yawn, 2012).  Increased SNS promotes vasoconstriction, increases platelet aggregation, decreases fibrinolysis, and increases pulse and blood pressure, while decreased PNS leads to reduced arterial pressure and cardiac output that may increase the risk of thrombosis due to sluggish flow and arterial wall collapse (Badimon & Vilahur, 2014; Ciarka, van de Borne, & Pathak, 2008). Numerous studies have illustrated the contribution of autonomic dysfunction for the development of cardiovascular diseases (Curtis & O’Keefe, 2002; Huikuri & Stein, 2013; Sroka, Peimann, & Seevers, 1997; Steptoe & Kivimäki, 2012). In the 1980s, studies had already shown that the low values of both time and frequency domain parameters of HRV (SDNN, ULF, VLF, LF) existed in MI survivors and was correlated with the poor prognosis (including death) after acute MI (Bigger et al., 1992; Kleiger et al., 1987; Huikuri & Stein, 2013). HF appears to be more depressed at the 55  early phase of acute MI with substantial improvement during recovery (Jokinen, Tapanainen, Seppänen, & Huikuri, 2003). Low HF and LF has been shown to be correlated with the development of the first cardiovascular event (Hillebrand et al., 2013).   2.3.6 Atrial Fibrillation and ANS Dysfunction Atrial fibrillation (AF) is one of the high risk cardiac sources for cardioembolic ischemic stroke (Adams et al., 1993; Goldstein et al., 2011; Meschia et al., 2014). All types of AF, including paroxysmal, persistent and permanent, are associated with around 5-fold increased risk of ischemic stroke (Goldstein et al., 2011; Meschia et al., 2014; Riccio et al., 2013; Wolf, Abbott, & Kannel, 1991). In the Framingham study, the risk of ischemic stroke attributable to AF increased from 1.5% at age 50–59, to 23.5% at age 80–89 (Wolf et al., 1991).  Histological studies have shown that the pulmonary veins where the AF impulses originate, are richly innervated by both sympathetic and parasympathetic nerves (Patterson, Po, Scherlag, & Lazzara, 2005). As early as 1978, Coumel et al. reported that the cardiac autonomic dysfunction might predispose patients to develop paroxysmal atrial fibrillation (PAF) (Coumel et al., 1978). Studies on HRV and AF have further determined the crucial role of ANS, with increased SNS and decreased PNS, on the development, progression and maintenance of AF (Amar, Zhang, Miodownik, & Kadish, 2003; Bettoni & Zimmermann, 2002; Herweg, Dalal, Nagy, & Schweitzer, 1998). Patients with the onset of AF have significantly lower values of HF, LF, VLF and TP and increased LF/HF, compared to those without AF occurrence (Jons et al., 2010; Lombardi et al., 2001; Perkiömäki et al., 2014). Moreover, it has been shown that both HF and LF values increase during the 24 hours before the onset of AF, while LF/HF increases during the 24 hours but decreases sharply at 5 minutes before PAF (Bettoni & Zimmermann, 2002; Herweg et al., 1998). 56  This result suggests a primary increase in SNS followed by short vagal predominance before the PAF (Bettoni & Zimmermann, 2002; Herweg et al., 1998). To inhibit SNS, the classic treatment for all types of AF is the administration of beta-blockers (Ali, 1997; Fuster et al., 2006, 2011; Healey et al., 2005).   2.3.7 Aging and ANS Dysfunction Aging is a chronic stressor to the body and leads to ANS alteration (Collins, 1997; Hotta & Uchida, 2010; Marigold, Arias, Vassallo, Allen, & Kwan, 2010). It is a well-established risk factor of ischemic stroke and also contributes to the generation of other risk factors for ischemic stroke such as hypertension, diabetes and cardiovascular diseases (Goldstein et al., 2011; Meschia et al., 2014). The risk of ischemic stroke doubles for each successive decade after age 55 (Brown, Whisnant, Sicks, O’Fallon, & Wiebers, 1996; Carandang et al., 2006; Manolio, Kronmal, Burke, O’Leary, & Price, 1996). Autonomic dysfunction in seniors is attributed to several main features associated with aging, such as loss of neurons, loss of axon branches, alterations in neurotransmitters, and degenerative changes in effector organs innervated by autonomic nerves (Collins, 1997; Hotta & Uchida, 2010; Marigold et al., 2010). Many clinical syndromes associated with aging are caused by inadequate autonomic responses to physiological stressors (Marigold et al., 2010). Cardiovascular autonomic reflexes show altered responsiveness with aging, including RSA, vagal baroreflex, cardiopulmonary reflexes, tachycardia, facial cooling bradycardia and cold pressor reflexes (Collins, Abdel-Rahman, Easton, Sacco, & Ison, 1996; Eckberg & Sleight, 1992). Increase in blood pressure and decrease in baroreflex function in the elderly may be indicators of high SNS and down-regulated PNS activity, which leads to increased ventricular instability and less 57  protective effects (Collins, 1997; Hotta & Uchida, 2010; Marigold et al., 2010). Many studies have shown that elderly people have significantly lower values in HF, LF and TP powers than young people (Antelmi et al., 2004; Korkushko, Shatilo, Plachinda YuI, & Shatilo, 1991; Stein, Barzilay, Chaves, Domitrovich, & Gottdiener, 2009; Yeragani, Sobolewski, Kay, Jampala, & Igel, 1997).   2.3.8 Genetic Factors and ANS Dysfunction  The genetic factor is a non-modifiable factor that may modify the stress to the body systems. It is associated with the development of ischemic stroke (Goldstein et al., 2006, 2011; Meschia et al., 2014). Meta-analysis of cohort studies has shown that a positive family history of stroke increases the risk of stroke by around 30% (OR, 1.3; 95% CI, 1.2–1.5; P<0.001) (Floßmann, Schulz, & Rothwell, 2004). Numerous studies have reported that family members of people with hypertension or diabetes have reduced PNS and increased SNS activity, indexed by low HF, LF TP and SDNN as well as increased LF/HF (De Angelis et al., 2001; Ferrara et al., 1988; Lindmark, Wiklund, Bjerle, & Eriksson, 2003; Maver, Strucl, & Accetto, 2004). These findings may imply that the basic ANS function as well as HRV values are influenced by family history.   2.3.9 Psychological Stress and ANS Dysfunction Psychosocial stress, caused by occupational, familial or life events for instance, is recognized as a potential contributor to an individual’s perceptions of stress (Campbell & Ehlert, 2012; Macleod et al., 2002; Suadicani, Andersen, Holtermann, Mortensen, & Gyntelberg, 2011) and possible interaction with the ability to cope with the specific demands (Kemeny, 2003). Various components of psychosocial stress, including self-perceived stress, stressful life events and poor coping ability to stress, are associated with an increased risk of ischemic stroke (Booth 58  et al., 2015; House, Dennis, Mogridge, Hawton, & Warlow, 1990; Jood, Redfors, Rosengren, Blomstrand, & Jern, 2009; O’Donnell et al., 2010; Surtees et al., 2007; Truelsen, Nielsen, Boysen, & Gronbaek, 2003). Previous studies reported that the perceived psychosocial stress measured by Perceived Stress Scale (PSS) score was associated with increased risk of fatal stroke (HR 1.45, 95 % CI: 1.19,1.78; P < 0.001) and total ischemic stroke (HR 1.40, 95 % CI: 1.00,1.97; P = 0.05) (Booth et al., 2015). High level of chronic psychological stress leads to continuous activation of the stress system with prolonged secretion of stress mediators such as CRH, norepinephrine and cortisol (Campbell & Ehlert, 2012; Chrousos, 1998; Macleod et al., 2002). This  eventually turns to overstimulation of SNS and suppression of PNS (Chrousos, 2009). It has been reported an inverse relationship between HF and normalized HF, and perceived stress (Dishman et al., 2000; Jorna, 1992; Miu, Heilman, & Miclea, 2009), and a LF/HF increase during a short-term psychological stress (Delaney & Brodie, 2000; Jorna, 1992). This indicates a lower cardiac vagal activity among people who perceived more psychological stress.  2.3.10 Sedentary Life and ANS Dysfunction Sedentary life is a traditional risk factor of ischemic stroke and also correlated with other risk factors of ischemic stroke such as atherosclerosis, cardiovascular disorders, hypertension, diabetes and hyperlipidemia (Adams et al., 2007; Furie et al., 2011; Goldstein et al., 2006; Johnston et al., 2006; Kernan et al., 2014; Sacco et al., 2006). Exercise can reduce BP (Haskell et al., 2007; Whelton, Chin, Xin, & He, 2002), improve endothelial function (Endres et al., 2003), reduce insulin resistance, improve lipid metabolism (Gordon et al., 2004; Williams et al., 2002), and may help reduce weight (Goodpaster et al., 2010). Regular physical activity has been reported to be 59  associated with a 10% to 30% reduction in the incidence of stroke and cardiovascular diseases in both men and women (Haskell et al., 2007; Lee, Folsom, & Blair, 2003; O’Donnell et al., 2010; Shiroma & Lee, 2010; Wendel-Vos, 2004). Studies focused on ANS and sedentary life have shown that sedentary life is associated with autonomic imbalance, mainly the decreased parasympathetic activity with showing the decreases in HF, pNN50 and RMSSD, while exercise may improve the autonomic function with increases in HRV (Boutcher & Stein, 1995; Carter, Banister, & Blaber, 2003; Melanson & Freedson, 2001; Rosenwinkel, Bloomfield, Arwady, & Goldsmith, 2001).   2.3.11 Smoking and ANS Dysfunction  Cigarette smoking is another important independent risk factor for ischemic stroke (Mast et al., 1998; Wolf, D’Agostino, Kannel, Bonita, & Belanger, 1988), and contributes to an increased risk for silent brain infarction (Howard et al., 1998; Putaala et al., 2009). A systematic review article showed that an overall relative risk of stroke was 1.5 times higher in smokers than in non-smokers (Shinton & Beevers, 1989). In the Cardiovascular Health Study, smoking was found to be associated with a substantially increased risk for recurrent stroke in the elderly (HR, 2.06; 95% CI, 1.39–3.56) (Kaplan et al., 2005). Lower HRV values (HF, HF norm, LF and TP) and higher LF/HF ratio are found in smokers than in nonsmokers (Barutcu et al., 2005; Hayano, Yamada, Sakakibara, & Al., 1990; Minami, Ishimitsu, & Matsuoka, 1999; Niedermaier et al., 1993).  2.3.12 Alcohol Consumption and ANS Dysfunction Although light to moderate alcohol consumption may be associated with a reduced risk of first-ever stroke (Meschia et al., 2014); heavy alcohol use, binge drinking, and acute alcohol ingestion are reported to increase the risk of the first stroke (Guiraud et al., 2010; O’Donnell et al., 60  2010; Sundell, Salomaa, Vartiainen, Poikolainen, & Laatikainen, 2008) and recurrent strokes (Ois et al., 2008). Whether through binge doses or high cumulative lifetime consumption, alcohol abuse is clearly deleterious for the cardiovascular system, increasing the incidence of total and cardiovascular mortality, coronary and peripheral artery disease, heart failure, hypertension, dyslipidemia, and diabetes mellitus and stroke (Fernández-Solà, 2015; Smyth et al., 2015). Many early studies have shown that people suffering from alcohol abuse had decreased HRV values (RMSSD, HF) and increased LF/HF ratio, compared to healthy controls (Koskinen, Virolainen, & Kupari, 1994; Mäki et al., 1998; Thayer, Hall, Sollers, & Fischer, 2006).  2.3.13 Recent Infections and ANS Dysfunction  Infection is one important acute trigger/stress, in the context of all conventional chronic risk factors/stress, that could trigger the development of ischemic stroke (Elkind, 2007; Guiraud et al., 2010). It is evident from various studies that infections, mainly those affecting the respiratory tract, induce a transient state of increased stroke susceptibility in the week following infection (McColl, Allan, & Rothwell, 2009; Paganini-Hill et al., 2003). This theory is supported by a number of plausible mechanisms that link systemic infection and inflammation with the “stroke prone” state (Elkind, 2007; Guiraud et al., 2010; Koton et al., 2004; Sharma et al., 2015; Tofler & Muller, 2006).   It is considered that infection possibly contributes inflammation to atherosclerotic plaque pathology (McColl et al., 2009), accelerating the maturation of plaques and promoting plaque instability and rupture via the activation of inflammatory cytokines (Emsley & Tyrrell, 2002; McColl et al., 2009). Furthermore, acute infection is shown to be associated with endothelia dysfunction (McColl et al., 2009; Vallance, Collier, & Bhagat, 1997). Plentiful studies have 61  demonstrated the relationship between HRV changes and infections (Barnaby et al., 2002; Buchman, Stein, & Goldstein, 2002; Goldstein et al., 1998; Korach et al., 2001; Schmidt, Werdan, & Müller-Werdan, 2001; Toweill, Sonnenthal, Kimberly, Lai, & Goldstein, 2000). Specifically, patients with infections report significantly decreased HF, LF and TP, as well as increased LF/HF, compared to the recovery states (p<0.05) (Garrard, Kontoyannis, & Piepoli, 1993; Piepoli et al., 1995; Toweill et al., 2000), and healthy controls (p<0.05) (Annane et al., 1999; Buchman et al., 2002; Korach et al., 2001). Moreover, decreased HRV (HF, LF and TP) and increased LF/HF ratio are identified in association with the stage of infection and with deterioration, which supports that HRV could be an indicator of illness severity (Barnaby et al., 2002; Buchman et al., 2002; Goldstein et al., 1998; Toweill et al., 2000).    2.3.14 Prior TIA and ANS Dysfunction Prior TIA confers a higher risk of secondary ischemic stroke, as described in section 2.1.5. A recent TIA is considered a source of acute stress added to the chronic stressors, and thus increases the risk of secondary ischemic events in patients. Autonomic dysfunction of different degrees likely exists in all ischemic stroke patients (Bassi, Colivicchi, Santini, & Caltagirone, 2007; Korpelainen, Sotaniemi, & Myllylä, 1999). As expected, sympathetic cardiovascular modulation was found to be increased in patients after stroke (Dütsch, Burger, Dörfler, Schwab, & Hilz, 2007). Post–acute stroke patients may be at an increased risk for cardiac disorders in the face of such unopposed sympathetic stimulation (Al-Qudah et al., 2015). Numerous studies have shown a parasympathetic cardiac deficit with decreased HRV (HF, LF and TP) in post–acute stroke irrespective of the side of the ischemia (Dütsch et al., 2007; Korpelainen et al., 1999; Korpelainen, Sotaniemi, Huikuri, & Myllyä, 1996; Naver et al., 1996; Raedt, Vos, & Keyser, 2015). Low HRV 62  is shown to be predictive of stroke mortality (Pozzati, Pancaldi, Di Pasquale, Pinelli, & Bugiardini, 1995). Additionally, it has been hypothesized that parasympathetic activation may carry some neuroprotective effects and enhance neurogenesis (Cheyuo et al., 2011; Mravec, 2010).   So far, I have laid out a number of well-identified risk factors for TIA and minor stroke and their relationship with ANS function. The exact impacts of these risk factors however, are difficult to quantify, given individual heterogeneity and personal susceptibility (genetic predisposition or protection, for instance). Furthermore, other unidentified or unaddressed risk factors may also contribute to stroke development, but they may not have been given appropriate consideration in determining an individual’s risk profile. While critical, it is difficult to estimate the risk of stroke occurrence or recurrence after initial TIA, and to clearly identify who is at a higher risk of developing secondary ischemic events, among those with a burden of chronic risk factors/stress.  The assessment of HRV measures the objective physiological response to all stress-related factors. For an individual, HRV values might represent the comprehensive effect of “multiple stressors” at a given point in time. From this perspective, would it be possible to use HRV data (as a marker of autonomic activity and adaptation) to identify the personalized risk of developing a second ischemic event after initial TIA or minor stroke? In the next chapter, I will explain the rationale of this study.63  Chapter 3: Rationale   The study rationale is based on the following logic: Risk factors = stressors. According to the definition of stress, any disturbance of homeostasis including physiological, psychological and environmental, are considered as stress sources (Chrousos & Gold, 1992). Therefore, many risk factors of ischemic stroke, including the traditionally identified ones, hard-to-measured ones and unidentified ones, are either chronic or acute stress sources. All these factors may be responsible for the development of ischemic stroke.    ANS regulates stress responses. ANS is the regulatory system for stress responses. The assessment of autonomic activity may provide a good way to represent the condition of stress modulation, and the level of stress. The risk factors for ischemic stroke and TIA, as stressors or stress-related conditions, are associated with ANS responses.   HRV is a measurement of ANS activity, and it thus may reflect stress. HRV is a precise measurement of ANS activity. Basal ANS function can be assessed through the absolute values of HRV. Dynamic ANS function can be assessed by the changes in HRV parameters. Assessing HRV values may provide a precise and valid assessment of the overall stress the individual faces at a given time. Assessing HRV changes may reflect the changes of stress overtime, increasing or decreasing. 64   HRV may reflect the risk factors of ischemic stroke. HRV assessment might capture the effects of conventional risk factors in ABCD2 tool and other stressors not well measured. HRV could then be, for one individual, a comprehensive marker of the impact of all stress-related risk factors of ischemic stroke.   Therefore, HRV parameters, as markers of multiple stressors, may be good predictors of the occurrence of a secondary ischemic event after TIA or minor stroke; the models based on these HRV predictors may show better prediction than ABCD2 tool.   Our study was aimed at establishing the possible use of HRV data to predict the occurrence of secondary ischemic event after TIA or minor stroke and improve the traditional ABCD2 tool. The theory/logic of this approach is presented in Figure 3.1.   65  Figure 3-1 The logic of the HRV-based stress predictive model                         All risk factors (stressors) have effects during the progression from TIA to the development of outcome events (ischemic stroke, TIAs, cardiovascular events, and vascular death). ANS is directly affected by the risk factors/stressors; on the other hand, dysfunctional ANS conversely contributes to the development of risk factors/stressors. The study is then aimed at identifying whether HRV parameters (as markers of ANS activity) can predict the occurrence of secondary outcome events.  66  Chapter 4: Prior Hypotheses on Identification of HRV Predictors  Before presenting the study objectives in the next chapter, we need first to describe the HRV parameters that will be used to predict the occurrence of a secondary ischemic event after TIA or minor stroke. In normal conditions, ANS activity has a circadian rhythm with PNS increasing during nighttime and SNS activating during daytime. Such circadian rhythm leads to a 24-hour HRV rhythm. In healthy situations, HF power that presents the PNS activity is increased during the night and relaxation period, such as napping. The morning time, especially the few hours after wake-up is the period of SNS activation, which leads to the decrease of HF (6 am-9 am) (Figure 4.1) (Ishida et al., 1997). HF then recovers after 9 am and remains comparatively stable with small fluctuations during daytime and early evening (around 9 am – 9 pm) (Figure 4.1). Earlier study have shown the 24-hour rhythm of different HRV parameters (HF, LF, total power) in healthy populations and in diseased populations (Figure 4.2) (Burger, Charlamb, & Sherman, 1999): people with diabetes and chronic stable angina show less day-night rhythm in HF, compared to healthy people.  67  Figure 4-1 24-hour rhythm of HF in healthy population     The figure is adopted from published article by Ishida (Ishida et al., 1997)    68  Figure 4-2 24-hour rhythm of HF, LF and TP in healthy people, patients with diabetes and with chronic angina    The figures are adopted from the published article by Burger (Burger et al., 1999) A. 24-hour rhythm of HF in healthy people, patients with diabetes and with chronic angina.  B. 24-hour rhythm of LF in healthy people, patients with diabetes and with chronic angina.  C. 24-hour rhythm of TP in healthy people, patients with diabetes and with chronic angina69  Based on the expected values and changes of HRV during 24-hour rhythm in both healthy and diseased people, we identified the types of HRV data we expected to use and several time periods of interest, i.e. static (absolute values of HRV parameters) and dynamic (HRV parameters changes over time)  • Regarding HRV parameters: HF is the primary HRV parameter we used in the study, because it is a precise indicator of PNS activity. Normalized HF which represents the proportion of PNS to ANS was also selected. TP as a marker of overall ANS activity was included in the assessment. Additionally, based on the description of HF and LF in Figures 4.1 and 4.2, as well as our explanation/presumption of HF+LF in section 2.2.3.1, HF+LF was selected as another potential maker of overall ANS function. Although as two classic HRV parameters, LF is considered to be regulated by both SNS and PNS without knowing the proportion of the effect of each subsystem; and VLF cannot be explained by only ANS. Their results are provided in Appendix-I (see Chapter 7 Results).   • Regarding time periods: We selected 9 am to 12 pm to represent “morning time” in order to avoid the sharp decrease in PNS and increase in SNS immediately after wake-up (Burger et al., 1999); 3 to 6 pm to represent afternoon time to avoid the effect of the nap (based on the habit of Chinese people); and 12 to 3 am to represent the night time, when there is most likelihood that patients are sleeping.  • Use of HRV absolute values: Figure 4.2 showed HRV parameters (HF, LF and TP) in patients with diabetes and chronic angina are consistent lower than in healthy people. It 70  reveals that the absolute values of HRV parameters are indicators of people’s health condition. Those with a lower absolute value of HRV may have a higher level of stress. We then postulated that lower HRV is associated with worse health conditions and thus higher risk of developing secondary ischemic events, after acute TIA or minor stroke.   • Use of HRV changes during daytime: From both Figure 4.1 and 4.2, in healthy people, HRV remains stable from late morning (after 9 am) to afternoon (around 6 pm). According to the uncoupling theory described in section 2.2.3.2, the decreased HRV signifies the diminished ANS responses; and this process is progressed with disease severity. It is therefore reasonable to state that the decreases in HRV from morning (9 am) to afternoon (6 pm) would indicate an increased stress and diminished ANS modulation, and perhaps the deterioration of health condition. We postulated that the decrease in HF (or TP, HF+LF) might be associated with a higher risk of developing a new ischemic event (stress-related). Conversely, organ recoupling indexed by increased HRV (explained in section 2.2.3.2) would represent the return of ANS modulation. It is then reasonable to consider that an increase in HRV during 9 am to 6 pm would indicate reduced stress with ANS adaptation recovery, and perhaps improvement of health conditions in TIA or minor stroke patients. This change might be associated with a lower risk of developing new ischemic events.   • Use of HRV changes between day and night: As illustrated in Figure 4.2, HRV demonstrates greater day-versus-night discrepancies among healthy population than among people with diabetes or chronic angina. We had reasons to believe that decreased amplitude of HRV changes between day and night may suggest a reduction in ANS 71  modulation (less restoration of ANS activity and less control of stress) and a higher risk of secondary ischemic events.   From these considerations, we identified the HRV parameters, which are the most important ones we considered, to predict the occurrence of ischemic events after TIA or minor stroke. The first selection of candidate HRV predictors included those that were related to HF, because they represent PNS activity and should be lower in patients under stress. We provide their definitions and expected association in our study below.  • The absolute value of HF is our first predictor. Although no study has investigated the value of HF on secondary ischemic events prediction after TIA or minor stroke, baseline absolute value of HF had been shown to be lower in stroke patients than in healthy people (Dütsch et al., 2007; Korpelainen et al., 1999; Korpelainen et al., 1996; Naver et al., 1996; Raedt et al., 2015). Therefore, we decided to test the absolute value of HF as our first predictor, with a lower value of HF associated with a greater burden of body stress, and higher risk of developing secondary ischemic events after the initial TIA or minor stroke episode. The same hypotheses applied to HF.nu, HF+LF and total power.  • The HF change from morning (9 am – 12 pm) to afternoon (3 pm – 6 pm) is our second predictor, with less increases or decreases being associated with a worse condition of patient, and higher risk of developing secondary ischemic events after initial TIA or minor stroke. The same hypotheses applied to HF.nu, HF+LF and total power.   72  • The daytime-nighttime changes of HF (difference between day and night, or between morning and night) is our third predictor, with a lower change being associated with worse condition of patient, and a higher risk of developing a secondary ischemic event after initial TIA or minor stroke. The same hypotheses applied to HF.nu, HF+LF and total power.  73  Chapter 5: Objectives and Hypotheses   One important aspect of our study was exploratory, given the limited information published concerning the application of HRV parameters in predicting secondary ischemic events after TIA or minor stroke. The overarching objective of the study was to select and assess the quality of HRV parameters for ischemic events prediction, and to develop HRV-based stress models. More specifically, the primary objective of this study was to identify the best HRV predictor, from the ones we selected a priori, to predict the ischemic events after TIA or minor stroke. The secondary objectives focused on exploring the respective value of assessing psychological stress for ischemic events prediction; and comparing HRV-based models with traditional ABCD2 tool. We selected several models for the best adoption in different settings. Finally, the exploratory objectives were aimed at finding appropriate cut-offs for the candidate HRV predictors and HRV-based models, as well as attempting to develop more complex exploratory models. Study objectives and corresponding hypotheses are depicted below.  5.1 Primary Objective and Hypothesis The primary objective was to identify the best HRV indicator (assessed within 48 hours from the initial events), which could be an independent predictor of developing a secondary ischemic event during the 90-day period after the initial acute TIA or minor stroke.  Based on the prior hypotheses on identification of HRV predictors (Chapter 4), the candidate HRV predictors we tested were: Moring HF, Daytime HF changes, Day-night HF changes, and other HRV parameters (HF norm, HF+LF, total power).   74  Primary hypotheses were: - HF parameter (either Morning HF, Daytime HF changes, or Day-night HF changes) would be the best predictor of a secondary ischemic event after acute TIA or minor stroke; with expected negative correlation.  - Other HRV parameters (HF norm, HF+LF and TP) will also be associated (negatively correlated) with the occurrence of secondary ischemic events.   5.2 Secondary Objectives and Hypotheses Two secondary objectives were investigated in the study: • To examine whether the level of perceived psychological stress PSS score is associated with an increased risk of ischemic events development in patients with acute TIA or minor stroke.  - Hypothesis: Patients with higher PSS score would have an increased risk of ischemic events after TIA or minor stroke, compared to patients with lower PSS score. • To develop HRV-based stress models and compare their predictive values with ABCD2 score. - Hypothesis: The area under curve (AUC) of receiver operating characteristic curve (ROC) of HRV-based stress models should be higher than the AUC of ABCD2 score.  5.3 Exploratory Objectives and Hypotheses Another two exploratory objectives were tested in the study.  • To identify specific cut-off points for HRV predictors and cut-off formulas for HRV-based models, depending on different clinical purposes that lead to different selections of 75  sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV). Defining cut-off points is one critical step towards the use of the predictive tools in clinical practice.  - Hypothesis: Specific cutoffs for HRV predictor and HRV-based models can help identify the high-risk group.   • To investigate the effect of using a combination of HRV absolute values and HRV changes in the same model to increase discrimination ability of the model to identify individuals’ risk of developing secondary ischemic events. - Hypothesis: Both HRV absolute values and HRV changes have specific independent predictive values on ischemic events. Their combination will increase the predictive ability of the model.  76  Chapter 6: Methods  6.1 Clinical Care and Patients Monitoring in Study Setting  The study was conducted in the Department of Neurology at Beijing Tiantan Hospital, Beijing, P.R. China. The normal clinical care and monitoring protocol in this particular study setting is important to describe in order to provide a clear picture to understand the study methodology.  All acute events are referred/guided to the Emergency Clinic. The Neurology Emergency Clinic in Beijing Tiantan Hospital is a 24-hour clinic that receives an average of 60 to 80 patients with neurological symptoms per 24 hours. The preliminary diagnosis in the Emergency Clinic is based on the evaluation of neurological symptoms and the result of non-enhanced CT scan. It normally takes 30 minutes to receive the CT scan results. When the CT scan and clinical report are returned to the neurologist, it takes 20 to 30 minutes for the neurologist to evaluate the condition including assessment of ABCD2 score of TIA patients, and to make preliminary decisions regarding diagnosis, urgent treatment and the need for hospitalization. Due to the limitation of plain CT scan accuracy, the preliminary diagnosis of TIA in the Emergency clinic is mostly based on the time-based diagnostic criteria: typical neurological ischemic symptoms and duration (<24 hours), medical history and neurological examination. Patients who need urgent care without hospitalization are sent to the “Observation Room” to receive such care; this “Observation Room” is simply equipped with responsible neurologists and nurses to attend to the patients. Patients who need to be hospitalized are triaged to the ward. If a bed is unavailable at that moment, the “Observation Room” also serves as a temporary place for patients to get urgent treatment and 77  wait for the available beds in the ward. The stay time in Observation Room can go from few hours to several days based on the health status of patients and availability of beds in the ward.  For those hospitalized, the normal hospitalization duration is 10-14 days. The medical care follows three-level-physician responsibility system. During hospitalization, each patient has one level-I physician (resident) who operates the care, including sending patients to clinical tests and imaging, collecting all test results, preparing the preliminary medical records and dealing with daily events, whenever necessary. The level-II physician (specialist) is fully responsible for the patient’s personal care, which includes seeing the patient during ward rounds every morning, prescribing drugs and examinations, revising the medical record, and evaluating the overall health condition of the patient. The level-III physician takes ward rounds every other day; double checks the prescriptions; solves difficult and/or urgent issues for a group of patients. Routine comprehensive clinical tests, including biochemical test, imaging examinations (MRI, MRA/CTA), and 24-hour Holter monitoring, are prescribed immediately after hospitalization (normally within 1 hour). Vessel examination such as TCD and CUS, cardiac function examination such as echocardiography, and other imaging tests for chest, abdomen and limbs may also be prescribed if needed.  The 24-hour Holter examination is a guideline-driven routine clinical test for all hospitalized patients who have a suspected ischemic stroke in order to classify the etiology. The requisitions are sent to the Holter Laboratory where the specialists are responsible for arranging the Holter monitoring in all hospital departments. The schedule of these tests is based on the health status of patients and the availability of the laboratory. Because cardiac patients are typically prioritized for Holter monitoring; it normally takes 2 to 7 days to schedule the Holter for patients in Department of Neurology. Patients who are scheduled will be sent to the Holter Laboratory to 78  be fitted with the Holter monitor between 8 am and 12 pm. The Holter monitor will be removed by nurses at 7 am on the following morning and sent back to the Holter Laboratory. The 24-hour Holter data are imported to the central computer in the Holter laboratory. The turnaround time of Holter report usually takes 1 to 2 business days. During our study, patients followed the routine medical care process, but received special support from the Holter Laboratory in scheduling study patients within a short timeframe to respect the study protocol; this will be described in section 6.5.3.1.   6.2 Study Design  This is a prospective cohort study on patients who developed an acute episode of TIA or minor stroke. After baseline assessment of clinical information and study variables within 48 hours of the initial event, patients were followed for 90 days with periodic contacts for assessing the occurrence of outcome events.  6.3 Study Population Patients were eligible for the study if they met all study inclusion criteria and did not have any exclusion criterion, as described below.   • Inclusion criteria: - Over 40 years old - Diagnosed with TIA or minor stroke   Minor stroke is defined as an acute episode of focal or global neurological dysfunction caused by brain, spinal cord, or retinal vascular injury as a result of infarction (Hicks et al., 2014, 2015; Sacco et al., 2013); and NIHSS less than 4 with 79  each item ≤ 1 and normal consciousness (Furie et al., 2011; Jauch et al., 2013). It is normally evaluated based on the persistent signs/symptoms of the ischemic event at the time of recruitment, or an acute ischemic brain lesion documented by CT scan or MRI.   TIA is defined as a transient episode of focal neurological dysfunction caused by brain, spinal cord, or retinal ischemia, without permanent cerebral infarction (Easton et al., 2009; Hicks et al., 2014, 2015). At the study setting, TIA is normally evaluated by the neurologist through a history or neurological examination according to the time-based diagnosis criteria (<24 hours) (Albers et al., 2002; Johnston, 2002) in Emergency Clinic; and revised through further imaging examinations according to tissue-based diagnosis criteria (Easton et al., 2009) in the wards (as describe in section 6.1).   - Ischemic symptoms occurred in the past 48 hours - To be able to understand and sign the informed consent •  Exclusion criteria:  - Had already developed an outcome event before hospitalization - Having lethal disease such as cancer  - Having severe cognitive disorder - Loss of consciousness due to any causes - Any reason that leads to patients cannot wear/start Holter monitor within 48 hours after the initial events - All other conditions that, in the opinion of the clinician may represent difficulties for the subjects’ full participation in the study. 80  6.4 Recruitment  The study screening and recruitment in Beijing Tiantan Hospital were conducted in the Neurology Emergency Clinic of Department of Emergency Medicine and the wards of Department of Neurology. The Neurology Emergency Clinic is open 24 hours daily. The study manager, Dr. Ling Guan (a licensed physician) conducted the initial screening in the Neurology Emergency Clinic, and evaluated the time eligibility in the Neurology ward. If the patient was eligible for the study, the study coordinator talked to him/her and explained the study to the families as well. To respect the time limitation of this study, the Holter Laboratory consented to schedule the eligible patients in advance (will be described in section 6.5.3.1) rather than following the normal schedule as described in section 6.1. If the patient could meet the time eligibility of wearing Holter monitor and was interested in participating in this study, he/she was asked to sign the consent form (this action is normally conducted in Holter Laboratory). A copy of the informed consent form was given to the patients for their reference. Because TIA and minor stroke episodes are emergent events and study participation had to be immediate to respect the 48 hours’ delay, the patients had a short window to consent. However, all information for the study was collected as part of normal care; the scheduling of Holter monitoring in advance potentially provided benefit to the patients.  6.5 Study Variables and Assessment The demographic and clinical information of study patients were collected. All study patients had their ABCD2 score computed and a timely 24-hour Holter monitoring. They also completed the perceived stress scale.   81  6.5.1 Demographic and Clinical Information Demographic and clinical information included age, gender, current symptoms and duration, preliminary and final diagnosis, historical and present illnesses including atrial fibrillation, hypertension, diabetes, dyslipidemia, cardiovascular diseases (coronary artery disease, MI and/or unstable angina) and cerebrovascular disease (TIAs and/or ischemic stroke), as well as medication use (drugs alone, or drugs with intravenous thrombolysis or endovascular treatment). Medical history was based on patients’ statement. Presence of atrial fibrillation, hypertension, diabetes, dyslipidemia and cardiovascular diseases were diagnosed by specialists (through symptoms and results of clinical tests) during hospitalization, and also patients’ reporting. All information was collected from medical charts and test reports. Self-reporting of smoking and alcohol intake in terms of dose per day and the number of years were recorded. The type of alcohol intake (Chinese white wine, wine, beer) was also recorded. The smoking and drinking status were categorized into three levels: never, low and high. “Never” refers to patients never had or never had a habit of this action (rare occasions). The high-level smoking was defined as more than 20 cigarettes per day for more than 20 years. The high-level drinking was defined as more than 100 ml Chinese white wine per day or equal alcohol of other types (300 ml wine or 800 ml beer), for more than 20 years. “Low-level” was excluding the “Never” or “High-level”. This categorization of smoking and drinking was developed based on several good reviews and studies on the mechanism and account of smoking/alcohol on cardiovascular/stroke risk (Fernández-Solà, 2015; Kauhanen, Kaplan, Goldberg, & Salonen, 1997; Toustads, 2006; Wannamethee, Shaper, Whincup, & Walker, 1995; Wolf et al., 1988), as well as the specific life habits of the Chinese population. 82  All applicable examination results, including neurological examination, biochemical tests, neuroimaging examination (plain brain CT scan in Emergency clinic, brain MRI, MRA/CTA when being hospitalized), vessel examinations (TCD, CUS), cardiac function tests (echocardiography, 24-hour Holter), were extracted from the Electronical Medical System. An electronic copy of the 24-hour Holter raw data file was available to extract individual’s parameters and conduct HRV analyses.  6.5.2 Assessment of ABCD2 Score The ABCD2 score (Appendix C) is a traditional predictive tool widely used in clinical and research. The ABCD2 score is a risk assessment tool designed to improve the prediction of a secondary ischemic stroke after an episode of TIA (Johnston et al., 2007a; Perry et al., 2011; Sanders et al., 2012; Wardlaw et al., 2015). Sometimes it is also used for minor stroke (Johnston et al., 2007a; Perry et al., 2011; Sanders et al., 2012; Wardlaw et al., 2015). The ABCD2 score is calculated by summing up points for five independent factors: age, blood pressure, clinical features, duration of symptoms, and diabetes (Johnston et al., 2007a). Age, blood pressure and diabetes count for 1 score each, and clinical feature and duration of symptoms for 2 scores each; scores for each item are added together to produce an overall score ranging between zero and seven (Johnston et al., 2007a). ABCD2 score can be used with the specific score (as continuous data) (Johnston et al., 2007a; Perry et al., 2011) or categorized into different risk levels: (i) 0-3 score as low-risk, 4-5 score as medium-risk and 6-7 score as high-risk (Cancelli et al., 2011; Coutts et al., 2008); or (ii) 0-3 score as low risk, and ≥ 4 score as high risk (Amarenco, Labreuche, & Lavallée, 2012). Two-level categorization is normally used in clinical research to recruit “high risk TIA patients” with ABCD2 ≥ 4 scores (Amarenco et al., 2017; Wang et al., 2013). The interrater 83  reliability of ABCD2 score was reported to be fair (kappa is around 0.6) in several studies (Ishida et al., 2013; Ishida, Kasner, & Cucchiara, 2015). The ABCD2 score is computed for all TIA patients at the time they come to the Emergency clinic in the study setting (described in section 6.1). For study needs, the ABCD2 score was also computed for our patients with minor stroke by the responsible neurologist. The study manager conducted the secondary evaluation. Disagreement between the responsible neurologist and the study manager were discussed to reach consensus.  6.5.3 Assessment of Autonomic Activity (Heart Rate Variability) In this study, we used HRV measurement to assess the ANS function because it is a valid, reliable, non-invasive, and convenient measurement of overall cardiac autonomic activity, as described in section 2.2.3. Each patient benefited from 24-hour Holter recording, which simplified testing our a priori hypotheses regarding the changes of ANS function over 24 hours. Since HRV data can be derivated from the 24-hour Holter recording, patients did not need to do any extra testing. HRV testing is therefore considered the most practical assessment. Finally, identification of high risk patients by using HRV tool has been reported in previous studies (Fyfe-Johnson et al., 2016; Huikuri & Stein, 2013; Mäkikallio et al., 2004; Tsuji et al., 1994), which enables us to compare our results with the published ones.   6.5.3.1 24-hour Holter monitoring As described in section 6.1, instead of the 2 to 7 days it takes normally to organize Holter monitoring for neurology patients, the study patients were scheduled at the earliest available date. If patients came to the hospital in the morning, they were scheduled to wear Holter that morning 84  (as soon as the requisition was received). If patients came in the afternoon or evening until midnight, they were given Holter in the following morning at 8 am. All study patients should wear/start Holter within 48 hours after initial events. If the time of starting Holter would exceed the time criteria, the patient was not eligible for the study (described in section 6.3). Holter monitors were removed by nurses at 7 am in the next morning and sent back to the Holter Laboratory. The 24-hour Holter data were uploaded to the central computer in the Holter lab and then were transferred to the study computer. The raw data of 24-hour Holter were extracted and transformed into voltage format. The sampling rate is 250 Hz.   6.5.3.2 HRV assessment  The 24-hour Holter data in voltage format was used for the analyses of HRV in the study software Acqknowledge 4.1 (BIOPAC Systems, Inc., Goleta, USA) (Appendix F), which can process the ECG data and automatically identify R peaks. For the method of HRV analyses, we used frequency-domain analysis which is considered as a unique, convenient, and precise procedure for assessing autonomic function (Kamath & Fallen, 1993; Malliani et al., 1994; Montano et al., 1994, 2009; Myers et al., 1986). We used non-parametric spectral analyses with FFT which has several advantages, such as less cost, simple algorithm and high processing speed (Task Force of ESC, 1996). In our study, the analyses were performed twice to ensure consistency. The main HRV frequency-domain parameters assessed in the study include:  • HF power (ms2) : reflecting PNS function (Task Force of ESC, 1996).  • HF norm (n.u.): presenting the HF proportion to the sum of HF and LF, which emphasizes the controlled and balanced behavior of the two branches of ANS (Task Force of ESC, 1996).  85  • HF+LF (ms2): a fraction of HRV that can be totally explained by autonomic modulation (Guan et al., 2014; Task Force of ESC, 1996). It is well known that HF is modulated by PNS only, while LF is regulated by both SNS and PNS without knowing the proportion of each branch affecting the LF power (Task Force of ESC, 1996). We then have reasons to believe that the sum of HF and LF is modulated by PNS and SNS (i.e. the overall ANS) without influence of other physiological factors. • Total power (ms2): the sum of HF, LF and VLF in the short-term (2-5 minutes) HRV recording (Task Force of ESC, 1996). It is a traditional HRV variable considered to reflect the overall ANS activity although its component VLF is influenced by many other physiological factors (Task Force of ESC, 1996) • LF and VLF were not included in the primary analyses. Their results are shown in Appendix I.  As described in Chapter 4, we took three hours in the morning 9 am - 12 pm to represent the “morning period” (also as the baseline), three hours in the afternoon 3 pm - 6 pm, to represent the “afternoon period”, and three hours during the night 12 am - 3 am to represent the “night period” for HRV analyses. Two types of recording were available for the frequency-domain analyses: short-term recording over 2-5 minutes, for short-term analyses at different time periods, and long-term recording over the entire 24 hours for long term analyses of HRV during the circadian cycle. We conducted the short-term analysis method for two reasons: (i) it reflects the HRV values in different time periods (such as, morning, or afternoon), and allows for assessment of the changes across two periods; (ii) not all patients complete the entire 24-hour recording which limits the circadian analysis.  86  For the analysis of each period, we developed the following algorithm: using the first 5-minute interval in each half an hour to analyse HRV. The HRV during a specific time period (i.e. morning, afternoon and night) is then calculated from the mean of 3 hours’ data. If the 5-min ECG waves contain 5% or higher ectopic waves (such as in patients with persistent or paroxysmal AF), this segment was not selected (Salo et al., 2001). The 5-min recordings with ectopy free were prior to be selected in order to ensure the quality of analysis. Ectopic beat was interpolated by the mean of six adjacent normal beats (three before and three after) (Salo et al., 2001). Definitions for the selected HRV parameters are described below.  • Morning HRV value: an average value of six 5-minutes fragments from 9 am to 12 pm • Afternoon HRV value: an average value of six 5-minutes fragments from 3 pm to 6 pm • Night HRV value: an average value of six 5-minutes fragments from 12 am to 3 am  • Day HRV value: an average of Morning HRV and Afternoon HRV • 24-hour (24-h) HRV value: an average of Day HRV and Night HRV • Daytime HRV changes (defined as “the percentage difference between morning value and afternoon value”): Afternoon HRV−Morning HRVMorning HRV∗ 100% • Day-night HRV changes (defined as “the percentage difference between daytime value and night value”: Night HRV−Day HRVDay HRV∗ 100% • Morning-night HRV changes (defined as “the difference between morning value and night value”): Night HRV−Morning HRVMorning HRV∗ 100% Note: “Changes of HF norm” is defined as the real differences between time periods rather than the percentage changes  87  6.5.3.3 Preliminary selection of HRV parameters  The selection of candidate HRV predictors was based on a series of hypotheses regarding the potential predictive values of different HRV variables and the practicability of data recording, as explained in Chapter 4. The selected primary predictors include four forms of absolute values (morning value, day value, night value and 24-h value) and three forms of changes (daytime changes, day-night changes, and morning-night changes), for HF, HF+LF, HF.nu, and total power (Figure 6.1). They were all tested in the modeling.  88  Figure 6-1 Selected HRV parameters    All definitions have been given in section 6.5.3.2. HRV absolute valuesMorning HRV (Baseline HRV):•Morning HF•Morning HF norm•Morning HF+LF•Morning TPDay HRV:•Day HF•Day HF norm•Day HF+LF•Day TPNight HRV: •Night HF•Night HF norm•Night HF+LF•Night TP24-h HRV: •24-h HF•24-h HF norm•24-h HF+LF•24-h TPHRV changesDaytime HRV changes•Daytime HF changes•Daytime HF norm changes•Daytime HF+LF changes•Daytime TP changesDay-night HRV changes•Day-night HF changes•Day-night HF norm changes•Day-night HF+LF changes•Day-night TP changesMorning-night HRV changes•Morning-night HF changes•Morning-night HF norm changes•Morning-night HF+LF changes•Morning-night TP changes89  6.5.4 Assessment of Perceived Psychological Stress         All study patients completed the validated 10-question PSS created by Sheldon Cohen (Appendix G). The PSS is the most widely used psychological instrument for measuring the degree to which situations in one’s life in the last month are appraised as unpredictable, uncontrollable, and overloaded perceived stress (Booth et al., 2015; Cohen, Kamarck, & Mermelstein, 1983; Jood et al., 2009). The PSS items are straightforward and the response alternatives are simple to grasp. In each case, respondents are asked how often in the last month they felt a certain way. The total score for PSS is 40, with the higher scores indicating higher levels of perceived stress. In general, normal populations have PSS score lower than 15; people having scores of 20 or higher are considered to be under high stress (Cohen, 1988; Cohen et al., 1983). Both internal consistency reliability (Cronbach’s alpha) and test-retest reliability (ICC) of the PSS score were evaluated at >0.70 in all 12 studies from the review article on psychometrics property of PSS score (Lee, 2013). In our study, the perceived stress scale was given to the study patients at the earliest convenient time after their hospitalization. The research physician was available to provide any explanation of the questions if needed.   6.6 Follow-Up  We followed all study patients for 90 ± 5 days. The scheduled contacts were: face to face contact during the first 7 days (hospitalization), face to face contact on the day of discharge (normally between day 10 to day 14), phone contact at the 30 ± 5 day and the 90 ± 5 day.  During follow up, the development of a secondary ischemic event was recorded, with all symptoms that may be related to ischemia and the occurrence time. Patients’ overall health condition and any change in their clinical status including medication usage were recorded. 90  Furthermore, patients were asked to remember any symptoms they developed and the occurrence time. When symptoms (such as facial palsy, limb weakness and visual problems, speech problems, chest pain, etc.) occurred between two scheduled contacts, patients were encouraged to contact the research physician immediately for advice. All available information regarding diagnostic workup, imaging reports, and final diagnoses was kept for study reference.  6.7 Outcome Events Identification  The outcome events are composed of different ischemic events: ischemic stroke, TIA, cardiovascular events, and cerebro- and cardio-vascular death within 90 days after the initial TIA or minor stroke. The definition of each outcome ischemic event is given in section 6.7.1. The validation process for each event recorded is described in section 6.7.2.  6.7.1 Definition of Ischemic Events  The definitions of outcome ischemic events are based on the Standardized Draft on Definitions for Cardiovascular and Stroke Endpoint Events in Clinical Trials (Hicks et al., 2010, 2014, 2015; Thygesen et al., 2012). These definitions have been used in many clinical studies (Amarenco et al., 2017; Furie et al., 2011; Jauch et al., 2013; Wang et al., 2013). The definition of each outcome event is described below.   • Cerebrovascular ischemic events include ischemic strokes and TIAs. - Ischemic stroke: Ischemic stroke is defined as an acute episode of focal or global neurological dysfunction caused by brain, spinal cord, or retinal vascular injury as a result of infarction (Hicks et al., 2010, 2014; Sacco et al., 2013). Ischemic stroke is normally evaluated based on the persistent signs/symptoms at the time of patient 91  showing up, or an acute ischemic brain lesion assessed by CT scan or MRI (Sacco et al., 2013). One of the following three conditions is identified as ischemic stroke outcome (Hicks et al., 2010, 2014). o Sudden onset of a new focal neurologic deficit, with clinical or imaging evidence of infarction lasting 24 hours or more, and not attributable to a non-ischemic cause (i.e., not associated with brain infection, trauma, tumor, seizure, severe metabolic disease, or degenerative neurologic disease). o A new focal neurologic deficit lasting for less than 24 hours and not attributable to a non-ischemic cause, but accompanied by neuroimaging evidence of new brain infarction. o Rapid worsening of an existing focal neurologic deficit lasting more than 24 hours and not attributable to a non-ischemic cause, accompanied by new ischemic changes on MRI or CT of the brain and clearly distinct from the index ischemic event. - TIA events: TIA is defined in previous text (Chapter 2, and section 6.3) as a transient episode of focal neurological dysfunction caused by brain, spinal cord, or retinal ischemia, without permanent cerebral infarction (Easton et al., 2009; Hicks et al., 2014, 2015). A TIA event refers to neurological deficit symptoms that are without evidences of acute focal infarction of brain, and are not attributable to a non-ischemic etiology (such as brain infection, tumor, seizure, severe metabolic disease, or degenerative neurological disease). Diagnosis has to be made by specialists. AHA/ASA recommends duration ≥24 h as an operational definition of ischemic stroke rather than TIA (Easton et al., 2009; Hicks et al., 2015). 92  • Cardiovascular ischemic events: the cardiovascular ischemic events include myocardial infarction and unstable angina. Details are provided in Appendix H. - Myocardial infarction: In general, myocardial infarction is defined as clinical syndrome where there is evidence of myocardial necrosis in a clinical setting consistent with acute myocardial ischemia. The detection of a rise and/or fall of cardiac biomarker values, and with at least one of the following can be diagnosed with myocardial infarction (Hicks et al., 2010, 2014):  o Symptoms of ischemia o Evidence from ECG: New or presumed new significant ST-segment-T wave changes or new left bundle branch block or development of pathological Q waves in the ECG o Evidence from imaging (echocardiography): new loss of viable myocardium or new regional wall motion abnormality o Identification of an intracoronary thrombus by angiography or autopsy.  - Unstable angina: Unstable angina can be diagnosed where the following are satisfied (Hicks et al., 2010, 2014): o Ischemic discomfort (angina, or symptoms thought to be equivalent) ≥10 minutes occurring at rest or in an accelerating pattern with frequent episodes associated with progressively decreased exercise capacity  o New or worsening ST or T wave changes on resting ECG, or definite evidence of inducible myocardial ischemia, or angiographic evidence, or need for coronary revascularization procedure for the presumed lesions.  o Negative cardiac biomarkers and no evidence to be diagnosed with MI 93  • Cerebro- and cardio-vascular death: cerebro- and cardio-vascular death includes death due to stroke, acute myocardial infarction, heart failure, cardiovascular procedures, cardiovascular hemorrhage or other cardio- and cerebro-vascular causes, and sudden cardiac death (Hicks et al., 2014; Thygesen et al., 2012). Details are provided in Appendix H.  6.7.2 Assessment and Validation of Outcome Events The assessment of study outcome events was performed under different situations:  • If the event was recorded in the medical charts: For the events that occurred during the hospitalization, the signs and symptoms were recorded in the medical charts (some of them also had imaging evidences). Confirmation was made by responsible neurologists. • If the event was identified by the study manager: For the events that occurred after discharge, the signs and symptoms were recorded by the study manager.  - If the symptoms were typical, the responsible neurologist could validate these cases as “outcomes”. - If the symptoms were atypical, or without imaging tests or other supportive tests, the diagnosis of an outcome ischemic event was only “suspected”. Two senior neurologists adjudicated these cases. They reviewed the cases independently and selected one of the descriptive terms - “definite”, “probable”, “possible or “unlikely”. If the classification was concordant, the category was selected. For discrepant evaluation, a discussion between the two neurologists and the study PI/manager was made to reach consensus.   Definitions of the assessment terms “Definite”, “Probable”, “Possible” and “Unlikely” are based on the Causality Assessment System proposed by the World 94  Health Organization Collaborating Centre for International Drug Monitoring, the Uppsala Monitoring Center (WHO-UMC https://www.who-umc.org/). In our study, “Definite” refers to “The cases that strictly meet the definition of outcome events, with definite aggregative ischemic symptoms and/or new ischemic symptoms, and NIHSS scores increased to higher than 4, with or without imaging evidences”; “Probable” refers to “The cases that meet the definition of outcome events, with aggregative ischemic symptoms and/or new typical ischemic symptoms, normally without imaging evidences”; “Possible” refers to “The symptoms of the cases can be considered as a secondary ischemic symptom”; and “Unlikely” refers to “The symptoms of the cases are not related to ischemic events”.   6.8 Study Outcomes  We had three levels of study outcomes: • Primary outcome:  - Identification of the best HRV parameter in predicting occurrence of outcome events after initial TIA or minor stroke  • Secondary outcomes: - Development of HRV-based predictive models  - Selection of the best stress model - Selection of the most practical model - Comparison between the selected HRV-based models and ABCD2 score   95  • Exploratory outcomes: - Identification of cutoffs for selected HRV predictors and HRV-based models - Development of exploratory models with a combination of HRV absolute values and HRV changes  6.9 Sample Size and Power Calculation  In initial study design, we planned to use cox-regression for the predictive analysis. We assumed that the risk rate for secondary ischemic events would be 15% within 90 days in patients with TIA or minor stroke (Chandratheva et al., 2009; Coull et al., 2004; Johnston et al., 2000; Kleindorfer et al., 2005; Rothwell & Warlow, 2005; Wu et al., 2007). There was no previous data of HRV assessment in predicting a secondary ischemic event after TIA/minor stroke; therefore, we used published information of ABCD2 score to estimate our sample size. It was shown in previous studies that under the standard treatment, the risk ratio between high ABCD2 ≥4 (15%) and ABCD2 <4 (3%) within 90 days was 5 (Cancelli et al., 2011; Coutts et al., 2008; Johnston et al., 2007b; Rothwell et al., 2005). We assumed the standard deviation (SD) of ABCD2 score was 1. Sample size calculation was based on having 90% power to find a significant (p<0.05) hazard ratio of 2 (conservative estimate) between high and low HRV groups (equal numbers of patients in each group), if such a difference exists. The minimum number of study patients needed was 148. This number would be corrected if more covariates added into the model. The general rule requires 5 to 10 more outcome events for each one more covariate added (Vittinghoff, Glidden, Shiboski, & McCulloch, 2005). Given that the event rate is assumed to be 15% in our study, at least 33 patients (33 * 15% = 5) are needed for every one more covariable added.  96  After data collection, we found that the time to event was short in most outcome events, i.e. most cases occurred in the first several days and very few cases happened after 14 days. We then decided to use logistic regression rather than cox-regression since we were more concerned about whether the event happened, rather than measuring the time to event. We recalculated the power of the test by using our real data. The power calculation for logistic regression followed the standard methods (Chow, Shao, & Wang, 2007; Hsieh, 1989). Based on the real data we obtained in the study, the power is 0.71 if using “Morning HF value” as the primary predictor. Following the same method, the power is higher than 0.9 if using “Daytime HF changes” as the primary predictor.   6.10 Statistical Analyses Demographic and clinical characteristics were described as categorical variables with showing the number and percentage of patients in each category. HRV values and changes were described with median and interquartile range (IQR). ABCD2 score and PSS score were presented with mean and 95% confidence intervals (CI). A p value of 0.05 was used in significance testing; an extended p level of 0.1 was also accepted based on the study purpose and variable importance. We used logistic regression modeling for the prediction analysis. Because the distribution of the absolute values of HRV parameters (HF, HF+LF and TP) is right skewed, natural logarithmic transformation (log) was used for HRV absolute values when establishing the models. ORs of HF, HF+LF and TP then indicate “how event risk changes for every 1 unit increase in log (HRV)”. For the changes of HRV parameters, ORs originally indicate “how event risk changes for every 1 unit increase (i.e. 100% increase) in HRV parameter”; for practical purpose, the ORs were calculated to represent a change of risk for every 10% increase in HRV parameters during the 97  defined time period. Different from HF, HF+LF and TP, HF norm represents a proportion (from 0 to 1) and follows the normal distribution in our study; therefore, no log transformation was used. OR of HF norm was to represent the changes of event risk for every 0.1 increase in HF norm.  Dummy variables were created for categorical variables in the regression model. We developed univariate models for ABCD2 score, HRV parameters, and PSS score, to compare the predictive values between HRV parameters, PSS score and ABCD2 score. Kaplan-Meier analysis was also conducted to compare the event-free curves across different levels of the predictor.  We also established bivariate and full models (with three predictors: HRV, ABCD2 score and PSS score) to test whether adding HRV and PSS score into the model can improve the predictive power of ABCD2 score. The best predictive model was selected based on the AUC of the ROC curve. The most practical model was selected based on the AUC and clinical application. The AUC comparisons were conducted between the HRV-based predictive models and traditional predictive tool ABCD2 score.  The cut-off values were identified for the selected HRV parameters; and the cut-off formulas were developed for the selected HRV models. The identification of the cutoffs was based on the general rule which is to maximize the Youden’s index (sensitivity + specificity – 1), i.e. to get the maximum sum of sensitivity and specificity (Youden, 1950).  Finally, we developed and evaluated more complex exploratory stress models with a combination of HRV absolute values and HRV changes, ABCD2 score, and PSS score.  All statistical analyses were conducted by using R 3.3.2. All models are notated as: Event ~ predictor A + predictor B + predictor C (if any). This model notation is presented in R software. 98  Chapter 7: Results  Patients recruitment and data collection were conducted in Beijing Tiantan Hospital affiliated to Capital Medical University from Oct 2014 to May 2016. Data analyses were conducted at The University of British Columbia in Vancouver. The study was approved by The University of British Columbia Children’s and Women’s Clinical Research Ethics Board (UBC C&W NUMBER: H14-00435) and Research Ethics Board of Beijing Tiantan Hospital affiliated to Capital Medical University (NUMBER: KY2014-015-03).   7.1 Recruitment A total of 206 patients with a recent episode of TIA or minor stroke (<48 hours) were recruited in Beijing Tiantan Hospital from October 2014 to February 2016. The Neurology Emergency Clinic in Beijing Tiantan Hospital is a 24-hour clinic that receives about 60 to 80 patients with neurological symptoms per 24 hours within which an average of 5 to 6 patients are diagnosed with possible TIA or minor stroke. However, only 4 to 5 per week (on average) developed their symptoms in the past 48 hours. During the 17 months’ recruitment period, approximately 3000 patients who were preliminarily diagnosed with TIA or minor stroke (no time limitation) were screened; the number of eligible patients (within 48 hours) was 347. The criteria for preliminary diagnosis of TIA or minor stroke in Emergency Clinic were described in section 6.1.  Among the 347 patients who met the time-window of 48 hours, 276 consented to be hospitalized (while 71 refused). Due to the limited number of beds in the ward, 261 could be hospitalized within 48 hours after the initial TIA or minor stroke events. A 24-hour Holter 99  examination was prescribed for all study patients (standard care). Among the 261 patients, 240 were scheduled to wear the Holter monitor in the nearest morning and within the time-window of 48 hours. Among these patients, 34 patients refused to have the Holter test or could not arrive to wear Holter monitor in time, or refused to participate in the study. Finally, 206 patients were recruited to the study. Among the recruited patients, 5 patients were excluded afterwards from the final data analyses due to the correction of the preliminary diagnosis in Emergency clinic or unavailable ECG data: 3 TIAs were corrected to the diagnosis of benign paroxysmal positional vertigo; 1 TIA was revised diagnosed to Meniere’s disease, 1 patient had permanent atrial fibrillation which made ECG data unusable for HRV analyses. A total of 201 patients were included in the final analyses. The recruitment rate is calculated as:  The recruitment rate = Number of patients being recruitedTotal number of eligible patients  = 206/347 = 59.4% The recruitment flow chart is shown in Figure 7.1.   100  Figure 7-1 Study recruitment flow chart                                                Eligible patients in the Emergency Clinic (N=347) Consent to be hospitalized (N=276)  Be able to have Holter test within 48 hours (N=240)  Be hospitalized within 48 hours after initial events (N=261)  Recruited in the study (N=206)  Be included in final data analyses (N=201)  Refuse to attend study, refuse to have Holter test, or delay the Holter recording (N=34)  Cannot have Holter test within 48 hours (N=21) Cannot get available beds in time (N=15)  Refuse to be hospitalized (N=71) 1. Corrected to TIA mimics including benign positional vertigo and Meniere’s disease (N=4)  2. HRV data unavailable due to permanent atrial fibrillation (N=1)  101  7.2 Identification and Validation of Outcome Events 7.2.1 Identification of Possible Outcome Events The composite outcome includes ischemic stroke, TIAs, cardiovascular ischemic events (myocardial infarction, unstable angina) and cerebro- or cardio-vascular death. Each outcome event is defined based on the “Standardized Definitions for Cardiovascular and Stroke End Point Events in Clinical Trials” (Hicks et al., 2014; Thygesen et al., 2012) that has been descibed in section 6.7.1.  During the 90±5 days’ follow up, 47 patients were recorded as developing symptoms that were compatible with a diagnosis of an ischemic event that would qualify as outcomes for the study. Among them, 16 cases whose clinical expression was difficult to interpret were sent to two experienced neurologists for review and adjudication.  7.2.2 Validation of Suspect Outcome Events Two experienced neurologists adjudicated the 16 cases to one of the following categories - “unlikely”, “possible”, “probable”, or “definite” (definitions are in section 6.7.1). • One made decisions of “unlikely” on 11 cases, “possible” on 4 cases and “probable” on 1 case. • The other made decisions of “unlikely” on 7 cases, “possible” on 7 cases, “probable” on 1 case and “definite” on 1 case.  Among the 16 cases being validated, the major disagreement was between “unlikely” or “possible” for the outcome of “TIAs occurrence”. The 11 cases determined as “unlikely” by the first neurologist included the 7 cases rated as “unlikely” by the second neurologist. By integrating 102  the judgement from the two experts and further discussion between two neurologists and study PI/manager, we conservatively removed all these 11 cases with at least one adjudication of “unlikely” from the 47 cases. The rest 5 cases (out of 16 cases reviewed) were included and determined as “possible” (for 4) or “probable” (for 1). The total number of outcome events of the study was 36, among which 24 cases were considered as “definite”, 8 were considered as “probable”, and 4 were considered as “possible”.   7.2.3 Description of Cases  A total of 36 cases (24 as “definite”, 8 as “probable”, 4 as “possible”) out of the 201 patients were identified as outcome events; the total event rate was 17.9%, with 11.9% (24/201) of definite cases and 15.9% (32/201) of definite and probable cases. Among the 36 outcome ischemic events, 26 (72%) were ischemic strokes, 7 (19%) were TIAs, 2 (6%) were cardiovascular problems (myocardial infarction or unstable angina), 1 (3%) was vascular death.  The 90-day survival curve is shown in Figure 7.2. 17 (8.5%) cases occurred in the first 48 hours after wearing Holter monitor, 28 (13.9%) occurred within 7 days, 30 (14.9%) in 14 days, 32 (15.9%) in 30 days, and 36 (17.9%) in 90 days.  103  Figure 7-2 The 90-day event-free curve     Baseline  48 hours 7 days 14 days 30 days 90 days Cumulative No. of events 0 17 28 30 32 36 Risk rate  0 8.5% 13.9% 14.9% 15.9% 17.9%  The 90-day event-free curve (solid line) shows the distribution of patients with event-free in 90 days after the initial TIA or minor stroke events. Dash lines represent the 95% CI of the event-free curve. Outcome events include ischemic stroke, TIAs, cardiovascular ischemic events (myocardial infarction, unstable angina), and cerebro- and cardio-vascular death.104  7.3 Demographic and Clinical Information  A total of 201 patients were included in the final analyses (age =59 ±10 years; male =163, female =38). No patient withdrew from the study. Demographic and clinical information are shown in Table 7.1, including age, gender, final diagnosis (TIA or minor stroke), TOAST classification, historical or present atrial fibrillation, hypertension, diabetes, dyslipidemia and cardiovascular diseases, a history of cerebrovascular disease, smoking and alcohol status, medication, as well as the recruitment time delay from initial event.  49 patients were preliminarily diagnosed with TIA in the Emergency Clinic, among them 9 were revised to minor stroke based on further imaging tests (MRI, MRA or CTA) after being hospitalized. Therefore, finally 40 patients (19.9%) were diagnosed as TIA and 161 patients (80.1%) were diagnosed as minor stroke. Based on TOAST classification, 157 were diagnosed with large-artery atherosclerosis source, 14 with cardioembolic source, 10 with lacunar, 5 with other causes and 15 with undetermined etiology. Among 201 patients, 17 (8.5%) had historical AF or were identified AF by 24-hour Holter monitoring during hospitalization; 113 (56.2%) had hypertension or were newly diagnosed with hypertension; 41 (20.4%) had or were newly diagnosed with diabetes; and 51 (25.4%) had or were newly diagnosed with dyslipidemia. Patients who had historical or present cardiovascular diseases counted for 22.9% of total patients. Patients who had a history of cerebrovascular disease (ischemic stroke or TIAs) counted for 24.4% of total patients.  105  Table 7.1 Demographics and clinical information  Characteristics Patients, N=201  No. (%) Age     < 60 years 101 (50.2)   ≥ 60 years 100 (49.8) Gender      Female  38 (18.9)   Male  163 (81.1) Final diagnosis      TIA 40 (19.9)   Minor stroke  161 (80.1) TOAST classification      Large-artery atherosclerosis 157 (78.0)   Cardioembolism  14 (7.0)   Lacunar  10 (5.0)   Other causes 5 (2.5)   Undetermined causes  15 (7.5) Atrial fibrillation      Yes 17 (8.5)   No 184 (91.5) Hypertension      Yes 113 (56.2)   No 88 (43.8) Diabetes     Yes 41 (20.4)   No 160 (79.6) Dyslipidemia     Yes 51 (25.4)   No 150 (74.6) Cardiovascular diseases     Yes  46 (22.9)   No  155 (77.1) History of cerebrovascular disease     Yes  49 (24.4)   No  152 (75.6) Smoking dose*years      Never  64 (31.8)   Low  38 (18.9)   High  99 (49.3) Drinking dose* years      Never  75 (37.3)   Low  48 (23.9)   High  78 (38.8) 106  Table 7.1 Continuous   Characteristics Patients, N=201  No. (%) Medication     Drugs alone  173 (86.1)   Drugs and operation 28 (13.9) Recruitment time delay from initial event     <12 hours 85 (42.3)   12-24 hours  63 (31.3)   24-48 hours  53 (26.4) Cardiovascular diseases include coronary artery disease, MI and unstable angina. A history of cerebrovascular diseases includes ischemic stroke and TIA. Atrial fibrillation includes both persistent and paroxysmal subtypes. Definitions of smoking and drinking are provided in section 6.5.1. Medical operation includes intravenous thrombolysis or endovascular treatment.   107  The comparisons of demographic and clinical information between the group with secondary ischemic events (N=36) and the group without event (N=165) are shown in Table 7.2. The age distribution between two groups was similar (60.8 vs. 58.5 years, p=0.22). Presence of hypertension, diabetes and dyslipidemia were significantly greater in the event group than in the event-free group (hypertension: OR=3.3, p=0.006; diabetes: OR=2.8, p=0.012; dyslipidemia: OR=2.2, p=0.043). However, AF, cardiovascular disease, a history of cerebrovascular disease, cigarettes smoking or alcohol consumption, were not associated with the development of secondary ischemic events. 108  Table 7.2 Comparisons of demographic and clinical information between event group and event-free group   Clinical and personal factors  Patients with events, N=36 Patients with event-free, N=165  Odds ratio (95% CI)  p value   No. (%) No. (%) Age > 60 years 20  (55.6) 80 (48.5) 1.3 (0.6, 2.8) 0.44 Gender of male 30 (83.3) 133 (80.6) 1.2 (0.5, 3.4) 0.71 Diagnosis of TIA 9  (25.0) 31 (18.8) 1.4 (0.6, 3.3) 0.40 Atrial fibrillation  3 (8.3) 14 (8.5) 1.0 (0.2, 3.2) 0.98 Hypertension  28  (77.8) 85 (51.5) 3.3 (1.5, 8.1) 0.006 Diabetes  13  (36.1) 28 (17.0) 2.8 (1.2, 6.0) 0.012 Dyslipidemia 14  (38.9) 37 (22.4) 2.2 (1.0, 4.7) 0.043 Cardiovascular disease 8  (22.2) 38 (23.0) 0.9 (0.4, 2.2) 0.92 History of cerebrovascular disease 10  (27.8) 39 (23.6) 1.2 (0.5, 2.7) 0.60 Smoking         Low dose-years, compared to never 6  (16.7) 32 (19.4) 0.9 (0.3, 2.6) 0.86   High dose-years, compared to never 19  (52.8) 80 (48.5) 1.1 (0.5, 2.7) 0.75 Drinking         Low dose-years, compared to never 11  (30.6) 37 (22.4) 1.7 (0.7, 4.4) 0.25   High dose-years, compared to never 14  (33.3) 64 (38.8) 1.3 (0.5, 3.1) 0.58 Medication         Drugs alone, compared to drugs and      operation 30 (83.3) 143 (86.7) 1.3 (0.5, 3.3) 0.60 Recruitment time (from initial events)         12-24 hours, compared to 24-48 hours 12 (33.3) 51 (30.9) 1.8 (0.7, 5.7) 0.26   <12 hours, compared to 24-48 hours 18 (50.0) 67 (40.6) 2.1 (0.8, 6.2) 0.14 Cardiovascular diseases include coronary artery disease, MI and unstable angina. A history of cerebrovascular diseases includes ischemic stroke and TIA. Atrial fibrillation includes both persistent and paroxysmal subtypes. Definitions of smoking and drinking are provided in section 6.5.1. Medical operation includes intravenous thrombolysis or endovascular treatment. Red font: significant p values (<0.05)109  7.4 HRV Parameters, PSS Score and ABCD2 Score The absolute values and changes of HRV parameters are presented with median and IQR in Table 7.3. Among the 201 patients, 3 patients in the event-free group had unavailable ECG data during nighttime; the number of patients for HRV analyses during night period was therefore 198 (36 in event group, 162 in event-free group). Other predictors included PSS score and ABCD2 score. The average ABCD2 score in all patients was 4.5 (95% CI: 4.3, 4.7). The average PSS score was 18.2 (95% CI: 17.7, 18.7). 110  Table 7.3 HRV, PSS score and ABCD2 score in study patients  Predictors Patients, N=201  HRV absolute values, median (IQR)     HF (ms2)       Morning HF value  49.3 (27.3, 100.9)     Day HF value  53.9 (29.1, 99.5)     Night HF value 129.7 (61.8, 249.6)     24-h HF value  90.6 (48.5, 176.7)    HF norm (n.u.)       Morning HF norm  0.27 (0.18, 0.39)     Day HF norm 0.28 (0.21, 0.40)     Night HF norm 0.43 (0.29, 0.58)     24-h HF norm 0.36 (0.27, 0.48)    HF+LF (ms2)       Morning HF+LF value  199.0 (107.9, 316.9)     Day HF+LF value 184.4 (110.4, 311.6)     Night HF+LF value 292.8 (186.0, 562.3)     24-h HF+LF value 251.0 (168.5, 410.6)    TP (ms2)       Morning TP value  570.1 (367.7, 818.3)     Day TP value 566.5 (346.5, 824.4)     Night TP value 932.5 (624.3, 1401.0)     24-h TP value 768.1 (540.1, 1133.0)  HRV changes, median (IQR)     HF changes (%)       Day HF changes 14.5% (-11.4%, 51.6%)     Day-night HF changes 101.6% (39.1%, 224.4%)     Morning-night HF changes 132.6% (43.4%, 310.2%)    HF norm changes (n.u.)       Daytime HF norm changes  0.043 (-0.021, 0.115)     Day-night HF norm changes  0.113 (0.009, 0.209)     Morning-night HF norm changes  0.130 (0.034, 0.246)    HF+LF changes (%)       Daytime HF+LF changes  -5.2% (-23.5%, 29.1%)     Day-night HF+LF changes  49.5% (4.4%, 121.7%)     Morning-night HF+LF changes 53.3% (7.2%, 137.5%) 111  Table 7.3 Continuous Predictors Patients, N=201  HRV changes, median (IQR)    TP changes (%)       Daytime TP changes -1.7% (-25.5%, 23.6%)     Day-night TP changes 72.7% (14.4%, 133.8%)     Morning-night TP changes 71.6% (18.2%, 132.7%)  PSS score, mean (95% CI) 18.2 (17.7, 18.7)  ABCD2 score, mean (95% CI) 4.5 (4.3, 4.7)  HF: high frequency; HF norm: normalized HF; HF+LF: sum of HF and low frequency (LF); TP: total power is the sum of HF, LF and very low frequency (VLF). Definitions of HRV parameters are provided in section 6.5.3.2. PSS score: perceived stress scale score; ABCD2 score: A for age, B for blood pressure, C for clinical feature, two D for duration of symptoms and diabetes. 112  The comparisons of HRV parameters, PSS score and ABCD2 score between event group and event-free group are shown in Table 7.4 and Figure 7.3.  For the absolute values of HRV parameters, compared to patients without a secondary ischemic event during 90 days, patients with secondary ischemic events had a significantly lower absolute value of Morning HF (38.1 vs. 51.7 ms2, p=0.03), Day HF (38.8 vs. 59.9 ms2, p=0,001), Day HF norm (0.23 vs. 0.29 n.u., p=0.01) and Daytime HF+LF (152.0 vs. 220.0 ms2, p=0.04) (Figure 7.3 A, B, C, D). The difference of Morning HF norm, 24-h HF, 24-h HF norm, 24-h HF+LF and 24-h TP between event group and event-free group were moderate (p value between 0.05 and 0.1).  For changes of HRV parameters, patients with a secondary ischemic event showed decreases in HF, HF norm, HF+LF and TP during daytime, while patients without outcome events had a significantly increased change in these four HRV parameters: Daytime HF changes: -4.6% vs. 18.2%, p<0.001; Daytime HF norm changes: 0.021 vs. 0.050 n.u., p=0.02; Daytime HF+LF changes: -22.1% vs. 2.9%, p=0,002; and Daytime TP changes: -9.8% vs. 1.8%, p=0.03 (Figure 7.3 E, F, G, H). However, patients with a secondary ischemic event had a higher Day-night change in HF and HF+LF, compared to patients without an outcome event (Day-night HF changes: 177.4% vs. 90.8%, p=0.01; Day-night HF+LF changes: 87.7% vs. 40.9%, p=0.03).  Additionally, patients in event group had a significantly higher PSS score and ABCD2 score, compare to patients in event-free group (PSS score: 20.2 vs. 17.8, p<0.001; ABCD2 score: 5.1 vs. 4.4, p<0.001).   113  Table 7.4 Comparisons of HRV, PSS score and ABCD2 score between event group and event-free group   Patients with events, N=36 Patients with event-free, N=165  p value HRV absolute values, median (IQR)        HF (ms2)          Morning HF value  38.1 (21.6, 65.8) 51.7 (27.9, 112.1) 0.03     Day HF value  38.8 (21.8, 50.5) 59.9 (29.6, 120.9) 0.001     Night HF value 116.4 (55.2, 146.1) 138.1 (62.4, 267.8) 0.15     24-h HF value  80.7 (46.6, 101.6) 100.9 (50.2, 187.9) 0.06    HF norm (n.u.)          Morning HF norm  0.21 (0.17, 0.35) 0.28 (0.19, 0.39) 0.10     Day HF norm  0.23 (0.16, 0.35) 0.29 (0.22, 0.42) 0.01     Night HF norm 0.38 (0.22, 0.56) 0.44 (0.30, 0.58) 0.32     24-h HF norm  0.32 (0.24, 0.40) 0.37 (0.27, 0.49) 0.06    HF+LF (ms2)          Morning HF+LF value  165.7 (100.7, 291.5) 210.0 (111.9, 326.8) 0.16     Day HF+LF value  152.0 (96.1, 232.5) 220.0 (114.3, 329.0) 0.04     Night HF+LF value 261.4 (209.2, 412.0) 300.4 (181.2, 586.5) 0.32     24-h HF+LF value  204.0  (169.7, 347.2) 264.0 (163.6, 467.8) 0.09    TP (ms2)          Morning TP value 457.0 (368.1, 755.1) 577.0 (367.7, 821.0) 0.30     Day TP value  424.2 (327.6, 755.2) 574.8 (349.8, 838.0) 0.17     Night TP value 900.6 (529.8, 1120.0) 958.9 (637.0, 1505.0) 0.13     24-h TP value  718.8 (481.9, 913.8) 772.7 (557.3, 1189.0) 0.10   114  Table 7.4 Continuous    Patients with events, N=36  Patients with event-free, N=165  p value HRV changes, median (IQR)        HF changes (%)          Daytime HF changes  -4.6% (-28.9%, 12.8%) 18.2% (-6.1%, 62.2%) <0.001     Day-night HF changes  177.4% (55.5%, 423.9%) 90.8% (35.4%, 197.2%) 0.01     Morning-night HF changes  154.6% (42.9%, 423.5%) 128.1% (43.6%, 277.6%) 0.35    HF norm changes (n.u.)          Daytime HF norm changes  0.021 (-0.043, 0.073) 0.050 (-0.019, 0.125) 0.02     Day-night HF norm changes 0.140 (0.020, 0.265) 0.108 (0.009, 0.207) 0.28     Morning-night HF norm changes 0.139 (0.036, 0.231) 0.128 (0.033, 0.246) 0.70    HF+LF changes (%)          Daytime HF+LF changes  -22.1% (-35.3%, 1.1%) 2.9% (-19.9%, 36.0%) 0.002     Day-night HF+LF  87.7% (22.7%, 170.1%) 40.9% (1.4%, 113.3%) 0.03     Morning-night HF+LF changes  79.8% (9.5%, 178.0%) 49.9% (7.2%, 126.9%) 0.28    TP changes (%)          Daytime TP changes -9.8% (-27.7%, 13.0%) 1.8% (-21.2%, 33.2%) 0.03     Day-night TP changes 90.9% (22.0%, 131.1%) 67.6% (14.4%, 137.2%) 0.64     Morning-night TP changes  91.3% (17.4%, 127.7%) 70.4% (19.1%, 133.4%) 0.99  PSS score, mean (95% CI) 20.2  (19.2, 21.2) 17.8 (17.3, 18.3) <0.001  ABCD2 score, mean (95% CI) 5.1  (4.6, 5.5) 4.4 (4.2, 4.6) <0.001  HF: high frequency; HF norm: normalized HF; HF+LF: sum of HF and low frequency (LF); TP: total power is the sum of HF, LF and very low frequency (VLF). Definitions of HRV parameters are provided in section 6.5.3.2. Red font: significant p value (<0.05)115  Figure 7-3 Boxplots for HRV parameters between event group and event-free group         116   A: Morning HF value in event and event-free groups B: Day HF value in event and event-free groups C: Day HF norm in event and event-free groups D: Daytime HF changes in event and event-free groups E: Daytime HF norm changes in event and event-free groups F: Daytime HF+LF changes in event and event-free groups G: Daytime TP changes in event and event-free groups Green boxes: event group; Yellow boxes: event-free group; y-axis: the value of HRV parameters; p<0.05 for all figures.                                                                   117  7.4.1 Univariate Predictive Models ABCD2 score was taken as either continuous or categorical variable based on published categorization (Amarenco et al., 2012; Cancelli et al., 2011; Coutts et al., 2008) in building the univariate model. If ABCD2 score was taken as a continuous variable, AUC of the model with ABCD2 alone was 0.63 (p=0.02) (Table 7.5). If ABCD2 score was taken as a categorical variable, the AUC was 0.57 (p=0.1) for the categorization of two levels: 0-3 scores and 4-7 scores; and 0.61 for the categorization of three levels: 0-3 scores (reference), 4-5 scores (p=0.28) and 6-7 scores (p=0.03) (Table 7.5).  PSS score was taken as a continuous variable. The AUC for the univariate model with PSS score alone was 0.72 (p<0.001) (Table 7.5). The models with HRV absolute value or HRV changes are shown in Table 7.5. The best predictor is “Daytime HF changes” which showed the highest AUC of 0.70 (p<0.001). Moreover, other HRV parameters that show a comparable predictive value include: Morning HF value (AUC: 0.61; p =0.03) and Day HF value (AUC: 0.67; p =0.001), Day HF norm (AUC: 0.64; p=0.01); Day HF+LF value (AUC: 0.61; p=0.03), Daytime HF norm changes (AUC: 0.62; p=0.01); Daytime HF+LF changes (AUC: 0.67; p =0.04) and Daytime TP (AUC: 0.61; p =0.03). The differences of AUCs between ABCD2 score (0.57 for categorical variable), and Daytime HF changes (0.70), and between ABCD2 score and PSS score (0.72), were significant (p <0.01 for both comparisons). However, if using ABCD2 score as a continuous variable (AUC of 0.63), there were no significant differences between the AUCs of ABCD2 score, Daytime HF changes, and PSS score. 118  Table 7.5 Univariate predictive models   Models Odds ratio  AUC  Odds ratio (95% CI) p value  Model with ABCD2 alone      If ABCD2 (continuous) 1.39 (1.07, 1.84) 0.01 0.63   If ABCD2   0.57     <4 1       ≥4 2.33 (0.92, 7.14) 0.10    If ABCD2   0.61     0-3 1       4-5 1.80 (0.66, 5.80) 0.28      6-7 3.37 (1.19, 11.14) 0.03   Model with PSS alone        PSS 1.23 (1.10, 1.38) <0.001 0.72  Model with HRV alone       HRV values (natural logarithmic     transformed) *        Morning HF value 0.68 (0.47, 0.96) 0.03 0.61     Day HF value 0.54 (0.37, 0.78) 0.001 0.67     Night HF value 0.80 (0.57, 1.22) 0.21 0.58     24-h HF value 0.72 (0.49, 1.04) 0.08 0.60         Morning HF norm  0.81 (0.60, 1.05) ** 0.12 0.59     Day HF norm  0.68 (0.49, 0.90) ** 0.01 0.64     Night HF norm 0.93 (0.76, 1.13) ** 0.49 0.55     24-h HF norm  0.79 (0.59, 1.02) ** 0.08 0.60         Morning HF+LF value 0.71 (0.46, 1.10) 0.13 0.57     Day HF+LF value 0.61 (0.39, 0.95) 0.03 0.61     Night HF+LF value 0.83 (0.54, 1.27) 0.39 0.55     24-h HF+LF value 0.71 (0.45, 1.11) 0.14 0.59         Morning TP value 0.77 (0.46, 1.30) 0.33 0.56     Day TP value 0.65 (0.38, 1.10) 0.11 0.57     Night TP value 0.66 (0.38, 1.12) 0.13 0.58     24-h TP value 0.59 (0.33, 1.05) 0.08 0.59     119  Table 7.5 Continuous   AUC: area under the curve * Odds ratio for every e (2.72) fold increase in HRV;  ** No logarithmic transformation for HF.nu, and odds ratio for every 0.1 increase in HF.nu values; † Odds ratio for every 10% or 0.1 increase in HRV parameters;  Red font: the highest AUC; Blue font: AUCs higher than 0.60 that are comparable to AUC of ABCD2 score, and p<0.05. Models Odds ratio  AUC  Odds ratio (95% CI) p value    HRV changes †        Daytime HF changes 0.83 (0.75, 0.91) <0.001 0.70     Day-night HF changes 1.01 (1.00, 1.02) 0.06 0.64     Morning-night HF changes 1.00 (1.00, 1.02) 0.11 0.55         Daytime HF norm changes  0.64 (0.45, 1.89) 0.01 0.62     Day-night HF norm changes  1.12 (0.94, 1.45) 0.16 0.55     Morning-night HF norm changes  1.06 (0.85, 1.31) 0.60 0.52          Daytime HF+LF changes 0.91 (0.82, 0.99) 0.04 0.67     Day-night HF+LF changes 1.00 (0.99, 1.02) 0.48 0.62     Morning-night HF+LF changes 1.00 (0.98, 1.02) 0.74 0.56         Daytime TP changes 0.89 (0.80, 0.98) 0.03 0.61     Day-night TP changes 1.00 (0.96, 1.02) 0.73 0.52     Morning-night TP changes 0.98 (0.95, 1.01) 0.32 0.50     120  7.4.2 Diagnostic Performance of ABCD2 Score, PSS Score and the Selected HRV Parameters  “Daytime HF changes” with the highest AUC of 0.70 and “Morning HF value” (AUC of 0.61) with shorter duration of recording period and simpler calculation were selected as optimal HRV-related predictors of outcome ischemic events. The diagnostic performance of using ABCD2 score, PSS score, Daytime HF changes or Morning HF value alone to predict outcome events was tested in the study.   7.4.2.1 ABCD2 score The univariate predictive model with ABCD2 alone is presented as equation 7-1. The diagnostic performance of using ABCD2 score to predict outcome events is shown in Table 7.6  Equation 7-1 𝑙𝑜𝑔 (𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦1 − 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦) =  −3.0909 + 0.3316 ∗ 𝐴𝐵𝐶𝐷2  Table 7.6 Diagnostic performance of ABCD2 score  Sensitivity Specificity PPV NPV Cut-off value of ABCD2 score 94% 12% 19% 91% 3 86% 27% 21% 90% 4 69% 50% 23% 88% 5 PPV: positive predictive value; NPV: negative predictive value121  7.4.2.2 PSS score The univariate predictive model with PSS alone is presented as equation 7-2. The diagnostic performance of using PSS score to predict outcome events is shown in Table 7.7.  Equation 7-2 𝑙𝑜𝑔 (𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦1 − 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦) =  −5.423 + 0.206 ∗ 𝑃𝑆𝑆  Table 7.7 Diagnostic performance of PSS score  Sensitivity Specificity PPV NPV Cut-off value of PSS score 97% 32% 24% 98% 16 89% 44% 26% 95% 17 81% 58% 30% 93% 18 67% 64% 29% 90% 19 61% 67% 29% 89% 20 PPV: positive predictive value; NPV: negative predictive value   7.4.2.3 Daytime HF changes The univariate predictive model with Daytime HF changes is presented as:  Equation 7-3 𝑙𝑜𝑔 (𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦1 − 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦) =  −1.338 + (−1.866) ∗ 𝐷𝑎𝑦𝑡𝑖𝑚𝑒 𝐻𝐹 𝑐ℎ𝑎𝑛𝑔𝑒𝑠  Figure 7.4 A and B show ROC curves of the model “Event ~ Daytime HF changes”. In Figure 7.4 A, different color on the right y-axis represents different cut-off point of “event probability” (i.e. 122  the risk of developing outcome events), from 0 to 0.58. The colorful band of right y-axis corresponds to the section with the same color on the ROC curve. Figure 7.4 B shows the cut-off point of “event probability” (0.172) based on the maximum Youden’s index (sensitivity + specificity - 1). By using 0.172 as the cut-off point of event probability, model “Event ~ Daytime HF changes” has sensitivity of 75.0%, specificity of 59.4%, PPV of 28.4% and NPV of 91.5%. The cut-off value of “Daytime HF changes” is calculated with equation 7-3. Given that the cut-off point of event probability is 0.172, the cut-off value of “Daytime HF change” is 0.125, i.e., 12.5% increase. The diagnostic performance of using Daytime HF changes to predict outcome events is shown in Table 7.8.  123  Figure 7-4 ROC curve for Daytime HF changes   A: ROC curve for Daytime HF changes. Different color on the right y-axis represents different cut-off point of event probability, from 0 to 0.58. The colorful band of right y-axis corresponds to the same colorful section on ROC curve. AUC of Daytime HF changes is 0.70. B: ROC curve for Daytime HF changes which shows the cut-off point of event probability is 0.172, based on the maximum Youden’s index (i.e. the maximum sum of sensitivity (75.0%) and specificity (59.4%)).  124  Table 7.8 Diagnostic performance of Daytime HF changes   Sensitivity Specificity PPV NPV Cut-off value of Daytime HF changes 89% 40% 24% 94% 28% 86% 41% 24% 93% 26% 81% 44% 24% 91% 23% 75% 59% 28% 92% 12% 69% 62% 28% 90% 8% PPV: positive predictive value; NPV: negative predictive value125  7.4.2.4 Morning HF value  The univariate predictive model with Morning HF value is presented as:  Equation 7-4 𝑙𝑜𝑔 (𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦1 − 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦) =  −0.053 + (−0.392) ∗ 𝑙𝑜𝑔 (𝑀𝑜𝑟𝑛𝑖𝑛𝑔 𝐻𝐹 𝑣𝑎𝑙𝑢𝑒𝑠)  Figure 7.5 A and B show the ROC curves of the model “Event ~ Morning HF value”. In Figure 7.5 A, the colorful band of right y-axis corresponds to the section with the same color on the ROC curve. The cut-off point of “event probability” is 0.169 based on the maximum Youden’s index. By using 0.169 as the cut-off point of event probability, model “Event ~ Morning HF value” has sensitivity of 69.4%, specificity of 52.7%, PPV of 24.3% and NPV of 88.8%. Based on equation 7-4, the cut-off value of log (Morning HF) is 3.928, i.e. the cut-off value of Morning HF absolute value is 50.9 ms2. The diagnostic performance of using Morning HF value to predict outcome events is shown in Table 7.9.  126  Figure 7-5 ROC curve for Morning HF value                                                                                                               A: ROC curve for Morning HF value. Different color on the right y-axis represents different cut-off point of event probability, from 0 to 0.42. The colorful band of right y-axis corresponds to the same colorful section on ROC curve. AUC of Morning HF value is 0.61. B: ROC curve for Morning HF value which the cut-off point of event probability is 0.169, based on the maximum Youden’s index (i.e. the maximum sum of sensitivity (69.4%) and specificity (52.7%)).   127  Table 7.9 Diagnostic performance of Morning HF value  Sensitivity Specificity PPV NPV Cut-off value of Morning HF values (ms2) 89% 30% 22% 93% 92 86% 32% 22% 91% 84 81% 34% 21% 89% 76 75% 40% 21% 88% 64 69% 53% 24% 89% 51 PPV: positive predictive value; NPV: negative predictive value128  7.4.3 Kaplan-Meier Curves for ABCD2 Score, PSS Score and the Selected HRV Predictors  In this section, Kaplan-Meier curves for different levels of ABCD2 score (Figure 7.6), PSS score (Figure 7.7), Daytime HF changes (Figure 7.8) and Morning HF values (Figure 7.9) are shown. The number of patients free of outcome events at different time period is shown below each figure. Log-rank test was used to compare the survival differences between different subgroups of ABCD2 score, PSS score, Daytime HF changes and Morning HF value, respectively.  ABCD2 score - ABCD2 score was categorized into two levels: <4 scores and ≥4 scores; or three levels: 0 to 3 scores, 4 to 5 scores, and 6 to 7 scores. The difference between two-level curves was not significant with a p value of 0.09 in log-rank test. For three levels, the difference was marginally significant with 0.06.  PSS score - PSS score was categorized into two levels by using the selected cut-off value of 18 scores (section 7.4.2.2). It was also categorized into three levels by the tertiles: <16 scores, 16 to 19 scores, and ≥20 scores. The differences between curves in both two-level and three-level categorization were significant (p<0.001). Daytime HF changes - Daytime HF change was categorized into two groups by the selected cut-off point of 12.5%. Further, it was categorized into three groups by the value of “0” and the median of the positive changes which is 40% increases. Therefore, the three categories were: HF increase ≥40%, HF increase <40%, and HF decreases. The differences between curves in both two-level and three-level categorization were significant (p<0.001). Morning HF value - Morning HF value was categorized into two levels by the selected cut-off point of 51 ms2, and into three levels by the tertiles: ≥70 ms2, 34 to 70 ms2, and < 34 ms2. 129  The event-free differences were significantly different in two-level (p=0.03) but not among three-level categorization (p=0.55).  130  Figure 7-6 Kaplan-Meier curves for ABCD2 score                                                                                                                 A: Kaplan-Meier curves for ABCD2 score by two levels: <4 scores and ≥4 scores. B: Kaplan-Meier curves for ABCD2 score in three levels: 0-3 scores, 4-5 scores and 6-7 scores. The number of patients with event-free at baseline, within 48 hours, 7 days, 14 days, 30 days and 90 days, as well as the number and rate of events are shown.131  Figure 7-7 Kaplan-Meier curves for PSS score                                                                                                                    A: Kaplan-Meier curves for PSS score by two levels: <18 scores and ≥18 scores. B: Kaplan-Meier curves for PSS score in three levels: <16 scores, 16-19 scores and ≥20 scores. The number of patients with event-free at baseline, within 48 hours, 7 days, 14 days, 30 days and 90 days, as well as the number and rate of events are shown.132  Figure 7-8 Kaplan-Meier curves for Daytime HF changes                                                                                                                    A: Kaplan-Meier curves for Daytime HF changes by two levels: ≥12.5% increases and <12.5% increases. B: Kaplan-Meier curves for Daytime HF changes in three levels: ≥40% increases, <40% increases and decreased HF. The number of patients with event-free at baseline, within 48 hours, 7 days, 14 days, 30 days and 90 days, as well as the number and rate of events are shown.133  Figure 7-9 Kaplan-Meier curves for Morning HF value                      A: Kaplan-Meier curves for Morning HF value by two levels: ≥51 ms2 and <51 ms2. B: Kaplan-Meier curves for Morning HF value in three levels: ≥70 ms2, 34 to 70 ms2, and <34 ms2. The number of patients with event-free at baseline, within 48 hours, 7 days, 14 days, 30 days and 90 days, as well as the number and rate of events are shown.                                                                          134  7.5 HRV-based Stress Predictive Models In this section, I first show the established bivariate and full predictive models to determine whether adding HRV and PSS score can improve the predictive value of ABCD2 score. Then I compare AUCs of the selected HRV-based predictive models with ABCD2 score alone.  7.5.1 Bivariate Predictive Models Bivariate models included the ones with HRV parameters and ABCD2 score, the one with PSS score and ABCD2 score, and the ones with selected HRV parameters and PSS score.  Among the models with HRV parameters and ABCD2 score, model “Event ~ Daytime HF changes + ABCD2” had the highest AUC of 0.74; this improvement in predictive value was significant (p=0.01, shown in Table 7.10), compared to ABCD2 score (0.63). Additionally, models “Event ~ Day HF value + ABCD2 (AUC=0.72)” and “Event ~ Day HF norm + ABCD2 (AUC=0.72)” also had significantly improved AUC compared to ABCD2 score (p <0.05).  Model “Event ~ PSS + ABCD2” showed a AUC of 0.74 which was significantly higher than the AUC of ABCD2 score (p=0.012, shown in Table 7.10). Furthermore, models with selected HRV variables (Morning HF and Daytime HF changes) and PSS score were established and tested: AUCs for model “Event ~ Morning HF + PSS” and model “Event ~ Daytime HF changes + PSS” were 0.73 and 0.80, respectively (Table 7.10).   7.5.2 Full Predictive Models Among the full predictive models with all three predictors added, the best one was “Event ~ Daytime HF changes + PSS + ABCD2” with an AUC of 0.82 (Table 7.10). The OR of Daytime 135  HF changes was 0.84 (95% CI: 0.75, 0.92) (p<0.001) which indicates a reduction of 16% in the risk of developing a secondary ischemic outcome event for every 10% increase in HF changes from morning to afternoon. Models with other HRV parameters which provided significantly improved AUC compared to ABCD2 score (p<0.05), and significant ORs included:  • Event ~ Morning HF value + PSS + ABCD2 (AUC =0.76); • Event ~ Day HF value + PSS + ABCD2 (AUC=0.79); • Event ~ Morning HF norm + PSS + ABCD2 (AUC=0.77); • Event ~ Day HF norm + PSS + ABCD2 (AUC=0.80); • Event ~ 24-h HF norm +PSS +ABCD (AUC=0.76); • Event ~ Day-night HF changes + PSS + ABCD2 (AUC=0.77); • Event ~ Daytime HF norm changes + PSS +ABCD2 (AUC=0.77) • Event ~ Day-night HF norm changes + PSS +ABCD2 (AUC=0.76); and • Event ~ Daytime TP changes + PSS + ABCD2 (AUC=0.77)  Two candidate models: “Event ~ Morning HF value + PSS + ABCD2” (as the most practical model) and “Event ~ Daytime HF changes + PSS + ABCD2” (as the best predictive model) were compared with ABCD2 score and bivariate models (Table 7.11). AUC of model “Event ~ Daytime HF changes + PSS + ABCD2” (0.82) was significantly higher than AUCs of all bivariate models and ABCD2 alone, except model “Event ~ Daytime HF changes + PSS” (AUC=0.80). AUC of model “Event ~ Morning HF value + PSS + ABCD2” (0.76) was significantly higher than AUC of ABCD2 score (0.76 vs. 0.63, p=0.02).   136  Table 7.10 Multiple stress models  Models Odds ratio for HRV parameters AUC Odds ratio (95% CI) p value Model with HRV + ABCD2      HRV values (natural logarithmic transformed) *        Morning HF value + ABCD2 0.64 (0.43, 0.92) 0.02 0.68     Day HF value + ABCD2 0.51 (0.33, 0.74) 0.001 0.72     Night HF value + ABCD2 0.79 (0.655 1.12) 0.19 0.66     24-h HF value + ABCD2 0.69 (0.46, 1.01) 0.06 0.67         Morning HF norm +ABCD2 0.76 (0.56, 1.01) ** 0.07 0.68     Day HF norm + ABCD2 0.62 (0.44, 0.84) ** 0.004 0.72     Night HF norm + ABCD2 0.94 (0.77, 1.15) ** 0.57 0.64     24-h HF norm + ABCD2 0.77 (0.57, 1.01) ** 0.06 0.68         Morning HF+LF value + ABCD2 0.71 (0.45, 1.10) 0.13 0.65     Day HF+LF value + ABCD2 0.60 (0.38, 0.94) 0.03 0.68     Night HF+LF value + ABCD2 0.79 (0.51, 1.24) 0.31 0.65     24-h HF+LF value + ABCD2 0.68 (0.42, 1.09) 0.11 0.66         Morning TP value + ABCD2 0.77 (0.45, 1.32) 0.35 0.64     Day TP value + ABCD2 0.65 (0.37, 1.10) 0.11 0.65     Night TP value + ABCD2 0.63 (0.36, 1.08) 0.10 0.65     24-h TP value + ABCD2 0.57 (0.31, 1.03) 0.06 0.66       HRV percentage changes †        Daytime HF change + ABCD2  0.84 (0.75, 0.92) <0.001 0.74     Day-night HF changes + ABCD2 1.01 (1.00, 1.02) 0.04 0.69     Morning-night HF changes + ABCD2 1.01 (1.00, 1.02) 0.14 0.65  137  Table 7.10 Continuous   Models  Odds ratio for HRV paramters AUC  Odds ratio (95% CI) p value     Daytime HF norm change + ABCD2  0.62 (0.43, 0.86)  0.01 0.68     Day-night HF norm changes + ABCD2 1.33 (1.01, 1.78)  0.046 0.64     Morning-night HF norm changes + ABCD2 1.09 (0.89, 1.34)  0.41 0.64         Daytime HF+LF changes + ABCD2 0.91 (0.83, 0.99) 0.04 0.67     Day-night HF+LF changes + ABCD2 1.00 (0.99, 1.02) 0.47 0.64     Morning-night HF changes + ABCD2 1.00 (0.98, 1.02) 0.80 0.63         Daytime TP changes + ABCD2 0.89 (0.80, 0.98) 0.03 0.67     Day-night TP changes + ABCD2 0.99 (0.96, 1.02) 0.65 0.64     Morning-night TP changes + ABCD2 0.98 (0.95, 1.01) 0.26 0.65     Models with PSS + ABCD2        PSS+ABCD2 1.21 (1.09, 1.36) †† <0.001 0.74     Models with selected HRV + PSS        Morning HF values + PSS 0.65 (0.45, 0.94) 0.02 0.73     Daytime HF changes + PSS 0.84 (0.75, 0.92) <0.001 0.80     Models with HRV + PSS + ABCD2      HRV values (natural logarithmic transformed) *        Morning HF value + PSS + ABCD2 0.64 (0.43, 0.94) 0.02 0.76     Day HF value + PSS + ABCD2 0.52 (0.34, 0.76) 0.001 0.79     Night HF value + PSS + ABCD2 0.81 (0.56, 1.16) 0.24 0.75     24-h HF value + PSS + ABCD2 0.71 (0.48, 1.05) 0.09 0.76  138  Table 7.10 Continuous   Models Odds ratio for HRV parameters AUC  Odds ratio (95% CI) p value      Morning HF norm + PSS + ABCD2 0.73 (0.53, 0.96) ** 0.03 0.77     Day HF norm + PSS + ABCD2 0.60 (0.42, 0.81) ** 0.002 0.80     Night HF norm + PSS + ABCD2 0.93 (0.76, 1.14) ** 0.51 0.75     24-h HF norm + PSS + ABCD2 0.75 (0.56, 0.99) ** 0.048 0.76         Morning HF+LF value + PSS + ABCD2 0.75 (0.47, 1.19) 0.23 0.74     Day HF+LF value + PSS + ABCD2 0.64 (0.40, 1.01) 0.06 0.76     Night HF+LF value + PSS + ABCD2 0.83 (0.53, 1.30) 0.41 0.75     24-h HF+LF value + PSS + ABCD2 0.73 (0.45, 1.17) 0.19 0.75         Morning TP value + PSS + ABCD2 0.81 (0.46, 1.40) 0.45 0.74     Day TP value + PSS + ABCD2 0.68 (0.39, 1.16) 0.16 0.74     Night TP value + PSS + ABCD2 0.64 (0.36, 1.12) 0.12 0.76     24-h TP value + PSS + ABCD2 0.59 (0.32, 1.08) 0.09 0.75     HRV percentage changes †        Daytime HF change + PSS + ABCD2  0.84 (0.75, 0.92) <0.001 0.82     Day-night HF changes + PSS + ABCD2 1.01 (1.00, 1.03) 0.03 0.77     Morning-night HF changes + PSS + ABCD2 1.01 (1.00, 1.02) 0.12 0.75         Daytime HF norm change + PSS + ABCD2  0.61 (0.42, 0.87) 0.01 0.77     Day-night HF norm changes + PSS + ABCD2 1.22 (0.99, 1.52) 0.07 0.75     Morning-night HF norm changes + PSS + ABCD2 1.11 (0.90, 1.37) 0.33 0.75  139  Table 7.10 Continuous   AUC: area under the curve * Odds ratio for every e (2.72) fold increase in HRV;  ** No logarithmic transformation for HF.nu, and odds ratio for every 0.1 increase in HF.nu values; † Odds ratio for every 10% (or 0.1) increase in HRV parameters;  †† Odds ratio for PSS score Red font: with the highest AUCs and significant p value; Blue font: other models with good AUCs and significant p values Models Odds ratio for HRV parameters AUC  Odds ratio (95% CI) p value      Daytime HF+LF changes + PSS + ABCD2 0.91 (0.83, 0.99) 0.04 0.76     Day-night HF+LF changes + PSS + ABCD2 1.01 (0.99, 1.02) 0.36 0.75     Morning-night HF+LF changes + PSS + ABCD2 1.00 (0.98, 1.02) 0.75 0.74         Daytime TP changes + PSS + ABCD2 0.90 (0.80, 0.98) 0.04 0.77     Day-night TP changes + PSS + ABCD2 0.99 (0.96, 1.02) 0.67 0.74     Morning-night TP changes + PSS + ABCD2 0.98 (0.94, 1.01) 0.23 0.76  140  Table 7.11 AUC test metrics for the selected models  p values ABCD2 (0.63) Morning HF +ABCD2 (0.68)   Daytime HF changes +ABCD2 (0.74) PSS +ABCD2 (0.74) Morning HF +PSS (0.73) Daytime HF changes +PSS (0.80) Morning HF +PSS +ABCD2 (0.76) Daytime HF changes +PSS +ABCD2 (0.82) ABCD2 (0.63)  NA 0.15 0.01 0.012 0.16 0.01 0.02 <0.001 Morning HF + ABCD2 (0.68)  NA 0.21 0.16 0.36 0.06 0.04 0.01 Daytime HF changes + ABCD2 (0.74)   NA 0.99 0.81 0.18 0.83 0.014 PSS + ABCD2 (0.74)    NA 0.75 0.16 0.59 0.02 Morning HF + PSS (0.73)     NA 0.12 0.24 0.05 Daytime HF changes + PSS (0.80)      NA 0.37 0.31 Morning HF + PSS + ABCD2 (0.76)       NA 0.14 Daytime HF changes + PSS + ABCD2 (0.82)        NA p value metrics for AUC test between every two selected models; Red font: significant p values141  7.6 The Best Stress Model and the Most Practical Model  From the comparison of AUC between different models (Table 7.11), “Event ~ Daytime HF changes +PSS +ABCD2” shows the highest AUC of 0.82 which was significantly higher than the AUC of ABCD2 score and ABCD2-based bivariate models. The model “Event ~ Daytime HF changes +PSS +ABCD2” was then considered as the best stress model. The model “Morning HF value +PSS +ABCD2” with the AUC of 0.76 was considered as the most practical model as it requires the shortest recording of HRV and the simplest calculation.   7.6.1 The Best Stress Model From the analyses of multiple stress models in section 7.5, we found the best stress model (BSM) “Event ~ Daytime HF changes + ABCD2 +PSS” which provides the highest predictive value of 0.82. The model is presented as:  Equation 7-5 𝑙𝑜𝑔 (𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦1 − 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦)=  −6.248 + (−1.722) ∗ 𝐷𝑎𝑦𝑡𝑖𝑚𝑒 𝐻𝐹 𝑐ℎ𝑎𝑛𝑔𝑒𝑠 + 0.186 ∗ 𝑃𝑆𝑆 + 0.296∗ 𝐴𝐵𝐶𝐷2  The OR for Daytime HF changes (every 10% increase) is 0.84 (95% CI: 0.75, 0.92) with a p value <0.001; OR for PSS score is 1.20 (95% CI: 1.08, 1.36), p=0.002; OR for ABCD2 score is 1.35 (95% CI: 1.02, 1.82), p=0.045. The “event probability” can be calculated with equation 7-6.  142  Equation 7-6  𝑃𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦=𝑒𝑥𝑝(−6.248 + (−1.722) ∗ 𝐷𝑎𝑦𝑡𝑖𝑚𝑒 𝐻𝐹 𝑐ℎ𝑎𝑛𝑔𝑒𝑠 +  0.186 ∗  𝑃𝑆𝑆 +  0.296 ∗  𝐴𝐵𝐶𝐷2)1 + 𝑒𝑥𝑝(−6.248 +  (−1.722)  ∗  𝐷𝑎𝑦𝑡𝑖𝑚𝑒 𝐻𝐹 𝑐ℎ𝑎𝑛𝑔𝑒𝑠 +  0.186 ∗  𝑃𝑆𝑆 +  0.296 ∗  𝐴𝐵𝐶𝐷2)  Figure 7.10 shows the ROC curve for the BCS Model. Based on the maximum Youden’s index (sensitivity + specificity – 1), the cut-off point of “event probability” is 0.175. By using this cut-off point of event probability, BSM has sensitivity of 80.6%, specificity of 71.5%, PPV of 38.2%, NPV of 94.4%. Given that the cut-off point of event probability is 0.175, the cut-off formula for the BSM is:   Equation 7-7 4.7 = (−1.722) ∗ 𝐷𝑎𝑦𝑡𝑖𝑚𝑒 𝐻𝐹 𝑐ℎ𝑎𝑛𝑔𝑒𝑠 + 0.186 ∗ 𝑃𝑆𝑆 + 0.296 ∗ 𝐴𝐵𝐶𝐷2  The diagnostic performance of the BSM is shown in Table 7.12.    143  Figure 7-10 ROC curves for the Best Stress Model  A: ROC curve for the BSM. Different color on the right y-axis represents different cut-off point of event probability, from 0 to 0.77. The colorful band of right y-axis corresponds to the same colorful section on ROC curve. AUC of the BSM is 0.82 B: ROC curve for the BSM which shows the cut-off point of event probability is 0.175, based on the maximum Youden’s index (i.e. the maximum sum of sensitivity (80.6%) and specificity (71.5%)).     144  Table 7.12 Performance of the Best Stress Model  Sensitivity Specificity PPV NPV Cut-off point of the event probability 89% 61% 33% 96% 0.132 86% 63% 34% 95% 0.148 81% 72% 38% 94% 0.177 75% 73% 38% 93% 0.192 69% 81% 45% 92% 0.228 64% 83% 45% 91% 0.245 61% 84% 45% 91% 0.249 PPV: positive predictive value; NPV: negative predictive value145  7.6.2 The Most Practical Model Because of the shortest HRV recording, model “Event ~ Morning HF value + ABCD2 + PSS” with AUC of 0.76, was selected as the most practical model (MPM) for clinical use. This model is presented as:  Equation 7-8 𝑙𝑜𝑔 (𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦1 − 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦)=  −4.980 + (−0.443) ∗ 𝑙𝑜𝑔(𝑀𝑜𝑟𝑛𝑖𝑛𝑔 𝐻𝐹 𝑣𝑎𝑙𝑢𝑒) + 0.197 ∗ 𝑃𝑆𝑆 + 0.295∗ 𝐴𝐵𝐶𝐷2  In the model, OR for log (Morning HF) is 0.64 (95% CI: 0.43, 0.94), p=0.023; OR for PSS score is 1.22 (95% CI: 1.09, 1.37), p<0.001; and OR for ABCD2 score is 1.34 (95% CI: 1.03, 1.79), p=0.036. The “event probability” is calculated with equation 7-9.  Equation 7-9 𝑃𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦=𝑒𝑥𝑝[−4.980 + (−0.443)  ∗  𝑙𝑜𝑔 (𝑀𝑜𝑟𝑛𝑖𝑛𝑔 𝐻𝐹 𝑣𝑎𝑙𝑢𝑒)  +  0.197 ∗  𝑃𝑆𝑆 +  0.295 ∗  𝐴𝐵𝐶𝐷2]1 + 𝑒𝑥𝑝[−4.980 +  (−0.443)  ∗  𝑙𝑜𝑔 (𝑀𝑜𝑟𝑛𝑖𝑛𝑔 𝐻𝐹 𝑣𝑎𝑙𝑢𝑒)  +  0.197 ∗  𝑃𝑆𝑆 +  0.295 ∗  𝐴𝐵𝐶𝐷2]  Figure 7.11 shows the ROC curve for the MPM. Based on the maximum Youden’s index, the cut-off point of “event probability” is 0.224. By using this cut-off point of event probability, the MPM has sensitivity of 66.7%, specificity of 81.8%, PPV of 44.4% and NPV of 91.8%. Given that the cut-off point of event probability is 0.224, the cut-off formula for the MPM is:  146  Equation 7-10 3.74 = (−0.443) ∗ 𝑙𝑜𝑔 (𝑀𝑜𝑟𝑛𝑖𝑛𝑔 𝐻𝐹 𝑣𝑎𝑙𝑢𝑒) + 0.197 ∗ 𝑃𝑆𝑆 + 0.295 ∗ 𝐴𝐵𝐶𝐷2  The diagnostic performance of the MPM is shown in Table 7.13.147  Figure 7-11 ROC curves for the Most Practical Model  A: ROC curve for the MPM. Different color on the right y-axis represents different cut-off point of event probability, from 0 to 0.65. The colorful band of right y-axis corresponds to the same colorful section on ROC curve. AUC of the MPM is 0.76 B: ROC curve for the MPM which shows the cut-off point of event probability is 0.224, based on the maximum Youden’s index (i.e. the maximum sum of sensitivity (66.7%) and specificity (81.8%)).    148  Table 7.13 Performance of the Most Practical Model  Sensitivity Specificity PPV NPV Cut-off point of the event probability 89% 39% 24% 94% 0.096 86% 42% 25% 93% 0.103 81% 49% 26% 92% 0.117 75% 59% 28% 92% 0.142 69% 76% 39% 92% 0.192 64% 82% 43% 91% 0.245 61% 83% 44% 91% 0.262 PPV: positive predictive value; NPV: negative predictive value149  7.7 Exploratory Stress Model with Combined HRV Variables Based on all previous description, we see that both Daytime HF changes and Morning HF values show good prediction of outcome events occurrences. Because they measure different dimensions/conditions of HRV, it is reasonable to develop a model with inclusion of both variables (the absolute values and day changes of HRV). The exploratory model may have two forms:  • Adding both absolute values and changes of HF into the predictive model; or • Creating a new variable with the combination of HF morning value and HF changes.   7.7.1 Exploratory Stress Model-1 Our exploratory stress model-1 (ESM-1) with both the Morning HF value and Daytime HF changes included in the model is presented as:  Equation 7-11 𝑙𝑜𝑔 (𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦1 − 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦)=  −4.025 + (−1.953) ∗ 𝐷𝑎𝑦𝑡𝑖𝑚𝑒 𝐻𝐹 𝑐ℎ𝑎𝑛𝑔𝑒𝑠 + (−0.606)∗ 𝑙𝑜𝑔(𝑀𝑜𝑟𝑛𝑖𝑛𝑔 𝐻𝐹 𝑣𝑎𝑙𝑢𝑒) + 0.185 ∗ 𝑃𝑆𝑆 + 0.316 ∗ 𝐴𝐵𝐶𝐷2  The AUC of ESM-1 is 0.84 which is significantly higher than AUC of ABCD2 score (0.84 vs. 0.63, p<0.001). The OR for Daytime HF changes (every 10% increase) is 0.82 (95% CI: 0.73, 0.91) with a p value <0.001; OR for log (Morning HF values) is 0.55 (95% CI: 0.35, 0.83), p=0.005; OR for PSS score is 1.20 (95% CI: 1.07, 1.37), p=0.003; and OR for ABCD2 score is 1.37 (95% CI: 1.02, 1.88), p=0.04. Figure 7.12 A shows that different cut-off points of event 150  probability are associated with different diagnostic performance. Figure 7.12 B shows the cut-off point of event probability (0.237) based on the maximum Youden’s index. By using 0.237 as the cut-off point of event probability, the ESM-1 has sensitivity of 69.4%, specificity of 84.8%, PPV of 50.0% and NPV of 92.7%. Given that the cut-off point of event probability is 0.237, the cut-off formula of ESM-1 is:   Equation 7-12 6.881 = (−1.953) ∗ 𝐷𝑎𝑦𝑡𝑖𝑚𝑒 𝐻𝐹 𝑐ℎ𝑎𝑛𝑔𝑒𝑠 + (−0.606) ∗ 𝑙𝑜𝑔(𝑀𝑜𝑟𝑛𝑖𝑛𝑔 𝐻𝐹 𝑣𝑎𝑙𝑢𝑒) + 0.185∗ 𝑃𝑆𝑆 + 0.316 ∗ 𝐴𝐵𝐶𝐷2  The diagnostic performance of ESM-1 is shown in Table 7.14.   151  Figure 7-12 ROC curves for Exploratory Stress Model-1  A: ROC curve for the ESM-1. Different color on the right y-axis represents different cut-off point of event probability, from 0 to 0.86. The colorful band of right y-axis corresponds to the same colorful section on ROC curve. AUC of the ESM-1 is 0.84. B: ROC curve for the ESM-1 which shows the cut-off point of event probability is 0.237, based on the maximum Youden’s index (i.e. the maximum sum of sensitivity (69.4%) and specificity (84.8%)).     152  Table 7.14 Performance of Exploratory Stress Model-1  Sensitivity Specificity PPV NPV Cut-off point of the event probability 89% 58% 31% 96% 0.110 86% 66% 36% 96% 0.145 81% 70% 37% 94% 0.165 75% 75% 40% 93% 0.179 69% 85% 50% 93% 0.240 64% 88% 53% 92% 0.273 61% 88% 54% 91% 0.295 PPV: positive predictive value; NPV: negative predictive value153  7.7.2 Exploratory Stress Model-2 ESM-2 was built with a new variable that combines the Daytime HF changes and Morning HF value. The cut-off points of Daytime HF changes (12.5%) and Morning HF values (50.9 ms^2) were used to define the new variable. For the practical consideration, we took HF absolute value of 50 ms2 and HF changes of 12%, as the thresholds. The three categories of the new variable are: • Low risk: Daytime HF increase ≥ 12% and Morning HF ≥ 50 ms2 • Medium risk: Daytime HF increase ≥ 12% and Morning HF < 50 ms2, or Daytime HF increase < 12% and Morning HF ≥ 50 ms2 • High risk: Daytime HF increase < 12% and Morning HF < 50 ms2  The association between the risk exposure and development of secondary ischemic events is shown in Table 7.15. Table 7.15 Association between risk exposure and development of secondary ischemic events  New variable Event group Event-free group Event rate High risk  16 31 34.0% Medium risk  19 81 19.0% Low risk  1 53 1.9%  The ESM-2 with the combined HRV was built with the use of dummy variables: High Risk was assigned to “2”, Medium Risk was assigned to “1” and Low Risk was the “0” (reference). The model is presented as:     154  Equation 7-13 𝑙𝑜𝑔 (𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦1 − 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦)=  −10.109 + 2.864 ∗ 𝑀𝑒𝑑𝑖𝑢𝑚 𝑟𝑖𝑠𝑘 + 3.678 ∗ 𝐻𝑖𝑔ℎ 𝑟𝑖𝑠𝑘 + 0.221 ∗ 𝑃𝑆𝑆+ 0.347 ∗ 𝐴𝐵𝐶𝐷2  The AUC for ESM-2 is 0.82 which is significantly higher than ABCD2 score alone (0.82 vs. 0.63, p<0.001). The OR between high risk, medium risk and low risk (reference) is 39.6 (95% CI: 6.8, 769.1; p=0.001) and 17.5 (95% CI: 3.2, 332.8; p=0.008). Figure 7.13 A shows that different cut-off points of event probability are associated with different diagnostic performance. The cut-off of event probability based on the maximum sum of sensitivity (83.3%) and specificity (65.5%) is 0.137 (Figure 7.13 B). The cut-off formula is:   Equation 7-14 8.269 = 2.864 ∗ 𝑀𝑒𝑑𝑖𝑢𝑚 𝑟𝑖𝑠𝑘 + 3.678 ∗ 𝐻𝑖𝑔ℎ 𝑟𝑖𝑠𝑘 + 0.221 ∗ 𝑃𝑆𝑆 + 0.347 ∗ 𝐴𝐵𝐶𝐷2  The diagnostic performance of the ESM-2 is shown in Table 7.16.  155  Figure 7-13 ROC curves for Exploratory Stress Model-2   A: ROC curve for the ESM-2. Different color on the right y-axis represents different cut-off point of event probability, from 0 to 0.79. The colorful band of right y-axis corresponds to the same colorful section on ROC curve. AUC of the ESM-2 is 0.82. B: ROC curve for the ESM-2 which shows the cut-off point of event probability is 0.137, based on the maximum Youden’s index (i.e. the maximum sum of sensitivity (83.3%) and specificity (65.5%)).  156  Table 7.16 Performance of the Exploratory Stress Model-2  Sensitivity Specificity PPV NPV Cut-off point of the event probability 89% 53% 29% 96% 0.116 86% 60% 32% 95% 0.132 78% 68% 35% 93% 0.153 75% 71% 36% 93% 0.164 69% 77% 40% 92% 0.217 64% 81% 42% 91% 0.250 61% 84% 45% 91% 0.276 PPV: positive predictive value; NPV: negative predictive value157  Chapter 8: Discussion and Conclusion  By assessing the predictive values of ANS changes for the occurrence of secondary ischemic events in TIA and minor stroke patients, using HRV parameters as proxy for ANS function and stress, this study observed interesting findings that correspond to our hypotheses: • Several HRV parameters (related to HF, HF norm, HF+LF and TP) yielded good predictive values of ischemic events after TIA or minor stroke. Compared to ABCD2 score, their AUC were superior or at least comparable (section 7.4.1, Table 7.5).  • Two HRV variables showed optimal predictive results: Morning HF value and Daytime HF changes (section 7.4.2). Based on the Youden’s index, the optimal cut-off value for Morning HF is 50 ms2; and for Daytime HF changes is 12.5%. • Several HRV-based stress predictive models were built. Compared to ABCD2 score, they all had significantly improved predictive power (section 7.5, Tables 7.10 and 7.11).  • Two models were selected: The Best Stress Model and the Most Practical model for clinical use (section 7.6). Their diagnostic performance was tested (Tables 7.12 and 7.13). The cut-off formulas were calculated based on the maximum sum of sensitivity and specificity.   In this chapter, I will discuss successively the predictive values of different HRV parameters: HF values, HF changes, HF+LF, and TP in section 8.1. Both expected and unexpected results will be discussed. Then I will attend to the comparison between HRV predictors and ABCD2 score in section 8.2. The use of PSS will also be discussed (section 8.3). In section 8.4, I will address the HRV-based predictive models mainly on the BSM and the MPM. The clinical relevance, unique aspects and strengths of the study will be described in section 8.5. Some 158  important issues, challenges and limitations of the study will be reported in section 8.6. After all, I will outline possible future directions in section 8.7.   8.1 Predictive Values of HRV Parameters Because of the exploratory nature of several elements in this study, we tested a series of HRV parameters and identified their predictive values on the occurrence of secondary ischemic events after TIA or minor stroke. This broad scope approach on the use of various HRV parameters delivers different perspectives and interpretations on the use of HRV in predicting ischemic events. We will in turn discuss HF values, HF changes, HF+LF and TP in this section.    ❖ HF values Compared to those without outcome events, patients with secondary ischemic events had significantly lower Morning HF values (38.1 vs. 51.7 ms2, p=0.03) and Morning HF norm (0.21 vs. 0.28 n.u.; p=0.10), Day HF (38.1 vs. 59.9 ms2, p=0.001) and Day HF norm (0.23 vs. 0.29 n.u., p=0.01), as well as 24-h HF (80.7 vs. 100.9 ms2, p=0.06) and 24-h HF norm (0.32 vs. 0.37 n.u., p=0.06); although several p values slightly exceeded the significance level. These six HF variables yield acceptable AUC in univariate models: Morning HF: 0.61; Morning HF norm: 0.59; Day HF: 0.67; Day HF norm: 0.64; 24-h HF: 0.60; 24-h HF norm: 0.60.  The use of HF absolute values is supported by the goal of assessing the level of chronic stress of an individual. The lower absolute and normalized values of HF in the morning and entire day indicate a decreased PNS activity (Task Force of ESC, 1996), and thus the impaired function of PNS to control stress response in event-group patients (Black, 2003; Chrousos, 2009; Schneiderman et al., 2005). The dysfunctional PNS response to stress may be caused by the high 159  level of overall stress that includes both chronic and acute stress (Black, 2003; Chrousos, 2009; Schneiderman et al., 2005), as explained in section 2.2. Our results show that the level of baseline stress assessed by Morning HF value may predict the occurrence of secondary ischemic events. Although no study, to our knowledge, has focused on acute TIA/minor stroke population to examine the association between decreased HF values and an increased risk of secondary ischemic events, our results are supported by earlier findings which showed that (i) patients with ischemic stroke had decreased HF (and also LF and TP) and dysfunctional ANS compared to healthy controls (Dütsch et al., 2007; Korpelainen et al., 1999; Korpelainen et al., 1996; Naver et al., 1996; Raedt et al., 2015); and (ii) lower HF was correlated with a higher risk of incident stroke in adults (Fyfe-Johnson et al., 2016). Korpelainen and colleagues showed that compared to control subjects, stroke patients had significantly lower values of HF (1076 vs. 248 vs. 314 vs. 328 ms2, p<0.05) from acute phase to 1 and 6 months later. Impaired HRV also correlated with the severity of neurological deficits and disability (Korpelainen et al., 1996). Dutsch et al. reported that HF powers were reduced in patients after right- and left-sided strokes, and LF/HF ratio was elevated in patients after right- sided stroke, when compared with controls (HF: 162 vs. 234 ms2, p < 0.05; LF/HF: 1.77 vs. 0.32, p<0.05) (Dütsch et al., 2007). Fyfe-Johnson et al. found a near significant association between the lowest HF quintile and a higher stroke risk (HR=1.7; p=0.06) among people with diabetes mellitus (Fyfe-Johnson et al., 2016). A recent review by Yperzeele probed the associations between decreased HRV or baroreceptor sensitivity, and stroke severity, early and late complications, dependency and mortality (Yperzeele et al., 2015).  Moreover, although few studies have explored the predictive values of HF in cerebrovascular diseases, ANS activity and HRV (HF, LF, LF/HF and TP) have been extensively studied in other clinical conditions especially on cardiac disorders including MI, unstable angina 160  and chronic heart failure (Fukuda, Kanazawa, Aizawa, Ardell, & Shivkumar, 2015; Huikuri et al., 1994; Huikuri & Stein, 2013; Marmar & Shivkumar, 2008; Tapanainen et al., 2002; Tsuji et al., 1996; Wennerblom et al., 2001). Tsuji et al. (Framingham Heart Study) found negative associations between HF values and the incidence of new cardiac events such as angina pectoris, myocardial infarction, coronary heart disease death, or congestive heart failure (HR =1.38 (95% CI: 1.03, 1.84; p=0.028), after adjusting for age, sex, cigarette smoking, diabetes, left ventricular hypertrophy and other relevant risk factors (Tsuji et al., 1996). Additionally, Thayer and colleagues conducted several reviews on the role of ANS/vagal function and the use of HRV in predicting the risk for cardiovascular disease and mortality (Thayer & Lane, 2007; Thayer et al., 2010). They used a broad range of indicators for vagal function including resting HR, heart rate recovery, HRV, and baroreflex sensitivity and found that decreased vagal function was associated with an increased risk for cardiovascular morbidity and mortality.  ❖ HF changes More interestingly, we found significant differences of changes in HF between the event and event-free groups. Patients developing outcome events show a negative change in HF (-4.6%), while patients lacking outcome events show a positive change (18.2%) (p<0.001). The comparison of Daytime HF norm changes between patients with and without outcome events (0.021 vs. 0.050, p=0.02) indicates that the HF proportion increases more from morning to afternoon in patients without outcome events. Daytime HF changes render the highest AUC of 0.70 among all HRV parameters and significant OR in a univariate stress model (Table 7.5).  Our results illustrate that the changes in HF (increase or decrease), as a measure of dynamic PNS function/activity (improvement or deterioration) in response to acute stress after initial TIA 161  or minor stroke, could be an indicator of changes in stress level over a period. Daytime HF change might be an indicator of the change (improvement or deterioration) of health status during a short time period and the most appropriate HRV parameter for ischemic events prediction. This assessment can also be seen as an indicator of the “rebound capacity” of an individual’s body, with those having a poor rebound capacity being at higher risk of developing secondary ischemic events.  A decreasing trend in HF indicates a suppression of PNS function/activity. This effect may limit the positive actions of parasympathetic activity (anti-stress, anabolic effect) and may be a sign of health compromise. The uncoupling theory proposed by Ellenby (Ellenby et al., 2001) supports this statement. The uncoupling theory suggests that during systemic inflammatory response (an uncontrolled stressful situation), the organ responsiveness to autonomic signaling is diminished with decreased HRV, and this trend progresses with the severity of the disease (Godin & Buchman, 1996). This theory implies that the diminished autonomic modulation with decreased parasympathetic activity and relative increased sympathetic activity leads to reduced HRV (Beauchaine, 2001). This can be interpreted as a sign of uncontrolled stress that reflects (and may also contribute to) the progression toward health deterioration, and predicts a poor prognosis (Ellenby et al., 2001; Godin & Buchman, 1996). In opposition to the “uncoupling” process, recovery from diseases is characterized by the recuperation of HRV, or “recoupling” of inter-organ communication with ANS modulation (Ellenby et al., 2001). “Recoupling” theory signifies the return of parasympathetic activity and rebalance of sympatho-vagal ratio. Such rebalancing of ANS is associated with an overall increase in HRV, indicating a stress adaptation, homeostasis restoration and thus a recovered or improved health condition (Pumprla et al., 2002; Stein & Kleiger, 1999).  162  Our study is the first to document the possible use of dynamic ANS parameters to predict the occurrence of ischemic events after TIA or minor stroke. These results are supported by other studies on cardiovascular diseases which showed that both time and frequency parameters of HRV were more depressed at the early phase of acute MI, and substantially improve during the recovery phase, or after cardiac rehabilitation (Jokinen et al., 2003; Malfatto et al., 1996, 1998; Pardo et al., 2000; Routledge, Campbell, McFetridge-Durdle, & Bacon, 2010; Sandercock, Grocott-Mason, & Brodie, 2007; Santos-Hiss et al., 2011).  Jokinen and colleagues reported that HF power was 214 ± 350 ms2 at 5-7 days after acute MI, and 333 ± 648 ms2 at recovery phase 12 months after MI; p<0.001 (Jokinen et al., 2003). Sandercock et al. showed that after 8 weeks of cardiac rehabilitation program, patients with post-MI had significant increases in: logHF (3.8 ± 1.0 vs. 4.3 ± 0.3, p=0.024), logLF (2.4 ± 1.2 vs. 3.0 ± 2.4, p=0.014) and SDNN (29 ± 15 vs. 35 ± 17 ms, p=0.015), as well as physiological and psychological improvement (Sandercock et al., 2007). Santos-Hiss et al reported that after 5 days of phase-I cardiac rehabilitation, patients with post-MI showed an increase in HF nu (from 35.9 ± 19.5 to 65.19 ± 25.4, p<0.05) (Santos-Hiss et al., 2011). These results demonstrate that an improvement of health condition or rehabilitation is correlated with increases in HRV parameters, especially the HF power.  On the other hand, decreased HF or HF norm has been shown to be correlated with poor disease prognosis, such as mortality after myocardial infarction (Bigger et al., 1992; Brateanu, 2015; Kleiger et al., 1987; Malik et al., 1989), deterioration of ischemic stroke (Hilz et al., 2011) and cardiac complications after ischemic stroke (Orlandi et al., 2000). In the early 1990s, Bigger et al. showed that compared to post-MI patients with high HF (above the cut-off of 20 ms2), the unadjusted RR of all-cause mortality, cardiac mortality and arrhythmic mortality for patients with low HF (under the cut-off) were 2.5, 2.6 and 2.4, respectively (p <0.05 for all) (Bigger et al., 1992). 163  Hilz et al. showed that HF was negatively correlated with the NIHSS score (a scoring system for quantifying stroke severity, Appendix A) in patients with ischemic stroke (p=0.014) (Hilz et al., 2011). This result indicates that the deterioration of ischemic stroke is associated with a progressive loss of overall autonomic modulation, decline in parasympathetic tone, and a progressive shift toward sympathetic dominance.   Considerations regarding Day-night changes of HF- Against our a priori hypotheses, the Day-night changes of HF was positively associated with the occurrence of outcome events: OR = 1.01 (95% CI: 1.00, 1.02), p=0.06. This result might be interpreted as patients with outcome events having more gains in HF during sleep compared to patients without outcome events, although the median night HF value was still higher in patients without events (Night HF value: 138.1 vs. 116.4 ms2, p=0.15; Night HF norm: 0.44 vs. 0.38 n.u., p=0.32). Previous studies on circadian changes of HRV in stroke patients showed that compared to healthy people, patients with ischemic stroke had a lower circadian changes in all HRV parameters (HF, LF, VLF and total power) (Korpelainen, Sotaniemi, Huikuri, & Lylä, 1997; Wennerblom et al., 2001). Our findings may be in part explained by the fact that the study patients all suffered TIA or minor stroke, therefore the day-night contrast between these patients (with and without a secondary ischemic event) may be less important than that between stroke patients and healthy controls. This unexpected association may also receive other interpretations: sleep disorder is a common phenomenon in the immediate post-stroke phase, mainly caused by infarcts in specific brain locations and other precipitating factors such as emotional disorders (depression, anxiety), psychological stress, post-stroke pain, drugs, medical complications and environmental factors (noise and light) in the hospital (Balami, Chen, Grunwald, & Buchan, 2011). Patients who had a higher level of stress as assessed by HRV, were 164  likely to have more sleep problems, and therefore at a possible risk for administration of sedating or hypnotic drugs in order to improve sleep, mood or behavior (Balami et al., 2011). These drugs would influence the ANS function and HRV (Gouin et al., 2015; Silke, Campbell, & King, 2002; Win, Fukayama, Kohase, & Umino, 2005). However, we did not have the information regarding the sleep quality, and we did not have a complete description of all drugs taken during this period (will be discussed in section 8.6 “Limitations” and section 8.7 “Future Direction”). Overall, we believe that our assessment of Night HRV values may be affected by many factors that are not well assessed, which makes this observation of paradoxical increased Day-night HF in patients who developed an ischemic event, difficult to interpret.   ❖ HF+LF and TP HF+LF is not a traditional parameter of HRV frequency-domain analyses (Task Force of ESC, 1996); however, based on the physiological connection between ANS and HF (regulated by PNS) and LF powers (mediated by PNS and SNS), we have reasons to believe that the sum of HF and LF represents the actions of PNS and SNS, thus may act as another indicator, in parallel or even superior to total power, to reflect the ANS activity more specifically (as explained in section 5.5.3.2). In our study, as expected, we demonstrated the predictive values of HF+LF and TP on ischemic events occurrence. Patients with a secondary ischemic event had lower values of Day HF+LF (152.0 vs. 220.0 ms2, p=0.04), 24-h HF+LF (204.0 vs. 264.0 ms2, p=0.09), and 24-h TP (718.8 vs. 772.7 ms2, p=0.1) compared to those without outcome events. Moreover, patients with outcome events had significantly lower changes in HF+LF and TP during the daytime, compared to those without outcome events (Daytime HF+LF changes: -22.1% vs. 2.9%, p=0.002; Daytime TP changes: -9.8% vs. 1.8%, p=0.03). More interestingly, we observed a higher AUC of HF+LF 165  compared to TP: Day HF vs. Day TP: 0.61 vs. 0.57; Daytime HF+LF changes vs. Daytime TP changes: 0.67 vs. 0.61. The predictive effectiveness of both HF+LF and TP supports the fact that ANS activity as a whole is affected by stress. This more accurate prediction of HF + LF (representing ANS function only) compared to TP (representing ANS and other factors expressed in VLF spectrum) may also be interpreted and supported by the whole stress hypothesis that states that assessing ANS function is a good way to determine stress status, rebound capacity (as described in section “HF change” above), hence health prognosis. We focused mostly on HF values and changes because of the strong correlations between HF and the occurrence of ischemic events; but all HRV parameters showed the same trends of association. Since HF and LF are components of HRV that are more specifically related to ANS activity/modulation (PNS and SNS branches), the increases in overall HF and LF reflect the improvement of overall ANS adaptation to stress, and may be markers of an enhanced or rehabilitated health condition. This inference can then explain the association between increased HF+LF and a reduced risk of developing a secondary ischemic event.  Our results are supported by the findings from measures of total power (measures of overall ANS activity) in cerebro- and cardio-vascular research (Eller, 2007; Hanss et al., 2008; Hillebrand et al., 2013; Iellamo, Legramante, Massaro, Raimondi, & Galante, 2000; Jokinen et al., 2003; Korpelainen et al., 1996; Mäkikallio et al., 2004; McLaren et al., 2005). Stroke patients were reported to have a significantly lower value of TP compared to healthy people, after adjusting for confounding factors: logTP: 6.31 ± 1.17 vs. 5.90 ± 0.96, p=0.032 (McLaren et al., 2005). Korpelainen et al. displayed an association between lower TP and poor post-stroke prognosis, such as neurological deficits and disability (p=0.006) (Korpelainen et al., 1996). Jokinen and colleagues found that all time and frequency-domain measurements of HRV increased significantly from the 166  acute phase of MI to the one-year recovery period: VLF power: 1037 ± 829 vs. 1380 ± 1105 ms2; LF power: 430 ± 447 vs. 632 ± 765 ms2; HF power: 214 ± 350 vs. 333 ± 648 ms2; p<0.01 for all HRV parameters (Jokinen et al., 2003). Hanss, et al. reported that total power <500 ms2/Hz was associated with high sensitivity and specificity for the prediction of hypotension and bradycardia (Hanss et al., 2008). Furthermore, Hillebrand et al. conducted a meta-analysis on HRV and the first cardiovascular event in populations without known cardiovascular disease, and showed that the pooled RR was 1.35 (95% CI: 1.10, 1.67, p<0.05), when comparing the lowest level to the highest level of SDNN (a time-domain parameter that represents overall ANS activity) (Hillebrand et al., 2013).   8.2 HRV Predictors versus ABCD2 Score In this study, we assessed the predictive power of ABCD2 score, as the reference in our study patients, (Table 7.5). The AUC of ABCD2 score tested in our study is consistent with previous studies on ABCD2 score that showed its AUC ranging from 0.57 to 0.7 (Chandratheva et al., 2011; Kiyohara et al., 2014; Perry et al., 2011; Sanders et al., 2012; Wardlaw et al., 2015). Some HRV predictors, including Morning HF, Day HF, Day HF norm and Day HF+LF, as well as Daytime changes of all tested HRV parameters, showed comparable or even better AUC (0.61 to 0.70) compared to ABCD2 score. Additionally, when adding HRV parameters into the model with ABCD2 score alone, the prognostic power of the model was improved to 0.74 (p=0.01).  There are no previous studies focused on exploring the prediction improvement by adding HRV to ABCD2 score. However, as described in Chapters 2 and 3, HRV parameters can be seen as indicators of the level of stress; we might compare our results of HRV with other stress indicators. Two recent studies that focused on the predictive value of the stress hormone 167  “copeptin” on cerebrovascular recurrence after TIA reported very similar results to ours (De Marchis et al., 2014; Katan et al., 2011). The study by Marchis et al. showed that after the addition of copeptin to the ABCD2 score, the AUC was improved significantly from 0.60 (95% CI, 0.46, 0.74) to 0.74 (95% CI, 0.60, 0.88), p=0.02 (De Marchis et al., 2014). The other study by Katan et al. showed that copeptin provided a significantly higher AUC to predict cerebrovascular re-events than the ABCD2 score (0.73 vs 0.43; p <0.01) and improved ABCD2’s predictive power to 0.77 (p=0.002) (Katan et al., 2011). These results support the usefulness of stress indicators on cerebrovascular ischemic event prediction in TIA or minor stroke patients. Despite the fact that ABCD2 score is a clinically recognized tool, the limited number of ANS parameters (or stressors) involved in score computation constrains its predictive power. Moreover, because TIA symptoms and signs usually have resolved by the time of clinical assessment, some of the ABCD2 items (clinical features and symptom duration) that rely on patients’ self-reporting, may be imprecise. HRV parameters however, as representative of a physiological phenomenon, are objective indicators of the overall stress in the individual.   8.3 The Use of PSS Score in the Model In our study, PSS score generated a predictive value of 0.7, the same as the best HRV predictor, Daytime HF changes. By adding PSS score into the predictive model, the AUC of the full model (with and ABCD2 score, PSS score and HRV) was significantly improved to 0.82. We found that the OR of PSS score is 1.20 (95% CI: 1.08, 1.36; p =0.002) in the BSM and 1.22 (95% CI: 1.09, 1.37; p <0.001) in the MPM. Our results are supported by previous study which showed that PSS score was associated with increased risk of fatal stroke (HR 1.45, 95 % CI: 1.19,1.78; P < 0.001) and total ischemic stroke (HR 1.40, 95 % CI: 1.00,1.97; P = 0.05) (Booth et al., 2015).  168  As described in section 2.3.9, psychological stress is a well-established risk factor for stroke development. Psychological stress can be assessed through different tools such as PSS score, self-reported stress events, single questions, major life events, and others (Booth et al., 2015; Henderson et al., 2013; Kornerup et al., 2010; Ohlin, Nilsson, Nilsson, & Berglund, 2004; Suadicani et al., 2011; Truelsen, Nielsen, Boysen, & Grønbæk, 2003). Regardless of the type of measurement, higher levels of psychological stress were associated with an increased hazard ratio of stroke outcomes (Booth et al., 2015; Henderson et al., 2013; Kornerup et al., 2010; Ohlin et al., 2004; Suadicani et al., 2011; Truelsen et al., 2003). Three reasons drove us to use the PSS score rather than other psychological stress tools in the study: (i) PSS is a simple questionnaire with only 10 questions that are easy to understand. Junior middle school education is sufficient to ensure proper understanding of the questionnaire (Cohen et al., 1983); (ii) using PSS to assess the perception of stress eliminates the effect of personality and coping ability with stressful events; and (iii) PSS does not record the specific stressful events and so protects patients’ privacy and therefore arguably results in less reporting bias.  Compared to presence of diabetes, hypertension and dyslipidemia that represent chronic stress, PSS references a specific time period of acute stress (for example one month prior to the assessment moment). Hospital stress, which potentially occurs at the end of the PSS measurement period, may however influence the responses.   8.4 The Use of Multiple Testing   Normally in walk-in clinics or emergency settings, doctors prefer simple tests to evaluate patients rather than computer-based multivariate models. Therefore, we attempted to set up cut-off values for the three tests: ABCD2 score, Morning/Baseline HF value, and PSS score to get 169  clinically relevant categories of sensitivity and specificity. We identified the number of patients who scored positive in one test (+), positive in two tests (++) and positive in three tests (+++). The results of the three tests are presented in Figure 8.1.  By considering the effectiveness and safety of urgent treatment of TIA/minor stroke patients (discussed in section 8.5 clinical relevance), we prefer a sensitivity of at least 80% when setting up the thresholds. We set a score of ABCD2 ≥ 4 for our threshold, as it is a well-recognized cut-off to identify TIA patients under high risk of recurrent ischemic stroke (Easton et al., 2009); it provides 86% sensitivity and 27% specificity in our study (Table 7.6). We set a score of PSS ≥18 to provide 81% sensitivity and 58% specificity (Table 7.7). Lastly, we set Morning HF value at ≤ 80 ms2 to offer 83% sensitivity and 33% specificity. The test performances for single tests (ABCD2 score, Morning HF and PSS score), combined two tests and three tests are shown in Table 8.1. Indeed, gain in sensitivity correlates with the loss of specificity. Physicians can choose the most meaningful approach, based on the clinical situation.170  Figure 8-1 Multiple testing by ABCD2 score, HRV and PSS score   Yellow shadow: those who get positive result if tested with ABCD2 score, i.e. ABCD2 ≥4  Blue shadow: those who get positive result if tested with Morning HF, i.e. Morning HF ≤80 ms2  Red shadow: those who get positive result if tested with PSS score, i.e. PSS ≥18 Overlapping area of yellow and blue: those who get positive in both ABCD2 score and Morning HF, i.e. ABCD2 ≥4 and Morning HF ≤80 ms2 Overlapping area of yellow and red: those who get positive in both ABCD2 score and PSS score, i.e. ABCD2 ≥4 and PSS ≥18 Overlapping area of blue and red: those who get positive in both Morning HF and PSS score, i.e. Morning HF ≤80 ms2 and PSS ≥18 Overlapping area of yellow, blue and red: those who get positive in all three tests, i.e. ABCD2 ≥4 and Morning HF ≤80 ms2 and PSS ≥18 171  Table 8.1 Testing performance of ABCD2 score, Morning HF value and PSS score, and combined tests  Tests with Results No. of patients with events No. of patients with event-free Sensitivity Specificity PPV NPV  ABCD2 score         ABCD2 ≥4 scores 31 120 86% 27% 21% 90%   ABCD2 <4 scores 5 45 Morning HF         MorningHF <80 ms2 30 111 83% 33% 21% 90%   MorningHF ≥80 ms2 6 54 PSS score         PSS ≥18 scores 29 69 81% 58% 30% 93%   PSS <18 scores 7 96 ABCD2 + Morning HF         High risk (two “+”) 25 81 69% 51% 24% 88%   Low risk (at least one “-”) 11 84 ABCD2 + PSS         High risk (two “+”) 24 47 67% 72% 34% 91%   Low risk (at least one “-”) 12 118 Morning HF + PSS         High risk (two “+”) 25 46 69% 72% 35% 92%   Low risk (at least one “-”) 11 119 ABCD2 + Morning HF + PSS         High risk (three “+”) 20 34 56% 79% 37% 89%   Low risk (at least one “-”) 16 131 PPV: positive predictive value; NPV: negative predictive value;  High risk (two “+”) refers to positive results in the two selected tests;  High risk (three “+”) refers to positive results in all three tests;  Low risk refers to at least one negative result in the selected tests (i.e. not all positive)172  8.5 HRV-based Stress Predictive Models The HRV-based predictive stress models involving HRV parameters, PSS score and ABCD2 score as predictors were described in section 7.5. Some models provide significant ORs (p<0.1) and higher AUCs (between 0.75 and 0.8) compared to ABCD2 score alone (0.63) (p<0.05) (Table 7.10). The use and interpretation of these models can follow the same logic as we developed for single HRV indicator, explained above. In this section, I will focus on our two selected models: the Best Stress Model and the Most Practical Model.  ❖ The Best Stress Model The BSM contains Daytime HF changes, PSS score and ABCD2 score (Equation 7-5) and thus is able to assess a broad spectrum of stressors (physiological/metabolic and psychological, chronic and acute). The BSM generates the best AUC of 0.82 (significantly higher than ABCD2 score, p<0.001) among all models, which indicates its excellent ability in predicting ischemic events. It has a comparable predictive performance to imaging testing which is considered as the most accurate predictive tool at present as describe in section 2.1.6 (Ay et al., 2009; Calvet et al., 2009; Merwick et al., 2010; Song et al., 2013). OR of 0.84 (95% CI: 0.75, 0.92, p<0.001) for Daytime HF changes indicates that for every 10% increase in HF from morning to afternoon, the risk of developing a secondary ischemic outcome event is reduced by 16% (95% CI: 8%, 25%). OR of 1.20 (95% CI: 1.08, 1.36, p =0.002) for PSS score indicates that for every 1 point increase 173  in PSS score, the risk of developing a secondary ischemic outcome event is increased by 20% (95% CI: 8%, 36%).  The predictive value of HF changes has been discussed in section 8.1. The AUC of the BSM was significantly higher than that of the ABCD2 score and ABCD2-based bivariate models (Table 7.11), but was not significantly superior to the model involving Daytime HF changes plus PSS (AUC=0.80). This implies that the addition of ABCD2 score into the model does not significantly improve model’s predictive power. However, since ABCD2 score is a traditional predictive tool comprising important items for evaluating a patient’s stress profile, we decided to keep ABCD2 score in the final predictive model.  The BSM can be used in any clinical or home settings. In order to test the dynamic changes of HRV, the Holter monitor has to be worn for an entire day (morning and afternoon), which delays decision-making by 5-6 hours and makes HF change computation more complex. Based on equation 7-5, the outcome events probability can be calculated for each individual by providing her/his specific profile (HF changes, PSS and ABCD2 score). Depending on the specific objective, different cut-off points of “event probability” can be selected (Table 7.12).   ❖ The Most Practical Model The MPM consists of Morning HF, PSS score and ABCD2 score (Equation 6-6). It significantly improved the predictive power of ABCD2 score from 0.63 to 0.76 (p=0.02). The OR 174  of 0.64 (95% CI: 0.43, 0.94, p=0.023) for Morning HF indicated that for every 2.72-fold (the natural constant “e” =2.71828) increases in Morning HF, the risk of developing an outcome event reduced by 36% (95% CI: 6%, 57%). The OR of 1.22 (95% CI: 1.09, 1.37, p<0.001) for PSS illustrates that for every 1 point increase in PSS score, the risk of developing an outcome event is increased by 22% (95% CI: 9%, 37%).  The MPM has several positive features: (i) it only requires assessing HRV in the morning which is less time consuming and more practical, and enables rapid decision making and urgent treatment; (ii) in order to test the predictive value of HRV, we started to record outcome events after Holter removal, which led to missing the cases that occurred during the Holter monitoring. Because of the shortest recording of HRV in the MPM, it may miss the least number of outcome cases; and (iii) it is an optimal representative of a patient’s stress profile that includes the effect of life-long chronic stressors and acute stress situation (section 2.2.4 and section 2.3). As the effects of chronic stressors are relatively stable in one person, morning HF absolute values represent a reliable marker of baseline stress level for an individual (Freed, Stein, Gordon, Urban, & Kligfield, 1994; Guijt et al., 2007).  This model can be used more easily in an emergency environment, where time is limited and physicians need to provide rapid assessment and management of patients. When using this model, the trade-off between sensitivity and specificity has to be considered depending on its intended purposes. In order to identify all “high risk” patients and administer the necessary urgent 175  treatment in the emergency environment, a high sensitivity may be more desirable. A low cut-off of disease probability (0.1) may be suggested, with a sensitivity of 90% and a specificity of 40% (Table 7.13).   ❖ The Exploratory Stress Model  We also constructed two exploratory stress models by combing HRV parameters (section 7.7). In the ESM-1 where we added both Morning HF value and Daytime HF changes into the model, the AUC was improved to 0.84. In the ESM-2, we created a new HRV variable by combining Morning HF value and Daytime HF changes; its AUC was 0.82. Both models had significantly higher AUCs compared to ABCD2 score (p<0.001). We defined two situations as the “medium risk level” in ESM-2: (a) HF increases ≥12% plus HF value < 50 ms2, and (b) HF changes <12% plus HF value ≥50 ms2, as it is hard to distinguish the effect differences between HF value and HF changes at exploratory level. Further, three categories drive more practical values for clinical use. The ORs between the high-risk, medium-risk, and low-risk groups indicate the excellent risk discriminative ability of the ESM-2; however, the single event in the low-risk group leads to unstable results with huge standard error and thus huge variance of ORs: 17.5 with 95% CI between 3.2 and 332.8 for medium-risk group (p=0.008), and 39.6 with 95% CI between 6.8 and 769.1 (p=0.001) for high-risk group, low-risk group as reference (Table 7.15).  176  Overall, ESM -1 provides excellent predictive power, while ESM -2 is simpler to use. The improvement in predictive power illustrates the importance of considering both HF baseline values and its changes during the day, as they collectively measure the baseline stress/health status (HF values) and the changes in stress/health condition (HF changes during the day). Both HF baseline values and HF changes are independently associated with the prediction of ischemic events. The exploratory models follow the current concept/logic that both chronic risk factors/stressors and acute triggers/stressors contribute to the occurrence of ischemic stroke, as described in Chapter 2 (Elkind, 2007; Furie et al., 2011; Kernan et al., 2014; Sacco et al., 2006; Sharma et al., 2015).  8.6 Clinical Relevance, Unique Aspects and Strengths Clinical relevance is related to the importance of the results with regard to clinical impact (Bhardwaj, Camacho, Derrow, Fleischer, & Feldman, 2004). As described in section 2.1.7, dual antiplatelet therapy, anticoagulative treatment, and operations (such as intravenous thrombolysis and endovascular treatment) may reduce the risk of early recurrent ischemic stroke and improve long-term outcome in survivors (Furie et al., 2011; Goldstein et al., 2006; Jauch et al., 2013; Kernan et al., 2014; Powers et al., 2015; Sacco et al., 2006; Sandercock et al., 2008; Wardlaw et al., 2009). However, they may also increase the risk of both minor bleeding and fatal intracranial hemorrhage (Diener et al., 2004; Geeganage et al., 2012; Miller et al., 2011). With HRV we get 177  more accurate identification of the group at high risk to provide appropriate treatment, therefore limiting the number of people exposed to the drug or operations without receiving any benefit.  In addition, our HRV-based predictive models show superior performances to ABCD2 score. For example, if sensitivity of 86% is taken for the model with ABCD2 score alone, the BSM and the MPM, the specificity, PPV and NPV are improved in the BSM and the MPM (Table 8.2). This indicates important clinical significances of the BSM and the MPM.  Table 8.2 Comparison of diagnostic performance between ABCD2 score, BSM and MPM   Sensitivity  Specificity  PPV NPV AUC ABCD2 score 86% 27% 21% 90% 0.63 Best Stress Model 86% 63% 34% 95% 0.82 Most Practical Model 86% 42% 26% 92% 0.76 Data are generated from Tables 7.6, 7.10, 7.12 and 7.13  Our study has a number of unique elements:  ❖ Stress and strokes  Our study is the first to address the association between stress and risk factors of ischemic stroke. By showing the association between risk factors of ischemic stroke (or stressors) and ANS modulation, we developed several stress models for predicting secondary ischemic events after TIA or minor stroke. These models include multiple dimensions of stress sources including chronic and acute, physiological and psychological. Our study contributes to developing the perspective of ischemic stroke being a stress-related condition. 178  ❖ Selection of HRV predictors We conducted comprehensive analyses of important HRV parameters (HF, HF.nu, HF+LF and total power) based on a priori hypotheses. We used different forms (absolute value and changes over time) of HRV variables and compared their predictive ability. Although no study has directly examined the values of HRV parameters on ischemic events prediction in patients with TIA or minor stroke, our choice of selected variables is supported by abundant literature from HRV studies in other domains. The logic and considerations in selecting the candidate HRV predictors would likely be directly applicable to HRV research in chronic diseases. We also identified the cut-off points for the best HRV predictor: 50 ms2 for Baseline HF values and 12.5% increase for Daytime HF changes. These thresholds, if verified, would represent the preliminary risk stratification for TIA/minor stroke patients.   ❖ Establishment of HRV-based models  We established several HRV-based stress predictive models, which showed an association between HRV parameters and the risk of ischemic events after TIA and minor stroke. We selected two models for different purposes: the BSM offers the best AUC of 0.82 which could be used widely if long-term recording of HRV is possible; and the MPM with the shortest HRV recording is accessible in emergency settings. We also developed two exploratory models by including both baseline and dynamic changes of ANS parameters, which proffer equal or even better predictive 179  power (0.82 and 0.84) compared to the BSM and the MPM. These results invite more research into the area of using both the baseline values of HRV parameters and their changes over time to predict the individual’s health status (baseline level) and rebound capacity (the change), and therefore the risk of future event.   ❖ Study design and conduct  We used both time-based and tissue-based diagnosis criteria for TIA. The secondary/confirmed diagnosis in the ward corrected the preliminary diagnosis in emergency setting. From the corrections, 9 TIAs were re-diagnosed as minor stroke, and 4 TIAs were re-diagnosed as other diseases mimicking TIA. Because we included both TIA and minor stroke patients, the double diagnostic process ensured the accuracy of diagnosis without causing the loss of patients.  We designed our study to test HRV during the acute phase (48 hours) post TIA and minor stroke, although such a limited time-window created difficulties in patient recruitment. This 48-hour criteria, however, reduced the heterogeneity of study patients to some extent. For example, if recruiting patients with TIA or minor stroke in 7 days, the ones recruited during the acute phase (such as 48 hours) may have different health conditions from the ones recruited at day 5 or day 7 after their initial events.  180  We limited our study to hospitalized patients in order to ensure that the quality of care between patients was comparable. An example can be seen in the use of 24-hour Holter monitoring, which was used as our primary study assessment. More comparable results were achieved for our data because the tests were performed on patients at the similar time of day, and nurses were available to help if lines fell off, largely reducing the possibility of missing crucial Holter data. Moreover, as the first several days after TIA or minor stroke are a high-risk period for the development of ischemic events, the hospitalized patients were constantly monitored by both the responsible physician and study manager during the first 7 days to guarantee the quality of outcome events assessment. At this exploratory level, our study aimed to examine the effectiveness of HRV in predicting secondary ischemic events after TIA and minor stroke. Our study was designed with strict inclusion and exclusion criteria in order to make the participants well described; the generalizability can be assessed in future research.    Overall, our study shows that assessing stress expression through HRV and its changes over a discrete time period may have clinical utility in identifying patients at high risk of developing an ischemic event after TIA or minor stroke. The HRV-based predictive models, especially HF power which represents PNS activity, possess a series of advantages: important predictive performance (significantly superior prediction to classic ABCD2 score), simple and 181  rapid use, potential cost savings, and applicability in a wide range of settings. These results however need to be verified in other studies before changes of practice recommended.  8.7 Important Issues, Challenges and Limitations  During the two-year data collection under a specific medical system, we faced several important issues and challenges that warrant discussion.   ❖ Difficulties in recruiting study patients The first constraint to our study concerned with the recruitment. Recruiting patients within the first 48 hours after their TIA or minor stroke events has always been a major issue for study recruitment. Moreover, access to 24-hour HRV recording within 48 hours further shortened the time-window for eligible enrollment, especially due to fixed delay of Holter wearing in normal clinical procedure (diagnosis, hospitalization, requisition), and due to Holter laboratory arranging tests only in the morning. Therefore, we could only recruit patients who came to the hospital within 40 hours (or even less) after the initial event. As described in section 7.1, we screened around 3000 patients who came for potential ischemic symptoms at the Neurology Emergency Clinic; only 347 (11%) fit under the defined time-window and diagnosed with TIA or minor stroke. This low rate indicates a need to educate the general population to sensitize people to the “benign symptoms” of ischemic events. Among the 347 eligible patients, 206 (59%) were finally recruited into the study. 182  This low recruitment rate was mostly attributed to the factors that result in exceeding the time window, such as the availability of beds, Holter appointment time, and conflicts between Holter wearing and MRI test which is always the priority for stroke patients (Figure 7.1). Additionally, we had to follow each of our patients for 3 months to assess the possible development of a secondary ischemic stroke; such period of follow-up time may reduce patients’ motivation to participate in the study.  ❖ Challenges in diagnosis of TIA and minor stroke The challenges of diagnosing TIA are commonly described in the literature (Amort et al., 2011; Barrett et al., 2008; Hand et al., 2006; Nadarajan et al., 2014; Schrock, Glasenapp, Victor, Losey, & Cydulka, 2012). The main diagnostic challenge is due to the fact that most symptoms and signs have resolved by the time of assessment (Schrock et al., 2012). There is no test for TIA, and as such the gold standard remains the assessment by a clinical expert as soon as possible (Schrock et al., 2012). The diagnosis relies heavily on the patient’s account of their history and on expert interpretation of that history (Schrock et al., 2012). Interrater agreement for the diagnosis of TIA between different stroke-trained physicians and emergency physician is poor (agreement coefficient of 0.35) (Castle et al., 2010; Koudstaal, Gerritsma, & van Gijn, 1989; Schrock et al., 2012). Even in a neurological emergency clinic, we still had 9 patients who were preliminarily 183  diagnosed with TIA and then later corrected to “minor stroke”, and 4 TIAs corrected to other diseases mimicking TIA.   ❖ Outcome event rate   The event rate of 17.9% (11.9% for “Definite”, and 15.9% for “Probable and Definite”) in our study is similar to the ones showed in other observational studies (Chandratheva et al., 2009; Coull et al., 2004; Johnston et al., 2000; Rothwell & Warlow, 2005), but higher than the ones reported in recent RCTs (Johnston et al., 2016; Wang et al., 2013). This discrepancy may be due to different populations and the different ways in which health outcomes are assessed in RCTs and in observational studies. RCTs are conducted under ideal conditions to determine the efficacy of a treatment or intervention (Anglemyer, Horvath, & Bero, 2014; Friedman, Furberg, & DeMets, 2010; Gordis, 2008). In order to create an ideal experimental condition, RCT designs follow a rigorous methodology including strict inclusion and exclusion criteria to recruit the “most appropriate” patients for assessing the effects of the intervention (Friedman, Furberg, & DeMets, 2010; Gordis, 2008). Patients in clinical trials usually have a better prognosis than other similar patients not included in the trials (Goyal et al., 2012). There are several reasons for this: in RCT patients get better medical care including easy access to the health system, receiving the best conventional treatment, frequent clinical assessment, and regular follow ups (Hannan, 2008; Friedman, Furberg, & DeMets, 2010; Goyal et al., 2012). Observational studies on the other hand 184  are conducted in “real world” scenarios to describe the population or determine the effectiveness of an intervention (Gordis, 2008; Rothman, 2012).  In addition, the potential risk of biases in observational studies is higher than in RCTs (Gordis, 2008; Rothman, 2012), although several methods are used, at both design and analysis levels, to control and reduce these biases (Gordis, 2008; Rothman, Lash, & Greenland, 2012). In our study, the respondant bias (due to errors of recall and/or reporting, or intention) may exist, when patients were discharged home and followed by telephone; however, these possible biases would likely be similar in patients with different stress profiles. There is also a possibility of misdiagnosis of the outcome events when the description is not clear or information is incomplete. To minimize this possibility, every event was evaluated carefully by two neurologists who were not aware of the patient’ stress profile, i.e. ABCD2 score, HRV parameters, and PSS score.   ❖ 24-hour Holter recording (HRV recording)  Holter recording is one of the most important processes in this study. Several issues are discussed.   • Extraction of Holter data The raw data from 24-hour Holter monitoring was stored in the central computer in the Holter laboratory and could not be read by other analytic software. We contacted the company and received permission to use the data for research purposes. They provided 185  software to transform the raw Holter data into voltage format. We checked the consistency between the ECG waves produced by the transformed data and the original ECG waves analysed by the complete set of analytic software of the 24-hour Holter monitor. Although not obtained directly, we believe that the transformed HRV data set is valid. The qualified transformed voltage data was then used to analyse HRV with Acqknowledge software 4.1.   • The definition of time period for HRV analysis As explained in Chapter 4, we used 9 am to 12 pm, 3 pm to 6 pm, and 12 am to 3 am to represent the morning period, afternoon period and night period, respectively. Such proxy definitions are based on: (i) Figure 4.1 and 4.2 which show that HRV is relatively stable from 9 am to 12 pm and from 3 pm to 6 pm; (ii) avoidance of “wake-up” effect (6 am to 9 am); (iii) avoidance of nap time (12 pm to 3 pm); and (iv) assumption of 12 am to 3 am as the sleeping time. These definitions although not validated are justifiable.   • Starting time of Holter monitoring The 24-hour Holter monitoring was administered to patients only in the morning period, as was consistent with hospital procedures. At the beginning of the study, because of the strict time criteria and low recruitment rate, the study manager arranged two Holter slots for patients who were hospitalized in the afternoon or early evening; this accommodated 5 186  patients. The resultant inconsistent Holter start time caused difficulties when comparing the baseline HRV values and changes. The 5 patients who had Holter monitoring in the afternoon or early evening were processed with using the nighttime data and the daytime data in the following day (still in the time window). All 5 patients were in the event-free group, therefore, the day-night change of HRV measurement was not influenced by these 5 patients.   • Calculation of HRV variables Twenty-one patients who came to the hospital within the 48 hours after the initial events were excluded, because they could not have Holter monitoring within the 48 hours. In our study, most patients were able to wear the Holter at 8 am or shortly after. The “Morning HRV” was calculated using the data from the start time to 12 pm. As explained in the previous section 6.5.3.2, we selected 5 minutes interval from each half hour to analyse HRV, so the “Morning HRV (normally from 9 am to 12 pm)” was calculated from the average of six 5-minute intervals. If the starting time was delayed, we used 5 or 4 fragments to calculate the mean instead of using 6 fragments. No patient received Holter after 10 am, so none had less than 4 fragments. Based on the 24-hour rhythm of HRV shown in Figures 4.1 and 4.2, we know that HRV stays relatively stable from 9 am to 12 pm. The average 187  HRV between 9 am to 12 pm is considered to be comparable to average HRV between 10 am to 12 pm.   • Generalizability of HRV cut-off  Although one unique aspect of this study was to identify the cut-off points for Morning HF values (50 ms2) and Daytime HF changes (12%) in order to conduct the primary risk stratification for TIA/minor stroke patients, this generalizability deserves caution. Our view is that such cutoffs should be attempted only in the context of considering the implications for a specific outcome and study population. Take acute MI as an example, Lanza et al. reported that patients surviving with an acute MI with HF <10 ms2 had a higher risk of cardiac death (RR=3.0, 95% CI: 1.1-8.2, p<0.02) and sudden death (RR=4.5, 95% CI: 1.3-15, p<0.02) (Lanza et al., 1998). In another study by Kleiger et al, over a 31 month follow up period, patients with SDNN <50 ms were reported to have 5.3 times higher risk of mortality compared to patients with SDNN >100 ms (Kleiger et al., 1987). Moreover, Rovere et al. found that having either SDNN <70 ms or BRS <3.0 ms/mmHg carried a significant risk of cardiac mortality: 3.2 (95% CI 1.42 – 7.36, p<0.05) and 2.8 (95% CI 1.24 - 6.16, p<0.05), respectively (Teresa et al., 1998). There is no universal consensus on an algorithm for determining the ideal cutoffs for different HRV measures for different outcomes, populations and studies (Huikuri & Stein, 2013). Our study identified the cut-188  off values (based on study data) to discriminate the group that had the highest risk of developing an ischemic event from the group estimated to have a lower risk. These cut-off points were calculated to maximize collective sensitivity and specificity (Youden’s index). Indeed, the cutoffs would change if we decide to give priority to sensitivity rather than the sum of sensitivity and specificity. See Figures 7.8 and 7.9 for a full picture. Our results need to be verified in other studies as we cannot exclude that some of the observed associations may be due to random effects.   ❖ Unavailability of HRV data in patients with permanent atrial fibrillation  HRV is defined as the variation between normal R-R intervals or correspondingly to the instantaneous heart rate, which is affected partly by an autonomic neural regulation of the cardiocirculatory system (Task Force of ESC, 1996). Any heartbeat that is not from the sinus node cannot be included in the HRV analysis  (Task Force of ESC, 1996). HRV is highly sensitive to artifact and errors; as low as 2% of the data will result in unwanted biases in HRV calculations (Citi, Brown, & Barbieri, 2012). To ensure accurate results, it is critical to manage artifacts and R-R errors appropriately prior to performing any HRV analyses (Citi et al., 2012). To respect the requirement of data quality, the specific section with AF waves on the ECG becomes unavailable for any types of HRV analyses.  189  In our study, we excluded one patient with ectopic complexes in most of the ECG waveforms - this patient was then diagnosed with permanent AF by the cardiologist during hospitalization. We were able to keep the 17 patients with paroxysmal or persistent AF, because the utilization of 24-hour recording and selection of 5-minute interval in every half an hour (described in section 6.5.3.2) provided us the availability to skip the 5-minute segment where the AF was present when analysing HRV. In our study, no deletion of data was used. Interpolations were performed for ectopic beats in some cases (Lippman et al., 1994). As we used the mean of 3 hours’ data to represent a period (i.e. the morning, the afternoon and the night), the skipping of 5-min or interpolations for some ectopic waves did not affect the reliability of HRV analyses (Salo et al., 2001). This calls to our attention the limitation of the use of HRV in any population with permanent AF and the caution of HRV utility in paroxysmal arrhythmia.   ❖ Assessment of patients’ status at night Another limitation in the study was the absence of assessment of the patients’ status at nighttime, such as the quality of sleep and the medication use at night. As explained in section 8.2, the study results on Day-night changes of HF were opposite from our hypothesis and other study results; however, the interpretation of the direction of this change was limited by the absence of reliable information regarding nighttime conditions.   190  ❖ Absence of association with some traditional risk factors In section 7.3, we compared personal and clinical variables between event and event-free groups, including age, gender, diagnosis, historical and present atrial fibrillation, hypertension, diabetes, dyslipidemia and cardiovascular diseases, a history of cerebrovascular diseases, cigarette smoking, alcohol consumption, medication and recruitment time. We did not find the classic association between some traditional risk factors and the outcome events, such as age, AF, cardiovascular diseases, a history of cerebrovascular diseases, smoking and alcohol consumption.  Several reasons may explain this finding. Some of these risk factors are difficult to measure. For instance, some cases of atrial fibrillation, especially the paroxysmal type, may not be identified by 24 hours’ Holter recording, which may affect the true association between AF and the risk of outcome ischemic events. Cardiovascular diseases are a combination of diseases, including coronary artery disease, myocardial infarction and unstable angina. However only myocardial infarction and coronary artery disease have been demonstrated to be associated with development of ischemic stroke (Curtis & O’Keefe, 2002; Huikuri & Stein, 2013; Sroka et al., 1997; Steptoe & Kivimäki, 2012; Touzé et al., 2005). A combination of diseases (without categorizing the type of cardiovascular disease) associated with unclear occurrence time may create imprecision in the assessment, which affects the true association. Also, given possibly different definitions of smoking/drinking status, the accurate measurement of dose and years of 191  smoking and alcohol consumption, and imprecise determination of their effects on human body may affect the association at stake. Quality of measurement should also be taken into consideration as another possible factor. All medical history and smoking and drinking status were assessed by patients’ self-reporting, which may be affected by recall and reporting biases. Moreover, because of the country cultures and customs, smoking and drinking are largely affected by gender, occupation and educational level. For example, in China female smoking or drinking is not common for cultural reasons, especially among the senior generations. Conversely, in males, heavy smoking and drinking, especially for the senior generations, are frequent. In our study, among 38 female patients only 7 (18%) reported smoking and 2 (5%) reported drinking. However, among 163 male patients, 130 (80%) were smokers and 124 (76%) were drinkers. Among the 36 patients who developed outcome events, none of 6 female cases reported smoking or drinking. All of these reasons may cause misclassifications that affect the associations with outcome events.  ❖ Reliability and validity of HRV assessment       In our study, the sampling rate for extracting ECG waves was 250 Hz, which meets the technical requirement for optimal sampling rate (250 to 500 Hz, or even higher) suggested by the Task Force Guideline (Task Force of ESC, 1996). However, another software with a higher 192  sampling rate (1000Hz for instance) may provide more precise ECG waves that make the HRV analysis more reliable.  Quality of HRV analysis: I personally conducted the HRV analyses. I received a special training and supervision from the Biopac Inc. seven years ago. After the training, I have conducted several HRV studies in my Master and PhD. During HRV analyses, one important procedure is to clean the ECG waves by removing ectopic and artifact waves. To ensure consistency in the process, I conducted the analysis of each individual in the study three times independently; each analysis was separated by at least 2 weeks. I also conducted the same analyses on a sample of patients using different sampling periods to consolidate the findings. We found that this second analysis was confirmatory of the initial one (less than 10% variation). This result is also a confirmation of the HRV parameters’ stability in one individual, over a short period of time. Consistency of HRV parameters’ assessment has been shown in other studies (Freed et al., 1994; Guijt et al., 2007; Massimo Piepoli et al., 1996).    HRV assessment as an indicator of ANS activity is well established. Several books now summarize the findings, showing that HRV is a valid assessment of ANS activity, especially HF for the assessment of PNS activity (Buijs & Swaab, 2013; Iwase, Hayano, & Orimo, 2017; Robertson, 2004). ANS is an important regulatory system for stress response, as explained in Chapters 2, 3, and 4. We thus consider that HRV values and HRV changes may represent the level of stress. We suggest further studies to consolidate our findings. 193  ❖ Study scope The main objective of our study is to establish proofs of the possible value of HRV parameters in ischemic events prediction after TIA or minor stroke. We selected several optimal HRV parameters and constructed some simple and efficient predictive models that can be used under different conditions. At an exploratory level, we expected to find that adding HRV and psychological stress could improve the predictive value of ABCD2 score. Therefore, some traditional risk factors were not included in the model, such as presence of dyslipidemia, history of cardiovascular and cerebrovascular diseases, obstructive sleep apnea, smoking and drinking status, sedentary life, nutrition, and recent infections. Based on the positive results of this study, larger confirmatory studies may be conducted. Having larger sample size would give more power to study other outcomes.   8.8 Future Directions Our study was the first to identify useful HRV predictors and construct HRV-based stress models with significantly improving the predictive power on ischemic events after TIA and minor stroke, compared to ABCD2 score. It opens an alternative path of research in this area. However, we remain cautious with regard to the use of these results in practice. We rather recommend that other studies verify all of the associations and models’ predictive values that have been identified in our study. Therefore, the first future direction will be the validation of our results in a larger 194  scale study in order to better understand the relationship between stress/HRV and ischemic stroke development. Several other directions are worth considering: • To enlarge the sample size to assess more potential predictors/stressors, such as dyslipidemia, history of cardiovascular and cerebrovascular diseases, obstructive sleep apnea, smoking and drinking status, physical inactivity, nutrition, recent infections and stress events. This can improve the predictive model by adding new variables. • To study on patients with “hyper-acute” TIA or stroke events, i.e. patients developing TIA or minor stroke in the past 12 or 24 hours. Capturing data from very early stage of the disease can help to understand how patients’ stress profile/health condition progresses after the occurrence of initial events.  • To have a long-term or repeated measurement of HRV, for instance 7 days. A long-term HRV recording can provide more information of how ANS function changes during the most high risk time period after the initial events. Also, given the rapid occurrence of ischemic events after the initial event, the follow-up may be reduced to one week or two weeks instead of 3 months, but with more careful assessment of HRV over time, as described above.  • To assess the predictive values of ANS activity and HRV indicators on long-term prognosis, such as 1 year after TIA or minor stroke patients. Repeated assessment of HRV 195  values in every month may help to understand the dynamic of stress, stress management and the occurrence of severe ischemic events. • To examine the association between HRV and different etiological subtypes (LAA, cardioembolism, lacunar, other causes and undetermined cause) of ischemic stroke.  • To test whether a specific intervention targeted on ANS to restore its balance can reduce the risk or delay the occurrence of outcome events. • For patients with possible TIA, to examine the values of HRV to distinguish TIA mimics from TIAs.  • For ischemic stroke patients, to assess the association between HRV parameters and post-stroke rehabilitation, including clinical function, quality of life, and emotional status. • For general or senior population, to assess the predictive values of ANS/HRV and stress profiles in the development of cerebrovascular ischemic accidents. 196  Bibliography Adams, H. ., Adams, H. ., Bendixen, B. ., Bendixen, B. ., Kappelle, L. ., Kappelle, L. ., … Marsh, E. . (1993). Classification of Subtype of Acute Ischemic Stroke: Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke, 23(1), 35–41. Adams, H. P., del Zoppo, G., Alberts, M. J., Bhatt, D. L., Brass, L., Furlan, A., … Wijdicks, E. F. M. (2007). Guidelines for the Early Management of Adults With Ischemic Stroke: A Guideline From the American Heart Association/American Stroke Association Stroke Council, Clinical Cardiology Council, Cardiovascular Radiology and Intervention Council, and the Atheros. Circulation, 115(20), e478–e534. AHA/ASA. (2017). Heart Disease and Stroke Statistics 2017 At-a-Glance. Akselrod, S., Gordon, D., Ubel, F. A., Shannon, D. C., Barger, A. C., & Cohen, R. J. (1981). Power Spectrum Analysis of Heart Rate Fluctuation: A Quantitative Probe of Beat-To-Beat Cardiovascular Control. Science, 213(4504), 220–222. Al-Qudah, Z. A., Yacoub, H. A., & Souayah, N. (2015). Disorders of the Autonomic Nervous System after Hemispheric Cerebrovascular Disorders: An Update. Journal of Vascular and Interventional Neurology, 8(4), 43–52. Albers, G., Caplan, L., & Easton, J. (2002). Transient ischemic attack- proposal for a new definition. New England Journal of Medicine, 347(21), 1713–1716. 197  Ali, I. (1997). Beta-blocker effects on postoperative atrial fibrillation. European Journal of Cardio-Thoracic Surgery, 11(6), 1154–1157. Amar, D., Zhang, H., Miodownik, S., & Kadish, A. H. (2003). Competing Autonomic Mechanisms Precede the Onset of Postoperative Atrial Fibrillation. Journal of the American College of Cardiology, 42(7), 1262–1268. Amarenco, P., Albers, G. W., Denison, H., Easton, J. D., Evans, S. R., Held, P., … Johnston, S. C. (2017). Efficacy and safety of ticagrelor versus aspirin in acute stroke or transient ischaemic attack of atherosclerotic origin: a subgroup analysis of SOCRATES, a randomised, double-blind, controlled trial. Lancet Neurology, 16(4), 301–310. Amarenco, P., Bogousslavsky, J., Caplan, L. R., Donnan, G. a., & Hennerici, M. G. (2009). New approach to stroke subtyping: The A-S-C-O (phenotypic) classification of stroke. Cerebrovascular Diseases, 27(5), 502–508. Amarenco, P., Labreuche, J., & Lavallée, P. C. (2012). Patients with transient ischemic attack with ABCD2 <4 can have similar 90-day stroke risk as patients with transient ischemic attack with ABCD2 ≥4. Stroke, 43(3), 863–865. Amort, M., Fluri, F., Schäfer, J., Weisskopf, F., Katan, M., Burow, A., … Engelter, S. T. (2011). Transient ischemic attack versus transient ischemic attack mimics: Frequency, clinical characteristics and outcome. Cerebrovascular Diseases, 32(1), 57–64. Angelis, C., Perelli, P., Trezza, R., Casagrande, M., Biselli, R., Pannitteri, G., … Farrace, S. 198  (2001). Modified autonomic balance in offsprings of diabetics detected by spectral analysis of heart rate variability. Metabolism, 50(11), 1270–1274. Anglemyer, A., Horvath, H. T., & Bero, L. (2014). Healthcare outcomes assessed with observational study designs compared with those assessed in randomized trials. The Cochrane Database of Systematic Reviews, 4(4). Annane, D., Trabold, F., Sharshar, T., Jarrin, I., Blanc, A. S., Raphael, J. C., & Gajdos, P. (1999). Inappropriate Sympathetic Activation at Onset of Septic Shock. American Journal of Respiratory and Critical Care Medicine, 160, 458–465. Antelmi, I., De Paula, R. ., Shinzato, A. ., Peres, C. ., Mansur, A. ., & Grupi, C. . (2004). Influence of age, gender, body mass index, and functional capacity on heart rate variability in a cohort of subjects without heart disease. The American Journal of Cardiology, 93(March), 381–385. Ay, H., Arsava, E. M., Johnston, S. C., Vangel, M., Schwamm, L. H., Furie, K. L., … Sorensen, A. G. (2009). Clinical- and imaging-based prediction of stroke risk after transient ischemic attack: the CIP model. Stroke, 40(1), 181–6. Badimon, L., & Vilahur, G. (2014). Thrombosis formation on atherosclerotic lesions and plaque rupture. Journal of Internal Medicine, 276(6), 618–632. Balami, J. S., Chen, R. L., Grunwald, I. Q., & Buchan, A. M. (2011). Neurological complications of acute ischaemic stroke. The Lancet Neurology, 10(4), 357–371. 199  Bamford, J., Sandercock, P., Dennis, M., Burn, J., & Warlow, C. (1990). A prospective study of acute cerebrovascular disease in the community: the Oxfordshire Community Stroke Project, 1981-86. Journal of Neurology, Neurosurgery & Psychiatry, 53(June), 16–22. Barnaby, D., Ferrick, K., Kaplan, D. T., Shah, S., Bijur, P., & Gallagher, E. J. (2002). Heart rate variability in emergency department patients with sepsis. Academic Emergency Medicine, 9(7), 661–670. Barrett, K. M., Levine, J. M., & Johnston, K. C. (2008). Diagnosis of stroke and stroke mimics in the emergency setting. Continuum Lifelong Learning in Neurology, 14(6), 13–27. Barutcu, I., Esen, A. M., Kaya, D., Turkmen, M., Karakaya, O., Melek, M., … Basaran, Y. (2005). Cigarette Smoking and Heart Rate Variability: Dynamic Influence of Parasympathetic and Sympathetic Maneuvers. Annals of Noninvasive Electrocardiology, 10(3), 324–9. Bassi, A., Colivicchi, F., Santini, M., & Caltagirone, C. (2007). Cardiac autonomic dysfunction and functional outcome after ischaemic stroke. European Journal of Neurology, 14(8), 917–922. Beauchaine, T. (2001). Vagal tone, development, and Gray’s motivational theory: Toward an integrated model of autonomic nervous system functioning in psychopathology. Development and Psychopathology, 13(2), 183–214. Benjamin, E. J., Blaha, M. J., Chiuve, S. E., Cushman, M., Das, S. R., Deo, R., … Muntner, P. 200  (2017). Heart Disease and Stroke Statistics—2017 Update: A Report From the American Heart Association. Circulation, 135. Berntson, G. G., Thomas Bigger, J., Eckberg, D., Grossman, P., Kaufmann, P., & Malik, M. (1997). Heart rate variablity: origins, methods, and interpretive caveats. Psychophysiology, 34, 623–648. Bettoni, M., & Zimmermann, M. (2002). Autonomic Tone Variations Before the Onset of Paroxysmal Atrial Fibrillation. Circulation, 105, 2753–9. Bhardwaj, S. S., Camacho, F., Derrow, A., Fleischer, A. B., & Feldman, S. R. (2004). Statistical Significance and Clinical Relevance. Archives of Dermatology, 140, 1520–1523. Bigger, J. T., Fleiss, J. L., Steinman, R. C., Rolnitzky, L. M., Kleiger, R. E., & Rottman, J. N. (1992). Frequency domain measures of heart period variability and mortality after myocardial infarction. Circulation, 85(1), 164–171. Black, P. H. (2003). The inflammatory response is an integral part of the stress response: Implications for atherosclerosis, insulin resistance, type II diabetes and metabolic syndrome X. Brain, Behavior, and Immunity, 17(5), 350–364. Booth, J., Connelly, L., Lawrence, M., Chalmers, C., Joice, S., Becker, C., & Dougall, N. (2015). Evidence of perceived psychosocial stress as a risk factor for stroke in adults: a meta-analysis. BMC Neurology, 15, 233. Bots, M. L., Elwood, P. C., Nikitin, Y., Salonen, J. T., Freire de Concalves, A., Inzitari, D., … 201  Grobbee, D. E. (2002). Total and HDL cholesterol and risk of stroke. EUROSTROKE: A collaborative study among research centres in Europe. Journal of Epidemiology and Community Health, 56(SUPPL. 1). Boutcher, S. H., & Stein, P. (1995). Association between heart rate variability and training response in sedentary middle-aged men. European Journal of Applied Physiology and Occupational Physiology, 70(1), 75–80. BPLT Trialists’ Collaboration. (2003). Blood Pressure Lowering Treatment Trialists’ Collaboration. Effects of different blood- pressure- lowering regimens on major cardiovascular events: results of prospectively-designed overviews of randomised trials. Lancet, 362(9395), 1527–35. Brateanu, A. (2015). Heart rate variability after myocardial infarction: what we know and what we still need to find out. Current Medical Research and Opinion, 31(10), 1855–1860. Brott, T., Jr, H. P. A., Olinger, C. P., Marler, J. R., Barsan, W. G., Biller, J., … Walker, M. (1989). Measurements of Acute Cerebral Infarction: a Clinical Examination Scale, 20, 864–871. Brown, R. D., Whisnant, J. P., Sicks, J. D., O’Fallon, W. M., & Wiebers, D. O. (1996). Stroke incidence, prevalence, and survival: secular trends in Rochester, Minnesota, through 1989. Stroke, 27(3), 373—380. Buchman, T. G., Stein, P. K., & Goldstein, B. (2002). Heart rate variability in critical illness and 202  critical care. Current Opinion in Critical Care, 8(4), 311–315. Buijs, R. M., & Swaab, D. F. (Eds.). (2013). Autonomic Nervous System. Elsevier. Burger, A. J., Charlamb, M., & Sherman, H. B. (1999). Circadian patterns of heart rate variability in normals, chronic stable angina and diabetes mellitus. International Journal of Cardiology, 71(1), 41–48. Callahan, A., Amarenco, P., Goldstein, L. B., Sillesen, H., Messig, M., Samsa, G. P., … Welch, M. (2011). Risk of Stroke and Cardiovascular Events After Ischemic Stroke or Transient Ischemic Attack in Patients With Type 2 Diabetes or Metabolic Syndrome. Archives of Neurology, 68(10), 1245. Calvet, D., Touzé, E., Oppenheim, C., Turc, G., Meder, J.-F., & Mas, J.-L. (2009). DWI lesions and TIA etiology improve the prediction of stroke after TIA. Stroke, 40(1), 187–92. Campbell, J., & Ehlert, U. (2012). Acute psychosocial stress: does the emotional stress response correspond with physiological responses? Psychoneuroendocrinology, 37(8), 1111–34. Cancelli, I., Janes, F., Gigli, G. L., Perelli, A., Zanchettin, B., Canal, G., … Valente, M. (2011). Incidence of transient ischemic attack and early stroke risk: Validation of the ABCD2 score in an Italian population-based study. Stroke, 42(10), 2751–2757. Caplan, L. R. (2009). Caplan’s Stroke: A Clinical Approach (4th ed.). Elsevier Canada. Carandang, R., Beiser, A., Kelly-hayes, M., Kase, C. S., Kannel, W. B., & Wolf, P. a. (2006). Trends in Incidence, Lifetime Risk, Severity, and 30-Day Mortality of Stroke Over the Past 203  50 Years. The Journal of American Medical Association, 296(24), 2939–2946. Carnethon, M. R., Golden, S. H., Folsom, A. R., Haskell, W., & Liao, D. (2003). Prospective investigation of autonomic nervous system function and the development of type 2 diabetes: the Atherosclerosis Risk In Communities study, 1987-1998. Circulation, 107(17), 2190–5. Carter, J., Banister, E., & Blaber, A. (2003). Effect of endurance exercise on autonomic control of heart rate. Sports Medicine, 33, 33–46. Carthy, E. R. (2014). Autonomic dysfunction in essential hypertension: A systematic review. Annals of Medicine and Surgery, 3(1), 2–7. Castle, J., Mlynash, M., Lee, K., Caulfield, A. F., Wolford, C., Kemp, S., … Olivot, J. M. (2010). Agreement regarding diagnosis of transient ischemic attack fairly low among stroke-trained neurologists. Stroke, 41(7), 1367–1370. Chan, J. M., Rimm, E. B., Colditz, G. A., Stampfer, M. J., & Willett, W. C. (1994). Obesity, fat distribution, and weight gain as risk factors for clinical diabetes in men. Diabetes Care, 17(9), 961–969. Chandratheva, A., Geraghty, O. C., & Rothwell, P. M. (2011). Poor performance of current prognostic scores for early risk of recurrence after minor stroke. Stroke, 42(3), 632–7. Chandratheva, A., Metha, Z., Geraghty, O. C., Marquardt, L., & Rothwell, P. M. (2009). Population-Based Study of Risk of Stroke in the First Few Hours after a TIA. Neurology, 80(4), 466. 204  Charmandari, E., Tsigos, C., & Chrousos, G. P. (2005). Neuroendocrinology of stress. Annual Review of Physiology, 67, 259–284. Cheyuo, C., Jacob, A., Wu, R., Zhou, M., Coppa, G. F., & Wang, P. (2011). The parasympathetic nervous system in the quest for stroke therapeutics. Journal of Cerebral Blood Flow and Metabolism, 31(5), 1187–95. Chow, S.-C., Shao, J., & Wang, H. (2007). Sample size calculations in clinical research (2nd ed.). Chapman and Hall/CRC. Christensen, J., Toft, E., Christensen, M., & Schmidt, E. (1999). Heart rate variability and plasma lipids in men with and without ischaemic heart disease. Atherosclerosis, 145, 181–186. Chrousos, G., & Gold, P. (1992). The Concepts of Stress and Stress System Disorders. The Journal of American Medical Association, 267, 1244–1252. Chrousos, G. P. (1998). Stressors, stress and neuroendocrine integration of the adaptive response. Annals of the New York Academy of Sciences, 851, 311–335. Chrousos, G. P. (2009). Stress and disorders of the stress system. Nature Reviews Endocrinology, 5(7), 374–81. Ciarka, A., van de Borne, P., & Pathak, A. (2008). Myocardial infarction, heart failure and sympathetic nervous system activity: new pharmacological approaches that affect neurohumoral activation. Expert Opinion on Investigational Drugs, 17(9), 1315–30. 205  Citi, L., Brown, E. N., & Barbieri, R. (2012). A real-time automated point-process method for the detection and correction of erroneous and ectopic heartbeats. IEEE Transactions on Biomedical Engineering, 59(10), 2828–2837. Cohen, S. (1988). Perceived stress in a probability sample of the United States. In S. Spacapan & S. Oskamp (Eds.), The Social Psychology of Health (pp. 31–67). Newbury Park, CA: Sage. Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A Global Measure of Perceived Stress. Journal of Health and Social Behavior, 24(4), 385–396. Collins, K., Abdel-Rahman, T., Easton, J., Sacco, P., & Ison, J. (1996). Effects of facial cooling on elderly and young subjects: interactions with breath-holding and lower body negative pressure. Clinical Science, 90, 485–92. Collins, K. J. (1997). Aging, disease and the autonomic nervous system. Reviews in Clinical Gerontology, 7(2), 119–126. Coull,  a J., Lovett, J. K., & Rothwell, P. M. (2004). Population based study of early risk of stroke after transient ischaemic attack or minor stroke: implications for public education and organisation of services. BMJ, 328(7435), 326. Coumel, P., Attuel, P., Lavallée, J., Flammang, D., Leclercq, J. F., & Slama, R. (1978). The atrial arrhythmia syndrome of vagal origin. Arch Mal Coeur Vaiss, 6(645–49). Coutts, S. B., Eliasziw, M., Hill, M. D., Scott, J. N., Buchan, A. M., & Demchuk, A. M. (2008). An improved scoring system for identifying patients at high early risk of stroke and 206  functional impairment after an acute transient ischemic attack or minor stroke. International Journal of Stroke, 3(February), 3–10. Cryer, P. E. (2013). Mechanisms of Hypoglycemia-Associated Autonomic Failure in Diabetes. New England Journal of Medicine, 369, 362–72. Curtis, B. M., & O’Keefe, J. H. (2002). Autonomic tone as a cardiovascular risk factor: the dangers of chronic fight or flight. Mayo Clinic Proceedings, 77(1), 45–54. De Marchis, G. M., Weck, A., Audebert, H., Benik, S., Foerch, C., Buhl, D., … Katan, M. (2014). Copeptin for the prediction of recurrent cerebrovascular events after transient ischemic attack: Results from the CoRisk study. Stroke, 45(10), 2918–2923. Delaney, J. P. A., & Brodie, D. A. (2000). Effects of Short-Term Psychological Stress on the Time and Frequency Domains of Heart-Rate Variability. Perceptual and Motor Skills, 91(2), 515–524. Dennis, M. S., Bamford, J. M., Sandercock, D., & Warlow, C. P. (1989). A Comparison of Risk Factors and Prognosis for Transient Ischemic Attacks and Minor Ischemic Strokes The Oxfordshire Community Stroke Project. Stroke, 20, 1494–1499. Diener, H., Bogousslavsky, J., & Brass, L. (2004). Aspirin and clopidogrel compared with clopidogrel alone after recent ischaemic stroke or transient ischaemic attack in high-risk patients (MATCH): randomised, double-blind, placebo-controlled trial. Lancet, 364(9431), 331–337. 207  Dishman, R. K., Nakamura, Y., Garcia, M. E., Thompson, R. W., Dunn, A. L., & Blair, S. N. (2000). Heart rate variability, trait anxiety, and percieved stress among physically fit men and women. International Journal of Psychophysiology, 37, 121–133. Dütsch, M., Burger, M., Dörfler, C., Schwab, S., & Hilz, M. J. (2007). Cardiovascular autonomic function in poststroke patients. Neurology, 69(24), 2249–55. Easton, J. D., Saver, J. L., Albers, G. W., Alberts, M. J., Chaturvedi, S., Feldmann, E., … Sacco, R. L. (2009). Definition and evaluation of transient ischemic attack:A Scientific Statement for Healthcare Professionals From the American Heart Association/American Stroke Association Stroke Council; Council on Cardiovascular Surgery and Anesthesia; Council on Cardiov. Stroke, 40(6), 2276–2293. Ebrahim, S., Sung, J., Song, Y., Ferrer, R., Lawlor, D. A., Ebrahim, S., … Smit, G. D. (2006). Serum cholesterol , haemorrhagic stroke , ischaemic stroke and myocardial infarction : Korean national health system prospective cohort study. BMJ, 333(7557), 22–25. Eckberg, D., & Sleight, P. (1992). Human baroreflexes in health and disease. Oxford: Oxford University Press. Elkind, M. (2006). Inflammation, Atherosclerosis, and Stroke. The Neurologist, 12, 140–148. Elkind, M. (2007). Why now? Moving from stroke risk factors to stroke triggers. Current Opinion in Neurology, 20, 51–57. Ellenby, M. S., McNames, J., Lai, S., & McDonald, B. A. (2001). Uncoupling and recoupling of 208  autonomic regulation of the heart beat in pediatric septic shock. Shock, 16(4), 274–277. Eller, N. H. (2007). Total power and high frequency components of heart rate variability and risk factors for atherosclerosis. Autonomic Neuroscience: Basic & Clinical, 131(1–2), 123–30. Emsley, H., & Tyrrell, P. (2002). Inflammation and infection in clinical stroke. Journal of Cerebral Blood Flow & Metabolism, 22, 1399 –1419. Endres, M., Gertz, K., Lindauer, U., Katchanov, J., Schultze, J., Schrok, H., … Laufs, U. (2003). Mechanisms of stroke prevention by physical activity. Annals of Neurology, 54(5), 582–590. Esler, M. (2014). Sympathetic nervous system moves toward center stage in cardiovascular medicine: from Thomas Willis to resistant hypertension. Hypertension, 63, e25–e32. Esler, M., Lambert, E., & Schlaich, M. (2010). Point: Chronic Activation of the Sympathetic Nervous System is the Dominant Contributor to Systemic Hypertension. Journal of Applied Physiology, 109(6), 1996–1998. Ewing, D., Martyn, C., Young, R., & Clarke, B. (1985). The Value of Cardiovascular Autonomic Function Tests: 10 years Experience in Diabetes. Diabetes Care, 8(5), 491–498. Fazeli, M. S., Pourrahmat, M.-M., Liu, M., Guan, L., & Collet, J.-P. (2016). The Effect of Head Massage on the Regulation of the Cardiac Autonomic Nervous System: A Pilot Randomized Crossover Trial. Journal of Aternative and Complementary Medicine, 22(1), 75–80. 209  Fernández-Solà, J. (2015). Cardiovascular risks and benefits of moderate and heavy alcohol consumption. Nature Reviews Cardiology, 12, 576–587. Ferrara, L., Moscato, T., Pisanti, N., Marotta, T., Krogh, V., Capone, D., & Mancini, M. (1988). Is the sympathetic nervous system altered in children with familial history of arterial hypertension? Cardiology, 75, 200–205. Floßmann, E., Schulz, U. G. R., & Rothwell, P. M. (2004). Systematic Review of Methods and Results of Studies of the Genetic Epidemiology of Ischemic Stroke. Stroke, 35(1), 212–227. Folson, A., Rasmussen, M., Chambless, L., Howard, G., Cooper, L., Schmidt, M., & Heiss, G. (1999). Prospective Associations of Fasting Insulin, Body Fat Distribution, and Diabetes With Risk of Ischemic Stroke. Diabetes Care, 22(7), 1077–1083. Ford, E. S., & Cooper, R. S. (1991). Risk factors for hypertension in a national cohort study. Hypertension, 18(5), 598–606. Franklin, R. M., Baynard, T., Weinstock, R. S., Goulopoulou, S., Carhart, R., Ploutz-Snyder, R., … Kanaley, J. a. (2008). Autonomic responses to physiological stressors in women with type 2 diabetes. Clinical Autonomic Research, 18(2), 66–73. Freed, L. A., Stein, K. M., Gordon, M., Urban, M., & Kligfield, P. (1994). Reproducibilfty of power spectral measures of heart rate variability obtained from short-term sampling periods. The American Journal of Cardiology, 74(9), 972–973. Freeman, R. (2006). Assessment of cardiovascular autonomic function. Clinical 210  Neurophysiology, 117(4), 716–730. Freeman, R., & Chapleau, M. W. (2013). Testing the autonomic nervous system. In Handbook of Clinical Neurology (Vol. 115, pp. 115–136). Friedman, L. M., Furberg, C. D., & DeMets, D. (2010). Fundamentals of Clincial Trials. Springer. Fruchart, J.-C. (2004). New Risk Factors for Atherosclerosis and Patient Risk Assessment. Circulation, 109(23_suppl_1), III-15-III-19. Fukuda, K., Kanazawa, H., Aizawa, Y., Ardell, J., & Shivkumar, K. (2015). Cardiac Innervation and Sudden Cardiac Death. Circulation Research, 116(12), 2005–2019. Furie, K. L., Kasner, S. E., Adams, R. J., Albers, G. W., Bush, R. L., Susan, C., … Wentworth, D. (2011). Guidelines for the Prevention of Stroke in Patients With Stroke or Transient Ischemic Attack: A Guideline for Healthcare Professionals From the American Heart Association/American Stroke Association. Stroke, 42, 227–76. Fuster, V., Rydén, L. E., Cannom, D. S., Crijns, H. J., Curtis, A. B., Ellenbogen, K. A., … Wann, L. S. (2011). 2011 ACCF/AHA/HRS Focused Updates Incorporated Into the ACC/AHA/ESC 2006 Guidelines for the Management of Patients With Atrial Fibrillation. Journal of the American College of Cardiology, 57(11), e101–e198. Fuster, V., Rydén, L. E., Cannom, D. S., Crijns, H. J., Curtis, A. B., Ellenbogen, K. a., … Wann, S. (2006). ACC/AHA/ESC 2006 Guidelines for the Management of Patients With Atrial 211  Fibrillation. Journal of the American College of Cardiology, 48(4), e149–e246. Fyfe-Johnson, A. L., Muller, C. J., Alonso, A., Folsom, A. R., Gottesman, R. F., Rosamond, W. D., … MacLehose, R. F. (2016). Heart Rate Variability and Incident Stroke: The Atherosclerosis Risk in Communities Study. Stroke, 47(6), 1452–1458. Garrard, C. S., Kontoyannis, D. a, & Piepoli, M. (1993). Spectral analysis of heart rate variability in the sepsis syndrome. Clinical Autonomic Research, 3(1), 5–13. Geeganage, C. M., Diener, H.-C., Algra, A., Chen, C., Topol, E. J., Dengler, R., … Bath, P. M. W. (2012). Dual or Mono Antiplatelet Therapy for Patients With Acute Ischemic Stroke or Transient Ischemic Attack. Stroke, 43(4), 1058–1066. Genest, J., McPherson, R., Frohlich, J., Anderson, T., Campbell, N., Carpentier, A., … Ur, E. (2009). 2009 Canadian Cardiovascular Society/Canadian guidelines for the diagnosis and treatment of dyslipidemia and prevention of cardiovascular disease in the adult – 2009 recommendations. Canadian Journal of Cardiology, 25(10), 567–579. Gidron, Y., Kupper, N., Kwaijtaal, M., Winter, J., & Denollet, J. (2007). Vagus-brain communication in atherosclerosis-related inflammation: A neuroimmunomodulation perspective of CAD. Atherosclerosis, 195(2), 1–9. Giles, M. F., Albers, G. W., Amarenco, P., Arsava, E. M., Asimos, A. W., Ay, H., … Rothwell, P. M. (2011). Early stroke risk and ABCD2 score performance in tissue- Vs time-defined TIA: A multicenter study. Neurology, 77(13), 1222–1228. 212  Giles, M. F., & Rothwell, P. M. (2007). Risk of stroke early after transient ischaemic attack: a systematic review and meta-analysis. Lancet Neurology, 6(12), 1063–72. Go, A. S., Mozaffarian, D., Roger, V. L., Benjamin, E. J., Berry, J. D., Blaha, M. J., … Turner, M. B. (2014). Heart Disease and Stroke Statistics - 2014 Update: A report from the American Heart Association. Circulation, 129(3), 1–268. Godin, P. J., & Buchman, T. G. (1996). Uncoupling of biological oscillators: a complementary hypothesis concerning the pathogenesis of multiple organ dysfunction syndrome. Critical Care Medicine, 24(7), 1107–16. Goff, D. C., Lloyd-Jones, D. M., Bennett, G., Coady, S., D’Agostino, R. B., Gibbons, R., … Wilson, P. W. F. (2014). 2013 ACC/AHA guideline on the assessment of cardiovascular risk: A report of the American college of cardiology/American heart association task force on practice guidelines. Circulation, 129(25 SUPPL. 1). Goldstein, B., Fiser, D. H., Kelly, M. M., Mickelsen, D., Ruttimann, U., & Pollack, M. M. (1998). Decomplexification in critical illness and injury: Relationship between heart rate variability, severity of illness, and outcome. Critical Care Medicine, 26(2), 1–11. Goldstein, L. B., Adams, R., Alberts, M. J., Appel, L. J., Brass, L. M., Bushnell, C. D., … Sacco, R. L. (2006). Primary prevention of ischemic stroke: a guideline from the American Heart Association/American Stroke Association Stroke Council. Circulation, 113(24), e873-923. Goldstein, L. B., Bushnell, C. D., Adams, R. J., Appel, L. J., Braun, L. T., Chaturvedi, S., … 213  Pearson, T. A. (2011). Guidelines for the primary prevention of stroke: A Guideline for Healthcare Professionals from the American Heart Association/American Stroke Association. Stroke, 42(2), 517–584. Goodpaster, B. H., Delany, J. P., Otto, A. D., South-paul, J. E., Thomas, S. B., Brown, J., & Lang, W. (2010). Effects of Diet and Physical Activity Interventions on Weight Loss and Cardiometabolic Risk Factors in Severely Obese Adults, 304(16), 1795–1802. Gordis, L. (2008). Epidemiology (4th ed.). Elsevier Canada. Gordon, N. F., Gulanick, M., Costa, F., Fletcher, G., Franklin, B. A., Roth, E. J., & Shephard, T. (2004). Physical Activity and Exercise Recommendations for Stroke Survivors: An American Heart Association Scientific Statement from the Council on Clinical Cardiology, Subcommittee on Exercise, Cardiac Rehabilitation, and Prevention; the Council on Cardiovascula. Circulation, 109(16), 2031–2041. Gouin, J. P., Wenzel, K., Boucetta, S., O’Byrne, J., Salimi, A., & Dang-Vu, T. T. (2015). High-frequency heart rate variability during worry predicts stress-related increases in sleep disturbances. Sleep Medicine, 16(5), 659–664. Goyal, J., Nuhn, P., Huang, P., Tyagi, P., Oh, D., Carducci, M. A., … Antonarakis, E. S. (2012). The effect of clinical trial participation versus non-participation on overall survival in men receiving first-line docetaxel- containing chemotherapy for metastatic castration-resistant prostate cancer. BJU Int., 110, E575-82. 214  Grassi, G. (2009). Assessment of sympathetic cardiovascular drive in human hypertension: achievements and perspectives. Hypertension., 54, 690–697. Grassi, G., Cattaneo, B. M., Seravalle, G., Lanfranchi, A., & Mancia, G. (1998). Baroreflex control of sympathetic nerve activity in essential and secondary hypertension. Hypertension, 31(1), 68–72. Grassi, G., Seravalle, G., Bertinieri, G., Turri, C., Dell’Oro, R., Stella, M., & Mancia, G. (2000). Sympathetic and reflex alterations in systo-diastolic and systolic hypertension of the elderly. Journal of Hypertension, 18, 587–593. Grau, A. J., Weimar, C., Buggle, F., Heinrich, A., Goertler, M., Neumaier, S., … Brandt, T. (2001). Risk Factors, Outcome, and Treatment in Subtypes of Ischemic Stroke. Stroke, 32, 2559–2566. Guan, L., Collet, J.-P., Ruitenbeek, A., Eibensteiner, S., & Kissoon, N. (2014). The Effect of Massage Intervention on Autonomic Activity in Critical Ill Children with Sepsis. In IN-CAM Symposium. Calgary, Canada. Guan, L., Collet, J. P., Yuskiv, N., Skippen, P., Brant, R., & Kissoon, N. (2014). The effect of massage therapy on autonomic activity in critically ill children. Evid Based Complement Alternat Med, 2014(656750). Guijt, A. M., Sluiter, J. K., & Frings-Dresen, M. H. W. (2007). Test-retest reliability of heart rate variability and respiration rate at rest and during light physical activity in normal subjects. 215  Archives of Medical Research, 38(1), 113–120. Guiraud, V., Amor, M. Ben, Mas, J.-L., & Touze, E. (2010). Triggers of Ischemic Stroke A Systematic Review. Stroke, 41, 2669–77. Hand, P. J., Kwan, J., Lindley, R. I., Dennis, M. S., & Wardlaw, J. M. (2006). Distinguishing Between Stroke and Mimic at the Bedside: The Brain Attack Study. Stroke, 37(3), 769–775. Hannan, E. L. (2008). Randomized Clinical Trials and Observational Studies. Guidelines for Assessing Respective Strengths and Limitations. Journal of the American College of Cardiology, 1(3), 211–217. Hanss, R., Renner, J., Ilies, C., Moikow, L., Buell, O., Steinfath, M., … Bein, B. (2008). Does heart rate variability predict hypotension and bradycardia after induction of general anaesthesia in high risk cardiovascular patients? Anaesthesia, 63(2), 129–135. Hansson, G. K. (2005). Inflammation and coronary artery disease. New England Journal of Medicine, 352, 1685–95. Hansson, G. K., Robertson, A.-K. L., & Söderberg-Nauclér, C. (2006). Inflammation and atherosclerosis. Annual Review of Pathology, 1(1), 297–329. Harris, K. F. (2004). Interactions Between Autonomic Nervous System Activity and Endothelial Function: A Model for the Development of Cardiovascular Disease. Psychosomatic Medicine, 66(2), 153–164. Haskell, W. L., Lee, I. M., Pate, R. R., Powell, K. E., Blair, S. N., Franklin, B. A., … Bauman, 216  A. (2007). Physical activity and public health: Updated recommendation for adults from the American College of Sports Medicine and the American Heart Association. Circulation, 116(9), 1081–1093. Hayano, J., Yamada, M., Sakakibara, Y., & Al., E. (1990). Short- and long-term effects of cigarette smoking on heart rate variability. American Journal of Cardiology, 65, 84–88. Healey, J. S., Baranchuk, A., Crystal, E., Morillo, C. A., Garfinkle, M., Yusuf, S., & Connolly, S. J. (2005). Prevention of atrial fibrillation with angiotensin-converting enzyme inhibitors and angiotensin receptor blockers: A meta-analysis. Journal of the American College of Cardiology, 45(11), 1832–1839. Heidland, U. E., & Strauer, B. E. (2001). Left Ventricular Muscle Mass and Elevated Heart Rate are Associated With Coronary Plaque Disruption. Circulation, 104, 1477–1482. Henderson, K. M., Clark, C. J., Lewis, T. T., Aggarwal, N. T., Beck, T., Guo, H., … Everson-Rose, S. A. (2013). Psychosocial distress and stroke risk in older adults. Stroke, 44(2), 367–372. Herweg, B., Dalal, P., Nagy, B., & Schweitzer, P. (1998). Power Spectral Analysis of Heart Period Variability of Preceding Sinus Rhythm Before Initiation of Paroxysmal Atrial Fibrillation. American Journal of Cardiology, 82, 869–874. Hicks, K. A., Hung, H. M. J., Mahaffey, K. W., Mehran, R., Nissen, S. E., Stockbridge, N. L., … Rosenfield, K. (2014). Standardized Definitions for Cardiovascular and Stroke End Point 217  Events in Clinical Trials. Draft Definitions for CDISC, August. Hicks, K. A., Hung, H. M. J., Mahaffey, K. W., Mehran, R., Nissen, S. E., Stockbridge, N. L., … Temple, R. (2010). Standardized definitions for end point events in cardiovascular trials. Circulation, 20, 1–37. Hicks, K. A., Tcheng, J. E., Bozkurt, B., Chaitman, B. R., Cutlip, D. E., Farb, A., … Targum, S. L. (2015). 2014 ACC/AHA Key Data Elements and Definitions for Cardiovascular Endpoint Events in Clinical Trials. Journal of the American College of Cardiology, 66(4), 403–469. Hillebrand, S., Gast, K., Mutsert, R., Swenne, C., Jukema, J., Middeldorp, S., … Dekkers, O. (2013). Heart rate variability and first cardiovascular event in populations without known cardiovascular disease : Meta-analysis and dose-response meta-regression. Europace, 15, 742–9. Hilz, M. J., Moeller, S., Akhundova, A., Marthol, H., Pauli, E., De Fina, P., & Schwab, S. (2011). High NIHSS values predict impairment of cardiovascular autonomic control. Stroke, 42(6), 1528–1533. Horenstein, R. B., Smith, D. E., & Mosca, L. (2002). Cholesterol predicts stroke mortality in the women’s pooling project. Stroke, 33(7), 1863–1868. Horn, E., & Lee, S. (1965). Electronic evaluations of the fetal heart rate patterns preceding fetal death: further observation. American Journal of Obstetrics and Gynecology, 87, 814–26. 218  Hotta, H., & Uchida, S. (2010). Aging of the autonomic nervous system and possible improvements in autonomic activity using somatic afferent stimulation. Geriatrics and Gerontology International, 10(SUPPL. 1), 127–136. House, A., Dennis, M., Mogridge, L., Hawton, K., & Warlow, C. (1990). Life events and difficulties preceding stroke. Journal of Neurology, Neurosurgery, and Psychiatry, 53, 1024–8. Howard, G., Wagenknecht, L. E., Cai, J., Cooper, L., Kraut, M. A., & Toole, J. F. (1998). Cigarette smoking and other risk factors for silent cerebral infarction in the general population. Stroke, 29(5), 913–7. Hsieh, F. Y. (1989). Sample size tables for logistic regression. Statistics in Medicine, 8(7), 795–802. Huikuri, H. V., Niemela, M. J., Ojala, S., Rantala, A., Ikaheimo, M. J., & Airaksinen, K. E. (1994). Circadian rhythms of frequency domain measures of heart rate variability in healthy subjects and patients with coronary artery disease. Effects of arousal and upright posture. Circulation, 90(1), 121–126. Huikuri, H. V., & Stein, P. K. (2013). Heart rate variability in risk stratification of cardiac patients. Progress in Cardiovascular Diseases, 56(2), 153–159. Huikuri, H. V, Jokinen, V., Syva, M., Nieminen, M. S., Airaksinen, K. E. J., Koistinen, J. M., … Niemela, K. O. (1999). Heart Rate Variability and Progression of Coronary Atherosclerosis. 219  Arteriosclerosis, Thrombosis, and Vascular Biology, 1979–1985. Iellamo, F., Legramante, J. M., Massaro, M., Raimondi, G., & Galante, A. (2000). Effects of a Residential Exercise Training on Baroreflex Sensitivity and Heart Rate Variability in Patients With Coronary Artery Disease. Circulation, 102(21), 2588–2593. Ishida, K., Kasner, S. E., & Cucchiara, B. (2015). Inter-rater Reliability and Misclassification of the ABCD2 Score after Transient Ischemic Attack. Journal of Stroke and Cerebrovascular Diseases, 24(6), 1174–1178. Ishida, K., Raser-Schramm, J. M., Wilson, C. A., Kasner, S. E., Mullen, M. T., Messe, S. R., & Cucchiara, B. (2013). Convergent validity and interrater reliability of estimating the ABCD 2 score from medical records. Stroke, 44(3), 803–805. Ishida, S., Nakagawa, M., Fujino, T., Yonemochi, H., Saikawa, T., & Ito, M. (1997). Circadian variation of QT interval dispersion: Correlation with heart rate variability. Journal of Electrocardiology, 30(3), 205–210. Iso, H., Jacobs, D. J., Wentworth, D., Neaton, J., & Cohen, J. (1989). Serum cholesterol levels and six-year mortality from stroke in 350,977 men screened for the multiple risk factor intervention trial. New England Journal of Medicine, 320, 904–910. Ivey, F. M., Ryan, A. S., Hafer-Macko, C. E., Garrity, B. M., Sorkin, J. D., Goldberg, A. P., & Macko, R. F. (2006). High prevalence of abnormal glucose metabolism and poor sensitivity of fasting plasma glucose in the chronic phase of stroke. Cerebrovascular Diseases, 22(5–220  6), 368–371. Iwase, S., Hayano, J., & Orimo, S. (Eds.). (2017). Clinical Assessment of the Autonomic Nervous System. Tokyo: Springer. Jauch, E. C., Saver, J. L., Adams, H. P., Bruno, A., Connors, J. J. B., Demaerschalk, B. M., … Wintermark, M. (2013). Guidelines for the Early Management of Patients With Acute Ischemic Stroke: A Guideline for Healthcare Professionals From the American Heart Association/American Stroke Association. Stroke, 44, 870–947. Johnson, E. O., Kamilaris, Kamilaris, T. C., Chrousos, G. P., & Gold, P. W. (1992). Mechanisms of stress: a dynamic overview of hormonal and behavioral homeostasis. Neuroscience and Biobehavioral Reviews, 16(2), 115–30. Johnston, S. C. (2002). Transient Ischemic Attack. New England Journal of Medicine, 347(21), 1687–1692. Johnston, S. C., Amarenco, P., Albers, G. W., Denison, H., Easton, J. D., Evans, S. R., … Wong, K. S. L. (2016). Ticagrelor versus Aspirin in Acute Stroke or Transient Ischemic Attack. New England Journal of Medicine, 375(1), 35–43. Johnston, S. C., Gress, D. R., Browner, W. S., & Sidney, S. (2000). Short-term Prognosis after Emergency Department Diagnosis of TIA. The Journal of American Medical Association, 284(22), 2901–2906. Johnston, S. C., Nguyen-Huynh, M. N., Schwarz, M. E., Fuller, K., Williams, C. E., Josephson, 221  S. A., … Albers, G. W. (2006). National Stroke Association guidelines for the management of transient ischemic attacks. Annals of Neurology, 60(3), 301–13. Johnston, S. C., Rothwell, P. M., Nguyen-Huynh, M. N., Giles, M. F., Elkins, J. S., Bernstein, A. L., & Sidney, S. (2007a). Validation and refinement of scores to predict very early stroke risk after transient ischaemic attack. Lancet, 369(9558), 283–292. Johnston, S. C., Rothwell, P. M., Nguyen-Huynh, M. N., Giles, M. F., Elkins, J. S., Bernstein, A. L., & Sidney, S. (2007b). Validation and refinement of scores to predict very early stroke risk after transient ischaemic attack. Lancet, 369(9558), 283–92. Jokinen, V., Tapanainen, J. M., Seppänen, T., & Huikuri, H. V. (2003). Temporal changes and prognostic significance of measures of heart rate dynamics after acute myocardial infarction in the beta-blocking era. American Journal of Cardiology, 92(8), 907–912. Jons, C., Raatikainen, P., Gang, U. J., Huikuri, H. V., Joergensen, R. M., Johannesen, A., … Thomsen, P. E. B. (2010). Autonomic dysfunction and new-onset atrial fibrillation in patients with left ventricular systolic dysfunction after acute myocardial infarction: A CARISMA substudy. Journal of Cardiovascular Electrophysiology, 21(9), 983–990. Jood, K., Redfors, P., Rosengren, A., Blomstrand, C., & Jern, C. (2009). Self-perceived psychological stress and ischaemic stroke: a case–control study. BMC Medicine, 7(53). Jorna, P. G. A. M. (1992). Spectral analysis of heart rate and psychological state: A review of its validity as a workload index. Biological Psychology, 34(2–3), 237–257. 222  Julius, S., Pascual, A., & London, R. (1971). Role of parasympathetic inhibition in the hyper-kinetic type of borderline hypertension. Circulation, 44, 413–418. Kamath, M., & Fallen, E. (1993). Power spectral analysis of heart rate variability: a noninvasive signature of cardiac autonomic function. Critical Reviews in Biomedical Engineering, 21(3), 245–311]. Kamath, M., & Fallen, E. (1995). Correction of the heart rate variability signal for ectopics and missing beats. Armonk: Futura, 75–85. Kaplan, R. C., Tirschwell, D. L., Longstreth, W. T., Manolio, T. a, Heckbert, S. R., Lefkowitz, D., … Psaty, B. M. (2005). Vascular events, mortality, and preventive therapy following ischemic stroke in the elderly. Neurology, 65(6), 835–42. Katan, M., Nigro, N., Fluri, F., Schuetz, P., Morgenthaler, N. G., Jax, F., … Christ-Crain, M. (2011). Stress hormones predict cerebrovascular re-events after transient ischemic attacks. Neurology, 76(6), 563–566. Kauhanen, J., Kaplan, G. A., Goldberg, D. E., & Salonen, J. T. (1997). Beer binging and mortality: results from the Kuopio ischaemic heart disease risk factor study, a prospective population based study. Bmj, 315(7112), 846–851. Kemeny, M. E. (2003). The Psychobiology of Stress. Current Directions in Psychological Science, 12(4), 124–129. Kennedy, J., Hill, M. D., Ryckborst, K. J., Eliasziw, M., Demchuk, A. M., & Buchan, A. M. 223  (2007). Fast assessment of stroke and transient ischaemic attack to prevent early recurrence (FASTER): a randomised controlled pilot trial. Lancet Neurology, 6(11), 961–969. Kernan, W. N., Ovbiagele, B., Black, H. R., Bravata, D. M., Chimowitz, M. I., Ezekowitz, M. D., … Wilson, J. A. (2014). Guidelines for the prevention of stroke in patients with stroke and transient ischemic attack: A guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke, 45(7), 2160–2236. Kiyohara, T., Kamouchi, M., Kumai, Y., Ninomiya, T., Hata, J., Yoshimura, S., … Kitazono, T. (2014). ABCD3 and ABCD3-I scores are superior to ABCD2 score in the prediction of short- and long-term risks of stroke after transient ischemic attack. Stroke, 45(2), 418–425. Kleiger, R. E., Miller, J. P., Bigger, J. T., & Moss, A. J. (1987). Decreased heart rate variability and its association with increased mortality after acute myocardial infarction. American Journal of Cardiology, 59, 256–267. Kleiger, R. E., Stein, P. K., & Bigger, J. T. (2005). Heart rate variability: Measurement and clinical utility. Annals of Noninvasive Electrocardiology, 10(1), 88–101. Kleindorfer, D., Panagos, P., Pancioli, A., Khoury, J., Kissela, B., Woo, D., … Broderick, J. P. (2005). Incidence and short-term prognosis of transient ischemic attack in a population-based study. Stroke, 36(4), 720–3. Kloet, E. R., Joëls, M., & Holsboer, F. (2005). Stress and the brain: from adaptation to disease. Nature Reviews Neuroscience, 6(6), 463–475. 224  Korach, M., Sharshar, T., Jarrin, I., Fouillot, J. P., Raphaël, J. C., Gajdos, P., & Annane, D. (2001). Cardiac variability in critically ill adults: influence of sepsis. Critical Care Medicine, 29(7), 1380–5. Korkushko, O. V, Shatilo, V. B., Plachinda YuI, & Shatilo, T. V. (1991). Autonomic control of cardiac chronotropic function in man as a function of age: assessment by power spectral analysis of heart rate variability. Journal of the Autonomic Nervous System, 32(3), 191–8. Kornerup, H., Osler, M., Boysen, G., Barefoot, J., Schnohr, P., & Prescott, E. (2010). Major life events increase the risk of stroke but not of myocardial infarction: results from the Copenhagen City Heart Study. European Journal of Cardiovascular Prevention & Rehabilitation, 17(1), 113–118. Kornitzer, M., Dramaix, M., & De Backer, G. (1999). Epidemiology of risk factors for hypertension: implications for prevention and therapy. Drugs, 57(0012–6667), 695–712. Korpelainen, J., Sotaniemi, K., & Myllylä, V. (1999). Autonomic nervous system disorders in stroke. Clinical Autonomic Research, 9(6), 325–333. Korpelainen, J. T., Sotaniemi, K. A., Huikuri, H. V., & Lylä, V. (1997). Circadian Rhythm of Heart Rate Variability Is Reversibly Abolished in Ischemic Stroke. Stroke, 28, 2150–2154. Korpelainen, J. T., Sotaniemi, K. a, Huikuri, H. V, & Myllyä, V. V. (1996). Abnormal heart rate variability as a manifestation of autonomic dysfunction in hemispheric brain infarction. Stroke, 27(11), 2059–2063. 225  Koskinen, P., Virolainen, J., & Kupari, M. (1994). Acute alcohol intake decreases short-term heart rate variability in healthy subjects. Clinical Science, 230, 225–230. Koton, S., Tanne, D., Bornstein, N. M., & Green, M. S. (2004). Triggering risk factors for ischemic stroke: a case-crossover study. Neurology, 63(11), 2006–2010. Koudstaal, P. J., Gerritsma, J. G., & van Gijn, J. (1989). Clinical disagreement on the diagnosis of transient ischemic attack: is the patient or the doctor to blame? Stroke, 20(2), 300–301. Kruyt, N. D., Biessels, G. J., Devries, J. H., & Roos, Y. B. (2010). Hyperglycemia in acute ischemic stroke: pathophysiology and clinical management. Nature Reviews Neurology, 6(3), 145–155. Kudat, H., Akkaya, V., Sozen, A., Salman, S., Demirel, S., Ozcan, M., … Guven, O. (2006). Heart Rate Variability in Diabetes Patients. Journal of International Medical Research, 34(3), 291–296. Kullo, I. J., Gau, G. T., & Tajik, A. J. (2000). Novel Risk Factors for Atherosclerosis. Mayo Clinic Proceedings, 75(4), 369–380. Kupari, M., Virolainen, J., Koskinen, P., & Tikkanen, M. (1993). Short-term heart rate variability and factors modifying the risk of coronary artery disease in a population sample. American Journal of Cardiology, 72, 897–903. Kurth, T., Everett, B. M., Buring, J. E., Kase, C. S., Ridker, P. M., & Gaziano, J. M. (2007). Lipid levels and the risk of ischemic stroke in women. Neurology, 68(8), 556–562. 226  Lambert, E., Straznicky, N., & Sari, C. (2013). Dyslipidemia is associated with sympathetic nervous activation and impaired endothelial function in young females. American Journal of Hypertension, 26, 250–56. Lanza, G. A., Guido, V., Galeazzi, M. M., Mustilli, M., Natali, R., Ierardi, C., … Maseri, A. (1998). Prognostic role of heart rate variability in patients with a recent acute myocardial infarction. The American Journal of Cardiology, 82(11), 1323–1328. Lavallée, P. C., Meseguer, E., Abboud, H., Cabrejo, L., Olivot, J.-M., Simon, O., … Amarenco, P. (2007). A transient ischaemic attack clinic with round-the-clock access (SOS-TIA): feasibility and effects. Lancet Neurology, 6(11), 953–60. Lawes, C. M. M., Bennett, D. A., Feigin, V. L., & Rodgers, A. (2004). Blood Pressure and Stroke: An Overview of Published Reviews. Stroke, 35(3), 776–785. Lee, E. H. (2013). Review of the Psychometric Evidence of the Perceived Stress Scale. Asian Nursing Research, 7, 160. Lee, M., Saver, J. L., Hong, K.-S., Song, S., Chang, K.-H., & Ovbiagele, B. (2012). Effect of pre-diabetes on future risk of stroke: meta-analysis. Bmj, 344(jun07 3), e3564–e3564. Lee, C. Do, Folsom, A. R., & Blair, S. N. (2003). Physical activity and stroke risk: A meta-analysis. Stroke, 34(10), 2475–2481. Leppälä, J. M., Virtamo, J., Fogelholm, R., Albanes, D., Olli, P., & Pressure, B. (1999). Different Risk Factors for Different Stroke Subtypes. Stroke, 30, 2535–2540. 227  Liao, D., Cai, J., Barnes, R., & Al., E. (1996). Association of cardiac autonomic function and the development of hypertension: the ARIC study. American Journal of Hypertension, 9, 1147–56. Liao, D., Cai, J., Brancati, F., & Al., E. (1995). Association of vagal tone with serum insulin, glucose, and diabetes mellitus—the ARIC Study. Diabetes Res Clin Pract, 30, 211–21. Liao, D., Sloan, R. P., Cascio, W. E., Folsom, A. R., Liese, A. D., Evans, G. W., … Sharrett, A. R. (1998). Multiple metabolic syndrome is associated with lower heart rate variability. The Atherosclerosis Risk in Communities Study. Diabetes Care, 21(12), 2116–22. Libby, P. (2002). Inflammation and Atherosclerosis. Circulation, 105(9), 1135–1143. Libby, P. (2012). Inflammation in atherosclerosis. Arteriosclerosis, Thrombosis, and Vascular Biology, 32(9), 2045–51. Licht, C. M. M., Vreeburg, S. A., Dortland, A. K. B., Giltay, E. J., Hoogendijk, W. J. G., DeRijk, R. H., … Penninx, B. W. J. H. (2010). Increased sympathetic and decreased parasympathetic activity rather than changes in hypothalamic-pituitary-adrenal axis activity is associated with metabolic abnormalities. The Journal of Clinical Endocrinology and Metabolism, 95(5), 2458–66. Lindmark, S., Wiklund, U., Bjerle, P., & Eriksson, J. W. (2003). Does the autonomic nervous system play a role in the development of insulin resistance? A study on heart rate variability in first-degree relatives of Type 2 diabetes patients and control subjects. Diabetic Medicine, 228  20(5), 399–405. Lippman, N., Stein, K. M., & Lerman, B. B. (1994). Comparison of methods for removal of ectopy in measurement of heart rate variability. American Journal of Physiology, 36, H411–H418. Lombardi, F., Colombo, A., Basilico, B., Ravaglia, R., Garbin, M., Vergani, D., … Fiorentini, C. (2001). Heart Rate Variability and Early Recurrence of Atrial Fibrillation After Electrical Cardioversion. Journal of the American College of Cardiology, 37(1). Luitse, M. J. A., Biessels, G. J., Rutten, G. E. H. M., & Kappelle, L. J. (2012). Diabetes, hyperglycaemia, and acute ischaemic stroke. Lancet Neurology, 11(3), 261–271. Lusk, B., & Lash, A. A. (2005). The Stress Response , Psychoneuroimmunology, and Stress Among ICU Patients. Dimensions of Critical Care Nursing, 24(1). Macleod, J., Davey Smith, G., Heslop, P., Metcalfe, C., Carroll, D., & Hart, C. (2002). Psychological stress and cardiovascular disease: empirical demonstration of bias in a prospective observational study of Scottish men. BMJ, 324, 1247–51. Mäki, T., Toivonen, L., Koskinen, P., Näveri, H., Härkönen, M., & Leinonen, H. (1998). Effect of ethanol drinking, hangover, and exercise on adrenergic activity and heart rate variability in patients with a history of alcohol-induced atrial fibrillation. The American Journal of Cardiology, 82(3), 317–322. Mäkikallio,  a M., Mäkikallio, T. H., Korpelainen, J. T., Sotaniemi, K. a, Huikuri, H. V, & 229  Myllylä, V. V. (2004). Heart rate dynamics predict poststroke mortality. Neurology, 62(10), 1822–6. Malfatto, G., Facchini, M., Bragato, R., Branzi, G., Sala, L., & Leonetti, G. (1996). Short and long term effects of exercise training on the tonic autonomic modulation of heart rate variability after myocardial infarction. European Heart Journal, 17(4), 532–538. Malfatto, G., Facchini, M., Sala, L., Branzi, G., Bragato, R., & Leonetti, G. (1998). Effects of cardiac rehabilitation and beta-blocker therapy on heart rate variability after first acute myocardial infarction. The American Journal of Cardiology, 81(7), 834–40. Malik, M., Farrell, T., Cripps, T., & Camm, A. (1989). Heart rate variability in relation to prognosis after myocardial infarction: selection of optimal processing techniques. European Heart Journal, 10, 1060–74. Malliani, A., Lombardi, F., & Pagani, M. (1994). Power spectrum analysis of heart rate variability: A tool to explore neural regulatory mechanisms. British Heart Journal, 71, 1–2. Malliani,  a., Pagani, M., Lombardi, F., & Cerutti, S. (1991). Cardiovascular neural regulation explored in the frequency domain. Circulation, 84(2), 482–492. Malpas, S. C., & Maling, T. J. (1990). Heart-rate variability and cardiac autonomic function in diabetes. Diabetes, 39(10), 1177–81. Mancia, G., & Grassi, G. (2014). The autonomic nervous system and hypertension. Circulation Research, 114(11), 1804–1814. 230  Manfrini, O., Pizzi, C., Viecca, M., & Bugiardini, R. (2008). Abnormalities of cardiac autonomic nervous activity correlate with expansive coronary artery remodeling. Atherosclerosis, 197(1), 183–189. Manolio, T. A., Kronmal, R. A., Burke, G. L., O’Leary, D. H., & Price, T. R. (1996). Short-term Predictors of Incident Stroke in Older Adults. Stroke, 27, 1470–1486. Marfella, R., Angelis, L. De, Nappo, F., Manzella, D., Siniscalchi, M., & Paolisso, G. (2000). Elevated plasma fatty acid concentrations prolong cardiac repolarization in healthy subjects. American Journal of Clinical Nutrition, 72, 727–30. Marigold, J. R. G., Arias, M., Vassallo, M., Allen, S. C., & Kwan, J. S. (2010). Autonomic dysfunction in older people. Reviews in Clinical Gerontology, 21(1), 28–44. Mark, A. (1996). The sympathetic nervous system in hypertension: a potential long-term regulator of arterial pressure. Journal of Hypertension, 14(suppl 5), S159–S165. Marmar, V., & Shivkumar, K. (2008). The Role of the Autonomic Nervous System in Sudden Cardiac Death. Progress Cardiovascular Disease, 50(6), 404–419. Marwah, R. S., Doux, J. D., Lee, P. Y., & Yun, A. J. (2007). Is atherosclerosis a neurogenic phenomenon? Medical Hypotheses, 69(4), 884–887. Mast, H., Thompson, J. L., Lin, I. F., Hofmeister, C., Hartmann, A., Marx, P., … Sacco, R. L. (1998). Cigarette smoking as a determinant of high-grade carotid artery stenosis in Hispanic, black, and white patients with stroke or transient ischemic attack. Stroke, 29(5), 231  908–912. Matz, K., Kerseztes, K., Tatschl, C., Nowotny, M., Dachenhausen, A., Brainin, M., & Tuomilehto, J. (2006). Disorders of Glucose Metabolism in Acute Stroke Patients. Diabetes Care, 29(4), 792–797. Maver, J., Struci, M., & Accetto, R. (2004). Autonomic nervous system in normotensive subjects with a family history of hypertension. Clinical Autonomic Research, 14, 369–375. Maver, J., Strucl, M., & Accetto, R. (2004). Autonomic nervous system activity in normotensive subjects with a family history of hypertension. Clinical Autonomic Research, 14(6), 369–75. McColl, B. W., Allan, S. M., & Rothwell, N. J. (2009). Systemic infection, inflammation and acute ischemic stroke. Neuroscience, 158(3), 1049–61. Mcewen, B. s. (2007). Physiology and neurobiology of stress and adaptation: central role of the brain. Physiological Reviews, 87, 873–904. McEwen, B. S. (1993). Stress and the Individual: Mechanism Leading to Disease. Archives of Internal Medicine, 153, 2093–2101. McLaren, A., Kerr, S., Allan, L., Steen, I. N., Ballard, C., Allen, J., … Kenny, R. A. (2005). Autonomic function is impaired in elderly stroke survivors. Stroke, 36(5), 1026–30. Medzhitov, R. (2008). Origin and physiological roles of inflammation. Nature, 454(7203), 428–35. 232  Melanson, E. L., & Freedson, P. S. (2001). The effect of endurance training on resting heart rate variability in sedentary adult males. European Journal of Applied Physiology, 85(5), 442–449. Merwick, Á., Albers, G. W., Amarenco, P., Arsava, E. M., Ay, H., Calvet, D., … Kelly, P. J. (2010). Addition of brain and carotid imaging to the ABCD2 score to identify patients at early risk of stroke after transient ischaemic attack: A multicentre observational study. The Lancet Neurology, 9(11), 1060–1069. Meschia, J. F., Bushnell, C., Boden-albala, B., Braun, L. T., Bravata, D. M., Chaturvedi, S., … Wilson, J. A. (2014). Guidelines for the Primary Prevention of Stroke A Statement for Healthcare Professionals From the American Heart Association/American Stroke Association. Stroke, 45. Micieli, G., & Cavallini, A. (2008). The autonomic nervous system and ischemic stroke: A reciprocal interdependence. Clinical Autonomic Research, 18(6), 308–317. Miller, D. J., Simpson, J. R., & Silver, B. (2011). Safety of thrombolysis in acute ischemic stroke: a review of complications, risk factors, and newer technologies. The Neurohospitalist, 1(3), 138–47. Minami, J., Ishimitsu, T., & Matsuoka, H. (1999). Effects of smoking cessation on blood pressure and heart rate variability in habitual smokers. Hypertension, 33, 586–90. Miu, A. C., Heilman, R. M., & Miclea, M. (2009). Reduced heart rate variability and vagal tone 233  in anxiety: Trait versus state, and the effects of autogenic training. Autonomic Neuroscience: Basic and Clinical, 145(1–2), 99–103. Mokdad, A. H., Ford, E. S., Bowman, B. A., Dietz, W. H., Vinicor, F., Bales, V. S., & Marks, J. S. (2015). Prevalence of Obesity, Diabetes, and Obesity-Related Health Risk Factors, 2001, 289(1), 2001–2004. Montano, N., Porta, A., Cogliati, C., Costantino, G., Tobaldini, E., Casali, K. R., & Iellamo, F. (2009). Heart rate variability explored in the frequency domain: a tool to investigate the link between heart and behavior. Neuroscience and Biobehavioral Reviews, 33(2), 71–80. Montano, N., Ruscone, T. G., Porta, A., Lombardi, F., Pagani, M., & Malliani, A. (1994). Power spectrum analysis of heart rate variability to assess the changes in sympathovagal balance during graded orthostatic tilt. Circulation, 90(4), 1826–1831. Mozaffarian, D., Benjamin, E. J., Go, A. S., Arnett, D. K., Blaha, M. J., Cushman, M., … Turner, M. B. (2016). Heart disease and stroke statistics-2016 update a report from the American Heart Association. Circulation, 133(4), e38–e48. Mravec, B. (2010). The role of the vagus nerve in stroke. Autonomic Neuroscience: Basic and Clinical, 158(1–2), 8–12. Myers, G., Martin, G., & Magid, N. (1986). Power Spectral Analysis of Heart Rate Varability in Sudden Cardiac Death: Comparison to Other Methods. IEEE Transactions on Biomedical Engineering, 33(12), 1149–1156. 234  Nadarajan, V., Perry, R. J., Johnson, J., & Werring, D. J. (2014). Transient ischaemic attacks: mimics and chameleons. Practical Neurology, 14(1), 23–31. Naver, H. K., Blomstrand, C., Wallin, B. G., Hans, K. M. D., Christian, M. D., Gunnar, B. M. D., … Gunnar, B. M. D. (1996). Reduced Heart Rate Variability After Right-Sided Stroke . Stroke, 27(2), 247–251. Niedermaier, O. N., Smith, M. L., Beightol, L. a., Zukowska-Grojec, Z., Goldstein, D. S., & Eckberg, D. L. (1993). Influence of cigarette smoking on human autonomic function. Circulation, 88(2), 562–571. NINDS Committee. (1990). Special report from the National Institute of Neurological Disorders and Stroke. Classification of cerebrovascular diseases III. Stroke, 21(4), 637–676. Nussinovitch, U., Cohen, O., Kaminer, K., Ilani, J., & Nussinovitch, N. (2012). Evaluating reliability of ultra-short ECG indices of heart rate variability in diabetes mellitus patients. Journal of Diabetes and Its Complications, 26(5), 450–453. O’Donnell, M. J., Denis, X., Liu, L., Zhang, H., Chin, S. L., Rao-Melacini, P., … Yusuf, S. (2010). Risk factors for ischaemic and intracerebral haemorrhagic stroke in 22 countries (the INTERSTROKE study): A case-control study. The Lancet, 376(9735), 112–123. Ohlin, B., Nilsson, P. M., Nilsson, J. A., & Berglund, G. (2004). Chronic psychosocial stress predicts long-term cardiovascular morbidity and mortality in middle-aged men. European Heart Journal, 25(10), 867–873. 235  Ois, A., Gomis, M., Rodríguez-Campello, A., Cuadrado-Godia, E., Jiménez-Conde, J., Pont-Sunyer, C., … Roquer, J. (2008). Factors associated with a high risk of recurrence in patients with transient ischemic attack or minor stroke. Stroke, 39(6), 1717–21. Orlandi, G., Fanucchi, S., Strata, G., Pataleo, L., Landucci Pellegrini, L., Prontera, C., … Murri, L. (2000). Transient autonomic nervous system dysfunction during hyperacute stroke. Acta Neurologica Scandinavica, 102(5), 317–21. Pagani, M., Lombardi, F., Guzzetti, S., Rimoldi, O., Furlan, R., Pizzinelli, P., … Piccaluga, E. (1986). Power spectral analysis of heart rate and arterial pressure variabilities as a marker of sympatho-vagal interaction in man and conscious dog. Circulation Research, 59(2), 178–193. Paganini-Hill,  a., Lozano, E., Fischberg, G., Perez Barreto, M., Rajamani, K., Ameriso, S. F., … Fisher, M. (2003). Infection and Risk of Ischemic Stroke: Differences Among Stroke Subtypes. Stroke, 34(2), 452–457. Pagkalos, M., Koutlianos, N., Kouidi, E., Pagkalos, E., Mandroukas, K., & Deligiannis, A. (2007). Heart rate variability modifications following exercise training in type 2 diabetic patients with definite cardiac autonomic neuropathy. British Journal of Sports Medicine, 42(1), 47–54. Palatini, P., & Julius, S. (2009). The role of cardiac autonomic function in hypertension and cardiovascular disease. Current Hypertension Reports, 11(3), 199–205. 236  Pardo, Y., Merz, C. N., Velasquez, I., Paul-Labrador, M., Agarwala, A., & Peter, C. T. (2000). Exercise conditioning and heart rate variability: evidence of a threshold effect. Clinical Cardiology, 23(8), 615–620. Patterson, E., Po, S., Scherlag, B., & Lazzara, R. (2005). Triggered firing in pulmonary veins initiated by in vitro autonomic nerve stimulation. Heart Rhythm, 2, 624–631. Pearson, T. A., Blair, S. N., Daniels, S. R., Eckel, R. H., Fair, J. M., Fortmann, S. P., … Taubert, K. A. (2002). AHA Guidelines for Primary Prevention of Cardiovascular Disease and Stroke: 2002 Update: Consensus panel guide to comprehensive risk reduction for adult patients without coronary or other atherosclerotic vascular diseases. Circulation, 106(3), 388–391. Perkiömäki, J., Ukkola, O., Kiviniemi, A., Tulppo, M., Ylitalo, A., Kesäniemi, Y. A., & Huikuri, H. (2014). Heart rate variability findings as a predictor of atrial fibrillation in middle-aged population. Journal of Cardiovascular Electrophysiology, 25(7), 719–724. Perry, J., Sharma, M., Sivilotti, M., Sutherland, J., Symington, C., Worster, A., & Émond, M. (2011). Prospective Validation of the ABCD2 score for patients in the emergency department with transient ischemic attack. Canadian Medical Association Journal, 1–9. Petit, J.-M., Bour, J.-B., Galland-Jos, C., Minello, A., Verges, B., Guiguet, M., … Hillon, P. (2001). Risk factors for diabetes mellitus and early insulin resistance in chronic hepatitis C. Journal of Hepatology, 35(2), 279–283. 237  Phillip A. Low. (2003). Testing the Autonomic Nervous System. Seminars in Neurology, 23(4), 407–422. Piepoli, M., Garrard, C. S., Kontoyannis, D. A., & Bernardi, L. (1995). Autonomic control of the heart and peripheral vessels in human septic shock. Intensive Care Medicine, 21(2), 112–119. Piepoli, M., Radaelli, A., Ponikowski, P., Adamopoulos, S., Bemardi, L., Sleight, P., & Coats, A. J. S. (1996). Reproducibility of heart rate variability indices during exercise stress testing and inotrope infusion in chronic heart failure patients. Clinical Science (Lond), 91, Suppl:87– 8. Pinna, G. D., Maestri, R., Cesare, A. Di, R Colombo, & Minuco, G. (1994). The accuracy of power-spectrum analysis of heart-rate variability from annotated RR lists generated by Holter systems. Physiological Measurement, 15(2). Pomeranz, B., Macaulay, R. J. B., Adam, D. A. N., Gordon, D., Kilborn, M., Clifford, A., … Assessment, H. B. (1985). Assessment of autonomic function by heart rate spectral analysis in humans. American Journal of Physiology, 248, H151–H153. Ponikowski, P., Piepoli, M., Amadi, A. A., Chua, T. P., Harrington, D., Volterrani, M., … Coats, A. J. (1996). Reproducibility of heart rate variability measures in patients with chronic heart failure. Clinical Science (London, England : 1979), 91(4), 391–398. Powers, W. J., Derdeyn, C. P., Biller, J., Coffey, C. S., Hoh, B. L., Jauch, E. C., … Yavagal, D. 238  R. (2015). 2015 American Heart Association/American stroke association focused update of the 2013 guidelines for the early management of patients with acute ischemic stroke regarding endovascular treatment: A guideline for healthcare professionals from the American. Stroke, 46(10), 3020–3035. Pozzati, A., Pancaldi, L. G., Di Pasquale, G., Pinelli, G., & Bugiardini, R. (1995). A transient decrease in heart rate variability in patients with sudden “ischemic” death during Holter monitoring. Giornale Italiano Di Cardiologia, 25(6), 725–32. Prabhakaran, S., Silver, A. J., Warrior, L., McClenathan, B., & Lee, V. H. (2008). Misdiagnosis of transient ischemic attacks in the emergency room. Cerebrovascular Diseases, 26(6), 630–635. Prencipe, M., Culasso, F., Rasura, M., Anzini, A., Beccia, M., Cao, M., … Fieschi, C. (1998). Long-term Prognosis After a Minor Stroke : 10-Year Mortality and Major Stroke Recurrence Rates in a Hospital-Based Cohort. Stroke, 29(1), 126–132. Pumprla, J., Howorka, K., Groves, D., Chester, M., & Nolan, J. (2002). Functional assessment of heart rate variability: Physiological basis and practical applications. International Journal of Cardiology, 84(1), 1–14. Purroy, F., Caballero, P. E. J., Gorospe, A., Torres, M. J., Álvarez-Sabin, J., Santamarina, E., … Masjuán, J. (2012). Prediction of early stroke recurrence in transient ischemic attack patients from the PROMAPA study: A comparison of prognostic risk scores. 239  Cerebrovascular Diseases, 33(2), 182–189. Putaala, J., Kurkinen, M., Tarvos, V., Salonen, O., Kaste, M., & Tatlisumak, T. (2009). Silent brain infarcts and leukoaraiosis in young adults with first-ever ischemic stroke. Neurology, 72(21), 1823–1829. Raedt, S. De, Vos, A. De, & Keyser, J. De. (2015). Autonomic dysfunction in acute ischemic stroke: An underexplored therapeutic area? Journal of the Neurological Sciences, 348(1–2), 24–34. Reed, G. W., Rossi, J. E., & Cannon, C. P. (2017). Acute myocardial infarction. Lancet, 389(10065), 197–210. Reiner, Ž., Catapano, A. L., De Backer, G., Graham, I., Taskinen, M. R., Wiklund, O., … Wood, D. (2011). ESC/EAS Guidelines for the management of dyslipidaemias. European Heart Journal, 32(14), 1769–1818. Riccio, P. M., Klein, F. R., Pagani Cassará, F., Muñoz Giacomelli, F., González Toledo, M. E., Racosta, J. M., … Sposato, L. a. (2013). Newly diagnosed atrial fibrillation linked to wake-up stroke and TIA: hypothetical implications. Neurology, 80(20), 1834–40. Robertson, D. (2004). Primer on the Autonomic Nervous System. (I. Biaggioni & G. Burnstock, Eds.) (second). San Diego: Elsevier Academic Press. Rosenwinkel, E. T., Bloomfield, D. M., Arwady, M. a, & Goldsmith, R. L. (2001). Exercise and autonomic function in health and cardiovascular disease. Cardiology Clinics, 19(3), 369–87. 240  Ross, R. (1999). Inflammation or Atherogenesis. New England Journal of Medicine, 340(2), 115–126. Rothman, K. J. (2012). Epidemiology. Oxford University Press. Rothman, K. J., Lash, T. L., & Greenland, S. (2012). Modern Epidemiology (3rd ed.). Wolters Kluwer. Rothwell, P., Giles, M., Chandratheva, A., Marquardt, L., Geraghty, O., Redgrave, J., … Mehta, Z. (2007). Effect of urgent treatment of transient ischaemic attack and minor stroke on early recurrent stroke (EXPRESS study): a prospective population-based sequential comparison. Lancet, 370(9596), 1432–42. Rothwell, P. M., Buchan, A., & Johnston, S. C. (2006). Recent advances in management of transient ischaemic attacks and minor ischaemic strokes. Lancet Neurology, 5(4), 323–331. Rothwell, P. M., Giles, M. F., Flossmann, E., Lovelock, C. E., Redgrave, J. N. E., Warlow, C. P., & Mehta, Z. (2005). A simple score (ABCD) to identify individuals at high early risk of stroke after transient ischaemic attack. Lancet, 366(9479), 29–36. Rothwell, P. M., & Warlow, C. P. (2005). Timing of TIAs preceding stroke: time window for prevention is very short. Neurology, 64(5), 817–820. Routledge, F. S., Campbell, T. S., McFetridge-Durdle, J. A., & Bacon, S. L. (2010). Improvements in heart rate variability with exercise therapy. Canadian Journal of Cardiology, 26(6), 303–312. 241  Rundek, T., Gardener, H., Xu, Q., Goldberg, R. B., Wright, C. B., Boden-Albala, B., … Sacco, R. L. (2010). Insulin resistance and risk of ischemic stroke among nondiabetic individuals from the northern Manhattan study. Archives of Neurology, 67(10), 1195–1200. Sacco, R. L., Adams, R., Albers, G., Alberts, M. J., Benavente, O., Furie, K., … Tomsick, T. (2006). Guidelines for prevention of stroke in patients with ischemic stroke or transient ischemic attack: a statement for healthcare professionals from the American Heart Association/American Stroke Association Council on Stroke: co-sponsored by the Council on C. Stroke, 37(2), 577–617. Sacco, R. L., Kasner, S. E., Broderick, J. P., Caplan, L. R., Connors, J. J., Culebras, A., … Vinters, H. V. (2013). An updated definition of stroke for the 21st century: A statement for healthcare professionals from the American heart association/American stroke association. Stroke, 44(7), 2064–2089. Salo, M. a, Huikuri, H. V, & Seppänen, T. (2001). Ectopic beats in heart rate variability analysis: effects of editing on time and frequency domain measures. Annals of Noninvasive Electrocardiology, 6(1), 5–17. Sandercock, G. R. H., Grocott-Mason, R., & Brodie, D. A. (2007). Changes in short-term measures of heart rate variability after eight weeks of cardiac rehabilitation. Clinical Autonomic Research, 17(1), 39–45. Sanders, L. M., Srikanth, V. K., Blacker, D. J., Jolley, D. J., Cooper, K. A., & Phan, T. G. 242  (2012). Performance of the ABCD2 score for stroke risk post TIA meta-analysis and probability modeling. Neurology, 79(10), 971–980. Santos-Hiss, M. D. B., Melo, R. C., Neves, V. R., Hiss, F. C., Verzola, R. M. M., Silva, E., … Catai, A. M. (2011). Effects of progressive exercise during phase I cardiac rehabilitation on the heart rate variability of patients with acute myocardial infarction. Disability and Rehabilitation, 33(10), 835–842. Sarwar, N., Gao, P., Kondapally Seshasai, S. R., Gobin, R., Kaptoge, S., Di Angelantonio, E., … Wormser, D. (2010). Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: A collaborative meta-analysis of 102 prospective studies. The Lancet, 375(9733), 2215–2222. Schlaich, M., Straznicky, N., Lambert, E., & Lambert, G. (2015). Metabolic syndrome: A sympathetic disease? Lancet Diabetes and Endocrinology, 3(2), 148–158. Schmidt, H. B., Werdan, K., & Müller-Werdan, U. (2001). Autonomic dysfunction in the ICU patient. Current Opinion in Critical Care, 7(5), 314–22. Schneiderman, N., Ironson, G., & Siegel, S. D. (2005). Stress and Health: Psychological, Behavioral, and Biological Determinants. Annual Review of Clinical Psychology, 1(1), 607–628. Schobel, H., Fischer, T., Heuszer, K., Geiger, H., & Schmieder, R. (1996). Preeclampsia - a state of sympathetic overactivity. New England Journal of Medicine, 335, 1480–1485. 243  Schrock, J. W., Glasenapp, M., Victor, A., Losey, T., & Cydulka, R. K. (2012). Variables associated with discordance between emergency physician and neurologist diagnoses of transient ischemic attacks in the emergency department. Annals of Emergency Medicine, 59(1), 19–26. Schroeder, E. B., Whitsel, E. A., Evans, G. W., Prineas, R. J., Chambless, L. E., & Heiss, G. (2004). Repeatability of heart rate variability measures. Journal of Electrocardiology, 37(3), 163–172. Schroeder, E., Liao, D., Chambless, L., Prineas, R., Evans, G., & Heiss, G. (2003). Hypertension, blood pressure, and heart rate variability: the Atherosclerosis Risk in Communities (ARIC) study. Hypertension, 42, 1106–11. Selvin, E., Steffes, M. W., Zhu, H., Matsushita, K., Wagenknecht, L., Pankow, J., … Brancati, F. L. (2010). Glycated Hemoglobin, Diabetes, and Cardiovascular Risk in Nondiabetic Adults. New England Journal of Medicine, 362(9), 800–811. Selye, H. (1955). Stress and disease. Science, 122(3171), 625–631. Sharma, A., Prasad, K., Padma, M. V., Tripathi, M., Bhatia, R., Singh, M. B., & Sharma, A. (2015). Prevalence of triggering factors in acute stroke: Hospital-based observational cross-sectional study. Journal of Stroke and Cerebrovascular Diseases, 24(2), 337–347. Shebuski, R. J., & Kilgore, K. S. (2002). Role of inflammatory mediators in thrombogenesis. Journal of Pharmacology and Experimental Therapeutics, 300, 729–735. 244  Sheehan, O. C., Kyne, L., Kelly, L. A., Hannon, N., Marnane, M., Merwick, A., … Kelly, P. J. (2010). Population-based study of ABCD2 score, carotid stenosis, and atrial fibrillation for early stroke prediction after transient ischemic attack: The North Dublin TIA study. Stroke, 41(5), 844–850. Sheehan, O. C., Merwick, A., Kelly, L. A., Hannon, N., Marnane, M., Kyne, L., … Kelly, P. J. (2009). Diagnostic usefulness of the ABCD2 score to distinguish transient ischemic attack and minor ischemic stroke from noncerebrovascular events: The North Dublin TIA study. Stroke, 40(11), 3449–3454. Shinton, R., & Beevers, G. (1989). Meta-analysis of relation between cigarette smoking and stroke. BMJ (Clinical Research Ed.), 298(6676), 789–94. Shiroma, E. J., & Lee, I.-M. (2010). Physical Activity and Cardiovascular Health: Lessons Learned From Epidemiological Studies Across Age, Gender, and Race/Ethnicity. Circulation, 122(7), 743–752. Silke, B., Campbell, C., & King, D. (2002). The potential cardiotoxicity of antipsychotic drugs as assessed by heart rate variability. Journal of Psychopharmacology, 16(4), 355–360. Singh, J., Larson, M., Tsuji, H., Evans, J., O’Donnell, C., & Levy, D. (1998). Reduced heart rate variability and new-onset hypertension: insights into pathogenesis of hypertension: the Framingham Heart Study. Hypertension, 32, 293–7. Smyth, A., Teo, K. K., Rangarajan, S., O’Donnell, M., Zhang, X., Rana, P., … Yusuf, S. (2015). 245  Alcohol consumption and cardiovascular disease, cancer, injury, admission to hospital, and mortality: A prospective cohort study. Lancet, 386(10007), 1945–1954. Sollers, J. J., Sanford, T. A., Nabors-Oberg, R., Anderson, C. A., & Thayer, J. F. (2002). Examining changes in HRV in response to varying ambient temperature. IEEE Engineering in Medicine and Biology, 21(4), 30–34. Song, B., Fang, H., Zhao, L., Gao, Y., Tan, S., Lu, J., … Xu, Y. (2013). Validation of the ABCD3-I score to predict stroke risk after transient ischemic attack. Stroke, 44(5), 1244–8. Sroka, K., Peimann, C. J., & Seevers, H. (1997). Heart rate variability in myocardial ischemia during daily life. Journal of Electrocardiology, 30(1), 45–56. Stein, P. K., Barzilay, J. I., Chaves, P. H. M., Domitrovich, P. P., & Gottdiener, J. S. (2009). Heart rate variability and its changes over 5 years in older adults. Age and Ageing, 38(2), 212–218. Stein, P. K., & Kleiger, R. E. (1999). Insights from the study of heart rate variability. Annual Review of Medicine, 50, 249–61. Steptoe, A., & Kivimäki, M. (2012). Stress and cardiovascular disease. Nature Reviews Cardiology, 9(6), 360–70. Straznicky, N., Lambert, E., Lambert, G., Masuo, K., Esler, M., & Nestel, P. (2005). Effects of dietary weight loss on sympathetic activity and cardiac risk factors associated with the metabolic syndrome. The Journal of Clinical Endocrinology & Metabolism, 90, 5998–6005. 246  Suadicani, P., Andersen, L. L., Holtermann, A., Mortensen, O. S., & Gyntelberg, F. (2011). Perceived psychological pressure at work, social class, and risk of stroke: a 30-year follow-up in Copenhagen male study. Journal of Occupational and Environmental Medicine, 53(12), 1388–95. Sundboll, J., Horvath-Puho, E., Schmidt, M., Pedersen, L., Henderson, V. W., Botker, H. E., & Sorensen, H. T. (2016). Long-term risk of stroke in myocardial infarction survivors: Thirty-year population-based cohort study. Stroke, 47(7), 1727–1733. Sundell, L., Salomaa, V., Vartiainen, E., Poikolainen, K., & Laatikainen, T. (2008). Increased Stroke Risk Is Related to a Binge Drinking Habit. Stroke, 39, 3179–84. Surtees, P., Wainwright, N., Luben, R., Wareham, N., Bingham, S., & Khaw, K. (2007). Adaptation to social adversity is associated with stroke incidence. Evidence from the EPIC-Norfolk Prospective Cohort Study. Stroke, 38, 1447–53. Surwit, R., & Feinglos, M. (1988). Stress and autonomic nervous system in type II diabetes-A hypothesis. Diabetes Care, 11, 83—85. Tapanainen, J. M., Thomsen, P. E. B., Køber, L., Torp-Pedersen, C., Mäkikallio, T. H., Still, A.-M., … Huikuri, H. V. (2002). Fractal analysis of heart rate variability and mortality after an acute myocardial infarction. The American Journal of Cardiology, 90(4), 347–52. Tarvainen, M. P., Laitinen, T. P., Lipponen, J. A., Cornforth, D. J., & Jelinek, H. F. (2014). Cardiac autonomic dysfunction in type 2 diabetes - effect of hyperglycemia and disease 247  duration. Frontiers in Endocrinology, 5(AUG), 1–9. Task Force of ESC. (1996). Heart Rate Variability : Standards of Measurement, Physiological Interpretation, and Clinical Use. Circulation, 93, 1043–1065. Taylor, J. A., Carr, D. L., Myers, C. W., & Eckberg, D. L. (1998). Mechanisms underlying very-low-frequency RR-interval oscillations in humans. Circulation, 98(6), 547–555. Teresa, M., Rovere, L., Jr, J. T. B., Marcus, F. I., Mortara, A., Schwartz, P. J., & Commentary, S. (1998). Baroreflex sensitivity and heart-rate variability in prediction of total cardiac mortality after myocardial infarction. Lancet, 351, 478–84. Thayer, J. F., Åhs, F., Fredrikson, M., Sollers, J. J., & Wager, T. D. (2012). A meta-analysis of heart rate variability and neuroimaging studies: Implications for heart rate variability as a marker of stress and health. Neuroscience and Biobehavioral Reviews, 36(2), 747–756. Thayer, J. F., Hall, M., Sollers, J. J., & Fischer, J. E. (2006). Alcohol use, urinary cortisol, and heart rate variability in apparently healthy men: Evidence for impaired inhibitory control of the HPA axis in heavy drinkers. International Journal of Psychophysiology, 59(3), 244–250. Thayer, J. F., & Lane, R. D. (2007). The role of vagal function in the risk for cardiovascular disease and mortality. Biological Psychology, 74(2), 224–242. Thayer, J. F., & Sternberg, E. (2006). Beyond heart rate variability: Vagal regulation of allostatic systems. Annals of the New York Academy of Sciences, 1088, 361–372. 248  Thayer, J. F., Yamamoto, S. S., & Brosschot, J. F. (2010). The relationship of autonomic imbalance, heart rate variability and cardiovascular disease risk factors. International Journal of Cardiology, 141(2), 122–131. Thygesen, K., Alpert, J. S., Jaffe, A. S., Simoons, M. L., Chaitman, B. R., & White, H. D. (2012). Third Universal Definition of Myocardial Infarction. Journal of the American College of Cardiology, 60(16), 1581–1598. Tofler, G. H., & Muller, J. E. (2006). Triggering of acute cardiovascular disease and potential preventive strategies. Circulation, 114(17), 1863–1872. Toustads, A. J. J. (2006). Cardiovascular risks associated with smoking: a review for clinicians. Eur Journal of Cardiovascular Prev Rehabil, 13(4), 507–14. Touzé, E., Varenne, O., Chatellier, G., Peyrard, S., Rothwell, P. M., & Mas, J.-L. (2005). Risk of myocardial infarction and vascular death after transient ischemic attack and ischemic stroke: a systematic review and meta-analysis. Stroke, 36(12), 2748–55. Toweill, D., Sonnenthal, K., Kimberly, B., Lai, S., & Goldstein, B. (2000). Linear and nonlinear analysis of hemodynamic signals during sepsis and septic shock. Critical Care Medicine, 28(6), 2051–7. Tracey, K. J. (2002). The inflammatory reflex. Nature, 420(6917), 853–9. Tracey, K. J. (2007). Physiology and immunology of the cholinergic antiinflammatory pathway. The Journal of Clinical Investigation, 117(2). 249  Truelsen, T., Nielsen, N., Boysen, G., & Gronbaek, M. (2003). Self-reported stress and risk of stroke. Stroke, 34, 856–62. Truelsen, T., Nielsen, N., Boysen, G., & Grønbæk, M. (2003). Self-reported stress and risk of stroke: The Copenhagen City Heart Study. Stroke, 34(4), 856–862. Tsigos, C., & Chrousos, G. P. (2002). Hypothalamic-pituitary-adrenal axis, neuroendocrine factors and stress. Journal of Psychosomatic Research, 53(4), 865–71. Tsuji, H., Larson, M. G. M., Venditti, F. J. F. J., Manders, E. E. S., Evans, J. J. C., Feldman, C. L. C., & Levy, D. (1996). Impact of Reduced Heart Rate Variability on Risk for Cardiac Events. Circulation, 94(11), 2850–2855. Tsuji, H., Venditti, F. J., Manders, E. S., Evans, J. C., Larson, M. G., Feldman, C. L., & Levy, D. (1994). Reduced heart rate variability and mortality risk in an elderly cohort. The Framingham Heart Study. Circulation, 90(2), 878–883. Ulrich-Lai, Y. M., & Herman, J. P. (2009). Neural Regulation of Endocrine and Autonomic Stress Responses. Nature Review Neuroscience, 10(6), 397–409. Vallance, P., Collier, J., & Bhagat, K. (1997). Infection, inflammation, and infarction: does acute endothelial dysfunction provide a link? Lancet, 349, 1391–1392. Vallbo, A. B. (2004). Microneurography: how the technique developed and its role in the investigation of the sympathetic nervous system. Journal of Applied Physiology, 96(4), 1262–1269. 250  Vittinghoff, E., Glidden, D. V, Shiboski, S. C., & McCulloch, C. E. (2005). Regression Methods in Biostatistics. (Springer, Ed.). Wang, Y., Minematsu, K., Wong, K. S. L., Amarenco, P., Albers, G. W., Denison, H., … Johnston, S. C. (2017). Ticagrelor in acute stroke or transient ischemic attack in Asian patients. Stroke, 48(1), 167–173. Wang, Y., Wang, Y., Zhao, X., Liu, L., Wang, D., Wang, C., … Johnston, S. C. (2013). Clopidogrel with Aspirin in Acute Minor Stroke or Transient Ischemic Attack. New England Journal of Medicine, 369(1), 11–19. Wannamethee, S., Shaper, A., Whincup, P., & Walker, M. (1995). Smoking cessation and the risk of stroke in middle-aged men. Journal of the American Medical Association, 274(2), 155–160. Wardlaw, J. M., Brazzelli, M., Chappell, F. M., Miranda, H., Shuler, K., & Sandercock, P. A. G. (2015). ABCD2 score and secondary stroke prevention. Neurology, 85, 1–8. Watson, T., Shantsila, E., & Lip, G. Y. (2009). Mechanisms of thrombogenesis in atrial fibrillation: Virchow’s triad revisited. Lancet, 373(9658), 155–166. Wellen, K. E., & Hotamisligil, G. S. (2005). Inflammation, stress, and diabetes. The Journal of Clinical Investigation, 115(5), 1111–1119. Wendel-Vos, G. (2004). Physical activity and stroke. A meta-analysis of observational data. International Journal of Epidemiology, 33(4), 787–798. 251  Wennerblom, B., Lurje, L., Karlsson, T., Tygesen, H., Vahisalo, R., & Hjalmarson A. (2001). Circadian variation of heart rate variability and the rate of autonomic change in the morning hours in healthy subjects and angina patients. International Journal of Cardiology, 79(1), 61–9. Whelton, S. P., Chin, A., Xin, X., & He, J. (2002). Effect of aerobic exercise on blood pressure: A meta-analysis of randomized, controlled trials. Annals of Internal Medicine, 136(7), 493–503. Whisnant, J. P., Wiebers, D. O., Fallon, W. M. O., Sicks, J. D., & Frye, R. L. (1996). A population-based model of risk factors for ischemic stroke. Neurology, 47, 1–6. Williams, M. A., Fleg, J., Ades, P., Chaitman, B., Miller, N., Mohiuddin, S., … Wenger, N. (2002). Secondary Prevention of Coronary Heart Disease in the Elderly (With Emphasis on Patients >75 Years of Age). Circulation, 105(14), 1735–1743. Win, N. N., Fukayama, H., Kohase, H., & Umino, M. (2005). The different effects of intravenous propofol and midazolam sedation on hemodynamic and heart rate variability. Anesthesia and Analgesia, 101(1), 97–102. Withers, G. S., & Toggweiler, J. R. (1993). The pathogenesis of atherosclerosis: a perspective for the 1990s. Nature, 363, 210–211. Witt, B. J., Ballman, K. V., Brown, R. D., Meverden, R. A., Jacobsen, S. J., & Roger, V. L. (2006). The incidence of stroke after myocardial infarction: A meta-analysis. American 252  Journal of Medicine, 119(4), 7–10. Witt, B. J., Jr, R. D. B., Jacobsen, S. J., Weston, S. A., & Yawn, B. P. (2012). A Community-Based Study of Stroke Incidence after Myocardial Infarction. Annals of Internal Medicine, 143(11). Wiysonge, C. S., Bradley, H. A., Volmink, J., Mayosi, B. M., & Opie, L. H. (2012). Beta-blockers for hypertension. Cochrane Database of Systematic Reviews, (8). Wolf, M., Varigos, G., Hunt, D., & Sloman, J. (1978). Sinus arrhythmia in acute myocardial infarction. The Medical Journal of Australia, 2(2), 52–53. Wolf, P. A., Abbott, R. D., & Kannel, W. B. (1991). Atrial Fibrillation as an Independent Risk Factor for Stroke : The Framingham Study. Stroke, 22, 983–988. Wolf, P. A., D’Agostino, R. B., Kannel, W. B., Bonita, R., & Belanger, A. (1988). Cigarette Smoking as a Risk Factor for Stroke. The Journal of American Medical Association, 259(7), 1025. Wu, C. M., McLaughlin, K., Lorenzetti, D. L., Hill, M. D., Manns, B. J., & Ghali, W. a. (2007). Early risk of stroke after transient ischemic attack. Archives of Internal Medicine, 167(22), 2417–2422. Yeragani, V. K., Sobolewski, E., Kay, J., Jampala, V. C., & Igel, G. (1997). Effect of age on long-term heart rate variability. Cardiovascular Research, 35(1), 35–42. Yi, C. X., Fleur, S. E., La Fliers, E., & Kalsbeek, A. (2010). The role of the autonomic nervous 253  liver innervation in the control of energy metabolism. Biochimica et Biophysica Acta, 1802(4), 416–431. Youden, W. J. (1950). Index for rating diagnostic tests. Cancer, 3(1), 32–35. Yperzeele, L., van Hooff, R. J., Nagels, G., De Smedt, A., De Keyser, J., & Brouns, R. (2015). Heart rate variability and baroreceptor sensitivity in acute stroke: a systematic review. International Journal of Stroke, 10(6), 796–800. Zygmunt, A., & Stanczyk, J. (2010). Methods of evaluation of autonomic nervous system function. Archives of Medical Science, 6(1), 11–18. 254  Appendices Appendix A  NIHSS Score National Institutes of Health Stroke Scale (NIHSS) score (https://www.ninds.nih.gov/) is a tool used by healthcare providers to objectively quantify the severity/impairment caused by a stroke. The NIHSS is composed of 11 items (item 1 with a, b, c; items 5 and 6 with a and b), each of which scores a specific ability between a 0 and 4. For each item, a score of 0 typically indicates normal function in that specific ability, while a higher score is indicative of some level of impairment. The individual scores from each item are summed in order to calculate a patient's total NIHSS score. The maximum possible score is 42, with the minimum score being a 0. The NIH Stroke Scale is shown below (https://www.ninds.nih.gov/Stroke-Scales-and-Related-Information)  255   256    257    258  259  Appendix B  TOAST Classification for Ischemic Stroke TOAST classification is a widely used etiological classification system in clinical practice and research (Adams et al., 1993). It classifies ischemic stroke into five subtypes based on their potential causes.  i. Large-artery atherosclerosis (LAA): Patients who have clinical and brain imaging findings of either significant (>50%) stenosis or occlusion of a major brain artery, presumably due to atherosclerosis are classified to large-artery atherosclerotic ischemic stroke. Clinical findings include cerebral cortical impairment (aphasia, neglect, restricted motor involvement), brain stem or cerebellar dysfunction. A history of intermittent claudication, TIAs in the same vascular territory, a carotid bruit, or diminished pulses help support the diagnosis of this subtype. Cortical or cerebellar lesions and brain stem or subcortical hemispheric infarcts greater than 1.5 cm in diameter on CT or MRI are considered to be of potential large-artery atherosclerotic origin. Diagnosis should exclude potential sources of cardiogenic embolism. The diagnosis of large-artery atherosclerotic stroke cannot be made if duplex or arteriographic studies are normal or show only minimal changes. This subtype takes 20-25% of total ischemic stroke. ii. Cardioembolism: Patients with arterial occlusions presumably due to an embolus arising in the heart are categorized into cardioembolic ischemic stroke. Based on the evidence of their relative propensities for embolism, cardiac sources are divided into high-risk group (such as mechanical prosthetic valve, mitral stenosis with atrial fibrillation, atrial fibrillation, left atrial/atrial appendage thrombus, sick sinus syndrome, recent myocardial infarction (<4 weeks), left ventricular thrombus, dilated cardiomyopathy, akinetic left ventricular segment, atrial myxoma, infective endocarditis) and medium-risk group (such 260  as mitral valve prolapse, mitral annulus calcification, mitral stenosis without atrial fibrillation, left atrial turbulence, atrial septal aneurysm, patent foramen ovale, atrial flutter, lone atrial fibrillation, bioprosthetic cardiac valve, nonbacterial thrombotic endocarditis, congestive heart failure, hypokinetic left ventricular segment, myocardial infarction (>4 weeks, <6 months)). At least one cardiac source for an embolus must be identified for a possible diagnosis of cardioembolic stroke. Evidence of a previous TIA or stroke in more than one vascular territory or systemic embolism supports a clinical diagnosis of cardiogenic stroke. Potential large-artery atherosclerotic sources of thrombosis or embolism should be eliminated. A stroke in a patient with a medium-risk cardiac source of embolism and no other cause of stroke is classified as a possible cardioembolic stroke. This subtype takes 25-30% of total ischemic stroke. iii. Lacunar source (small-vessel occlusion): Lacunar stroke results from occlusion of small penetrating arteries which provide blood to the brain's deep structures. The patient should have one of the traditional clinical lacunar syndromes and should not have evidence of cerebral cortical dysfunction. A history of diabetes mellitus or hypertension supports the clinical diagnosis. The patient should also have a normal CT/MRI examination or a relevant brain stem or subcortical hemispheric lesion with a diameter of less than 1.5 cm. Potential cardiac sources for embolism should be absent, and evaluation of the large extracranial arteries should not demonstrate a stenosis of greater than 50% in an ipsilateral artery. This subtype takes 20-25% of total ischemic stroke. iv. Other determined etiology: This category includes patients with rare causes of stroke, such as nonatherosclerotic vasculopathies, hypercoagulable states, or hematologic disorders. Diagnostic tests such as blood tests or arteriography should reveal one of these 261  unusual causes of stroke. Cardiac sources of embolism and large-artery atherosclerosis should be excluded by other studies. This subtype takes approximate 5% of total ischemic stroke. v. Undetermined etiology (cryptogenic): In some conditions, the cause of a stroke cannot be determined with any degree of confidence. This category includes a condition that no probable etiology is determined despite an extensive evaluation, a condition that no cause is found but the evaluation was cursory. This category also includes patients with two or more potential causes of stroke so that the physician is unable to make a final diagnosis. For example, a patient with a medium-risk cardiac source of embolism who also has another possible cause of stroke identified would be classified as having a stroke of undetermined etiology. Other examples would be a patient who has atrial fibrillation and an ipsilateral stenosis of 50%, or the patient with a traditional lacunar syndrome and an ipsilateral carotid stenosis of 50%. This subtype takes 20-25% of total ischemic stroke. 262  Appendix C  ABCD, ABCD2, ABCD3 and ABCD3-I Scores ❖ ABCD score (P. M. Rothwell et al., 2005) Risk factors   Points Score  Age    ≥ 60 years  1  Blood pressure   Systolic BP ≥ 140 mm Hg OR Diastolic BP ≥ 90 mm Hg  1  Clinical features of TIA (choose one)   Unilateral weakness with or without speech impairment    Speech impairment without unilateral weakness  2 1  Duration    TIA duration ≥ 60 minutes   TIA duration 10-59 minutes  2 1   Total ABCD score  0-6    ❖ ABCD2 score (Johnston et al., 2007a) Risk factors  Points Score   Age    ≥ 60 years  1  Blood pressure   Systolic BP ≥ 140 mm Hg OR Diastolic BP ≥ 90 mm Hg  1  Clinical features of TIA (choose one)   Unilateral weakness with or without speech impairment OR    Speech impairment without unilateral weakness  2 1  Duration    TIA duration ≥ 60 minutes   TIA duration 10-59 minutes  2 1  Diabetes  1   Total ABCD2 score  0-7       263  ❖ ABCD3 score (Merwick et al., 2010) Risk factors   Points Score  Age    ≥ 60 years  1  Blood pressure   Systolic BP ≥ 140 mm Hg OR Diastolic BP ≥ 90 mm Hg  1  Clinical features of TIA (choose one)   Unilateral weakness with or without speech impairment OR    Speech impairment without unilateral weakness  2 1  Duration    TIA duration ≥ 60 minutes   TIA duration 10-59 minutes  2 1  Diabetes  1  Dual TIA 2   Total ABCD3 score   0-9   ❖ ABCD3-I score (Merwick et al., 2010) Risk factors  Points Score   Age    ≥ 60 years 1  Blood pressure   Systolic BP ≥ 140 mm Hg OR Diastolic BP ≥ 90 mm Hg 1  Clinical features of TIA (choose one)   Unilateral weakness with or without speech impairment OR    Speech impairment without unilateral weakness  2 1  Duration    TIA duration ≥ 60 minutes   TIA duration 10-59 minutes  2 1  Diabetes  1  Dual TIA 2  Imaging: ipsilateral ≥50% stenosis of internal carotid artery 2  Imaging: acute diffusion-weighted imaging hyperintensity 2   Total ABCD3-I score  0-13   264  Appendix D  The General Picture of Autonomic Nervous System  The autonomic nervous system has two subsystems which are sympathetic nervous system and parasympathetic nervous system. The schematic diagram of the ANS is shown in Figure D.1. The sympathetic innervation of the peripheral organs is derived from the efferent preganglionic fibers, whose preganglionic cell neurons are located in the intermediolateral horn of the spinal cord between T1 and L2 or L3. The sympathetic ganglia are adjacent to the spine and consist of the vertebral (sympathetic chain) and prevertebral ganglia. The postganglionic fibers are long fibers which run from the sympathetic ganglia to effector organs, including the smooth muscle of blood vessels, viscera, lungs, scalp, pupils, heart and glands (sweat, salivary, and digestive). The major SNS neurotransmitter is norepinephrine. Most postganglionic sympathetic fibers are adrenergic fibers while the preganglionic fibers are cholinergic (Robertson, 2004).  The nerve fibers of the PNS are the cranial nerves primarily the vagus nerve, and the lumbar spinal nerves. The preganglionic cell bodies of PNS are located in the brain stem with the 3rd, 7th, 9th, and 10th (vagus) cranial nerves and sacral portion of the spinal cord at S2 and S3. The vagus nerve contains about 75% of all parasympathetic fibers. As the parasympathetic ganglia are located in the effector organs and the postganglionic fibers are short, only 1 or 2 mm, the parasympathetic system thus can produce specific, localized responses in effector organs, such as blood vessels of the head, neck, and thoracoabdominal viscera, lacrimal and salivary glands, smooth muscle of glands and viscera and muscles of the pupil. The major neurotransmitter for PNS is acetylcholine. All the preganglionic fibers cholinergic fibers, as well as all postganglionic parasympathetic fibers and some postganglionic sympathetic fibers (Robertson, 2004).   265  Figure D.1: Schematic diagram of the autonomic nervous system  @2007 Encyclopedia Britannica   266  Appendix E  Heart Rate Variability Measurement  ❖ Time domain analysis (Task Force of ESC, 1996) The simplest method to perform is the time domain measure. With this method, either heart rate at any point in time or the intervals between successive normal complexes are determined. Normal-to-normal (NN) intervals are detected on the continuous electrocardiographic (ECG) record. NN intervals are intervals between adjacent QRS complexes resulting from sinus node depolarizations. Also, the instantaneous heart rate can be determined by this method. Time-domain variables can be calculated include the mean NN interval, the mean heart rate, the difference between the longest and the shortest NN interval, the difference between night and day heart rate, and others. The variations in instantaneous heart rate secondary to respiration, tilt, Valsalva maneuver, can be also used. Two methods are used in time-domain analysis: statistical methods and geometrical methods.  • Statistical measures can be calculated from the series of instantaneous heart rate or cycle intervals recorded over 24 hours. There are two classes of measures: (i) those derived from direct measurements of NN intervals or instantaneous heart rate; (ii) those derived from the differences between NN intervals. These variables may be calculated from the analysis of smaller segments of ECG or derived from the total electrocardiographic recording (24 hours).  • Geometrical measures: three general approaches are used in geometric methods: (i) a basic measurement of the geometric pattern is converted into the measure of HRV; (ii) the geometric pattern is interpolated by a mathematically defined shape and then the parameters of this mathematical shape are used; and (iii) the geometric shape is classified into several pattern-based categories which represent different classes of HRV. 267  Selected time-domain measures of HRV are listed in the following Table E.1. Approximate correspondence of time-domain and frequency-domain measures applied to 24-Hour ECG recordings is listed in Table E.2.   268  Table E.1: Selected Time-domain measures of HRV (Task Force of ESC, 1996)     Table E.2: Approximate correspondence of Time-domain and Frequency-domain measures applied to 24-Hour ECG recordings (Task Force of ESC, 1996)  269  ❖ Non-linear analysis (Task Force of ESC, 1996) Non-linear analysis is determined by complex interactions of hemodynamic, electrophysiological and humoral variables, as well as by autonomic and central nervous regulations. It has been speculated that analysis of HRV based on the methods of non-linear dynamics might elicit valuable information for the physiological interpretation of HRV and for the assessment of the risk of sudden death. The parameters which have been used to measure non-linear properties of HRV include 1/f scaling of Fourier spectra, H scaling exponent, and Coarse Graining Spectral Analysis (CGSA). For data representation, Poincarè plot, low-dimension attractor plots, singular value decomposition, and attractor trajectories have been used. For other quantitative descriptions, the correlation dimension, Lyapunov exponents, and entropy have been employed. 270  Appendix F  Acqknowledge Software https://www.biopac.com/product/acqknowledge-software/ 271  Appendix G  Perceived Stress Scale   272  273  Appendix H  Definition of the Outcome Events  The definitions of myocardial infarction, unstable angina, and vascular death are based on the Standardized Draft on Definitions for Cardiovascular and Stroke Endpoint Events in Clinical Trials (Hicks et al., 2010, 2014, 2015; Thygesen et al., 2012).   ❖ Myocardial Infarction • Criteria for acute myocardial infarction  The term acute MI should be used when there is evidence of myocardial necrosis in a clinical setting consistent with acute myocardial ischemia. Under these conditions any one of the following criteria meets the diagnosis for MI: - Detection of a rise and/or fall of cardiac biomarker values (preferably cardiac troponin [cTn]) with at least one value above the 99th percentile upper reference limit (URL) and with at least one of the following: o Symptoms of ischemia. o New or presumed new significant ST-segment-T wave (ST-T) changes or new left bundle branch block (LBBB) o Development of pathological Q waves in the ECG o Imaging evidence of new loss of viable myocardium or new regional wall motion abnormality o Identification of an intracoronary thrombus by angiography or autopsy - Cardiac death with symptoms suggestive of myocardial ischemia and presumed new ischemic ECG changes or new LBBB, but death occurred before cardiac biomarkers were obtained, or before cardiac biomarker values would be increased. 274  - Percutaneous Coronary Intervention (PCI)-related MI is arbitrarily defined by elevation of cTn values (>5 * 99th percentile URL) in patients with normal baseline values (≤99th percentile URL) or a rise of cTn values >20% if the baseline values are elevated and are stable or falling. In addition, either (i) symptoms suggestive of myocardial ischemia or (ii) new ischemic ECG changes or (iii) angiographic findings consistent with a procedural complication, or (iv) imaging demonstration of new loss of viable myocardium or new regional wall motion abnormality are required. - Stent thrombosis associated with MI when detected by coronary angiography or autopsy in the setting of myocardial ischemia and with a rise and/or fall of cardiac biomarker values with at least one value above the 99th percentile URL. - Coronary Artery Bypass Grafting (CABG)- related MI is arbitrarily defined by elevation of cardiac biomarker values (>10 * 99th percentile URL) in patients with normal baseline cTn values (≤99th percentile URL). In addition, either (i) new pathological Q waves or new LBBB, or (ii) angiographic documented new graft or new native coronary artery occlusion, or (iii) imaging evidence of new loss of viable myocardium or new regional wall motion abnormality • Criteria for prior myocardial infarction Any one of the following criteria meets the diagnosis for prior MI: - Pathological Q waves with or without symptoms in the absence of non-ischemic causes - Imaging evidence of a region of loss of viable myocardium that is thinned and fails to contract, in the absence of a non-ischemic cause - Pathological findings of a prior MI 275  ❖ Unstable angina Unstable angina can be diagnosed if matching all the following criteria: • Symptoms of myocardial ischemia at rest (chest pain or equivalent) or an accelerating pattern of angina with frequent episodes associated with progressively decreased exercise capacity • At least one of the following: - New or worsening ST or T wave changes on resting ECG o Transient ST elevation (duration < 20 minutes) New ST elevation at the J point in two anatomically contiguous leads with the cut-off points: ≥ 0.2 mV in men (> 0.25 mV in men < 40 years) or ≥ 0.15 mV in women in leads V2-V3 and/or ≥ 0.1 mV in other leads. o ST depression and T-wave changes New horizontal or down-sloping ST depression ≥ 0.05 mV in two contiguous leads; and/or new T inversion ≥ 0.1 mV in two contiguous leads. The above ECG criteria illustrate patterns consistent with myocardial ischemia. It is recognized that lesser ECG abnormalities may represent an ischemic response and may be accepted under the category of abnormal ECG findings. - Definite evidence of myocardial ischemia on myocardial scintigraphy (clear reversible perfusion defect), stress echocardiography (reversible wall motion abnormality), or MRI (myocardial perfusion deficit under pharmacologic stress) that is believed to be responsible for the myocardial ischemic symptoms/signs. - Angiographic evidence of ≥ 70% lesion and/or thrombus in an epicardial coronary artery that is believed to be responsible for the myocardial ischemic symptoms/signs 276  - Need for coronary revascularization procedure (PCI or CABG) during the same hospital stay. This criterion would be fulfilled if the admission for myocardial ischemia led to transfer to another institution for the revascularization procedure without interceding home discharge • No evidence of acute myocardial infarction  ❖ Cerebro- and cardio-vascular Death  Cardiovascular death includes sudden cardiac death, death due to acute MI, death due to heart failure, death due to stroke, death due to other cardiovascular causes (for example, dysrhythmia unrelated to sudden cardiac death, pulmonary embolism, cardiovascular intervention, aortic aneurysm rupture, or peripheral artery disease), and deaths for which there was no clearly documented non-cardiovascular cause (presumed CV death). Death due to intracranial hemorrhage (including fatal hemorrhagic stroke) will be considered CV death. Additionally, CV deaths will be sub-classified by coronary heart disease (CHD) death and non-CHD death. CHD death includes sudden cardiac death, death due to acute MI, and the subset of death due to other cardiovascular causes that are secondary to a coronary revascularization procedure. 277  Appendix I  Analyses of LF and VLF   LF is mediated by a complex mixture of SNS and PNS modulation. The physiological underpinnings of the VLF component are less well defined as it represents the influences of the peripheral vasomotor and renin-angiotensin systems, thermoregulation, hormonal changes, as well as other uncontrolled factors (Berntson et al., 1997; Kleiger et al., 2005; Sollers et al., 2002; Task Force of ESC, 1996; Taylor et al., 1998). However, despite some ambiguity as to the precise underlying mechanisms of VLF and LF oscillations in HRV, they are used indices in cardiovascular research and have reasonable prognostic relationships with overall cardiovascular morbidity and mortality (Bigger et al., 1992).  The comparisons of LF and VLF between patients with outcome event and patients with event-free are shown in Table I.1. Predictive models based on LF and VLF are shown in Table I.2.   Table I.1 Comparisons of LF and VLF between event group and event-free group HRV Parameters Patients  p value   Total, N=201 Event group, N=36 Event-free group, N=165   HRV absolute values (ms2)            Morning LF value  141.8  (75.3, 226.4) 120.4 (65.8, 187.4) 142.6 (76.9, 226.7) 0.30     Afternoon LF value 128.1 (61.9, 219.6) 108.3 (56.0, 151.0) 131.9 (63.9, 245.5) 0.08     Day LF value  127.7 (72.7, 225.8) 103.6 (76.3, 199.8) 139.2 (68.9, 238.4) 0.15     Night LF value 157.9 (92.1, 280.2) 151.1 (92.0, 219.7) 162.5 (92.1, 286.0) 0.39     24-h LF value  157.4 (89.6, 256.7) 133.5 (89.3, 206.4) 161.7 (90.6, 266.5) 0.18      Morning VLF value  363.9 (235.0, 512.0) 316.2 (224.9, 523.6) 371.6 (242.9, 506.4) 0.51     Afternoon VLF value 341.0 (216.9, 561.8) 311.7 (188.6, 547.6) 341.0 (230.4, 561.8) 0.35     Day VLF value  358.9 (234.8, 513.7) 296.1 (225.0, 515.4)  365.6 (235.0, 513.7) 0.36     Night VLF value 627.3 (354.9, 924.6) 507.6 (289.9, 693.4) 645.5 (371.2, 996.3) 0.08     24-h VLF value  488.7 (340.9, 718.6) 444.2 (301.5, 606.6) 498.1 (349.2, 742.2) 0.12  278  Table I.1 Continuous   HRV Parameters Patients  p value  Total, N=201 Event group, N=36 Event-free group, N=165  HRV changes (%),             Daytime LF changes  -12.5% (-33.8%, 28.2%) -23.4% (-43.4%, 6.5%) -7.7% (-31.0%, 30.7%) 0.06     Day-night LF changes  23.2% (-21.3%, 78.2%) 29.2% (-11.5%, 123.9%) 22.8% (-21.9%, 67.2%) 0.34     Morning-night LF changes  16.8% (-29.3%, 91.7%) 12.5% (-29.5%, 114.0%) 17.5% (-29.3%, 78.3%) 0.86              Daytime VLF changes  0.3% (-28.6%, 29.3%) -17.6% (-31.8%, 5.9%) 3.2% (-25.2%, 30.4%) 0.06     Day-night VLF changes  73.1% (13.8%, 151.9%) 76.7% (16.2%, 124.9%) 72.5% (13.8%, 159.1%) 0.54     Morning-night VLF changes  70.6% (7.7%, 149.7%) 51.6% (8.5%, 126.6%) 74.6% (7.7%, 159.5%) 0.23  Data are shown as median (IQR) LF: low frequency; VLF: very low frequency; Definitions of HRV variables are provided in section 6.5.3.2;  OR: odds ratio; CI: confident interval; AUC: area under the curve; p value for Wilcoxon rank-sum test; IQR: interquartile range  279  Table I.2 Predictive models based on LF and VLF Models  Odds ratio for HRV parameters AUC  Odds ratio (95% CI) p value  Models with HRV alone      HRV values (natural logarithmic transformed)         Morning LF value 0.80 (0.53, 1.21) 0.29 0.56     Day LF value 0.75 (0.49, 1.12) 0.16 0.58     Night LF value 0.83 (0.55, 1.25) 0.37 0.55     24-h LF value 0.76 (0.48, 1.18) 0.22 0.57     Morning VLF value 0.84 (0.49, 1.42) 0.51 0.54     Day VLF value 0.72 (0.42, 1.24) 0.24 0.55     Night VLF value 0.64 (0.39, 1.04) 0.08 0.59     24-h VLF value 0.59 (0.33, 1.03) 0.07 0.58     HRV changes †        Daytime LF changes 0.97 (0.90, 1.03) 0.44 0.60     Day-night LF changes 1.00 (0.98, 1.01) 0.93 0.55     Morning-night LF changes 1.00 (0.98, 1.01) 0.92 0.49     Daytime VLF changes  0.95 (0.88, 1.02) 0.21 0.61     Day-night VLF changes  0.98 (0.95, 1.01) 0.23 0.47     Morning-night VLF changes  0.97 (0.94, 1.00) 0.10 0.56     Models with HRV + PSS + ABCD2 score      HRV values (natural logarithmic transformed)         Morning LF value + PSS + ABCD2 0.88 (0.57, 1.36) 0.56 0.74     Day LF value + PSS + ABCD2 0.81 (0.53, 1.25) 0.35 0.74     Night LF value + PSS + ABCD2 0.84 (0.55, 1.29) 0.43 0.75     24-h LF value + PSS + ABCD2 0.80 (0.50, 1.27) 0.34 0.75     Morning VLF value + PSS + ABCD2 0.85 (0.49, 1.49) 0.58 0.74     Day VLF value + PSS + ABCD2 0.73 (0.41, 1.28) 0.27 0.74  280  Table I.2 Continuous  Models  Odds ratio for HRV parameters AUC  Odds ratio (95% CI) p value      Night VLF value + PSS + ABCD2 0.61 (0.36, 1.02) 0.06 0.76     24-h VLF value + PSS + ABCD2 0.73 (0.41, 1.28) 0.06 0.75       HRV changes         Daytime LF change + PSS + ABCD2  0.97 (0.91, 1.03) 0.40 0.75     Day-night LF changes + PSS + ABCD2 1.00 (0.97, 1.01) 0.97 0.74     Morning-night LF changes + PSS + ABCD2 1.00 (0.98, 1.01) 0.90 0.74     Daytime VLF change + PSS + ABCD2  0.95 (0.87, 1.02) 0.20 0.76     Day-night VLF changes + PSS + ABCD2 0.98 (0.94, 1.01) 0.19 0.76     Morning-night VLF changes + PSS + ABCD2 0.97 (1.94, 1.00) 0.06 0.77  LF: low frequency; VLF: very low frequency; Definitions of HRV variables are provided in section 6.5.3.2;  AUC: area under the curve; CI: confident interval; p value for Wilcoxon rank-sum test;   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0357024/manifest

Comment

Related Items