Drugs and pacemakers for vasovagal, carotid sinus and situational syncope (Review)

Transcription

1 Drugs and pacemakers for vasovagal, carotid sinus and situational syncope (Review) Romme JJCM, Reitsma JB, Black CN, Colman N, Scholten RJPM, Wieling W, Van Dijk N This is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library 2013, Issue 5

2 T A B L E O F C O N T E N T S HEADER ABSTRACT PLAIN LANGUAGE SUMMARY SUMMARY OF FINDINGS FOR THE MAIN COMPARISON BACKGROUND OBJECTIVES METHODS RESULTS Figure Figure Figure ADDITIONAL SUMMARY OF FINDINGS DISCUSSION AUTHORS CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES CHARACTERISTICS OF STUDIES DATA AND ANALYSES Analysis 1.1. Comparison 1 Vasovagal syncope. Beta-blockers vs. standard treatment., Outcome 1 Proportion of patients with one or more syncopal episode(s) during follow-up Analysis 1.2. Comparison 1 Vasovagal syncope. Beta-blockers vs. standard treatment., Outcome 2 Proportion of patients with syncopal episodes induced by provocation testing Analysis 2.1. Comparison 2 Vasovagal syncope. Beta-blockers vs. placebo., Outcome 1 Proportion of patients with one or more syncopal episode(s) during follow-up Analysis 2.2. Comparison 2 Vasovagal syncope. Beta-blockers vs. placebo., Outcome 2 Proportion of patients with syncopal episodes induced by provocation testing Analysis 2.3. Comparison 2 Vasovagal syncope. Beta-blockers vs. placebo., Outcome 3 Proportion of patients with side effects Analysis 3.1. Comparison 3 Vasovagal syncope. Alpha-adrenergic agents vs. placebo., Outcome 1 Proportion of patients with syncopal episodes induced by provocation testing. Treatment initiation before provocation testing Analysis 3.2. Comparison 3 Vasovagal syncope. Alpha-adrenergic agents vs. placebo., Outcome 2 Proportion of patients with syncopal episodes induced by provocation testing. Treatment initiation before provocation testing Analysis 3.3. Comparison 3 Vasovagal syncope. Alpha-adrenergic agents vs. placebo., Outcome 3 Proportion of patients with syncopal episodes induced by provocation testing. Treatment initiation after the occurrence of symptoms Analysis 3.4. Comparison 3 Vasovagal syncope. Alpha-adrenergic agents vs. placebo., Outcome 4 Proportion of patients with side effects Analysis 4.1. Comparison 4 Vasovagal syncope. Selective serotonin reuptake inhibitors vs. placebo., Outcome 1 Proportion of patients with syncopal episodes induced by provocation testing Analysis 4.2. Comparison 4 Vasovagal syncope. Selective serotonin reuptake inhibitors vs. placebo., Outcome 2 Proportion of patients with one or more syncopal episode(s) during follow-up Analysis 5.1. Comparison 5 Vasovagal syncope. ACE-inhibitors vs. placebo., Outcome 1 Proportion of patients with syncopal episodes induced by provocation testing Analysis 6.1. Comparison 6 Vasovagal syncope. Anticholinergic agents vs. placebo., Outcome 1 Proportion of patients with syncopal episodes induced by provocation testing. Treatment initiation after the occurrence of symptoms Analysis 7.1. Comparison 7 Vasovagal syncope. DDD-pacemaker vs. conventional standard therapy., Outcome 1 Proportion of patients with one or more syncopal episode(s) during follow-up Analysis 7.2. Comparison 7 Vasovagal syncope. DDD-pacemaker vs. conventional standard therapy., Outcome 2 Average time elapsed from start of an intervention until the first episode of syncope Analysis 8.1. Comparison 8 Vasovagal syncope. Active DDD-pacemaker vs. non-active pacemaker.., Outcome 1 Proportion of patients with one or more syncopal episode(s) during follow-up i

3 Analysis 8.2. Comparison 8 Vasovagal syncope. Active DDD-pacemaker vs. non-active pacemaker.., Outcome 2 Proportion of patients with minor physical trauma due to syncope Analysis 8.3. Comparison 8 Vasovagal syncope. Active DDD-pacemaker vs. non-active pacemaker.., Outcome 3 Proportion of patients with side effects Analysis 9.1. Comparison 9 Vasovagal syncope. DDD-pacemaker vs. DDI-pacemaker., Outcome 1 Proportion of patients with one or more syncopal episode(s) during follow-up ADDITIONAL TABLES APPENDICES WHAT S NEW CONTRIBUTIONS OF AUTHORS DECLARATIONS OF INTEREST SOURCES OF SUPPORT DIFFERENCES BETWEEN PROTOCOL AND REVIEW NOTES INDEX TERMS ii

4 [Intervention Review] Drugs and pacemakers for vasovagal, carotid sinus and situational syncope Jacobus JCM Romme 1, Johannes B Reitsma 2, Catherine N Black 1, Nancy Colman 3, Rob JPM Scholten 4, Wouter Wieling 5, Nynke Van Dijk 6 1 Department of Clinical Epidemiology, Biostatistics and Bioinformatics, Academic Medical Center, Amsterdam, Netherlands. 2 Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht, Netherlands. 3 Department of Cardiology, Academic Medical Center, Amsterdam, Netherlands. 4 Dutch Cochrane Centre, Academic Medical Center, Amsterdam, Netherlands. 5 Department of Internal Medicine, Academic Medical Center, Amsterdam, Netherlands. 6 Department of General Practice / Family Medicine, Academic Medical Center, Amsterdam, Netherlands Contact address: Nynke Van Dijk, Department of General Practice / Family Medicine, Academic Medical Center, Meibergdreef 9, Amsterdam, 1105 AZ, Netherlands. n.vandijk@amc.uva.nl. Editorial group: Cochrane Heart Group. Publication status and date: Stable (no update expected for reasons given in What s new ), published in Issue 5, Review content assessed as up-to-date: 21 August Citation: Romme JJCM, Reitsma JB, Black CN, Colman N, Scholten RJPM, Wieling W, Van Dijk N. Drugs and pacemakers for vasovagal, carotid sinus and situational syncope. Cochrane Database of Systematic Reviews 2011, Issue 10. Art. No.: CD DOI: / CD pub3. Background A B S T R A C T Neurally mediated reflex syncope is the most common cause of transient loss of consciousness. In patients not responding to nonpharmacological treatment, pharmacological or pacemaker treatment might be considered. Objectives To examine the effects of pharmacological therapy and pacemaker implantation in patients with vasovagal syncope, carotid sinus syncope and situational syncope. Search methods We searched the Cochrane Central Register of Controlled Trials (CENTRAL) on The Cochrane Library (Issue 1, 2008), PubMed (1950 until February 2008), EMBASE on OVID (1980 until February 2008) and CINAHL on EBSCOhost (1937 until February 2008). No language restrictions were applied. Selection criteria We included parallel randomized controlled trials and randomized cross-over trials of pharmacological treatment (beta-blockers, fludrocortisone, alpha-adrenergic agonists, selective serotonine reuptake inhibitors, ACE inhibitors, disopyramide, anticholinergic agents or salt tablets) or dual chamber pacemaker treatment. Studies were included if pharmacological or pacemaker treatment was compared with any form of standardised control treatment (standard treatment), placebo treatment, or (other) pharmacological or pacemaker treatment. We did not include non-randomized studies. 1

5 Data collection and analysis Two reviewers independently assessed the risk of bias. Using a standardised data extraction form, they extracted characteristics and results of the various studies. In a consensus meeting they discussed any disagreements that had occurred during data extraction. If no agreement could be reached, a third reviewer was asked to make a decision. Summary estimates with 95% confidence intervals of treatment effect were calculated using relative risks, rate ratios or weighted means differences depending on the type of outcome reported. Main results We included 46 randomized studies, 40 on vasovagal syncope and six on carotid sinus syncope. No studies on situational syncope matched the criteria for inclusion in our review. Studies in general were small with a median sample size of 42. A wide range of control treatments were used with 22 studies using a placebo arm. Blinding of patients and treating physicians was applied in eight studies. Results varied considerably between studies and between types of outcomes. For vasovagal syncope, the occurrence of syncope upon provocational head-up tilt testing was lower upon treatment with beta-blockers, ACE-inhibitors and anticholinergic agents compared to standard treatment. For carotid sinus syncope, the occurrence of syncope upon carotid sinus massage was lower on midodrine treatment compared to placebo treatment in one study. Authors conclusions There is insufficient evidence to support the use of any of the pharmacological or pacemaker treatments for vasovagal syncope and carotid sinus syncope. Larger studies using patient relevant outcomes are needed. P L A I N L A N G U A G E S U M M A R Y Drugs and pacemakers for transient loss of consciousness Neurally mediated reflex syncope (including fainting) is the most common cause of transient loss of consciousness. It is caused by a sudden decrease in blood pressure and/or lowering of heart rate. The main treatment goal therefore is to increase blood pressure and heart rate. In most patients, this can be achieved by non-pharmacological treatment measures (e.g. adequate fluid and salt intake, physical counterpressure manoeuvres). In patients not responding to this treatment, pharmacological or pacemaker treatment might be considered. We investigated the effectiveness of these treatments for different subtypes of neurally mediated reflex syncope, namely vasovagal syncope (fainting), carotid sinus syncope (fainting due to pressure on the neck) and situational syncope (fainting when passing urine of faeces or swallowing). Where data were available, we determined the treatment effectiveness for different outcome measures including occurrence of syncope, amount of (pre-)syncopes per year during follow-up and quality of life. We included 46 randomized studies, 40 on vasovagal syncope and six on carotid sinus syncope. No studies on situational syncope matched the criteria for inclusion in our review. Studies in general were small with a median size of 42 patients. A wide range of control treatments were used with 22 studies using placebo control treatment. Blinding of patients and treating physicians was applied in eight studies. The type of outcomes reported by studies varied considerably with only 16 studies reporting on non-provoked recurrences during follow-up. As a consequence of all these differences, results varied considerably between studies and between types of outcomes. In some studies significant results were reported for one type of outcome, but not for other outcomes. We conclude that there is insufficient evidence either to support or to refute the use of any of the pharmacological or pacemaker treatments for vasovagal syncope and carotid sinus syncope. 2

6 S U M M A R Y O F F I N D I N G S F O R T H E M A I N C O M P A R I S O N [Explanation] Comparison treatment Syncope burden Proportion of patients with syncope during follow-up Time to syncopal recurrence Pre-syncope burden Proportion of patients with syncope upon headup tilt testing Beta-blockers vs. standard treatment Riskratio0.36(95%CI )(3 studies, 129 patients)(guan 1999; Lin 2000; Ventura 2002) 0 vs. 4.8 episodes/year (p-value not reported) (1 study, 32 patients)(lin 2000) Risk ratio 0.75 (95% CI ) (2 studies, 145 patients)(di Girolamo 1998; Lin 2000) Note:heterogeneous treatment effects(p-valuechi 2 -test= 0.02;I 2 =82%) Beta-blockers vs. placebo Median 2 vs. 0(p= 0.22) (1 study, 50 patients) Rateratio1.33(95%CI )(1study,68 patients)(madrid 2001) Risk ratio 1.07 (95% CI )(3 studies, 320 participants)(madrid 2001; Sheldon 2006; Theodorakis 2006) Median 1.1 vs pre-syncope/year(p= 0.35)(1 study, 208 patients)(sheldon 2006) Rate ratio 0.92 (95% CI )(1 study, 62 patients)(theodorakis 2006) Odds ratio 0.36 (95% CI )(5 studies, 173 patients)(eldadah 2006; Flevari 2002; Kluger 1998; Mahanonda 1995; Theodorakis 2006) Alpha-adrenergic agonists vs. standard treatment. Rate ratio 3.20 (95% CI )(1 study, 61 patients)(perez-lugones 2001) Risk ratio 0.22 (95% CI )(1 study, 61 patients)(perez-lugones 2001) Hazard ratio 0.31(95% CI )(1study,61 patients)(perez-lugones 2001) Risk ratio 0.99 (95% CI )(1 study, 113 patients)(di Girolamo 1998) Alpha-adrenergic agonists vs. placebo Riskratio1.00(95%CI )(1 study, 126 patients)(raviele 1999) Hazard ratio 1.00(95% CI )(1 study, 126 patients)(raviele 1999) Rate ratio 1.47 (95% CI )(1 study, 126 patients)(raviele 1999) I. Treatment initiation before provocational headup tilt test. Odds ratio 0.54 (95% CI )(5 studies, 292 patients)(di Girolamo 1998; Kaufmann 2002; Moya 1995; Raviele Proportion of patients with side effects Side effects occurred in intervention group. Control group: not assessed. (2 studies, 81 patients) (Guan 1999; Lin 2000) 12vs. 7(p-valuenot reported)(1 study, 208 patients)(sheldon 2006) Odds ratio 5.31 (95% CI )(2 studies, 80 patients)(flevari 2002; Madrid 2001) At least one patient with side effects in both treatmentgroups(1study,30 patients)(moya 1995) Odds ratio 1.60 (95% CI )(3 studies, 220 patients)(ammirati 2000; Raviele 1999; 3

7 Selective serotonin reuptake inhibitors vs. placebo Rate ratio 0.65 (95% CI )(1 study, 63 patients)(theodorakis 2006) Risk ratio 0.39 (95% CI )(2 studies, 131 patients)(di Girolamo 1999; Theodorakis 2006) ACE-inhibitors vs. standard treatment Riskratio2.00(95%CI )(1 study, 32 patients)(lin 2000) 1999; Ward 1998). Subgroup Etilefrine treatment: odds ratio 0.94 (95% CI ) (3 studies, 264 patients) (Di Girolamo 1998; Moya 1995; Raviele 1999). Subgroup midodrine treatment: odds ratio (95% CI ) (2 studies, 28 patients) (Kaufmann 2002; Ward 1998) II. Treatment initiation upon symptom occurrence during provocational head-up tilt test. Odds ratio 0.02 (95% CI )(2 studies, 105 patients)(ammirati 2000; Raviele 2005) Ward 1998) Rate ratio 0.16 (95% CI )(1 study, 63 patients)(theodorakis 2006) Risk ratio 0.75 (95% CI )(2 studies, 131 patients)(di Girolamo 1999; Theodorakis 2006) Risk ratio % CI )(1 study, 68 patients)(di Girolamo 1999) Rate ratio 0.67 (95% CI )(1 study, 32 patients)(lin 2000) Risk ratio 0.54 (95% CI )(1 study, 32 patients)(lin 2000) Intervention group: 6/16 (38%) Control group: not reported. (1 study, 32 patients)(lin 2000) 4

8 ACE-inhibitors vs. placebo Disopyramide vs. placebo Anticholinergic agents vs. standard treatment Anticholinergic agents vs. placebo Fludrocortison vs. any treatment DDD-pacemaker vs. standard treatment Rate ratio 1.27 (95% CI )(1 study, 42 patients)(sutton 2000) No occurrence of syncope in both treatment groups (1 study, 22 patients)(morillo 1993) Risk ratio 0.42 (95% CI )(1 study, 45 patients)(guan 1999) Riskratio0.20(95%CI )(3 studies, 116 patients)(connolly 1999; Flammang 1999; Sutton Hazard ratio 0.17(95% CI ) (3 studies, 116 patients)(connolly 1999; Flammang 1999; Rate ratio 0.45 (95% CI )(1 study, 54 patients)(connolly 1999) Risk ratio 0.04 (95% CI )(2 studies, 72 patients)(zeng 1998a; Zeng 1998b) Risk ratio 1.50 (95% CI )(1 study, 22 patients)(morillo 1993) Treatment initiation upon symptom occurrence during provocational head-up tilt test. Risk ratio 0.66 (95% CI )(2 studies, 142 patients)(lee 1996; Santini 1999) Note: heterogeneous treatment effects(p-value Chi 2 -test< ;i 2 = 94%) Risk ratio 0.96 (95% CI )(1 study, 42 patients)(sutton 2000) Riskratio7.00(95%CI )(1study,48 patients)(zeng 1998b) Intervention group: 5/20 (25%) Control group: not reported. (1 study, 45 patients) (Guan 1999) One patient(1/ 28(4%)) stopped scopolamine treatment ahead of time because of side effects (1 study, 58 patients)(lee 1996) Risk ratio (95% CI )(1 study, 54 patients)(connolly 1999) 5

9 2000) Sutton 2000) DDD-pacemaker versus non-active pacemaker Riskratio0.89(95%CI )(2 studies, 129 patients)(connolly 2003; Raviele 2004) Hazard ratio 0.73(95% CI )(1 study, 29 patients)(raviele 2004) 13 vs. 16 per 100 days (p-value not reported) (1 study, 100 patients) (Connolly 2003) DDD-pacemaker vs. DDIpacemaker 0vs.1.2syncopes(SD0. 8) during 18 months of follow-up(1 study, 25 patients)(occhetta 2004) Riskratio0.07(95%CI )(3 studies, 69 patients)(ammirati 1998; Deharo 2003; Occhetta 2004) Hazard ratio 0.05(95% CI )(1 study, 26 patients)(occhetta 2004) Rate ratio 0.50 (95% CI )(1 study, 23 patients)(deharo 2003) Risk ratio 0.40 (95% CI )(1 study, 20 patients)(ammirati 1998) Riskratio1.77(95%CI )(2 studies, 129 patients)(connolly 2003; Raviele 2004) 6

10 B A C K G R O U N D In neurally mediated reflex syncope a reflex is triggered that induces vasodilatation and/or bradycardia, resulting in a systemic hypotension and cerebral hypoperfusion (Moya 2009). It is by far the most common cause of transient loss of consciousness both in general practice and in clinical settings (Colman 2004; Soteriades 2002). The three main types of neurally mediated reflex syncope are vasovagal syncope, carotid sinus syncope and situational syncope (Moya 2009). Vasovagal syncope is characterised by premonitory signs and provoked by triggering events such as prolonged standing, emotion or pain (Moya 2009). Carotid sinus syncope is caused by stimulation of the carotid sinus baroreceptors, which may lead to hypotension or bradycardia or both. Situational syncope is defined as loss of consciousness at the onset of, during or directly after swallowing, defecation, micturition or coughing. The life-time prevalence of neurally mediated reflex syncope in the general population ranges between 30 and 50% for people up to 60 years of age (Ganzeboom 2003; Ganzeboom 2006). Vasovagal syncope accounts for the vast majority (75%) of episodes of reflex syncope in patients under 60 years of age that come to medical attention (Linzer 1997; Van Dijk 2008). Twenty-one percent of reflex syncopal episodes are caused by situational syncope and about 4% by carotid sinus syncope (Linzer 1997). Although the prognosis of neurally mediated reflex syncope is usually benign, it can be disabling and have a profound effect on quality of life (Van Dijk 2006a). Treatment of neurally mediated reflex syncope is aimed at ensuring an adequate blood supply to the brain (Moya 2009). Treatments options might however vary considerably among different types of neurally mediated reflex syncope. Most patients with a diagnosis of vasovagal syncope and situational syncope can be treated by counselling and advice. Avoiding or ameliorating the triggering event is a crucial treatment component for situational syncope, but is also applicable to patients with vasovagal syncope (Moya 2009; Wieling 2004). For both conditions, recognition of premonitory signs are also important as well as increasing fluid and salt intake (Wieling 2004). Patients with recognizable premonitory signs and symptoms are instructed to apply counterpressure manoeuvres (e.g. leg-crossing, muscle-tensing) (Van Dijk 2006b). Tilttraining has been put forward to prevent the recurrence of neurally mediated syncope (Ector 1998). Recent studies indicate that this intervention might only be effective in highly motivated patients (Foglia-Manzillo 2004; Gurevitz 2007), limiting widespread use of this intervention. The treatment of carotid sinus syncope depends upon the nature of its condition (Healey 2004). Pharmacological treatment is advocated for the vasodepressor form; pacemaker treatment for mixed and bradycardic forms. Pharmacological and pacemaker treatment can also be considered for vasovagal syncope and situational syncope upon failure of the aforementioned non-pharmacological treatment measures (Brignole 2004). Many pharmacological interventions have been proposed for treating neurally mediated reflex syncope but the evidence to suggest their effectiveness is limited to small studies, sometimes showing conflicting results. Medications used to prevent syncope include beta-adrenergic blockers, alpha-adrenergic agonists (e.g. etilephrine, midodrine), ACE-inhibitors, anticholinergic agents, anti-arrhythmic agents (e.g. disopyramide), serotonine re-uptake inhibitors (e.g. paroxetine) and medication for fluid retention, such as salt tablets and fludrocortisone (Benditt 1999; Calkins 1999; Moya 2009). The (presumed) mechanisms of actions of these pharmacological agents are discussed below (see Appendix 1). Since it is unknown whether salt tablets are effective to prevent neurally-mediated reflex syncope in addition to changes in lifestyle, we considered salt tablets as pharmacological treatment in our review. Pacemakers are also sometimes used in cases of neurally mediated reflex syncope with severe bradycardia or asystole (Kuriachan 2008). Apart from the clinical effectiveness, it is also important to address the issue of cost-effectiveness of pharmacological and pacemaker treatment for neurally mediated reflex syncope. In the Netherlands, the average treatment costs per year vary between EUR 6 and 2614 (USD 8 and 3570) for the suggested pharmacological treatment measures (College 2009). The costs associated with the implantation of a dual-chamber pacemaker system are about EUR 1145 (USD1560) per year. Costs might however vary considerably across countries and types of health care systems. Though pharmacological and pacemaker treatment are frequently prescribed to large numbers of patients suffering from recurrent neurally-mediated reflex syncope, little is known about the clinical effectiveness of these expensive treatments. Therefore, the aim of this review was to determine the effects of both pharmacotherapy and pacemaker therapy in patients with different forms of neurally mediated reflex syncope. Description of the condition Syncope A temporary loss of consciousness due to inadequate blood flow to the brain (generalised cerebral ischaemia). Vasovagal syncope Vasovagal syncope is diagnosed if precipitating events such as fear, severe pain, emotional distress, or prolonged standing are associated with prodromal symptoms (e.g. lightheadedness, sweating, 7

11 feeling nauseous) or when a vasovagal response is observed during tilt-table testing (Brignole 2004; Moya 2009). A vasovagal response is defined as a sudden drop of blood pressure, heart rate or both in association with syncope or near syncope. Carotid sinus syncope The diagnosis of carotid sinus syncope is based on either a typical history (spontaneous carotid sinus syncope as a consequence of pressure in the carotid area or turning of the head) or on the reaction to carotid sinus massage (induced carotid sinus syncope) (Brignole 2004). Situational syncope A typical history of situational syncope is defined as loss of consciousness occurring during or immediately after micturition, defaecation, coughing or swallowing (Brignole 2004; Moya 2009). Types of participants We included studies in which participants above 16 years of age were diagnosed with neurally mediated reflex syncope. The three types of neurally mediated reflex syncope that were addressed in this review are vasovagal syncope, carotid sinus syncope and situational syncope. Studies also reporting on patients with other causes of syncope like autonomic failure, medications inducing transient loss of consciousness, cardiac rhythm disturbance (e.g. complete heart block, ventricular tachycardia) or structural heart disease were included only if the results of patients with neurally mediated reflex syncope were reported separately. Furthermore, studies including patients with a transient ischaemic attack (TIA) or epilepsy were excluded. Types of interventions Why it is important to do this review Many pharmacological and pacemaker interventions have been implemented in clinical practice in order to reduce (pre-)syncopal recurrence in patients with neurally mediated reflex syncope. Though here is much debate on the effectiveness of these interventions, a systematic review has not been forthcoming. This Cochrane systematic review presents evidence which may help facilitate decision making for physicians and other health care workers involved in treating patients with neurally mediated reflex syncope. Moreover, it might lead to the development of new and better studies aiming to improve treatment for neurally mediated reflex syncope. Pharmacotherapy The following drugs were considered for inclusion in our review: beta-blockers (both selective and non-selective), fludrocortisone, alpha-adrenergic agonists, selective serotonine reuptake inhibitors (SSRIs), ACE-inhibitors, disopyramide, anticholinergic agents and salt tablets. The presumed mechanisms of action of these pharmacological interventions are stated in Appendix 1. Studies were included if drugs were compared with standard treatment, placebo treatment, counselling and advice, with any another drug, or with pacemaker therapy. Standard treatment consisted of any treatment physicians would offer to patients not including the intervention under investigation. Pacemaker therapy O B J E C T I V E S To determine the effects, both beneficial and harmful, of pharmacological therapy and/or pacemaker implantation in patients with neurally mediated reflex syncope. M E T H O D S Criteria for considering studies for this review Types of studies We included parallel randomized controlled trials, and randomized cross-over trials in our review. Pacemaker therapy is hypothesized to counteract the cardioinhibitory component of the vasovagal response (Sud 2007). There are different types of pacemakers, which can be classified by a three-letter abbreviation (Bernstein 2002). The first letter stands for the heart chambers paced (Atrial, Ventrical, Dual), the second for the heart chambers sensed (Atrial, Ventrical, Dual, or none (O)), the third for the response (Inhibited, Triggered, Dual (I and T) or none (O)). Sometimes a fourth letter is added to denote the presence or absence of rate modulation. In this review, we included all different types of dual-chamber cardiac pacemakers, either with or without rate hysteresis, or rate drop response sensing, provided that a comparison was made with no therapy or counselling and advice or pharmacotherapy, another type of pacemaker therapy or with pacemaker off. Studies in which concomitant treatment had been applied, were accepted if these treatments were applied in both arms of the trial. We included studies irrespective of the length of follow-up. 8

12 Types of outcome measures The general aim of this review was to determine the effectiveness of pharmacological and pacemaker treatment for neurally-mediated reflex syncope. We extracted data on several clinical outcomes, putting patient relevant outcomes higher in hierarchy. Primary outcomes Average number of episodes per year per patient (syncope burden). Secondary outcomes 1. Proportion of patients with one or more syncopal episode(s) during follow-up. 2. Average time elapsed from start of an intervention until the first episode of syncope. 3. Quality of life, measured with a validated questionnaire. 4. Cumulative incidence of minor and major physical trauma due to syncope. 5. Number and severity of side effects. 6. Average number of episodes of near syncope. 7. Proportion of patients with syncopal episodes induced by provocation testing or carotid sinus massage. Search methods for identification of studies Electronic searches A comprehensive search of the Cochrane Central Register of Controlled Trials (CENTRAL) on The Cochrane Library (Issue 1, 2008), PubMed (1950 until February 2008), EMBASE on OVID (1980 until February 2008), and CINAHL on EBSCOHost (1937 until February 2008) was performed (see Appendix 2). No language restrictions were applied. Searching other resources Reference lists of included studies and review articles were reviewed to identify potential relevant citations. Inquiries regarding other published or unpublished studies known and/or supported by the authors of the primary studies were made so that these results could be included in this review. Finally, personal contact with colleagues, collaborators and other trialists working in the field of neurally mediated syncope was made to identify potentially relevant studies. Data collection and analysis Selection of studies Publications identified from the search were assessed for eligibility for inclusion by two reviewers independent of each other. Firstly, a selection was made based on title. Next, the selected articles were assessed for eligibility based on both title and abstract. If any doubt occurred with respect to an article s relevance, the full text of the reference was obtained. Disagreement was resolved by consensus or third party adjudication. Two reviewers independently, read the full text of retained studies and included trials that met the inclusion criteria. A pre-defined form was used for this task. Articles finally selected for the review were checked to avoid including duplicates. Records of this were kept as advised in the QUOROM statement (Moher 1999). A consensus meeting was used to deal with any disagreements arising in the selection of the articles. If no agreement could be reached, a third reviewer decided whether the paper had to be included. Data extraction and management Characteristics of the various studies were extracted independently by two reviewers according to a standardised data extraction form. Information was collected regarding: 1. Design of study: parallel randomized controlled trial (RCT) or randomized cross-over trial. 2. Characteristics of the patients: age, gender, (lifetime) number of syncopal episodes, number of episodes of syncope in previous year, history of physical trauma due to syncope; setting (primary, secondary, or tertiary care setting) 3. Type, dose and duration of interventions and control interventions. 4. Outcomes as addressed by the original authors. From each study, we extracted data for as many appropriate outcome measures as possible. In case of premature stopping of allocated treatment or loss to follow-up, we extracted data according to an intention-to-treat-principle, if present in the study report. In other cases, we extracted data as presented in the study report. A consensus meeting was used to deal with any disagreements arising in data extraction. If no agreement could be reached, a third reviewer decided which data of a given paper could be extracted. For vasovagal syncope, carotid sinus syncope and situational syncope separate Summary of findings (Sof) tables were created. Assessment of risk of bias in included studies Risk of bias was assessed by two reviewers independent of each other (Altman 2008). Any disagreement between reviewers was resolved by consensus. In case no consensus could be reached a third reviewer assessed the risk of bias of the study. In studies where it was unclear whether the risk of bias criteria had been met, the reviewers tried to contact the first author to get additional information. 9

13 If the risk of bias in an included study was very high according to the Cochrane assessment criteria for risk of bias, we did not proceed to data extraction. Assessment of reporting biases The presence of publication bias was assessed using funnel plots if sufficient studies were retrieved for the primary outcome measure. Data synthesis Since the aetiology of the different subtypes of neurally mediated reflex syncope is quite different (Moya 2009), we presented the studies for vasovagal syncope, carotid sinus syncope and situational syncope in different sections. Within these sections, studies were categorized by pharmacological or pacemaker treatment modality. Next, for each treatment modality, studies were classified depending on the type of control treatment. Thus, all studies were classified at three levels: 1) subtype of neurally mediated reflex syncope, 2) pharmacological or pacemaker treatment modality, 3) treatment comparison. Next, for each study, summary estimates of treatment effect (with 95% confidence intervals (95% CI)) were calculated for each outcome measure provided that study data were available for a particular treatment comparison. For dichotomous outcomes risk ratios (RR) were calculated. For continuous outcomes we calculated the weighted mean difference (WMD) or weighted standardised mean difference (SMD), when appropriate. For data with discrete numbers we determined the rate ratio by dividing the mean number of syncopal episodes by the number of person years of followup. For time-to-event data we calculated the hazard ratio with the help of the P value of the log rank statistic (Tierney 2007). In order to combine a continuous outcome measure of included parallel and cross-over studies we determined the WMD as described by Curtin et al (Curtin 2002b). For the pooling of a dichotomous outcome measure from studies with and without a cross-over design we calculated a combined odds ratio (OR) according to the Becker-Balagtas method (Curtin 2002a; Elbourne 2002). In case the results of cross-over trials were presented as in parallel studies and no additional information was provided by contacting the corresponding authors of these studies, we assumed that there was no correlation between observations on the same patient. This means that such studies received less weight in the analyses because the effect estimate would have been more precise in case of a positive correlation. Results of clinically and statistically homogeneous studies (i.e. studies for which the participants, interventions, outcome measures and timing of the follow-up measurements were considered to be similar) were combined using a fixed effect model. In case of statistical heterogeneity (as assessed by visual inspection of the forest plots, the Q-statistic and the I 2 statistic (Higgins 2003)) and the availability of at least five studies, a random effects model was used. In any other case, no pooling was performed. Subgroup analysis and investigation of heterogeneity If results of two or more studies were available for a particular outcome measure, we presented the results of parallel and crossover studies in different subgroups. If there was little variation in the results of these studies, we determined a combined treatment effect. We did not calculate a combined treatment effect if there was substantial variation between study results. Sensitivity analysis If sufficient studies were retrieved for a certain outcome measure, sensitivity analyses were performed to examine the robustness of the results using the following quality criteria: concealment of allocation (yes/no/unclear), blinded outcome assessment (yes/no/ unclear) and completeness of follow-up (>=80%/ <80%/ unclear). R E S U L T S Description of studies Results of the search With the electronic search strategy we obtained 14,844 publications (3015 in MEDLINE, 540 in CENTRAL, 6640 in CINAHL and 4649 in EMBASE). After removal of duplicates 12,409 publications remained. Based on title, we considered 701 publications eligible. Of these, 91 publications remained after assessment based on title and -if available- abstract. We added 13 publications to this selection based on searching for relevant congress abstracts, important clinical studies, and study protocols (Ahmed 1991; Brignole 2006a; Brignole 2007; Garcia-Alberola 2002; Griebenow 1991; Morley 1983; STAND; Raviele 2004; Sheldon 2006; Stein 1994; Sutton 2000;Takata 2002; Williamson 1994). Fifty-eight of these 104 publications matched the criteria for inclusion in our review. Forty-six of these 58 publications were original study reports (see Included studies) and were subjected to further analysis. Twelve of the included publications were not suitable for further analysis because they were double publications of an already included study report (5), abstracts/preliminary study reports (3), or study protocols (4). Thirty-nine publications were excluded (see Excluded studies). We were unable to classify 4 publications (see Characteristics of studies awaiting classification). This was because we were unable to find Japanese translators for one study (Betsuyaku 1995) or were unable to find the full study report of three references (Ammirati 1999; Elstermann 1981; Ventura 2001). Three studies were still ongoing (Brignole 2007; Raj 2006; STAND)(see Characteristics of ongoing studies). A QUORUM flow diagram represents the selection of publications relevant for our review (see Figure 1). 10

14 Figure 1. QUORUM statement flow diagram 11

15 Included studies Of the 46 included original publications (see Characteristics of included studies), 42 were found in MEDLINE, two were identified in CENTRAL, one in EMBASE and one in CINAHL. Fortytwo studies were written in English, three in Chinese and one in Italian. We found 40 studies on vasovagal syncope, six studies on carotid sinus syncope and none on situational syncope. Of the included studies, 32 were parallel randomized controlled trials, the 14 remaining studies were randomized cross-over trials (Biffi 1997; Brignole 1988; Deharo 2003; El-Bedawi 1994; Eldadah 2006; Flevari 2002; Kaufmann 2002; Kurbaan 2000a; McIntosh 1997; Moore 2005; Morillo 1993; Moya 1995; Raviele 2005; Ward 1998). Vasovagal syncope The treatment of vasovagal syncope was investigated in 40 studies (2028 patients). In 29 studies drug treatment was investigated and in 10 studies pacemaker treatment. One study compared drug treatment (beta-blocker) with pacemaker treatment. In some of the studies on vasovagal syncope more than 2 treatment modalities were compared. Carotid sinus syncope Of the 6 studies (358 patients) on carotid sinus syncope, one investigated pharmacological treatment (Moore 2005) and the remaining pacemaker treatment (Brignole 1988; Brignole 1992b; Claesson 2007; Kenny 2001; McIntosh 1997). Situational syncope No studies on situational syncope matched the criteria for inclusion in our review. More detailed information about the studies considered suitable for quality assessment and -if possible- data extraction can be found in the Included Studies Section of this review. Excluded studies As stated above, 39 of the final selection of 104 publications have been classified as excluded (see Excluded studies). Twenty-six of these publications were not randomized. The other publications were found to be: abstracts of which we were unable to retrieve a full study report or any usable data from the authors after contacting them (5), study report meeting the criteria for exclusion (3), protocols to studies without a retrievable final study report (2), case reports (2), and a review (1). For more information about the specific reason for exclusion of a publication, we refer the reader to the Characteristics of excluded studies-section of our review (see Characteristics of excluded studies). Risk of bias in included studies The individual scores on all risk of bias items of the included studies and a summary of the scores are presented in Figure 2 and Figure 3. Figure 2. Methodological quality graph: review authors judgments about each methodological quality item presented as percentages across all included studies. 12

16 Figure 3. Methodological quality summary: review authors judgments about each methodological quality item for each included study. 13

17 Sequence generation The randomization method was reported in 19 of the 46 included studies. Ten of these studies were randomized by a computer-generated randomization list (Ammirati 2001; Di Girolamo 1999; Kaufmann 2002; Mahanonda 1995; Occhetta 2004; Raviele 1999; Sutton 2000; Ward 1998; Zeng 1998a; Zeng 1998b). The other studies were randomized by a randomization table (Brignole 1992a; Brignole 1992b; Raviele 2004), telephone (Connolly 1999), cards in envelopes (Claesson 2007) or any kind of central process (Connolly 2003; Guan 1999; Kenny 2001; Lin 2000). In 27 studies the randomization method was not described in sufficient detail to make a judgement. Allocation The allocation was concealed in 9 of the 46 included studies ( Ammirati 2001; Claesson 2007; Connolly 1999; Connolly 2003; Mahanonda 1995; Occhetta 2004; Raviele 1999; Sutton 2000; Ward 1998). In the other 37 studies, insufficient information was available to make a clear judgement. Blinding We made separate judgements for patients, physicians/personnel and outcome accessors. In 10 studies none of them were blinded. In 7 of these studies blinding was not possible, because pacemaker treatment was compared with non-pacemaker treatment (Ammirati 2001; Brignole 1992b; Claesson 2007; Connolly 1999; Flammang 1999; Kenny 2001; Sutton 2000). In the other studies, pharmacological treatment was compared with non-pharmacological standard treatment (Di Girolamo 1998; Perez-Lugones 2001; Ventura 2002) and no placebo was used. Incomplete outcome data Incomplete outcome data had been adequately addressed in 29 studies. In nine studies incomplete outcome data had not been adequately addressed. In 7 of these 9 studies 20% or more of the patients dropped out of the study (Madrid 2001; Mahanonda 1995; McIntosh 1997; Raviele 1999; Sheldon 2006; Sutton 2000; Theodorakis 2006); in the other two studies no information was given about incomplete outcome data (Kurbaan 2000a; Moya 1995). The risk of bias was unclear in eight studies. Selective reporting 37 Studies were free of selective outcome reporting. The risk of this bias was unclear in four studies (Brignole 1988; Brignole 1992a; Morillo 1993; Raviele 2005). We judged selective outcome reporting to be present in five studies (El-Bedawi 1994; Kurbaan 2000a; McIntosh 1997; Moore 2005; Sheldon 2006). In one of these studies the result of the tilt test was not reported (Kurbaan 2000a). In three of the studies it was unclear whether real syncope was present (El-Bedawi 1994; McIntosh 1997; Moore 2005). In one of the studies, the outcome measure quality of life was not reported in the study report, though this had been specified in the study protocol (Sheldon 2006). Other potential sources of bias We made efforts to retrieve missing information. If we were unable to get the full reports of eligible publications through our university library, we contacted the first authors of these publications in order to get the full study report. In some cases, this approach was successful. In 12 of the 14 cross-over studies that were included in our review no information was available about the correlation between responses to different treatment sequences (Biffi 1997; Brignole 1988; Deharo 2003; El-Bedawi 1994; Eldadah 2006; Flevari 2002; Kaufmann 2002; Kurbaan 2000a; McIntosh 1997; Moore 2005; Morillo 1993; Raviele 2005). We contacted all corresponding authors of these 12 studies in order to obtain this information. Some authors replied. Af few of them provided some additional information. However, none of these authors provided information about patients responses to different treatment sequences. In general, the responses to our requests were very limited. If we would have been able to get more information about the actual performance of studies, the risk of bias table would be more complete. We judged that 33 studies were free of other sources of bias. We detected other potential sources of bias in 13 studies. In three of these studies selection bias might have occurred (Brignole 1992a; Kurbaan 2000a; Morillo 1993). In one study, it remained unclear whether the primary condition of interest was vasovagal syncope or orthostatic hypotension (Ward 1998). Except for the treatment under investigation, patients in the intervention and control group were treated or were likely to have been treated differently in 5 studies (performance (Ammirati 1998; Brignole 1992b; Connolly 1999; Connolly 2003; Raviele 2005). Five studies were terminated ahead of time (Ammirati 2001; Connolly 1999; Occhetta 2004; Raviele 1999; Raviele 2004). More details can be found in the risk of bias tables of these respective studies (see Included studies). In four studies the risk of bias was unclear (El-Bedawi 1994; Kaufmann 2002; Perez-Lugones 2001; Zeng 1998a). Based on the risk of bias assessment we considered three included studies to be insufficiently valid and applicable for data-extraction (68 participants; Brignole 1992a; El-Bedawi 1994; Kurbaan 2000a). The reasons for these decisions are explained in the Risk of bias tables of the respective studies (see Characteristics of included studies). This implies that though these three studies were in- 14

18 cluded, no data were extracted from these studies. Data were extracted from 43 of the 46 included studies Patients participated in these remaining 43 studies. The trial sizes in the different studies ranged from 8 to 208 participants with a median sample size of 48 (interquartile range 25 to 65). The mean age of patients within these trials was 49.8 years of age (range 13 to 86). On average, 45% of the study participants were men. The studies were undertaken between 1992 and 2006, mainly in developed countries. As far as the duration of the wash-out period was reported in cross-over trials, it varied between 1 to 2 hours to 7 days. The mean follow-up time across the included studies was 11.3 ± 1.7 months. Effects of interventions See: Summary of findings for the main comparison Pharmacological and pacemaker treatment for vasovagal syncope.; Summary of findings 2 Pharmacological and pacemaker treatment for carotid sinus syncope. Vasovagal syncope Beta-blockers Comparison with standard treatment Beta-blocker treatment was compared with conventional therapy in four studies (Table 1)(Di Girolamo 1998; Guan 1999; Lin 2000; Ventura 2002). The occurrence of syncope during followup was lower for patients that received beta-blocker treatment compared to patients that received standard treatment (risk ratio 0.36; 95% CI 0.21 to 0.62)(see Analysis 1.1, Table 1). In one study, the number of pre-syncopal recurrences per year was 0 for patients that received atenolol treatment and 4.8 for patients that received standard treatment (Lin 2000). The occurrence of syncope upon provocation by head-up tilt testing was lower for patients that received beta-blocker treatment compared to patients that received standard treatment (risk ratio 0.75 (95% CI 0.58 to 0.97)(see Analysis 1.2, Table 1). The occurrence of side effects was determined in two studies (Guan 1999; Lin 2000). Patients treated with atenolol experienced side effects such as: bradycardia, palpitations, lightheadedness. In both studies, the occurrence of side effects in patients that received standard treatment was not assessed. Comparison with placebo treatment Beta-blocker treatment was compared with placebo treatment in eight studies (491 patients). However, one of these studies was considered insufficiently valid and applicable for data extraction (Brignole 1992a). In the remaining studies, no significant differences were found with respect to the average number of syncopal episodes per patient per year (syncope burden), syncopal episode(s) during follow-up or upon tilt testing, and average number of episodes of near syncope (see Table 1, Analysis 2.1 and Analysis 2.2). The occurrence of side effects was investigated in four studies (Flevari 2002; Madrid 2001; Mahanonda 1995; Sheldon 2006). No major differences were found, as far as analysis is possible for these studies (see Analysis 2.3, Table 1). Reported side effects on beta-blocker treatment were: fatigue, presyncope, insomnia, depression, edema, bradycardia and headache. Reported side effects on control treatment were: alterations of the libido, fatigue, presyncope, nausea and depression. One of these studies reported that side effects were more common in patients treated with beta-blockers compared to patients treated with placebo (p< 0.05)(Madrid 2001). Quality of life was assessed in three studies (see Table 1). Comparison between these studies for the outcome measure quality of life was not possible. Propranolol, nadolol and placebo treatment all improved patients self-assessed well-being (p< ; but no comparative data were available)(flevari 2002). In one study, quality of life improved in more patients treated with atenolol (71%) compared to patients treated with placebo (29%; P = 0.02)(Mahanonda 1995). By contrast, quality of life was lower after beta-blocker treatment than after placebo treatment in another study (mean difference 0.60; 95% CI 0.20 to 1.00)(see Table 1)(Theodorakis 2006). Comparison with other pharmacological treatment In 10 studies beta-blocker treatment was compared with other kinds of pharmacological treatment, namely: other beta-blocker treatment (Flevari 2002; Haghjoo 2006), clonidine (Biffi 1997), domperidon, dihydroergotamine and cafedrine (Brignole 1992a), scopolamine (Guan 1999), disopyramide (Nakagawa 1998), enalapril (Lin 2000), etilefrine (Di Girolamo 1998), verapamil (Jhamb 1996) and fluoxetine (Theodorakis 2006). The study comparing atenolol, domperidon, dihydroergotamine and cafedrine treatment was found to be insufficiently valid and applicable for data extraction (Brignole 1992a). In the aforementioned studies no significant differences were observed with respect to syncope burden, pre-syncope burden, or occurrence of syncope during follow-up (see Table 1). The proportion of patients experiencing syncope upon provocational head-up tilt testing did also not differ significantly between propranolol and several other kinds of pharmacological treatment (see Table 1)(Di Girolamo 1998; Flevari 2002; Haghjoo 2006; Nakagawa 1998; Theodorakis 2006). Similar findings were obtained in a study comparing atenolol and enalapril (Lin 2000). However, the proportion of patients experiencing syncope upon provocational head-up tilt testing was significantly different in the studies that compared metoprolol treatment with clonidine (Biffi 1997) or verapamil (Jhamb 1996)(see Table 1). With respect to the occurrence of side effects, no significant differences were observed between beta-blocker treatment and comparative treatment (see Table 1). Though quality of life improved 15

19 in all patients treated with propranolol, nadolol and placebo (p< ; but no comparative data were available)(flevari 2002), quality of life was worse in patients treated with propranolol compared to patients treated with fluoxetine (see Table 1)(Theodorakis 2006). Comparison with pacemaker treatment One study comparing beta-blocker treatment (atenolol) with DDD-pacemaker treatment observed that the occurrence of syncope during follow-up was higher in patients treated with atenolol compared to patients treated with a DDD-pacemaker (12 vs. 2; risk ratio 5.87; 95% CI 1.39 to 24.80)(see Table 1)(Ammirati 2001). However, with respect to the time to syncopal recurrence, minor physical trauma due to syncope and side effects no significant differences were observed (see Table 1)(Ammirati 2001). Major physical trauma did not occur in any of the patients studied. Reported side effects were: palpitations (pacemaker group) and fatigue, depression, anxiety and impotence (beta-blocker group). Alpha-adrenergic agonists Comparison with standard treatment Two studies compared treatment with alpha-adrenergic agonists with standard treatment (see Table 2)(Di Girolamo 1998; Perez-Lugones 2001). No significant differences were observed with respect to the syncope burden and occurrence of syncope upon head-up tilt testing (see Table 2). However, the proportion of patients experiencing syncope during follow-up was lower during treatment with an alpha-adrenergic agonist (midodrine) than conventional treatment (19.4% vs. 86.7%; risk ratio 0.22; 95% CI 0.11 to 0.46)(see Table 2)(Perez-Lugones 2001). The time to the first syncopal recurrence was also longer in patients treated with alpha-adrenergic agonists (hazard ratio 0.31; 95% CI 0.16 to 0.62)(see Table 2). Finally, quality of life (Endicott Questionnaire) was higher after midodrine treatment compared to conventional treatment (mean difference 20.00; 95% CI to 22.57)(see Table 2)(Perez-Lugones 2001). Comparison with placebo Treatment with alpha-adrenergic agonists was compared with placebo treatment in 7 studies (see Table 2). In a single study, no significant differences were found with respect to the proportion of patients experiencing syncope during followup, the median time to recurrence of syncope as well as the number of episodes of near syncope per patient (see Table 2)(Raviele 1999). From all studies, data about the proportion of patients with syncopal episodes induced by provocation testing could be derived. The studies fall into 2 categories: 1) treatment initiation before the provocational head-up tilt test (Di Girolamo 1998; Kaufmann 2002; Moya 1995; Raviele 1999; Ward 1998) or 2) treatment initiation upon the occurrence of signs and symptoms of an impending syncopal episode during provocation testing (Ammirati 2000; Raviele 2005)(see Table 2). We discuss the results for treatment initiation before the provocational head-up tilt test first. In general, the occurrence of syncope upon provocation testing after the initiation of treatment was not different for treatment with alpha-adrenergic agents compared to placebo treatment (odds ratio 0.54; 95% CI 0.24 to 1.20)(see Analysis 3.1, Analysis 3.2). For these studies, the results did not differ considerably between parallel randomized controlled trials and randomized cross-over trials (P value Chi 2 test = 0.28; I 2 = 13%)(see Analysis 3.1). The treatment effects within subgroups by type of alpha-adrenergic treatment (etilefrine or midodrine) differed considerably (P value Chi 2 test = ; I 2 = 91.0%)(see Analysis 3.2). Though no statistically significant differences were observed for etilefrine treatment (see Analysis 3.2, Subgroup1), the occurrence of syncope upon head-up tilt testing was lower in patients treated with midodrine compared to patients that received placebo treatment (odds ratio 0.12; 95% CI 0.04 to 0.36)(see Comparison Analysis 3.2, Subgroup 2). In studies in which treatment was started after the occurrence of symptoms, the occurrence of syncopal episodes induced by provocation testing was also lower in patients treated with alpha-adrenergic agents compared to patients that received placebo treatment (odds ratio 0.02, 95% CI 0.01 to 0.08)(see Analysis 3.3). One study evaluated the occurrence of trauma due to syncope (Raviele 1999). Major trauma did not occur in any of the patients; the occurrence of minor trauma was not assessed (see Table 2). The occurrence of side effects was investigated in 5 studies (Ammirati 2000; Moya 1995; Raviele 1999; Raviele 2005; Ward 1998). No significant differences were found with respect to the occurrence of side effects (see Analysis 3.4; Table 2). Reported side effects in patients receiving treatment with alpha-adrenergic agonists were: goose bumps, tingling, chills, agitation, depression, anxiety, insomnia, gastro-intestinal discomfort, palpitations, hypertension. Reported side effects in patients receiving placebo treatment were: gastro-intestinal discomfort, palpitations, hypertension. One study reported that psychiatric side effects such as agitation, depression, anxiety and insomnia were exclusively observed in patients treated with etilefrine (20% vs. 0%)(Raviele 1999). Quality of life was assessed in one study and was found to be improved upon treatment with an alpha-adrenergic agonist (midodrine)(no comparative data were available)(ward 1998). Comparison with other pharmacological treatment Alpha-adrenergic agonist (etilefrine) and beta-blocker (propranolol) treatment were compared in one study (Di Girolamo 1998). No significant differences were observed with respect to the occurrence of syncope upon provocation testing (see Table 2). Selective serotonin reuptake inhibitors Comparison with placebo In two studies comparing treatment with selective serotonin reuptake inhibitors and placebo treatment, no significant differences were found with respect to the syncope burden, pre-syncope 16

20 burden and occurrence of syncope induced by provocation testing (see Analysis 4.1; Table 3)(Di Girolamo 1999; Theodorakis 2006). However, the occurrence of syncope during follow-up was lower during treatment with selective serotonin reuptake inhibitors than during placebo treatment (risk ratio 0.39, 95% CI 0.20 to 0.76)(see Analysis 4.2; Table 3). The occurrence of side effects and quality of life were not significantly different during treatment with a selective serotonin reuptake inhibitor and placebo treatment (see Table 3)(Di Girolamo 1999). Reported side effects during paroxetine treatment were: severe recurrent headaches, transient sexual dysfunction, nausea and diarrhea (Di Girolamo 1999). In the same study, headache was reported by one patient in the placebo group. Study treatment had to be discontinued because of side effects in one patient treated with paroxetine and in none of the patients treated with placebo. Comparison with other pharmacological treatment One study also compared the effectiveness of fluoxetine treatment with beta-blocker treatment (propranolol)(n = 63 patients for this comparison)(theodorakis 2006). No significant differences were found with respect to the syncope and pre-syncope burden, the proportion of patients with one or more syncope during follow-up and the occurrence of syncope induced by provocation testing (see Table 3). However, quality of life was better in patients treated with selective serotonin reuptake inhibitors compared to beta-blocker treatment (mean difference -0.70; 95% CI to -0.30)(see Table 3). ACE-inhibitors Comparison with standard treatment In one study, comparing ACE-inhibitor treatment (enalapril) and standard treatment, no significant differences were observed with respect to the occurrence of syncope during follow-up and presyncope burden for patients that received enalapril treatment and patients that received standard treatment (see Table 4)(Lin 2000). However, the occurrence of syncope upon head-up tilt testing was lower for patients treated with enalapril compared to patients that received standard treatment (risk ratio 0.54 (95% CI 0.29 to 0.98)(see Table 4). Six of the 16 patients that received enalapril treatment experienced side effects (Lin 2000). These were: dry cough, low blood pressure, palpitations and lightheadedness. The occurrence of any side effects in patients that received standard treatment was not reported. Comparison with placebo treatment Two studies compared treatment with an ACE-inhibitor (enalapril) with placebo treatment (Zeng 1998a; Zeng 1998b). The proportion of patients with syncopal episodes induced by provocation testing was lower during enalapril treatment compared with placebo treatment (risk ratio 0.04, 95% CI 0.01 to 0.29)(see Analysis 5.1; Table 4). In one study, no significant differences were observed with respect to the occurrence of side effects (see Table 4)(Zeng 1998b). In this study, one patient treated with enalapril discontinued treatment because of dry cough. Comparison with other pharmacological treatment ACE-inhibitor treatment (enalapril) was compared with betablocker treatment (atenolol) in one study (Lin 2000). No significant differences were observed with respect to the occurrence of syncope during follow-up, syncope upon provocational head up tilt testing and the occurrence of side effects during follow-up (see Table 4). Side effects that occurred during both atenolol treatment and enalapril treatment were: lightheadedness and palpitations (Lin 2000). Patients treated with atenolol also experienced bradycardia. A dry cough and low blood pressure were also experienced as side effects of enalapril treatment. Disopyramide Two studies evaluated the effectiveness of disopyramide treatment for vasovagal syncope (Morillo 1993; Nakagawa 1998). One of these studies compared disopyramide treatment with placebo treatment (Morillo 1993). No syncopal episodes occurred during follow-up for both oral treatment modalities. The occurrence of syncope upon provocational head-up tilt testing was not significantly different for disopyramide treatment and placebo treatment (see Table 5). During disopyramide treatment patients in this study reported side effects such as dry mouth, urinary hesitancy or retention. Dry mouth was also reported as a side effect during placebo treatment. No significant differences were observed between disopyramide treatment and propranolol treatment with respect to the proportion of patients experiencing syncope induced by provocation testing (see Table 5)(Nakagawa 1998). Anticholinergic agents Comparison with standard treatment In one study, the recurrence of syncope during follow-up was lower in patients treated with scopolamine compared to patients that received standard treatment (risk ratio 0.42; 95% CI 0.18 to 0.95)(see Table 6)(Guan 1999). Twenty-five per cent of patients treated with scopolamine experienced side effects, such as: severe dry mouth, palpitations, blurred vision (Guan 1999). The occurrence of any side effects of patients that received standard treatment was not reported. Comparison with placebo treatment The effectiveness of anticholinergic agents versus placebo treatment for vasovagal syncope was evaluated in two studies (Lee 1996; Santini 1999). In both studies, treatment was only initiated when signs and symptoms occurred upon provocation testing. For anticholinergic agents there was a significant reduction in occurrence of syncope induced by provocation testing with treatment initiation after the occurrence of symptoms (risk ratio 0.66; 95% CI 17