Christine M. Albert, MD, MPH; Lawrence Rosenthal, MD; Hugh Calkins, MD; Jonathan S. Steinberg, MD; Jeremy N. Ruskin, MD; Paul Wang, MD, MPH; James E. Muller, MD, MPH; Murray A. Mittleman, MD, MPH

J Am Coll Cardiol.  2007;50(23) ©2007 Elsevier Science, Inc.

Abstract and Introduction

Abstract

Objectives: This study examined the risk of implantable cardioverter-defibrillator (ICD) shocks for ventricular tachycardia (VT) or ventricular fibrillation (VF) associated with driving.
Background: Concerns regarding VT/VF occurring during driving are the basis for driving restrictions in ICD patients; however, limited data are available to inform recommendations.
Methods: This study used a prospective nested case-crossover design to compare the risk of ICD shock for VT/VF both during and up to 60 min after an episode of driving as compared with that during other activities among 1,188 ICD patients enrolled in the TOVA (Triggers of Ventricular Arrhythmia) study.
Results: Over a median follow-up of 562 days, there were 193 ICD shocks for VT/VF with data on exposure to driving before ICD shock. The absolute risk of ICD shock for VT/VF within 1 h of driving was estimated to be 1 episode per 25,116 person-hours spent driving. The ICD shocks for VT/VF were twice as likely to occur within 1 h of driving a car as compared with other times (relative risk [RR] 2.24, 95% confidence interval [CI] 1.57 to 3.18). This risk was specific for shocks for VT/VF and occurred primarily during the 30-min period after driving (RR 4.46, 95% CI 2.92 to 6.82) rather than during the driving episode itself (RR 1.05, 95% CI 0.48 to 2.30).
Conclusions: Although the risk of ICD shock for VT/VF was transiently increased in the 30-min period after driving, the risk was not elevated during driving and the absolute risk was low. These data provide reassurance that driving by ICD patients should not translate into an important rate of personal or public injury.

Introduction

Sudden death from cardiac causes has been known to occur during driving, but it is unclear whether a causal relationship exists. In one study,^[1] 71 (5%) of 1,348 coronary heart disease deaths occurred while driving. Driving a car can be a moderately stressful activity for some, and psychological stress is believed to trigger life-threatening ventricular arrhythmias.^[2-4] Driving a car is also accompanied by exposure to particulate air pollution,^[5] and a large number of epidemiologic studies have found an association between short-term increases in particulate air pollution and cardiovascular morbidity and mortality.^[6-8] Recently, acute exposure to traffic has also been associated with onset of myocardial infarction (MI),^[9] but to our knowledge, no studies have examined the impact of driving on ventricular arrhythmias.

Because driving an automobile is not only a quality-of-life issue,^[10] but also is considered by many to be a basic necessity, knowledge about the incidence and risk of ventricular arrhythmias during driving is of direct clinical and public health importance. It is a well-recognized concern that patients with implantable cardioverter-defibrillators (ICDs) may experience life-threatening ventricular arrhythmias while driving, and the resultant loss of consciousness may result in injury to the patient, other passengers, and bystanders.^[11] Physicians who care for these patients must balance these risks when making recommendations regarding driving. However, there are limited data to adequately inform recommendations.^[12] Data from the AVID (Antiarrhythmics Versus Implantable Defibrillators) trial reported that 8% of patients experienced an ICD shock while driving,^[13] although it was unclear from these data whether a causal relationship existed. The TOVA (Triggers of Ventricular Arrhythmias) study presented a unique opportunity to examine these risks and further define the relationship between driving a car and the occurrence of ventricular arrhythmias among susceptible patients with ICDs.

Methods

Patient Population

The TOVA study is a multicenter prospective cohort study designed to identify lifestyle and psychological triggers of ICD shock for ventricular tachycardia (VT) and/or ventricular fibrillation (VF). From June 1, 2000, to March 21, 2003, 1,188 patients with ICDs were enrolled at the time of implantation or through the ICD clinics at 31 participating centers in the U.S. To qualify for enrollment, patients had to meet the 1998 American College of Cardiology/American Heart Association guidelines for ICD implantation^[14] (Class I to IIB indications), which included the following: 1) cardiac arrest; 2) spontaneous sustained ventricular tachycardia; 3) primary or recurrent syncope with inducible VT refractory to drug therapy; 4) nonsustained ventricular tachycardia with coronary disease, prior MI, left ventricular dysfunction, and inducible sustained VT or VF; or 5) familial or inherited condition with high risk of life-threatening arrhythmia.

Study Design

We used a nested case-crossover design to quantify the relative risk (RR) of an ICD shock for VT/VF during or up to 60 min after an episode of driving as compared with the risk during other activities (Fig. 1). The case-crossover study design assesses the change in risk of an acute event during a brief hazard period during and/or after exposure to a transient risk factor.^[15] For each individual ICD shock, the prospectively collected prior habitual frequency of driving for that participant served as the control information, and a self-matched analysis was conducted.

Figure 1. 

Hazard Periods for Case-Crossover Analysis. The green-shaded portion of the timeline represents the hazard period during or after a driving episode. If an implantable cardioverter-defibrillator (ICD) discharge occurred during this window, the patient was considered exposed in the primary analysis. In secondary analyses, we examined the relative risk of ICD discharge within the 3 separate shaded components of the hazard period: the period during driving (A), the first 30-min interval (B), and the second 30-min interval (C) after a driving episode.

     

Assessment of Control Exposures

At entry into the study, the study staff at each clinical site collected information on baseline clinical characteristics from the patient's medical record using a structured data abstraction form and interviewed patients regarding demographics, cardiac symptoms, lifestyle habits, and medications via standardized questionnaires. During this baseline interview, participants also were asked to report their usual frequency of participating in various activities, experiencing certain emotions, and exposure to other potential triggers. Updated information on usual frequency of exposures was collected with similar questionnaires at 6 and 12 months of follow-up.

Exposure to driving was ascertained with the following questions at baseline and at 6 months and 12 months of follow-up: "Since implantation or over the past 6 months, on average, how often do you drive a car?" and "On average, how long do you drive each time?" Responses to these questions were open ended. Calculation of exposed person-time depended on the hazard period examined. In the primary analysis, exposed person-time was calculated by adding 60 min to the reported usual duration of driving and multiplying this sum by the reported frequency of driving in the past 6 months. When examining the components of the hazard period, only the time spent in that particular portion of the hazard period was considered exposed. To calculate unexposed person-time, exposed person-time in hours was subtracted from the number of hours in the 6-month interval or the amount of time since enrollment if the shock occurred within the first 6 months of the study.

Assessment of Case Exposures

After entry into the study, patients were followed up for ICD discharges by the study and clinical staff at participating institutions. To allow the widest range of patients to be included in the study, we did not pre-specify how the ICD should be programmed, but instead chose to defer to physician judgment to use clinical standards of care. Patients were asked to call the staff within 72 h of experiencing an ICD shock. Participants were then interviewed about lifestyle and psychological exposures occurring within the minutes, hours, and days preceding ICD shock. To minimize bias in ascertainment, interviewers and patients were not informed of the hypothesized hazard period. Exposure to driving was ascertained with the following question: "Before the device discharged, when was the last time you drove?" There were 9 possible response categories ranging from "at the time of the shock" to "5 or more days before." Participants who reported driving at the time of, one-half hour before, or 1 h before the shock were considered exposed.

Case/end Point Confirmation

Stored electrograms were downloaded from the patient's ICD by the clinical centers at the patient's first post-shock visit and during routine clinic visits on a 3- to 6-month basis. Electrograms were then reviewed at the central core laboratory by electrophysiologists blinded to exposure status. The date and time of discharge, the type and number of therapies, and the presence and type of underlying arrhythmia were determined. The primary analysis included only appropriate ICD shocks for ventricular arrhythmias, but secondary analyses were also performed for inappropriate shocks.

Statistical Analyses

Means or proportions of baseline demographics and clinical characteristics were computed for the entire cohort across 4 categories of increasing driving frequency, and the significance of associations was tested using the Mantel-Haenszel chi-square test for trend for categorical variables and linear regression for continuous variables.

For the case-crossover analysis, each ICD shock serves as an individual case and the control information is based on the usual amount of driving reported on the most recent questionnaire before the shock. The analysis of case-crossover data is similar to that of a crossover experiment but with varying exposure times, in which each individual's risk is assessed during periods of exposure and nonexposure and each individual serves as both a case and control (self-matching). We computed the incidence rate ratio as a measure of RR for the association between driving and episodes of VT/VF using the Mantel-Haenszel estimator for sparsely distributed person-time data using each individual as the stratifying variable. In this analysis, each subject's person-time was categorized into periods of exposure and nonexposure as previously described.^[15-17] The data were then stratified by each individual shock and were analyzed using methods for cohort studies with sparse data in each stratum.^[17]

The pre-specified hazard period for the primary analysis was the time spent driving and up to 60 min after each episode. In secondary analyses, we examined the risk during driving and 30-min intervals after a driving episode (Fig. 1). Several sensitivity analyses were also performed. To explore the possibility of intraindividual confounding, sensitivity analyses excluding patients who were exposed to at least moderate levels of exertion (≥ 5 metabolic equivalents [METs]) as quantified on a scale of 1 to 6 METs^[18] or moderate levels of anger within 1 h of ICD shock were performed. To investigate the sensitivity of our result to repeated shocks in the same patient, we repeated the analysis only including the first shock experienced by each patient. Finally, because patients are unlikely to drive during the nighttime hours between 12 AM to 6 AM when the risk of an ICD shock is lowest,^[19,20] we performed a sensitivity analysis that excluded this nighttime period from the case-crossover analysis. In this analysis, we conservatively assume that none of the patients were exposed to driving during this time period. Potential modifiers of the RR were evaluated by comparing RRs in categories of the modifier and then comparing these RRs with the chi-square test for homogeneity.^[21] All reported p values are 2-sided. Statistical analyses were performed using SAS software version 9.1 (SAS Institute Inc., Cary, North Carolina).

Results

Among the 1,188 patients enrolled in the TOVA study, 1,106 patients provided information on their driving habits at entry into the study ( Table 1 ). The majority of patients (80%) reported driving a car at least once per week. Overall, participants reported spending a median of 3.8 h/week (interquartile range 1.2 to 7.7 h/week) or 2.3% (interquartile range 0.7% to 4.6%) of their time driving a car. Those who drove more frequently were younger, more educated, and more likely to be men or Caucasian. They also tended to be healthier and were less likely to have diabetes, hypertension, congestive heart failure, or a left ventricular ejection fraction lower than 30%. With respect to current recommendations regarding driving, patients who recently underwent ICD implantation tended to drive less (p = 0.08). However, of those with recent implants, 75.1% reported driving at least once per week and 38.6% were driving more than 2 times per day. Patients who underwent prophylactic ICD were less likely to abstain from driving (p = 0.04).

Over a median follow-up of 562 days (interquartile range 375 to 882 days), 414 total ICD shocks occurred in the 1,188 patients enrolled in the TOVA study. Of these, 324 (74.6%) were for VT or VF translating into an absolute risk of experiencing a shock for VT or VF of 1 per 56,260 person-hours spent in the study. The post-shock interview was completed within 72 h for 272 of these shocks (66%) occurring among 161 patients. Of these 272 shocks, 113 patients (70.2%) received 1 shock, 22 patients (13.6%) received 2 shocks, 16 (9.9%) patients received 3 to 4 shocks, and 10 patients (6.2%) received more than 4 shocks. Information on exposure to driving was available for 259 of these shocks (95%), and these events were included in the case-crossover analysis.

Of the 259 total ICD shock episodes included in the analysis, 193 (74.5%) were documented to be for an episode of VT or VF, of which 44 (22.8%) occurred within 1 h of driving a car among 23 patients. Of these shocks, 8 were preceded by lightheadedness and 3 resulted in syncope. Using the case-crossover method, the RR of an ICD shock for VT/VF within 1 h of driving was elevated at 2.24 (95% confidence interval [CI] 1.57 to 3.18, p < 0.001) compared with the risk at other times during the study. This elevation in the risk seemed to be specific for ventricular arrhythmias because there was no elevation in the risk of inappropriate shocks (n = 55, RR = 1.07, 95% CI 0.49 to 2.33). To put these RRs into perspective, if one assumes that the 2.24-fold elevation in risk observed in the case-crossover analysis was generalizable to the entire cohort, this would still translate into a low overall absolute risk of ICD shock for VT/VF (1 episode per 25,116 person-hours spent driving).

We then examined the RR of VT/VF within 3 components of the hazard period (Fig. 1): the period during driving (A), the first 30-min interval (B), and the second 30-min interval (C) after a driving episode. Of the 44 ICD shocks for VT/VF that occurred within 1 h of driving, 7 occurred during driving, 30 occurred 30 min immediately after driving, and 7 occurred in the last 30-min period. When these hazard periods were examined separately, the risk of VT/VF was only significantly elevated in the 30-min period immediately after an episode of driving (RR 4.46, 95% CI 2.92 to 6.82, p < 0.001). Importantly, the risk of VT/VF was not significantly elevated during an episode of driving (RR 1.05, 95% CI 0.48 to 2.30). In addition, none of the shocks for VT or VF that occurred during driving resulted in lightheadedness or syncope, and only 1 resulted in an automobile accident.

We then repeated the aforementioned case-crossover analysis only including the first shock experienced by each patient to evaluate how repeated shocks in the same patient may have influenced the results. In this sensitivity analysis, the results for all analyses remained essentially unchanged. The RR of ICD shock for VT/VF within 1 h of driving was still elevated at 2.25 (95% CI 1.38 to 3.66), and the risk continued to be confined to the 30-min period following driving (RR 4.27, 95% CI 2.36 to 7.75) rather than during the driving episode itself (RR 1.31, 95% CI 0.50 to 3.46). Again there was no elevation in the risk of inappropriate ICD shocks during or after driving.

To explore the possibility that the delayed elevation in risk of VT/VF could be attributable to another triggering activity linked in time to driving, we examined the specific activity participants were engaged in at the time of ICD shock for those that occurred in the 30-min hazard period after driving. This examination revealed 7 episodes (23.3%) in which patients were engaged in physical exertion of at least 5 METs at the time of ICD discharge for VT/VF as compared with 15 episodes among those not exposed to driving in the 30 min before the ICD shock (9.9%; p = 0.05). Those shocked within the 30-min period after driving were also more likely to have experienced at least moderate levels of anger as compared with those who were not (13.3% vs. 4.2%; p = 0.07). Therefore, we performed a sensitivity analysis that excluded all shocks during which the participant was exposed to moderate levels of exertion (≥ 5 METs) or anger within 1 h before the shock (n = 44). In this analysis, the RR within 1 h of driving was essentially unchanged (RR 1.95, 95% CI 1.29 to 2.96, p = 0.002) and the delayed peak in risk at 30 min was attenuated but remained significant (RR 3.41, 95% CI 2.02 to 5.78, p < 0.001). With respect to other activities participants were engaged in at the time of shock, none of these seemed to be known triggers. Specifically, 6 occurred before or during a meal and another 5 occurred during shopping. The remainder (n = 11) occurred at rest.

Because several studies have reported nadirs in the incidence of ICD shocks during the nighttime sleeping hours between 12 AM and 6 AM (19,20), and because patients are unlikely to drive during these sleeping hours, we performed a sensitivity analysis excluding this time period from the case-crossover analysis. Again, the RR within 1 h of driving remained significantly elevated (RR 1.93, 95% CI 1.35 to 2.77, p < 0.001) and the delayed peak in risk at 30 min was still present (RR 4.24, 95% CI 2.71 to 6.62, p < 0.001).

We then explored our data for possible effect modification using stratified analyses ( Table 2 ). None of the potential modifiers outlined in Table 2 reached statistical significance; however, our power to detect such interactions was limited. There was a weak trend for patients with recent implants (< 6 months) and those who had experienced prior ICD discharges to have a higher risk of ICD shock for VT/VF within 1 h of driving. There was no apparent difference in the risk elevation for those who had their ICD implanted as secondary or primary prevention. Interestingly, patients with a higher baseline left ventricular ejection fraction and those without congestive heart failure also tended to be at higher risk. Of note, there was no evidence that beta-blockers significantly modified risk.

Discussion

In this multicenter prospective cohort of 1,188 patients with ICDs, 80% of all participants and 75% of those within 6 months of their implantation reported driving a car at least once per week. Among participants who received ICD shocks, an ICD shock for VT or VF was twice as likely to occur within 1 h of driving a car as compared with during other activities or rest (RR 2.24, 95% CI 1.57 to 3.18, p < 0.001). This elevation in risk occurred primarily during the 30-min period after driving rather than during the driving episode itself and also seemed to be specific for ICD shocks attributable to VT/VF.

To our knowledge, our study is the first to examine the association between driving a car and the onset of ventricular arrhythmias. These results for ventricular arrhythmias in ICD patients are roughly consistent with the 2.9-fold increased risk of MI observed within 1 h of exposure to traffic reported in a recent case-crossover study.^[9] This prior study also found the risk to be higher after compared to during the exposure, and the majority of the exposure to traffic was through riding in cars. It was unknown whether the participants were driving or were passengers in cars when exposed, but elevated risks of MI also were noted after exposure to public transportation, suggesting that the association may not be entirely attributable to the driving act itself. Because we did not collect information on average time spent as a passenger in a car, we do not know whether the risk of VT/VF would be similarly elevated if the participants in our study were not driving.

This delayed elevation in risk found in both studies might be explained if an induction effect or activation of an intermediary mechanism were required to precipitate ventricular arrhythmias or MI. Alternatively, driving might increase the susceptibility to other triggering activities. Perturbations of the autonomic nervous system by stress or anger brought on by driving and/or exposure to particulate air pollution are possible mechanisms that could potentially underlie both of the observed associations. Exposure to particulate matter, which is estimated to be 2-fold greater while riding in a car than on a bike,^[4] has been consistently associated with short-term (minutes) and longer-term (hours) reductions in heart rate variability compatible with withdrawal of vagal tone.^[22-24] The resulting imbalance in autonomic tone could then lead to an increased susceptibility to ventricular arrhythmias or to acute plaque rupture and/or thrombosis.^[25]

In addition to the possible mechanistic implications, there are also clinical and public health implications of these data. First, in this multicenter study, although the amount of time ICD patients spent driving was somewhat lower than estimates for the general population (2.3% vs. 4.0% of person-time, respectively),^[1,26] most patients resumed driving within 6 months of ICD implantation. Despite guidelines recommending a period of abstention from driving after an episode of life-threatening ventricular arrhythmia,^[1,26,27] the majority of patients do not comply with this recommendation.^[13,28-30] The observed trends toward higher risks of ICD shock for VT/VF among patients with recent implants and prior discharges in association with driving concur with guidelines.^[1,26,27]

It is important to emphasize that the RR was not significantly elevated during the driving episode. Therefore, the absolute risk of experiencing an ICD shock for VT/VF during driving should be similar to that observed at other times, which in this population was quite low (1 per 56,260 person-hours spent in the study). Based on these data, driving among ICD patients would not be expected to translate into a significant excess risk of injury to the patient or society from automobile accidents. In agreement with this postulated low absolute risk, there was only 1 shock for VT/VF during driving that resulted in an automobile accident in this cohort. Other studies using differing methodologies have come to similar conclusions.^[13,30]

There are several limitations to our study. Our observational study cannot prove causality, and the association between driving and ICD shock for VT/VF could, at least in part, be caused by another activity or exposure that is associated in time with driving. From our limited examination of the types of activities patients were engaged in, possible associations between driving and subsequent exposure to exertion and anger were found. In sensitivity analyses excluding ICD shocks during which the participant was also exposed to exertion or anger, the delayed elevation in risk was attenuated slightly but remained elevated, suggesting that these activities do not explain the entire association between driving and ICD shocks for VT/VF. Also, if patients tended to drive at certain times of the day, circadian variations in risk of VT/VF could account for part of the association observed. However, the relationships persisted when nighttime hours were excluded from the analysis, and the discrete increase in risk in the 30-min period after driving without any elevation in the risk during driving would be difficult to explain on the basis of circadian variation alone.

Another limitation of our data is that the assessment of driving was by self-report and was not anonymous. Patients may have been less likely to report that a shock occurred while driving for fear that driving privileges would be curtailed, which could lead to an underestimation of the risk during driving. Also, if a patient were driving against medical advice, he or she may tend to minimize estimates of driving at baseline and follow-up interviews, which could lead to an overestimation of the risk. However, given the high frequency of reported driving in this cohort, severe underreporting is unlikely. Furthermore, not all patients resumed driving, and those who did tended to be healthier. Therefore, these results may not apply to all patients with ICDs.

Finally, although we made every attempt to obtain interview data on all patients within 72 h of the ICD shock, this proved to be difficult when patients failed to alert the participating center when a shock occurred. If the activity at the time of shock influenced whether the patient noticed or reported a shock, this could also potentially bias our risk estimates. This seems most plausible for shocks that might have occurred during sleep and thus gone unnoticed. However, the percentage of shocks that occurred between the hours of midnight to 6 AM was slightly greater among those with interview data versus those without (15.0% vs. 11.2% respectively, p = 0.24), which argues against this possibility.

In summary, these prospective data in a large multicenter cohort of patients with ICDs suggest that the risk of VT or VF is transiently increased in the 30-min period after driving in susceptible patients. The lack of elevation in risk during the driving episode, along with the expected low absolute risk, provides some reassurance that driving by ICD patients should not translate into an important rate of personal or public injury. The mechanism behind the delayed effect is unclear and may imply an induction effect and/or synergistic effects with other triggers. Whether these risks would extend to other patient populations or to the general population is unknown, and further study is required to evaluate the underlying mechanisms.

Myerburg RJ, Davis JH. The medical ecology of public safetyI: sudden death due to coronary heart disease. Am Heart J 1964;68:586-595. Lampert R, Joska T, Burg MM, Batsdord WP, McPherson CA, Jain D. Emotional and physical precipitants of ventricular arrhythmia Circulation 2002;106:1800-1805. Kop WJ, Krantz DS, Nearing BD, et al. Effects of acute mental stress and exercise on T-wave alternans in patients with implantable cardioverter defibrillators and controls Circulation 2004;109:1864-1869. Steinberg JS, Arshad A, Kowalski M, et al. Increased incidence of life-threatening ventricular arrhythmias in implantable defibrillator patients after the World Trade Center attack J Am Coll Cardiol 2004;44:1261-1264. Rank J, Folke J, Jespersen PH. Differences in cyclists' and car drivers' exposure to air pollution from traffic in the city of Copenhagen Sci Total Environ 2001;279:131-136. Pope 3rd CA, Burnett RT, Thurston GD, et al. Cardiovascular mortality and long-term exposure to particulate air pollution: epidemiological evidence of general pathophysiological pathways of disease Circulation 2004;109:71-77. Brook RD, Franklin B, Cascio W, et al. Expert Panel on Population and Prevention Science of the American Heart Association Air pollution and cardiovascular disease: a statement for healthcare professionals from the Expert Panel on Population and Prevention Science of the American Heart Association Circulation 2004;109:2655-2671. Dominici F, Peng RD, Bell ML, et al. Fine particulate air pollution and hospital admission for cardiovascular and respiratory diseases JAMA 2006;295:1127-1134. Peters A, von Klot S, Heier M, et al. Cooperative health research in the Region of Augsburg Study GroupExposure to traffic and the onset of myocardial infarction. N Engl J Med 2004;351:1721-1730. Jung W, Luderitz B. Quality of life and driving in recipients of the implantable cardioverter-defibrillator Am J Cardiol 1996;78:51-56. Larsen GC, Stupey MR, Walance CG, et al. Recurrent cardiac events in survivors of ventricular fibrillation or tachycardia: implications for driving restricts JAMA 1994;271:1335-1339. Epstein AE, Miles WM, Benditt DG, et al. Personal and public safety issues related to arrhythmias that may affect consciousness: implications for regulation and physician recommendations Circulation 1996;94:1147-1166. Akiyama T, Powell JL, Mitchell LB, Ehlert FA, Baessler C, Antiarrhythmics versus Implantable Defibrillators Investigators Resumption of driving after life-threatening ventricular tachyarrhythmia N Engl J Med 2001;345:391-397. Gregoratos G, Cheitlin, MD, Conill A, et al. ACC/AHA guidelines for implantation of cardiac pacemakers and antiarrhythmia devices: executive summary—a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee on Pacemaker Implantation) J Am Coll Cardiol 1998;31:1175-1209. Maclure M. The case-crossover design: a method for studying transient risks effects on the risk of acute events Am J Epidemiol 1995;133:144-153. Mittleman MA, Maclure M, Robbins JM. Control sampling strategies for case-crossover studies: an assessment of relative efficiency Am J Epidemiol 1995;142:91-98. Greenland S, Robbins JM. Estimation of a common effect parameter from sparse follow-up data Biometrics 1985;41:55-68. Albert CM, Mittleman MA, Chae CU, Lee I-M, Hennekens CH, Manson JE. Triggering of sudden death from cardiac causes by vigorous exertion N Engl J Med 2000;343:1355-1361. Kozak M, Krivan L, Semrad B. Circadian variations in the occurrence of ventricular tachyarrhythmias in patients with implantable cardioverter defibrillators Pacing Clin Electrophysiol 2003;26:731-735. Englund A, Behrens S, Wegscheider K, Rowland E, European 7219 Jewel Investigators Circadian variation of malignant ventricular arrhythmias in patients with ischemic and nonischemic heart disease after cardioverter defibrillator implantation J Am Coll Cardiol 1999;34:1560-1568. Rothman KJ, Greenland S. Modern EpidemiologyPhiladelphia, PA: Lippincott-Raven; 1998. pp. 253-279. Gold DR, Litonjua A, Schwartz J, et al. Ambient pollution and heart rate variability Circulation 2000;101:1267-1273. Magari SR, Hauser R, Schwartz J, Williams PL, Smith TJ, Christiani DC. Association of heart rate variability with occupational and environmental exposure to particulate air pollution Circulation 2001;104:986-991. Liao D, Duan Y, Whitsel EA, et al. Association of higher levels of ambient criteria pollutants with impaired cardiac autonomic control: a population-based study Am J Epidemiol 2004;159:768-777. Willich SN, Maclure M, Mittleman M, Arntz H-R, Muller JE. Sudden cardiac death: support for a role of triggering in causation Circulation 1993;87:1442-1450. Consensus Conference, Canadian Cardiovascular Society Assessment of the cardiac patient for fitness to drive Can J Cardiol 1992;8:406-412. Jung W, Anderson M, Camm AJ, et al. Recommendations for driving of patients with implantable cardioverter defibrillators Eur Heart J 1997;18:1210-1219. Conti JB, Woodard DA, Tucker KJ, Bryant B, King LC, Curtis AB. Modification of patient driving behavior after implantation of a cardioverter defibrillator Pacing Clin Electrophysiol 1997;20:2200-2204. Maas R, Ventura R, Kretzschmar C, Aydin A, Schuchert A. Syncope, driving recommendations, and clinical reality: survey of patients BMJ 2003;326:21. Curtis AB, Conti JB, Tucker KJ, Kubilis PS, Reilly RE, Woodard DA. Motor vehicle accidents in patients with an implantable cardioverter-defibrillator J Am Coll Cardiol 1995;26:180-184.

Funding Information

This study was supported by funding from the National Heart, Lung, and Blood Institute (R01 HL041016) and from the Guidant Foundation, Guidant Cardiac Rhythm Management. Dr. Albert has received research grant support from Boston Scientific and St. Jude Medical. Dr. Rosenthal has received research grant support from Boston Scientific, Reliant Pharmaceuticals, St. Jude Medical, Sanofi-Aventis, and Medtronic; honoraria as part of speaking engagements; and serves as a consultant for Boston Scientific. Dr. Calkins has received honoraria from Boston Scientific and Medtronic. Dr. Steinberg has received research support from Boston Scientific, Medtronic, and St. Jude Medical; has received honoraria from Boston Scientific and Medtronic; and serves as a consultant for Boston Scientific. Dr. Ruskin is a consultant with Medtronic. Dr. Wang has received research grant support, participation on the speakers' bureaus, and honorarium from Boston Scientific, Medtronic, and St. Jude Medical.

Abbreviation Notes

CI = confidence interval; ICD = implantable cardioverter-defibrillator; MET = metabolic equivalent; MI = myocardial infarction; RR = relative risk; VF = ventricular fibrillation; VT = ventricular tachycardia

Reprint Address

Dr. Christine M. Albert, Center for Arrhythmia Prevention, Division of Preventive Medicine and Cardiovascular Division, Brigham and Women's Hospital, 900 Commonwealth Avenue East, Boston, Massachusetts 02215-1204. Email: calbert@partners.org