WHAT IS ALREADY KNOWN ON THIS TOPIC

Patients have historically been asked to cease driving for 1 month after implantation of a primary prevention implantable cardioverter-defibrillator (ICD) and for up to 6 months after a secondary prevention ICD.

Current driving restrictions are based on underpowered studies that rely on convenience samples, lack appropriate control groups and lack objective data on subsequent motor vehicle crashes.

WHAT THIS STUDY ADDS

The current study examined 22 years of population-based administrative health and driving data for 9373 ICD patients and 28 119 age- and sex-matched controls.

Researchers found that real-world crash risk was lower after ICD implantation than crash risk among matched controls, likely because ICD recipients reduced their road exposure in response to contemporary postimplantation driving restrictions.

Traffic contraventions and crashes during the typical period of restricted driving indicate that adherence to postimplantation driving restrictions was inconsistent.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICYIntroduction

Clinicians and policymakers have long been concerned that cardioverter-defibrillator implantation increases the risk of sudden incapacitation while driving.1 2 Driver incapacitation might result from arrhythmogenic syncope or from the unexpected delivery of a shock, events that might be more common in the first few weeks following implantation.3 One autopsy case report vividly describes a fatal motor vehicle crash that occurred after the driver was incapacitated by a ventricular arrhythmia and implantable cardioverter-defibrillator (ICD) shock, yet the true prevalence of this kind of crash remains unknown.4

Clinical guidelines from the USA, Canada and Europe recommend temporary cessation of driving after ICD implantation (online supplemental item S1).5–8 Non-commercial driving restrictions vary by jurisdiction but patients in Canada and the USA have historically been asked to cease driving for 1 month after primary prevention ICD implantation and for up to 6 months after secondary prevention ICD implantation. These driving restrictions potentially prevent ICD-related crashes but can also cause social isolation, depression, loss of independence, consequences for employment and impaired quality of life.9 10 Patients and families sometimes describe driving restrictions as the hardest part of living with an ICD, and adherence to restrictions is limited.10 11 Accordingly, European and recently published Canadian guidelines have shortened the duration of driving restrictions.7 8

Postimplantation driving restrictions are based on studies with major methodological limitations. Most studies are underpowered for traffic injuries and fatalities.12 Many studies underestimate crash risk because self-reporting of driving behaviour and crash involvement can be subject to recall and social desirability biases.11–13 Other studies extrapolate from baseline risks, ignoring evidence that driving itself may influence the likelihood of arrhythmia.13–17 Studies that draw from randomised trial cohorts or convenience samples tend to under-represent females, racial minorities, socioeconomically marginalised individuals, patients with multimorbidity and recipients of primary prevention ICDs.11–13 18 Most studies have limited clinical utility because they predate modern ICD programming and cardiovascular drug therapy, both of which markedly reduce the incidence of arrhythmias and shocks.12 17 18 As a result, it remains unclear whether postimplantation driving restrictions are effective and beneficial.

To inform clinical decision-making, we used population-based administrative health and driving data and a retrospective cohort design to examine crash risk in the first 6 months after ICD implantation relative to crash risk among age-matched and sex-matched controls.

MethodsStudy setting

Our study was set in British Columbia (BC), Canada, a province with a land area four times larger than the UK and 40% larger than Texas. Almost 90% of BC’s 5 million residents live in metropolitan areas. Publicly funded universal health insurance provides access to medical care (including ICD implantation) that is free at the point of service. The Insurance Corporation of British Columbia (ICBC) provides mandatory basic automobile insurance for all vehicles registered in BC and maintains licensing data for all 3.2 million drivers. There were 9.9 traffic fatalities and 613 traffic injuries per billion vehicle-kilometres at study midpoint.19

Health data

We used Cardiac Services BC’s cardiac device registries to identify first ICD implantation date and indication (primary or secondary prevention of sudden cardiac death). We addressed missing indication by pooling 40 multiply imputed datasets using a validated predictive algorithm (online supplemental item S2).20 We used unique health insurance numbers to deterministically link registry data to population-based administrative health data that include hospitalisation records, physician fee-for-service claims, outpatient prescription drug fills and death records (online supplemental item S2).21–23 Comorbidities were considered present when encoded in any of the 25 diagnosis fields for ≥1 hospitalisation or any of five diagnosis fields for ≥2 physician visits in a 3-year look-back interval (online supplemental item S3). Active prescription medication use was established using dispensation date and days supplied from all prescriptions filled at any community pharmacy in BC.

Driving data

We obtained province-wide population-based driving history and crash data from ICBC. The Traffic Accident System captures data on all police-attended crashes in BC and we used it to identify fatal crashes. The ICBC Claims File includes all crashes that resulted in an insurance claim involving a vehicle registered in BC, and we used it to identify injury crashes because police attendance at non-fatal crashes is discretionary. Both sources identify the driver of every crash-involved vehicle. Baseline driver history was established using a 3-year look-back interval. We linked health and driving data using a previously established probabilistic linkage based on name and birthdate, with linkage rates ≥95%.22

Study cohort

We identified individuals with a first ICD implantation between 1 January 1997 and 31 October 2019, including primary and secondary prevention ICDs and those with biventricular leads or cardiac resynchronisation functions. We excluded patients with a prior ICD who only underwent generator or lead replacement. Each ICD patient was matched without replacement to three controls based on sex and birth year. Residents registered with BC’s provincial health insurance plan in the calendar year of their match’s ICD implantation were eligible as controls. ICD implantation date served as the index date in each matched set.

Primary analysis

The primary outcome was involvement as a driver in a motor vehicle crash in the first 6 months after ICD implantation (or after index date for controls; figure 1). Drivers with ICDs and their matched controls were followed from baseline (t0=index date) until first eligible crash or until the first censoring event: (1) death, (2) licence suspension or expiry for >30 days, (3) subsequent hospitalisation for >30 days, (4) subsequent ICD procedure (eg, lead or generator change, device upgrade, explant), (5) completion of 6-month follow-up or (6) study end (31 October 2019). Individuals without a valid BC driver licence at baseline are presented in table 1 but censored from analysis at t0, thus contributing no person-time to the analysis; this is analytically identical to excluding these patients from the cohort. The primary analysis used a Cox proportional hazards model to compare crash-free survival in ICD patients and controls after adjusting for potential confounders. The final model adjusted for age and sex; index year; number of overnight hospital stays and number of physician clinic visits in the past year; active vehicle insurance policy in the past 3 years (≥90% vs ≤89% of days); number of non-alcohol-related traffic contraventions (ie, violations) in the past 3 years; and any crash in the past 3 years. The proportional hazards assumption was tested using scaled Schoenfeld residuals (online supplemental items S4 and S5).

Figure 1

Cohort study schematic. Individuals who received a first implantable cardioverter-defibrillator (ICD) implantation between 1 January 1997 and 31 October 2019 were matched with three control individuals from the general population. We used a 3-year look-back from implantation date to establish baseline health and driving features. The primary analysis assessed the risk of crash in the first 6 months after the ICD implantation.

Table 1

Baseline characteristics

Additional analyses

We examined crash risk among clinically relevant subgroups defined by age, sex, rural residence, Charlson Comorbidity Index ≥2, cardiovascular disease, implantation era (1997–2005 vs 2006–2012 vs 2013–2019) and ICD indication. We used a piecewise Cox proportional hazards model with dummy variables for ‘time period from t0’ (0–30, 31–90, 91–180, 181–365 and 366–730 days since index date) to examine temporal variation in crash risk. We plotted weekly and monthly crash risk among ICD and control patients before and after the index date. We examined traffic contraventions before and after the index date as a potential surrogate for road exposure (ie, hours or miles of driving per week).22

Data were deidentified before release to investigators and were rarely missing (online supplemental item S2). Analysis occurred between January 2022 and November 2023 using R V.4.0. All inferences, opinions and conclusions drawn are those of the authors and do not reflect the opinions or policies of the Data Stewards. It was not possible to involve patients or the public in the design, or conduct, or reporting, or dissemination plans of our research.

Results

The study cohort consisted of 37 492 individuals, including 9373 patients with ICD implantation and 28 119 controls (median age (Q1, Q3), 66 years (56, 73); 19% female; 52% urban residents; race data unavailable; table 1; online supplemental item S6). Prior to imputation, 39% of ICDs were primary prevention ICDs, 36% were secondary prevention ICDs and 26% had missing/unknown indication (online supplemental item S7). As expected, baseline cardiovascular comorbidities were far more common among individuals with ICDs than among controls (eg, congestive heart failure was 23-fold more common; myocardial infarction, 18-fold; syncope, 14-fold; atrial fibrillation/flutter, 10-fold). Baseline medication use exhibited a similar pattern (eg, loop diuretics, 14-fold more common; beta blockers, 5-fold; other antiarrhythmics, 18-fold). About 70% of the cohort had an active vehicle insurance policy in the 3 years prior to index date. About a fifth of the cohort did not hold an active driver licence on the index date; as prespecified they were included in table 1 to better describe all ICD recipients but censored at t0 and contributed no at-risk person-time to subsequent analyses (online supplemental item S8).

In the first 6 months after the index date, 296 of 9373 individuals with ICD implantation and 1077 of 28 119 matched controls crashed prior to a censoring event, suggesting ICD implantation was associated with a reduced risk of subsequent crash (crude proportion, 3.2% vs 3.8%; crude incidence rate, 8.5 vs 10.5 crashes per 100 person-years; aHR, 0.71; 95% CI 0.61 to 0.83; p<0.001; figure 2, online supplemental items S9 and S10). About one-fifth of all crashes resulted in an injury or fatality (online supplemental item S11). Among the 139 police-attended crashes occurring in the first 6 months after the index date, officers only rarely reported ‘illness/fatigue’ as a contributing factor for either ICD patients or controls (<5 crashes in each of exposed and control groups; online supplemental item S12).

Figure 2

Kaplan-Meier survival curves. Patients with implantable cardioverter-defibrillator (ICD) implantations are depicted in the upper line in red. Controls are depicted in the lower line in black. The curves separate early then run nearly parallel, suggesting reduced road exposure in the first month after implantation largely drives the reduced crash risk after ICD implantation.

In a subgroup analysis, 122 of 4669 individuals receiving a secondary prevention ICD and 531 of 14 006 controls crashed prior to a censoring event during a 6-month follow-up interval (crude proportion, 2.6% vs 3.8%; crude incidence rate, 7.2 vs 10.5 crashes per 100 person-years; aHR, 0.57; 95% CI 0.45 to 0.72; p<0.001; online supplemental item S13). Similarly, 18 of 4704 individuals receiving a primary prevention ICD and 96 of 14 113 controls crashed prior to a censoring event during a 1-month follow-up interval (crude proportion, 0.4% vs 0.7%; crude incidence rate, 5.8 vs 10.6 crashes per 100 person-years; aHR, 0.48; 95% CI 0.26 to 0.89; p=0.022; online supplemental item S15). The results of other subgroup and sensitivity analyses were generally consistent with the main analysis (figure 3, online supplemental items S13–S16).

Figure 3

Forest plot of subgroup analyses. According to a prespecified analysis, interaction terms were used to evaluate whether the relationship between implantable cardioverter-defibrillator (ICD) implantation and crash risk was modified by sex, implantation era and ICD indication. We found that the relationship between ICD implantation and subsequent crash was significantly different among primary prevention ICD recipients compared with among secondary prevention ICD recipients (interaction term p value 0.03). Neither sex nor implantation era significantly modified the relationship between ICD implantation and crash. The latter finding is interesting in light of the improvements in cardiovascular drug therapy and ICD programming that occurred over the study interval.

The piecewise time interval analysis found that ICD recipients had a crash rate lower than controls in the first month after primary prevention ICD implantation and in the first 6 months after secondary prevention ICD implantation, intervals that correspond to contemporary driving restrictions (figure 4, online supplemental item S14). Crash rates among ICD recipients never exceeded those among controls in any tested interval in the first year after ICD implantation. Plots of weekly and monthly crash risk suggest no dramatic increase in crashes immediately after typical driving restrictions end (online supplemental item S15).

Figure 4

Relative crash rates in distinct time intervals after implantable cardioverter-defibrillator (ICD) implantation. Forest plot of relative crash rates for different time intervals after ICD implantation (0–30, 30–90, 90–180, 180–365 and 365–730 days since index date; estimates for each interval are shown near the last day of the corresponding interval). Adjusted HRs comparing crash rates among individuals with primary prevention ICD implantation to crash rates among controls are depicted in green; those comparing individuals with secondary prevention ICD implantation to controls are depicted in blue. Main findings are: (1) crash rates are lower among ICD recipients than among controls in the first 30 days after primary prevention ICDs and in the first 180 days after secondary prevention ICDs, corresponding to the typical duration of contemporary driving restrictions; (2) crash rates among ICD recipients never exceed those among controls in any tested interval in the first 2 years after ICD implantation.

Relative to the subset of controls with an active license at t0, the subset of ICD patients with an active licence at t0 had more traffic contraventions in the 3 years before implantation (crude proportion, 21% vs 19%; crude relative risk (cRR), 1.10; 95% CI 1.05 to 1.16; p<0.001) but fewer contraventions in the 6 months after implantation (crude proportion, 2.9% vs 4.3%; cRR, 0.68; 95% CI 0.59 to 0.78; p<0.001), suggesting that a reduction in road exposure or more conservative driving after ICD implantation might explain the observed reduction in crashes (online supplemental item S17). However, 98 drivers received a moving vehicle traffic contravention in the 6 months after implantation of a secondary prevention ICD (compared with 166 in the prior 6 months), implying that road exposure remained at 59% of baseline during the typical period of restricted driving. Similarly, seven drivers received a moving vehicle traffic contravention in the month after implantation of a primary prevention ICD (compared with 26 in the prior month), implying road exposure remained at 27% of baseline during the typical period of restricted driving.

Discussion

We examined a population-based retrospective cohort of 37 492 drivers over a 22-year study interval using real-world driving data and found that crash risk in the first 6 months after ICD implantation was substantially lower than among matched controls. Our findings provide indirect evidence that these patients reduced their aggregate road exposure but did not cease driving in the months after ICD implantation, strongly suggesting inconsistent adherence to contemporary postimplantation driving restrictions.11 Our novel findings provide strong empirical evidence that could be used to make postimplantation driving restrictions less burdensome to patients while continuing to protect other road users.14 15

We believe our results suggest that policymakers should liberalise current postimplantation driving restrictions, perhaps reducing the postimplantation driving restriction to 1 week after primary prevention ICDs and to <3 months after secondary prevention ICDs. These suggestions are in line with European and recently published Canadian guidelines.8 Less onerous restrictions reduce barriers to ICD implantation and improve quality of life among recent ICD recipients, who our data suggest are likely to exhibit an annual per capita crash rate similar to that of the general population. Adherence to restrictions might be improved if shorter restrictions are more persuasively supported by data on real-world crash risks.11 24 25 However, crash rates should be carefully monitored after these changes are made. It is possible, for example, that a fourfold increase in the relative risk of crash while driving in the weeks after ICD implantation is currently masked by a parallel 75% reduction in road exposure, and that more permissive driving restrictions will increase traffic risks for ICD recipients and other road users. Population-based monitoring of crash rates could be substantially strengthened by research that links driver monitoring technology (eg, driver-facing cameras, vehicle speed and braking data) to electrocardiographic tracings from the ICD itself to establish whether a given crash was plausibly related to cardiovascular incapacitation.26

Our study more than doubles the total number of drivers with ICD implantation in the published literature, but our results should still be considered in the context of prior studies (online supplemental item S18). A small single-centre case series from the 1990s with a median follow-up of 38 months found that 6% of 171 drivers reported crashing after implantation of a secondary prevention ICD.18 The Antiarrhythmics versus Implantable Defibrillators (AVID) trial found that 627 enrolled drivers self-reported a subsequent annual incidence of crash lower than among the US general population (3.4% vs 4.9%).12 Among the 295 trial participants who resumed driving after secondary prevention ICD implantation, 8% reported a shock while driving but there were no reports of shocks that resulted in a crash.12 24 A case–crossover analysis of study data from the Triggers of Ventricular Arrhythmia (TOVA) trial evaluated 259 ICD shocks, finding that there was only one shock for every 25 116 person-hours of driving and concluding that driving itself does not appear to increase the risk of ventricular arrhythmia.17 A multicentre prospective cohort study in Ireland with a mean follow-up of 27 months found that only eight (3.3%) of 241 ICD recipients reported receiving a shock while driving, five of which were reported to result in minor crashes.27 An Italian survey of 213 ICD patients found an annual reported crash rate of only 1.1%; investigators deemed only one of 11 crashes to be potentially related to an ICD.28 Another survey of 2741 Danish residents over a median follow-up of 2.3 years uncovered only five reported shocks while driving, only one of which was reported to result in a crash.11

Our study has many strengths. It is sufficiently powered and includes an appropriate control group. A straightforward population-based cohort design generated both absolute and relative risks while accounting for competing risks such as licence suspension and death. We focused on individuals with an active driver licence and used objective instead of self-reported crash data. Use of objective crash data also avoided the many untestable assumptions and extrapolations inherent to estimates based on the Canadian Cardiovascular Society’s widely used risk-of-harm formula.10 Granular health data captured comorbidities, medication use and the indication for ICD implantation. Unlike prior studies, we accounted for baseline driving habits (eg, prior crashes, traffic contraventions; these may also partly account for baseline road exposure) and could distinguish subsequent crash involvement as a driver from crash involvement as a passenger. Our study interval spanned more than two decades and encompassed the modern ICD era in which antitachycardia pacing and other innovations markedly reduced the incidence of shocks.14 15

The major limitation is the lack of a direct measure of road exposure. We partially addressed this by using moving vehicle contraventions as a surrogate for road exposure, an approach that is reasonable but has not been validated. It is also possible that patients more conscientiously follow road rules in the first weeks after ICD implantation. There are other limitations. Some control drivers might have undergone ICD implantation in another jurisdiction, but this was probably extremely rare. Individuals driving despite a restriction might have avoided reporting a crash to the police or to the insurer, but this is likely rare for serious crashes. We lacked granular clinical details for some patients including left ventricular ejection fraction and New York Heart Association class. Administrative data might underestimate the prevalence of some comorbidities and medications. We focused on all-cause crashes because this outcome is important to patients and clinicians, but we could not definitively identify the crashes that were caused by cardiovascular incapacitation. Patients at greater risk of shock may have voluntarily surrendered their driver licence, leading to over-representation of patients with comparatively less severe conditions. We lacked data on ICD deactivation, but this is typically associated with transition to palliative care and elective driving cessation and would not be expected to significantly impact results.29 We did not study commercial drivers. The specific physician driving advice provided to each patient remains unknown. Our findings may not generalise to other jurisdictions. Many of these shortcomings could be addressed by a large, multicentre randomised controlled trial that directly measures empirical road exposure and compares crash risk among ICD recipients assigned to a more lenient postimplantation driving restriction to crash risk among those assigned to a more stringent driving restriction. The feasibility of such a trial might be limited by the relative rarity of serious crashes in the first few weeks after ICD implantation.

Medical driving restrictions can significantly limit quality of life. In this population-based retrospective cohort study, we found that crash risk in the 6 months following an ICD implantation was lower than among control drivers. These findings have the potential to strengthen current clinical practice and improve fitness-to-drive policy by making postimplantation driving restrictions less burdensome to patients.

Data availability statement

Data may be obtained from a third party and are not publicly available. Access to data provided by the Data Stewards is subject to approval, but can be requested for research projects through the Data Stewards or their designated service providers. All inferences, opinions and conclusions drawn are those of the authors and do not reflect the opinions or policies of the Data Stewards.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

The University of British Columbia Clinical Research Ethics Board approved the study and waived requirements for individual consent (H16-02043).