Restrictions on driving and work for spells involving dizziness or transient loss of consciousness

Timothy C. Hain, MD Page last modified: January 8, 2017

There are many conditions that can make driving or work dangerous -- for example, syncope (fainting), seizures, and strong vertigo spells such as sometimes occur in Meniere's disease.

According to Barbic (2014) The 10 most dangerous jobs based on fatalities are as follows:

Job Description Main cause of death
Timber and logging workers Contacts with objects and equipment
Fisherman and related Transportation
Aircraft pilot and flight engineers Transportation
Structural iron and steel Contact with objects and equipment
Farmers and ranchers Transportation
Roofers and Lineman Falls
Electric power-line installers

Harmful substances/environment

Drivers and truck drivers Transportation
Refuse workers Transportation
Military and police Transportation
Construction laborers Falls
Firefighters Fires and explosions
Helpers, construction trades Falls
Grounds maintenance workers Falls

As can be seen, driving (transportation) and falls are large sources of death. In an attempt to put this into perspective, syncope or transient loss of consciousness is nowhere near the biggest risk factor for driving accidents. Sagberg (2005) analyzed 4448 crashes in Norway and noted that the biggest risk was unmedicated diabetes (risk 3.08), history of myocardial infarction (1.77), using glasses while driving (1.26), myopia (1.22), insomnia (1.87), tiredness (1.36), anxiety (3.15), feeling depressed (2.43) and taking antidepressants (1.7) and stroke (1.93). It is apparent here that psychological factors (such as anxiety, depression, tiredness) dominate the list. Not on the list is dementia and visual disturbance other than glasses -- for example cataracts - -we would think these would also greatly increase risk.

Nevertheless unpredictable spells involving loss of consciousness or dizziness can greatly increase risk associated with ordinary activities such as driving, flying, or just walking down the street. Raj (2009), from the Mayo clinic, reported that roughly 10% of patients with syncope had it during driving. Folino et al stated that 3-10% of patients with syncope experience episodes while driving (2012). Of course, this statistic should logically be interpreted in view of how much the individual drives after they have an event. In Raj's study, syncope during driving was distributed similarly to syncope when not driving - -about 37% from "neural" or vasovagal syncope, and 11.8% due to cardiac arrhythmia.

Bhatia et al (1999) stated that "syncope and injury while driving in patients with neurocardiogenic syncope is rare. " We see this is logical - given that there is a clear connection between an adverse event - -lets say having blood drawn -- and syncope, and that "vasovagal" is not just a synonym for "unknown", the risk of having another one during driving would seem very low. Sorajja et al (2009) reported, apparently using the same data set as Raj (2009), that syncope reoccurred in 72 patients of 381 who had syncope while driving, 7 of which were more than 12 months after the initial evaluation. Thus the chance of reoccurrence of syncope (not necessarily while driving) is roughly 20%. Logically then, the risk of having another syncope during driving should be equal to the product of time spent driving, and chance of recurrent syncope. Lurie et al (1999) surveyed 66 cardiologists regarding their recommendations regarding resumption of driving in vasovagal fainters. In asymptomatic patients with a positive tilt table test, negative on "treatment" (whatever that is), the mean recommended period without driving was 54 days. The most popular (mode) recommendation was 0 days (which makes sense). Perhaps the longer recommendations express some ambiguity between vasovagal and cardiac causes of faint, as one would want to be more cautious when one doesn't know the diagnosis. Another way of looking at it however is that this survey is simply documenting irrational treatment behavior among cardiologists, as we do not see how a tilt table test (measuring autonomic function) is related to emotional fainting (such as from a blood draw), and we also do not see how beta-blocker treatment has anything to do with risk of another emotional faint.

Guzman and Morillo (2015) suggested that "The presence of structural heart disease (reduced ejection fraction, previous myocardial infarction, significant congenital heart disease) potentially leads to high risk and should determine driving restrictions pending clarification of underlying heart disease and etiology of syncope. " In other words, syncope with high mortality should also have driving restrictions. Miles (1997) offered a contrary opinion and stated that "available data and expert consensus suggest that most patients with arrhythmias can return to driving with a relatively low risk of harm to themselves and others, that is, a risk within the limits deemed acceptable by society. " Larsen et al (1994) suggested that "All survivors of VT or VF should refrain from driving during the first month after hospital discharge when the hazard for events that could impair their ability to drive is greatest. Our data would support restricting driving for most patients until the eighth month after hospital discharge, when risk becomes lowest." Lerecouvreux et al (2005) commented that patients in France. with implanted ICD, often ignored advice from their physicians to abstain from driving for 3-6 months after implantation.

Folino et al (2012) examined risk of syncope in 40 persons who had syncope while driving, and compared to 50 subjects who had syncope in other contexts, using tilt table tests. We wonder a bit about the rationale for doing this as one is not tilted up while driving. The patients with syncope while driving had a higher prevalence of hypertension and positive tilt table tests They reported "Our study confirms a good prognosis for patients who experience syncope while driving, and indicates that more accurate risk stratification is needed in subjects aged over 50 years, who have had more than 4 episodes of loss of consciousness and vasodepressive reactions during UTT." We suspect that the value here is identifying heart disease.

Josephson et al (2007) suggested that some syncope patients are misdiagnosed as epilepsy, causing "unnecessary restrictions" on driving and unemployment. This points out that certain types of transient LOC (such as epilepsy) have much tighter legal controls (at least in Canada) than the others (such as cardiac arrhythmia).

Defining risk in a mathematical sense

Sorajja and Shen (2010) proposed using an algorithm originally developed by the Canadian Cardiovascular Society (1992) that took into account both the planned activity and the riskiness of the medical condition. They suggested:

"When the cause of syncope is unknown, the response to treatment is uncertain (such as treating the neurocardiogenic/vasovagal syncope), or ICD discharge is possible, driving recommendations are based on the estimation of "risk of harm while driving" and the general consensus on the threshold of "acceptable risk of harm." The annual risk of harm while driving can be estimated by the following formula: driving time (%) x vehicle type (commercial to private) x annual risk of syncope or incapacitation x probability of injury or accident. " This is the same formula as previously proposed by the Canadian Cardiovascular Society discussed in more detail below.

Note that Sorajja and Shen imply that "vasovagal" syncope is a variant of "unknown". Also, there are still quite a few judgment calls in this algorithm. But it is a start.

Barbic et al (2014) also suggested "quantitative model" to "stratify risk". They stated that two main variables to establish are the risk of syncope recurrence, and the expectation of harm if syncope occurs.

They presented a Risk of Harm Formula, as first published by the Canadian Cardiovascular Society (1992).

RH = TD x V x SCI x AC.

  • RH indicates the acceptable risk of harm (0.00005 chance per year of serious injury or death, from an event).
  • TD is the driving time over the day, or distance driven in a given time period (0.04) -- just 4% of the day, and 25% (0.25) for commercial drivers.
  • V is the value depending on the type of vehicle (0.28 for private cars), larger vehicles (> 3.5 tons) have a V of 1.0 . According to Simpson et al (2004), Loss of control of a heavy truck or passenger-carrying vehicle results in a more devastating accident than loss of control of a private automobile. Truckers are involved in only approximately 2% of all road accidents but in approximately 7.2% of all fatal accidents. (Simpson et al, 2004, who quotes Parsons (1986))
  • SCI is the risk of "sudden cardiac incapacitation". It seems to us that the "Cardiac" is portion is not needed here -- this formula is really talking about incapacitation in general.
  • AC is the probability that such an event, e.g. syncope, will result in a fatal or injurious accident (0.02).

Lets look at these numbers more carefully.

Where did the Risk of Harm value -- RH come from ? A reasonable estimate of RH is the real world. The average driver has a 4-5% risk of a crash every year(Vernon et al, 2002; Des Toupes, 2011). There usually are options of driving less, driving more carefully, etc. Thus it appears that society, on average, accepts a 4-5% risk of an automobile crash. Of course, this is not the same as accepting a 4-5% risk of death. According to The National Center for Health Statistics (2013), the one-year odds of dying in an auto accident for a car occupant is 1/47,718. This is 0.00002, roughly half of the "acceptable risk of harm" figure from The Canadian Cardiology society. Thus it would seem that the "RH" parameter is a proxy for acceptable risk of death, while driving. The CCS's recommended maximum RH is roughly 2 times the baseline death-rate from auto accidents. Why pick 2 times rather than 10 times the death-rate ? It would seem to us that individuals might reasonably choose to accept a larger RH, in return for (for example), an ability to continue working, or perhaps to drive to visit their children.

In order to get to the desired value of RH, it appears the other numbers were "guesstimated". In support of this conjecture, the source for these other numbers seems to include only a few papers or specialty society wisdom. Lets consider the "TD", "V" and the AC parameters.

The TD parameter is the time driving. It is set at 0.04 -- meaning about 1 hour of driving/day. Presumably 30 min to work, 30 min back. Not unreasonable, but certainly very variable depending on where one lives, where one works, and other activities that might include driving. For commercial drivers, the figure of 25% was derived by the Canadian Cardiovascular society from a paper by Parsons published in 1986 (Simpson et al, 2014). The paper by Parsons included roughly 220 cases. 25% is 6 hours. This estimate of time driving seems a little low to us for commercial drivers. The basis for this estimate, a single article by Parsons published in 1986, also seems weak. Why would commercial drivers work less than 8 hours/day ? Joly et al (1993) reported an average of 8.7 hours/day for commercial truck drivers. One would also think that there would be trade-off between speed (higher risk) and time (higher risk). Where is the current data ?

The SCI parameter -- risk of sudden cardiac incapacitation -- is set at a baseline of 1%/year -- is essentially a value judgement of several Canadian societies. Perhaps correct, or perhaps not. SCI is assumed to be 1% for the general population. One incapacitating attack/100 years. According to Simpson et al (2004), this number was extrapolated from the Canadian guidelines for commercial drivers, which were those of several specialty societies. In other words, SCI was basically chosen by the Canadian specialty societies, because they thought it was reasonable. Of course, someone needs to mark a "line in the sand".

The AC parameter -- expressing the risk that an event will be lethal to the driver, other road users, or bystanders was set at 0.02. This is the probability that a dangerous event during driving, expressed by SCI, will kill someone. According to Simpson et al (2004), quoting 4 sources, "fewer than 2% of reported incidents of driver sudden death or loss of consciousness have resulted in injury or death to other road users or bystanders". AC was set to 2% for all drivers. We would wonder here -- how about injury to the driver, injury to the vehicle, and how fast are people driving (i.e. highway or local ?). We don't see how one who suddenly loses consciousness (e.g. heart stops) could possibly avoid an accident. On the other hand, there are almost certainly variants of loss of consciousness that occur slowly enough to allow the person to pull over. There are many missing details.

For seizures, RH in the more liberal legal jurisdictions appears to be set at roughly 0.001 -- i.e. one death from an auto accident every 100 years.

It appears to us that there was a "working back process" from commercial vehicle (set at 1), as it seems to us that the numbers are adjusted to make sure that a single event in a commercial driver stops future commercial driving. In other words, the RH formula is a policy statement dressed up with mathematics.


Nevertheless, it is always good to have place to start, one might also generalize this formula to other activities, which might allow one think about risk of harm in other contexts than driving or cardiac syncope.

Well anyway, Barbic et al (2014) stated that using the formula above, if the acceptable RH is 0.00005 (1/20,000/year) then the yearly acceptable SCI is 22%. The inference was that if the acceptable SCI is < 22%, then patients may be returned to driving -- an SCI of 20% is 1 spell every 5 years. This is a very conservative and safe standard, but as it would eliminate the possibility of driving for anyone with a spell in the last 5 years, and because of this it seems rather unlikely to be followed. If one looks more carefully at the underlying numbers, it also seems incongruous with real world experience as the baseline accident rate in the general population is about 4%/year (Vernon et al, 2002), about 100 times greater than the RH statistic.

Risk of harm table

A risk calculator using the RH formula is here.

Practically, the legal standard for the acceptable frequency of transient disabling events such as seizures, for non-commercial drivers in the US, ranges between 3-24 months of being seizure free (Winston and Jaiser, 2012). Drivers with epilepsy have modestly higher rates of crashes and at-fault crashes (1.47 to 2.39). This data comes from a large study in Utah, which of course may not reflect the situation in other jurisdictions (Vernon et al, 2002) We can use this to estimate the range of legally acceptable RH. At one end is 4 seizures/year, or SCI=400%, or at the other, one seizure every other year - SCI=50%. The RH for these assumptions is shown above. This table implies that society accepts a roughly 20 times greater risk of harm, i.e. RH, for patients with epilepsy than the Canadian Cardiology Society recommended for fainting spells. If we do the acceptable harm (RH) calculation for 4 spells/year, given AC=0.02 (which is questionable), this results in a RH of 0.000896 chance per year of serious injury or death. Or in other words, about one chance/1000 of an accident/year from seizures, given that they occur every 3 months. As real world data suggests that society accepts a 4% risk of crash in ordinary drivers every year, and the real risk data is very different (see below), the RH risk goal seems to us to be hugely different than actual experience.

Another way of looking at this is that if the RH formula works, then the risk ratio between normal persons and persons at risk, predicted by the RH formula should match the measured risk. This does not appear to be anywhere close to the situation for seizures. The data from the very large Utah study (about 2 million drivers) suggests a roughly 2-fold greater risk of a crash in seizure patients (Vernon et al, 2002). From the same study, the risk of a crash in controls/year is about 4%/year, thus suggesting that the risk of a crash in persons with seizures is about 8%. Another relevant statistic is that auto insurance companies process claims for collisions in roughly 5% of drivers/year. These numbers are hugely greater than the RH calculation, and obviously the simple RH formula as proposed by the Canadian Cardiovascular Society doesn't match real world experience. Other factors must present. Our thought is that the RH formula neglects to consider baseline crash risk from behaviors such as texting while driving, or even talking while driving, as well as of events beyond the drivers control.

Specifically, to account for the real world data, we suggest that the AC figure is wrong for seizures (perhaps also so for other conditions). It is hard to imagine a generalized seizure that didn't result in an accident. For 4 seizures/year (SCI=4), and for a probabilty of 100% (AC =1) rather than 0.04, i.e. every seizure results in incapacitation and a crash, this could account for these numbers. There also might reasonably be an effect on driving judgement from seizure medications -- many of which make it harder to think -- which could in this model be done by increasing SCI again. These are conjectures that would certainly benefit from more study. They also have clinical implications as clinicians might reasonably adjust advice to patients.

With this in mind, now lets compute the RH for commercial drivers. If we set driving time to 6 hours/day (almost certainly too low), and use the 1.0 "V" commercial vehicle parameter, this works out to a RH for a healthy commercial driver of 0.00005. Or in other words, exactly at the maximum RH the Canadian Cardiology Society suggests is acceptable. Following this logic, any syncope event at all in a commercial driver should result in unacceptable risk (for the Canadian Cardiology Society), although as discussed above, society seems to accept roughly 100 times more risk for non-commercial drivers with seizures. If one uses the 1/1000 acceptable criterion instead (the set-point for seizures), then one event every 5 years (an SCI of 0.2) would meet this requirement. So logically then, any transient disabling event in a commercial driver with a significant chance of recurrence, should result in loss of their commercial license. As noted above, many of the assumptions in the formula above seem very arbitrary, and undoubtedly vary from person to person. In commercial drivers, the TD parameter is not adjustable. On the other hand, commercial drivers have less accidents/mile (and presumably time driving) than do non-commercial drivers, so the baseline AC parameter should reasonably be lower for commercial drivers.We are not convinced that commercial drivers handle sudden neurological events such as fainting or dizziness much better than non-commercial drivers.

For dizzy spells, there is no loss of consciousness, but there may be a loss of control. This would be parameter AC. We are dubious that AC (2% chance of a dizzy spell resulting in an accident) should be any less than 2%. Thus again, it would seem that one incapacitating dizzy spell every 3 months converts into an estimate of one harmful spell/1000. Certain types of dizzy spells -- e.g. drop attacks -- would make parameter AC 100%, and thus imply that driving should not be allowed unless the condition causing the drop attack is permanently stopped (e.g. with a destructive procedure). On the other hand, if dizzy spells were treated similarly as seizures, where one seizure/3 months is acceptable for driving in some jurisdictions, one would think that dizzy spells would be generally safer, and would convert into less risk.

For seizures, parameter AC might reasonably be 100% for a "tonic clonic" seizure -- which would again imply that just a single TC seizure would be enough, from this formula, to "pull" the non-commercial driver's license. On the other hand, Sheth et al (2004) argued that "Fatal driver crashes due to seizures are uncommon. This finding supports the current public policy of permitting patients whose seizures are controlled to drive. " From the logic developed above, the risk of crashes in seizure patients is about 2 times the risk of crash in controls, and this added risk does appear to be tolerated by society.

An implication of the formula above, is that commercial airline pilots as well as their regulators might need to be more assured that there will be no repeat of a transient loss of consciousness than individuals whose main exposure to risk is driving 2 blocks to church every Sunday. Similarly, commercial drivers or bus drivers might need to have better evidence of a low risk of passing out or becoming dizzy than ordinary drivers. This seems very logical to us, but of course it is bound to create disagreement as well as has legal ramifications. Practically in the US, it has been our observation that pilots on medications associated with risky conditions, such as seizure medications, are restricted from commercial flying. This can create a "catch-22" situation where a pilot might choose to go untreated in order to return to flying.

Most dangerous activities (such as driving and being a pilot) are regulated by law. In the United States, private driving and flying is regulated at a state level, while commercial driving and flying is regulated at the federal level (Sakaguchi and Li, 2013).

Sensible persons with episodic loss of consciousness self-restrict their driving (i.e. they reduce TD). Older persons in particular often stop driving due to their medical conditions, sometimes including syncope, although very few have their license revoked (Campbell et al, 1993). On the other hand, Akiyama et al (2001) commented that "Most patients with ventricular tachyarrhythmias resume driving early. Although it is common for them to have symptoms of possible arrhythmia while driving, accidents are uncommon and occur with a frequency that is lower than the annual accident rate of 7.1 percent in the general driving population of the United States." Note that this figure of 7.1% for "accidents" is different than the 1% rate of "crashes" quoted by Vernon et al (2002) in Utah. We would speculate that this means that while patients resume driving after a "spell", they reduce their driving (as is very sensible). Furthermore, driving is intrinsicially very risky, and the added risk from events is often disregarded as being a "drop in the bucket". The more dangerous persons might be those with no "common sense" or perhaps with some degree of suicidality, who drive or fly anyway.

Work risks.

According to Barbic (2014), between 16-25% of syncope occurs during work activity. Certain occupations have more risk than others. Barbic (2014) noted that the European society for cardiology suggested that professional drivers treated for arrhythmias need verification of the effectiveness of their treatment before resuming work. One would imagine the same would be true for commercial pilots. ICD implant rules out professional driving. (Task force, 2009). In general, Barbic stated that return to commercial driving after syncope should be avoided for at least a year (2014).

After unexplained syncope, the European task force suggested that return to professional driving should occur only "after diagnosis and appropriate therapy is established" (2009). In other words, never as long as there is no cause. One would imagine that this might motivate professional drivers to find a "cause" for syncope.

In the US, the US dept. of Transportation Federal Motor Carrier Safety administration states that a driver may be certified to drive if he/she is "at low risk for syncope/near syncope". This is very vague.

Work environments vary to a greater extent than driving, and there are a great many more variables to consider. According to Barbic (2014), syncope while working in a high place with a risk of fall is about 60 times more dangerous to the worker than working in an office. Harm to 3rd parties is about 250 times greater for ambulance drivers/emergency vehicle drivers than for office workers. Driving a truck in a construction area is dangerous both to the driver as well as third parties. Similarly, syncope is more dangerous in surgeons.