Units of Risk

Introduction

The two most common units of risk used in traffic casualty studies are: fatalities per billion kilometres (F/BnKM); and fatalities per million hours’ use (F/MHU).

You will notice that these units both contain enormous numbers: millions of hours, billions of KM. How can this be? Well, most forms of travel are fairly safe by everyday standards. That is why deaths in road crashes seem to be “someone else’s problem” for most people most of the time. We all know of people who were killed in road crashes. They stick in the mind not least because they are comparatively rare – and much rarer nowadays than forty years ago.

The exception is riding motorcycles. The reputation for danger surrounding motorbikes is real – a point I make with regret, having once been a die-hard aficionado. With great acceleration, speed and agility, motorcycles offer excitement and camaraderie, but they are unforgiving machines. Errors of judgement, road faults, mistakes by drivers; all have far greater consequences on motorbikes than in other forms of travel.

Whereas, walking, cycling and driving involve risks in the same general range, as we shall see. Undoubtedly all of these types of travel are dangerous for the naive, the heedless, the intoxicated, the elderly, or if the traffic situation is inherently conflicting, but they are otherwise “safe” by everyday standards for sensible people in sensible circumstances.

We’ll look at the two main units of risk each in more detail.

Fatalities per Billion Kilometres

This has traditionally been the favoured expression of risk in personal travel. It would appear sensible to compare risks on the basis of distance travelled. After all, most trips in life are over a set distance: the trip to work; to the local supermarket; to school; to the pub.

Let us consider a risk of two fatalities per billion kilometres, or one fatality per five hundred million KM. How can such an abstracted concept be related to everyday life? To drive 500 million kilometres would take more than 1,900 years, assuming 18 hours of typical driving every single day. How can anyone get killed driving?

You can get killed driving, but the risk is not high. First of all, 2 F/BnKM does not mean you will only die if you drive 500 million KM. It means that every KM you drive incurs a risk of death of 1 in 500 million. If you drive a million KM in your life (a pretty typical distance), the chance of being killed is (about) 1 chance in 500, or 0.2%. This would not appear to be a high risk.

Now let us consider a city of a million people, each of whom drives for an hour a day (say 40KM on average). How often would a person get killed if the average risk were 2 F/BnKM? Well, there are 40 million KM driven every single day, so we would expect one death every 12.5 days (say once every two weeks). Suddenly, a risk of 2 F/BnKM does not seem so small. You can appreciate that risk as perceived by the individual is quite a different matter from risk as viewed by state institutions.

Let us look at some real world data. Here are results for risk in walking, cycling, motorcycling and driving for Great Britain in recent years:

GB Modes By F per BnKM

Chart 1: Risk in Personal Travel GB (based on distance travelled 2006-2015 averages)

According to the logic of traditional road safety, driving is obviously far safer than any other way of getting about. To minimise road deaths, we must provide for cars as much as possible and discourage the riding of motorcycles. If people must walk or cycle, then cycling is slightly preferable to walking, although neither should be encouraged.

This caricature is not much wide of the mark. During the post-war era up to (roughly) the end of the last century, transport officials regarded walking as tolerable (well, at least for adults) but cycling was a barbarous relic that would be left to die. Cycling was ignored in transport planning. Dual carriageways built in the 1930’s with cycle tracks were widened for the post-war motoring boom; that was the end of the cycle tracks.  In the official mindset, cyclists were a non-conformist minority not in the great drive to the New Jerusalem of traffic jams in the sky. Like non-conformist minorities down the ages, their interests were ignored or even held in contempt.

Perhaps I come across as sour. I do not think I am being unfair. When I began advocating for positive cycling policies back in the mid 1990s, the above paragraph reflects the wall of bigoted ignorance presented by officialdom. Others had been dealing with it for decades.

Gradually, there has been a change in thinking. A force of evidence against the car-focused lifestyle and in favour of broader choices of transport had been gathering for decades. The 2009 Cycle Safety review by the Transport Research Laboratory seems to have been influential. In addition, and I am going to blow my own trumpet here, I am believe my investigations of risk in cycling have been helpful in easing the idiotic attitude that “more cycling means more danger”. Probably the most critical factors have been the obesity crisis and research showing that the risks of cycling are trivial relative to the health benefits. (see this page).

You must be wondering what social attitudes have to do with units of risk. The relevance is that as the value of walking and cycling became respected, so the measures of transport risk changed.

Fatalities per Million Hours’ Use

To consider risk purely on the basis of risk per distance travelled ignores the great variety of average speeds offered by different types of vehicle. It also ignores the way individuals decide how they will travel. Personal travel budgets are based on time, not distance. To quote from the 2015 National Travel Survey:

Compared to the 1970s, the number of trips and time spent travelling per person per year have remained broadly stable. The number of trips has decreased by 4% while the time spent travelling has increased by 4%.

Transport surveys show that one hour per day is the average personal travel budget. Motor traffic has increased because more people spend more time in cars, not because people spend more time travelling. This means they spend less time getting exercise from walking or cycling. This is part of the reason why obesity has become a problem (although there are other factors). Evidently reducing car dependency means increasing hours of travel by foot, bicycle and public transport. This means the risk per hour is the metric that matters.

This kind of thinking has been gradually accepted in Britain, at least at the upper levels of the state.

Let us get a feel for this unit, fatalities per million hours’ use. Suppose I face a risk of one fatality per million hours’ use (1 F/MHU). Should I worry? It all depends on how much of it I do. A few hours per year would not matter. However, normalization means getting it every day. Therefore, we have to consider an exposure of, say, half an hour per day. That is, I spend half of my travel budget in this activity.

If I am active for 70 years, the chance of death in this activity will be about 1 in 80. Cancer and heart disease cause almost half of all deaths, so it still looks like a remote risk relative to other hazards of life.

On the other hand, in a city of a million people, each taking half an hour per day of this activity, we would expect one death every two days, or 182 deaths per year. Suddenly, 1 F/MHU does not seem such a small risk. In fact, by the standards of personal travel, it is not a small risk, but a large one. These are some typical transport risks expressed as F/MHU, based on recent British experience:

GB-modes-by-F-per-MHU

Chart 2: Risk in Personal Travel GB (based on time spent travelling)

This chart shows more starkly than Chart 1 the high risks of riding motorcycles relative to even the most dangerous drivers of all: young males. The risks of walking and cycling no longer appear so out of line with driving.  However, it is a fundamental iniquity that driving is still generally safer than walking or cycling, whilst being a major component of the danger imposed on active modes of travel. It is a fundamental policy challenge to address this.

Chart 2 shows national averages, which are a crude measure of risk. Such all-encompassing averages do not express differences in the populations. The population that drives is not the same as that which cycles. More thorough risk analysis needs to drill down inside the averages to look at specific groups, such as males aged 17-20, or females aged 70+. These sub-averages are still averages, but they provide a closer like-for-like comparison between modes of travel.

I hope you have found this page useful. Please return to the main article for more information on Risk in Cycling.