My “proper” job is working in risk management for (usually) large banks around the world. Maths and statistics therefore interest me a lot, so for this piece, I thought it would be fun to apply some simple risk analysis to food and lifestyle-related issues.
The basis for one such type of analysis had been done a while ago. Around 1979, Ronald A. Howard of Stanford University invented a measure called the micromort which is a unit used to express the risk of a one in a million chance of dying due to engaging in some activity.
Micromorts change with age, so for example, a person aged 45 runs a risk of six micromorts (six in a million) of dying in bed while a 90-year old person has a daily risk of 463 micromorts of not waking up alive. Interestingly, a newborn baby runs a risk of 430 micromorts of not surviving the first day after birth, almost the same as a 90-year old person never getting up again.
The good news about micromorts is that they reset every risk period, which may be a day, a year, an event or after certain quantity thresholds. So if a person survived a day, then all that day’s daily micromorts would be reset to zero when he/she wakes up the next day.
The bad news is that during the course of a risk period, micromorts are cumulative. Therefore, if someone rode a motorbike 60 miles (10 micromorts) to go for a swim (12 micromorts), then that person runs a total daily risk of 22 micromorts (assuming he/she does not do anything else risky). One might consider various activities as additions to a bucket of risk which one is obliged to carry around, and practically everything you do adds some (normally very small) amount of risk to the bucket.
Somewhat worryingly, micromorts are also transferable, often without you even knowing about it. In many cases, a person’s individual risk bucket has a profound effect on everyone relying on that person, or who may be in contact with that person.
For example, if Mr X is driving people around when inebriated, every passenger of Mr X will have their buckets of risk micromorts raised to be at least the same level of risk as his. And complete strangers on the same road who do not even know Mr X will also have their buckets of micromorts significantly increased.
How micromorts are calculated
The issue, of course, is how one can use micromorts in a helpful manner. This is already being done commercially, but first it helps to know how micromorts are calculated. A common example is based on the number of deaths resulting from the use of anaesthetics (but not the actual surgical procedure). This works out at one death per 100,000 applications of anaesthetics which therefore gives a risk of 10 micromorts (1,000,000 / 100,000).
As for commercial usage of micromorts, a prime example would be cars with added safety features, like extra air bags or reinforced roll bars. There is often a statistical trick applied, which uses relative differences; for example, the death risk in a car with five air bags in a 50 kph head-on collision may be 5,000 micromorts whereas the death risk in a car with only three air bags may be 10,000 micromorts.
This may mean the five-airbag car being advertised as being “twice as safe” as the three-airbag car, even the death risk is actually 0.5% vs 1%. So people will pay thousands more for a car with two more airbags which is only 0.5% safer than another car.
As with risk management in banks, there are surprises when analysing actual micromorts. For example, many people might consider riding horses somewhat dangerous, but each ride on a horse is only rated at 0.5 micromorts while a long 17 mile walk is rated as twice as risky at one micromort. Skydiving is just as risky as riding a motorcycle for a distance of 60 miles – both are rated at 10 micromorts, and both activities are less risky than swimming, which is rated at 12 micromorts.
Regarding food, there are very low risks involved when measured using micromorts, mainly because it is difficult to establish a cause of death specifically due to one normal food episode (unless it is food poisoning). The risk of dying due to a steak dinner is rated at one micromort, but only after eating steak at least 100 times before. With bananas, the same risk of one micromort also applies, though only after eating 1,000 bananas earlier in your life. So in terms of food risk, micromorts are often too inexact to be of practical use and we therefore need to use another interesting statistical measure called microlives for food-related risks. Microlives will be covered in the next part.
To conclude the discussion about micromorts, it should be noted they can be helpful in other ways. For example, if one is forced to make a choice between swimming in a sea where sharks have been seen once in a while or running in a marathon, then despite what you might think, it is actually safer to do the swim because the chances of getting killed by a shark is only 0.125 micromort whereas a marathon has a risk of seven micromorts, which means that a marathon run is 56 times more likely to kill you.
But hopefully you would have noticed the statistical trick in that last statement, as in reality the chances of getting killed by a shark is 0.0000125% while the chances of dying during a marathon is 0.0007%.
Both are statistically extremely unlikely, but it has to be said that both are also possible, though death by shark is less probable.
Similarly, rock-climbing (three micromorts) is less risky than scuba-diving (five micromorts), travelling by train for 6,000 miles is as safe as flying in a jet for 1,000 miles or driving by car for 230 miles (all one micromort).
Curiously, one micromort of risk will also occur if you walk 17 miles, or cycle 20 miles, or ride a motorcycle for only 6 miles.
At this point, please understand that micromorts are based on statistics gathered from large populations and therefore not a measure of any individual’s personal risks when engaging in any activity.
The data used to derive micromorts will also vary from country to country. For example, the 0.5 micromort for horse riding is based on UK data where a safety helmet is normally required to ride a horse.
Horse-riding would therefore be more risky in countries where helmets are not mandatory. Murder risk is only 10 micromorts in the UK per year but 48 micromorts in the USA. Also it expresses only death rates, and ignores incidences such as serious injuries.
Micromorts also express risk generically. For example, swimming as an activity covers swimming in pools, lakes, rivers and oceans, but a high-level micromort for swimming does not necessarily express where deaths are more likely.
It is of course possible to calculate micromorts for specific situations; e.g. every attempt to climb Mount Everest has a risk of 37,932 micromorts.
Regardless of the above caveats, micromorts are a useful way to override our propensity to be irrational in our perception of risk. If data is available, it is extremely easy to calculate micromorts for any specific activity.
By providing a context for risk, micromorts allows us to be clear about the risks worth taking, thereby allowing us to balance what we enjoy doing with the risks associated with such activities. Micromorts also provide a basis to assess how much we are willing to pay to reduce/avoid risk in certain situations, and/or avoid getting into a car with a drunk driver and transferring his risks to you.
In the next part, we will discuss microlives and how they offer a thought-provoking framework for quantifying the risk in our diets and lifestyles.