One is often told eating five portions of fruit and vegetables daily is “good” for you, but then winning a nice prize is also “good” and you would feel it immediately.
They are not the same thing of course, but it is worth quantifying what “good” means in dietary and lifestyle terms if one wishes to know whether it is worth following any specific advice which affects what you eat and do. And one interesting way is to express things in terms of lifespan changes.
Using micromorts is too imprecise as they are targetted towards measuring sudden life-ending probabilities. So to gauge factors that affect life expectancy improvements or degradation, a measure called the “microlife” was proposed by David Spiegelhalter in the Christmas edition of the British Medical Journal in 2012.
The concept is simple, and involves dividing a theoretical human lifespan of 57 years into one million microlives, resulting in individual microlives of 30 minutes each.
Measurements of harmful or beneficial factors are then calibrated in microlives; for example, smoking a packet of cigarettes will lose around -10 microlives, or a statistically shortened lifespan of five hours.
Regarding the five portions of fruit and vegetables, consuming them will result in +4 microlives each time, or two hours added to a statistical lifespan. On the reverse side, eating red meat causes a loss of -1 microlife per 85g portion.
Microlives, unlike micromorts, do not reset and are cumulative for life, so eating five portions of fruit and vegetables plus 85g of red meat will result in a net balance of +3 microlives to be carried forward forever.
The databases used to calculate microlives, as for those used for micromorts, can only calculate statistics derived from large population pools and therefore cannot be used to precisely estimate any individual’s changes in life expectancy as people have wildly differing idiosyncratic factors such as genetics, habitation, occupations, hobbies, food, etc.
Investigating population-based microlives throws up interesting observations. For example, just being born as a man results in -4 microlives being lost daily, resulting in an average lifespan of -3.7 years shorter than a woman.
Geography is also a factor. If a man is a resident of Russia, then he loses -21 microlives a day compared to a Swedish man, but Russian women only lose -9 microlives daily relative to Swedish women.
Generational improvements in healthcare and food quality are also reflected in microlives. People living in 2010 add on +15 microlives a day compared to people in 1910, just by being alive.
More relevantly, being obese today knocks off -3 microlives each day for both men and women.
Exercise also affects microlives. Twenty minutes of moderate exercise adds +2 microlives per session but every additional 40 minutes of exercise after that only adds +1 microlife. But lounging lazily on a sofa for 2 hours will cost -1 microlife.
If you are a drinker, statistics indicate a curious +1 microlife benefit for the first 10 grams of alcohol but after that, one would lose -0.5 microlife per every subsequent 10 grams of alcohol, up to the sixth drink, after which there is no data. Drinking two to three cups of coffee a day also adds +1 microlife.
Extrapolation (or projection into the future) of microlives can also provide a statistical estimation of a person’s remaining lifespan, which can then be applied as a particularly stark way of quantifying the impact of diet and lifestyle. In short, given a person’s collection of current risks expressed in microlives and projecting these risks into the future, we can arrive at a statistical “effective age”.
This is the age of a typically healthy person with the same remaining total of microlives. For example, a person aged 40 may have an effective age of 50 due to accumulated bad habits and loss of microlives over many years. Similarly, a person aged 55 may have an effective age of only 50, due to picking up microlives because of good habits.
Graphically, effective age may be expressed as follows:
Can we do more?
A paper published in 1985 by Morris and Temple detailed how spirometry (lung function) tests were used to gauge the reduced capacity of lungs affected by smoking. They found the forced expiratory volume of air from the lungs for one second was the factor which provided the lowest error when estimating lung age.
From this data, they were able to inform patients about how their lungs performed compared with lungs of older people as an indication of how much their lungs had “aged” because of smoking.
Mainly, at the time they were trying to deter people from smoking rather than devising a statistical methodology for estimating lung health.
However, from reading the paper, one can easily deduce that by using mass population statistical data on the factors that affect human organs, it would also be possible to calculate the effective ages of the individual organs of a person.
The health of important organs, such as the heart, lungs, brain and liver, can be statistically deduced, sometimes remotely (without medical intervention), thus saving costs and resources.
From individual results, health services can then efficiently target patients at higher risk levels by suggesting changes to diets, lifestyles and/or medications.
Fortunately, there are several large epidemiological studies which have provided useful statistical data. One notable source of statistical data comes from the European Prospective Investigation into Cancer and Nutrition (Epic).
The use of statistical models to establish effective ages for organs and overall health is now standard practice in many Western clinics, where results from spirometry tests, blood tests (blood liver markers), ECGs, mental quizzes, etc, are used to estimate the effective ages of major organs and assist doctors in prescribing appropriate treatments.
However, like risk management, using effective age is not that simple. An interesting observation based on historical mortality rates is that the curve is not linear (ie, death rates are not a straight line chart).
The risk of dying actually decreases each year from birth during childhood, reaching its lowest point at the age of 10.
Then there is a sharp increase in the chances of dying between the ages of 15 and 25, and from that point, mortality rates are more or less linear, with death probabilities increasing at roughly 10% a year, doubling every seven years or so.
Risk and rate advancement periods
While older people die more frequently, what is also true is old age by itself is not a cause of death. People die only when they succumb to some disease or trauma.
Also many diseases do not occur at the same age in every person. The mean age for contracting Parkinson’s Disease (PD) is over 60 years, and 75% of strokes occur past the age of 65, as examples. Also most diseases are more likely to kill older patients than younger ones.
A paper on this issue was published in 1993 by Brenner, Gefeller and Greenland, which established the concept of “risk and rate advancement periods” (RAP).
This was designed as a method to measure the impact of disease by estimating the effective age of a person with a disease relative to the age of an older, healthy person without the disease. What could be simpler, eh?
Although RAP is quite widely used, it turns out to be very far from a simple black-and-white generic tool.
RAP does help establish ages when certain diseases are more likely but it fails as a measure of the mean survival period; for example, PD has no consistent impact on life expectancy as people with PD die from a variety of different complications, some of which relatable to the treatments and/or medications used.
Hence RAP is not as simple as projecting effective ages as many diseases have confounding and compounding effects, unless it is an absolute stone-cold killer such as Ebola.
Adjusting lifespan estimates due to various diseases is still an ongoing, complex statistical challenge.
I hope you found this elementary review of food and life risks entertaining.