Tuesday, July 8, 2014

Have we reached peak longevity?

The July 4 episode of the excellent BBC Radio 4 program, More or Less, addressed a question that demographers and public health researchers have been pondering in recent years: Will upcoming generations be the first to die at younger ages than their parents?  It has been accepted wisdom for a long time that each generation is healthier than the previous one, and life expectancies have risen for generations.  This was largely the result of declining infant and childhood mortality rates until perhaps the 1950's, but in most wealthy countries infant and childhood mortality are now so low that further decreases can't make very much difference in life expectancy; if life expectancy is to continue to improve, it will have to come from improved health at later ages.

But now, given the increasing prevalence of obesity in so many populations, more specifically childhood obesity, the concern is that we're eating ourselves right out of these life expectancy gains.  So More or Less took a look at the numbers.

It turns out that, yes, the obesity rate is higher, but the death rates from obesity related conditions such as heart attacks and stroke have been going down steeply.  Sir Richard Peto, Professor of Medical Statistics and Epidemiology at Oxford, said that while obesity has been rising sharply in Britain, the chances of dying from coronary heart disease-related conditions 30 years ago was 16%, at 1980 death rates, but 4% at 2010 death rates.  That is, it has declined by a factor of 4.


[1] 2009 rates; http://www.worldobesity.org/what-we-do/policy-prevention/ via Wikipedia


The trend is the same globally, the program notes.  Mexico is now the fattest country in the world but the probability of dying from obesity-related conditions at age 70 or younger in 1970 was 50%, but it is now 25%.  This is what's called the Obesity Paradox -- the fatter we get, the less likely it is to kill us.

What's responsible for this?  Well, obesity causes cardiovascular conditions but, so the thinking goes, treatment is better than it was 30 or 40 years ago, so while obese people may have chronic illnesses, they aren't as likely to die from them.  Perhaps they are getting medical care for heart or renal failure, or hypertension and type 2 diabetes, conditions associated with obesity, while thinner people don't see their doctors for anything, and thus are more likely to die.

By this reasoning, though, it's healthier to be ill!  And while the obesity paradox looks real, there's some question as to whether it is instead a statistical artifact.  The sample sizes were small in the studies that showed a benefit to being chronically ill and obese; the diagnoses of heart failure don't all match; the thin controls may be sicker than the obese subjects; when the effects of chronic conditions of obesity become life-threatening, the patient may have lost a lot of weight and no longer be obese (thus, obesity and weight gain themselves may indicate better health); BMI and fat patterning, two measures of obesity, may not be measuring the same thing, and so forth.  (See Habbu et al., e.g., for consideration of the obesity paradox.)

Whether or not the obesity paradox is real, the drop in death rates certainly is.  As Peto said, unless your country is chronically at war, or suffering an HIV epidemic, or you drink gallons of vodka, your probability of dying before age 50 is half what it was 40 years ago.  Or better -- in Iran the probability has fallen from 36% to 6%.

Death rates have fallen in rich countries -- by 1 or 2 percent a year since 1900, according to the National Bureau of Economic Research -- because of clean water, vaccination, antibiotics and other life-saving drugs, safer vehicles, the drop in smoking, improved obstetrics, safer food.  Public health measures, by and large.

Still, if half or more of the population of many countries is overweight or obese, and obesity is the cause of many chronic illnesses, what's going on?  Perhaps it's that it's not actually so risky to be overweight, as a paper in the Journal of the American Medical Association by Flegal et al. reported last year.  This was a meta-analysis of papers reporting the relative mortality risks associated with normal weight, overweight, and obesity, as defined by body mass index (BMI).  The authors found that obesity was associated with higher risk of mortality relative to normal weight, while overweight was associated with significantly lower risk.  It is healthier to be a bit fat than it is to be thin or obese. Perhaps the reserve energy is good when one is ill.

The paper was the subject of much controversy (e.g., discussed here), largely because everyone knows that being overweight is bad for your health.  But apparently there are many ways to be thin and unhealthy, and many ways to be "overweight" and healthy.  Does this answer the question of whether death rates will continue to decline, even as we get fatter?  No, because we haven't solved the looming antibiotic resistance problem, we can't predict which emerging infectious diseases will become pandemic but it's likely that something will, we don't yet know the health or agricultural consequences of climate change, though we can predict there will be some, and so on.

Measures of longevity
Life expectancy is the average age at death at a given age, and that's important, and must be stipulated.  Life expectancy at birth is affected by infant mortality rates.  Life expectancy at age 12 (say, beginning of adolescence) is average age at death of those who have escaped childhood mortality.  And so on.  So we must be aware of more than just chance of dying, but of dying after a specified age as the starting point.

Another important fact is that when we remove one trait -- say, obesity -- as a cause of disease and that disease becomes later or rarer, we automatically increase the risk of death (though perhaps at later ages) from other causes.  That's because if you escape, say, heart disease, you live to get cancer or dementia or arthritis -- or perhaps in the not so distant future, strep throat or gonorrhea.  So all of these health vs longevity statistics need to be viewed with care.

But at least we do know that smoking is still bad for you.

No comments: