Books for Long Life
Can reading books provide you a longer life? Is reading itself a key to better survival? A recent study from Yale argues that the active reading of books (more so than periodicals, sorry,) may give you a better chance to continue reading for a long, long time.
The research went back to the Health and Retirement Study to check the survival over a dozen years of a group of older people. Using a Cox proportional hazards model to figure out the expected lifespan of that population and looking at the 80% expected survival point, book readers possessed a 23% survival advantage, to non-readers. But was this result real?
It’s well known that better educated people live longer. Wealthier people live longer. Women live longer – and proportionately read books more than men. Married people live longer. Would the survival advantage of reading books wash out when you statistically controlled for sex, race, educations, comorbid medical conditions, self-rated health, wealth, depression, and marital status?
Correlation Versus Causation
Epidemiology, the study of disease rates in population, is hard for many to understand. It’s even harder to do.
Consider this “fact” – 85% of Americans die in bed. Do beds kill people?
The immediate answer of course is no. People who get sick end up in bed. The people who die will disproportionately be found in bed before they leave this earth. Beds are not killing people.
But somehow that thought gets left in the mind. It remains as a suggestion. And suggestions, as we’ve learned in this political season, are powerful. The appearance of a relation makes some of us think – what if? What if there might be some kind of relationship? And many of these “associations” are powerful emotionally. For example, one national politician suggested that the father of another was somehow “involved” in the Kennedy assassination. How? A figure in an old photograph of John Kennedy “looked like him.” So a mental “connection” was made between a “political” photographic identification, a Hispanic presidential candidate, and the assassination fifty three years ago of John Kennedy.
This political season has been rife with such “facts” that are analytically the opposite of true. Fortunately, the reading of books can inculcate in many people the analytic skills that can identify the intentional use of such “facts.”
Confounding Confounders
Associations that mask the relevant causative variables are called “confounders” in epidemiology. Indeed, they literally confound.
Over thirty years ago, a major public health policy recommendation came from the national Nurses’ Study, which followed tens of thousands of nurses over decades to try and identify their main causes of mortality.
What the researchers “discovered” was that hormone replacement therapy (HRT) for post-menopausal women was associated with a 50% decreased risk of heart disease.
Thus was a revelation. The main killer of women could be cut in half by using HRT. It was, as they sometimes say in medicine, a “no brainer.”
Except it was wrong – dead wrong.
The models used in predicting survival had not fully account of the effect of “volunteer bias.” Volunteers, in any medical study, are materially different than non-volunteers. Often they are healthier, better educated, more excited by new techniques and options. In other words, they often see much better results.
It turned out the women that women “volunteers” for hormone replacement therapy were a far healthier group from the start than those who did not use HRT. They were far more likely to eat their vegetables, exercise, engage in a host of habits that lead to longer life. In reality, HRT is now known to be associated with a higher risk of heart disease – as well as breast and endometrial cancer.
Epidemiologists recognize there are hundreds of variables relevant to their studies, of which they can track only a few. Beyond that, there are thousands of variables – like the effect of the microbiome – that nobody knows enough about to even track. So you do the best you can. So what happened to those “confounding” variables in the the reading study?
Reading for Health
In the Yale reading study, the most powerful known variables were controlled for. And when they were controlled for, with the adjustment of all the covariates, there was still a 20% survival risk advantage for book readers – a much higher result than for those who read periodicals. More interestingly, the advantage in the statistical models appeared to be mediated through cognition – better cognitive ability in the book readers appeared related to longer life. And that, at least, is a result that accords with theories of how book reading might lead to longer life.
Bottom Line
Epidemiology as performed today inevitably cannot know all the relevant variables associated with healthy outcomes. But you can know some. In the Yale Study, reading books – the long conversation between author and reader – was associated with longer life. Is it because book readers are more analytically inclined? Is it because the habits of book reading are associated with other, as yet not well identified, health improving habits? Does thinking long and hard lead to better health behaviors? Or is it just that people learn more reading books, and can use what they learn to promote longer lives?
Future studies will figure some of this out. But in the meantime, get yourself to the library. It may prove a decidedly healthy place to visit.