Here is how life expectancy at birth has varied over time in the United States :
Throughout human history and prehistory, from Stone Age times to about 1880, life expectancy at birth was generally 35 to 40 years. In the United States in 1880, it was 39.4 years. But beginning in 1880 life expectancy began a dramatic rise, increasing by five months per year and reaching 68.2 years in 1950. After 1950 the pace of the increase in life expectancy slowed to 2.2 months per year in the period between 1950 and 1980 and 1.8 months per year between 1980 and 2010.
This makes it look like all is peaches and cream—or gas and gingerbread as Henry James would have said. Life expectancy keeps getting longer. We must be getting healthier.
But some diseases are becoming more prevalent. Obesity for instance :
The fraction of Americans who are overweight or obese rose rapidly from the late 1970s to the early 2000s and may still be rising.
Obesity tends to be followed in time by other diseases. About one-fifth of the obese, for instance, go on to develop diabetes, typically about 15 years after the onset of obesity :
Image credit: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3014900/figure/F3/
Obesity and diabetes aren’t the only diseases to become more prevalent recently. Diseases from asthma to depression have also become more common.
It may seem paradoxical that lifespan is increasing while health span—the number of years of life lived free of serious disease—seems to be decreasing. One possible explanation is that modern medicine is helping disease patients delay death; this is plausibly the case for cardiovascular disease. But there is another, perhaps more likely, explanation.
Many diseases have long incubation periods. We know that smoking causes lung cancer, but usually only after a number of decades. Similarly, sustained eating of an obesogenic diet may cause obesity after perhaps 10 years, diabetes after perhaps 25 years.
If diseases have long incubation periods, what about death? How long do we spend dying?
There is reason to believe that health impairment suffered at any time during life may shorten lifespan. Indeed, healthfulness in youth is correlated with length of life.
One remarkable illustration of this was found by The Nun Study, which compared the emotional content of handwritten autobiographies written by 180 Catholic nuns in their youth to life outcomes 60 years later. The nuns whose autobiographies were the most positive had 60 percent lower mortality between ages 75 and 95. 
In the young especially, happiness and optimism are marks of good health. It appears that nuns who were healthy as young adults were much more likely to live long lives than nuns who were less healthy in their youth.
Obesity may develop over 10 years and diabetes over 25, but dying seems to be a lifelong process. To find the causes of death, we may have to look over the full lifespan—currently about 80 years.
Modeling the Healthfulness of the Environment
If that’s the case—if we must reach back over a period of 80 years to find the causes of death, but over a shorter period to find the causes of disease—then it’s easy for disease rates and lifespan to go in opposite directions. All we need is for the health environment to have been improving until recently, but then to have begun worsening.
If lifespan depends on the average healthfulness of the environment—diet, lifestyle, social structure, medical care, and other factors—over the last 80 years, then we can work backwards from life expectancy to estimate the healthfulness of our past environment.
We made a little Excel model and got this result:
The purple curve is actual life expectancy. The blue curve is our estimate of how healthful the environment was that year. And the green curve is the average of how healthful the environment had been over a previous human lifetime.
The green curve fits the purple curve very well. It is consistent with the health environment improving steadily from about 1860 to 1950, but then plateauing at a constant level of healthfulness since.
If there have been no improvements in the health environment since 1950, then we should expect life expectancy to continue rising until about 2030. At that date, everyone will have lived their whole lives in the modern healthful environment, and life expectancy should be stable thereafter at about 80 years.
However, the recent rises in obesity, diabetes, and other diseases suggest that maybe the health environment has worsened a bit since the 1970s. In that case, American life expectancy may begin to decrease, perhaps as early as 2020 or 2030.
We’ll discuss the implications of these observations in coming posts. What factors are most important for lifespan and health? Why did the health environment improve so dramatically in the century up to 1950, and why is it stable (possibly even worsening) now?
More: The maximum human lifespan is about 120 years. We increased lifespan by 40 years once before, in the period between 1880 and 1950. Can we do it again, and enable everyone to live to 120 years? What are the causes of ill health and shortened lifespan that we are still not addressing?
Answers to these questions may have large implications for personal health practice, the practice of medicine, and public policy. For example, the great growth of biomedical science, the pharmaceutical industry, and of health care has occurred since 1950. U.S. health care expenditures have risen from 5 percent of GDP in 1960 to 17.9 percent of GDP in 2010. What if these expenditures are having a negligible effect on our health and life expectancy?
There may be room for great improvements in our personal and society-wide health practices. Let’s explore.
 Historical Statistics of the United States, Millennial Edition, volume 1, series Ab644; World Bank, World Development Indicators, http://data.worldbank.org/.
 CDC/NCHS, Health, United States, 2008, Figure 7, http://www.ncbi.nlm.nih.gov/books/NBK19623/. Data from the National Health and Nutrition Examination Survey.
 Zhou SS et al. B-vitamin consumption and the prevalence of diabetes and obesity among the US adults: population based ecological study. BMC Public Health. 2010 Dec 2;10:746. http://pmid.us/21126339.
 Danner DD et al. Positive emotions in early life and longevity: findings from the nun study. J Pers Soc Psychol. 2001 May;80(5):804-13. http://pmid.us/11374751.