2 2 Things Come Apart

2 Things Come Apart. BY ITS END, much of the optimism of the twentieth century had faded. Towns and cities in the heartland of America that used to produce steel, glass, furniture, or shoes, and that are fondly remembered by people in their seventies as having been great places to grow up, had been gutted, their factories closed and shops boarded up.

In the wreckage, the temptations of alcohol and drugs lured many to their deaths. Most of these stories are never told. Stigma often removes the cause of death from obituaries when suicide, overdose, or alcoholism is involved. Addiction is seen as a moral weakness, not a disease, and it is believed that its effects are best covered up.

Exceptions are made when a famous chef kills himself or a music icon overdoses on fentanyl, or when the death is shocking to the community—for example, as reported by Congresswoman Ann McLane Kuster, “in a little town called Keene New Hampshire. There’s not a quieter place on this earth, and a beloved high school teacher, mother of three children, died of a heroin overdose.”1 Each story is real and tragic, but it needs to be considered in perspective.

When we look at the numbers, all the numbers, we see an even bigger, more frightening, and tragic story. The events that reach the media are selected for their news value, celebrities get attention, and the firsthand accounts of addiction or attempted suicide often come from those who are accustomed to writing about their experiences.

Spectacular and unusual deaths—upper-class suicides and drug deaths—are exhaustively reported; those of ordinary people rarely make headlines, although they too leave behind devastated families and friends. Today’s events are news; long-term trends are yesterday’s news, which usually means not news at all.

Deaths from lung cancer, heart disease, or diabetes are not news in and of themselves—lung cancer is not like Ebola or AIDS, though it takes many more lives—and we find out about them only incidentally when we read obituaries. Without the numbers to make comparisons, we don’t know whether we are looking at an event, like a plane crash or a terrorist attack, where the deaths are few but shocking and

newsworthy, or an epidemic, like Ebola or SARS, which terrified many but killed few, or whether we are dealing with something much larger, something that actually threatens the public health and upends a century of progress in human health.

All deaths in the US are reported to the authorities, and the information is assembled by the Centers for Disease Control and Prevention (CDC) in Atlanta.

When someone dies, a great deal of information is collected on the death certificate, including, for the last thirty years, the highest level of education attained. The CDC has a website, charmingly called CDC Wonder, where much of this information is readily available.

The death certificates themselves, with confidential information (such as name and social security number) removed, can also be downloaded and examined. It is with these data that we begin.

They are every bit as distressing as the stories.

American Exceptionalism, Breaking with the Past, and Leaving the Herd: The Facts

We saw in the previous chapter that the mortality rate for midlife whites in the US was 1,500 per 100,000 in 1900, and that by 2000 it had fallen to 400 per 100,000. We now follow this group into the twenty-first century.

We can also look at other countries around the world that, like the United States, are rich in terms of income per head and that share and implement the scientific and medical knowledge that is common across such countries.

Those countries showed rapid declines in midlife mortality after 1945, and as in the US, the decline was particularly rapid after 1970. In almost all wealthy countries, mortality rates for those aged forty-five to fifty-four declined at an average rate of 2 percent per year from the late 1970s to 2000.

FIGURE 2.1. Age-adjusted mortality rates, ages 45–54, for US white non-Hispanics (USW), France, the United Kingdom, and Sweden, and a predicted mortality rate for USW, a counterfactual that assumes the mortality rate for USW would continue falling at 2 percent per year after 1998. Authors’ calculations using

CDC data and the Human Mortality Database.

Figure 2.1 shows what happened. We call this the “things come apart” picture. Midlife mortality continued to decline in France, Britain, and Sweden; other rich countries, not shown, display similar progress. An entirely different pattern emerged for US white non-Hispanic Americans.

Not only did whites not keep pace with mortality declines in other countries, but mortality for them stopped falling altogether and began to rise.

The future that we might have predicted for white Americans in midlife,2 based on what had happened in the twentieth century, is shown here by the thick dotted line. Over time, white mortality pulled markedly away from what was seen in other wealthy countries, and what we might have predicted its path to be.

Something important, awful, and unexpected is happening. But is it just white men and women in middle age, or are other age-groups affected too?

Is it men more than women, or women more than men? And what about other groups?

Is it focused in one part of the country, or much the same everywhere? And, above all, why is it happening? As we shall see, the alcohol, suicide, and opioid epidemics are an essential part of the story, but we need to discuss a few other

issues before we get there. In chapter 1, when we showed falling midlife mortality through the twentieth

century, we noted that other age-groups also benefited. But the reversal in figure 2.1 is not universally shared. As we shall see, while there have been similar changes in mortality trends for younger age-groups, mortality among the elderly continued to fall as it had done at the end of the twentieth century.

We shall explore this a good deal further as we go, and we shall see that the reversal has begun to affect the youngest elderly too.

In figure 2.1 we switched from all whites to non-Hispanic whites, a narrower category for which data did not exist for most of the twentieth century.

Hispanics, who are much poorer on average than non-Hispanics, have lower mortality rates than non-Hispanics, and their progress kept pace with that in other countries; their mortality rates look like those for Britain over this period. African Americans have higher mortality rates than any of the groups or countries shown in the picture, but their rate of mortality decline has been faster than for any of the groups or countries shown here.

The midlife gap between US black and white mortality fell dramatically between 1990 and 2015, after which point the decline in midlife black mortality also came to an end, likely linked to opioids, as we shall see.

The story of racial differences in mortality is an important one, and we shall later argue that the differences between black and white mortality rates can be reconciled once we look carefully at the history. The differences have less to do with what than with when.

These differences in mortality by race and ethnicity are far from fully understood, but they have existed for many years.

For African Americans, there is widespread agreement that the worse outcomes, like so many other important outcomes, are tied to long-standing discrimination, as well as to poorer access to high-quality medical care.3 The superior longevity of Hispanics over non- Hispanic whites has been much researched but not fully explained.

It is worth noting that other groups, such as Asian Americans, do better still, better than either Hispanics or whites. As to the recent trends, which have been so different across the three main groups, we will return to them repeatedly throughout the book, though we should confess from the start that we shall find much that is not easy to explain.

Figure 2.1 is drawn for men and women together, which is always potentially misleading. Women have lower mortality rates than men throughout life, and so they live longer, about five years longer in the US.

Men and women suffer from different diseases, and to different extents from the same diseases and behaviors: men, for example, are three to four times more likely than women to kill

themselves. But the turnaround—from continual progress in the twentieth century to stalled progress, or even regression, in the twenty-first—has happened to both men and women in midlife, though the reversal is somewhat larger for women than men.

Even so, the gaps between whites in the US and other countries and between US whites and what we might have expected are large for both men and women, so that the figure does not mislead by taking men and women together.4

One measure of the importance of the white mortality reversal is to compare what actually happened with the trend shown by the dotted line.

The gap between the two lines shows the difference in mortality rates in each year, from which we can calculate for each year how many people aged forty-five to fifty- four died who would have been alive had late twentieth-century progress continued.

When we add up those numbers from 1999, the critical point where the turnaround began, to 2017, we get a very large total: 600,000 deaths of midlife Americans who would be alive if progress had gone on as expected. One immediate point of reference is the approximately 675,000

Americans who have died from HIV/AIDS since the beginning of the epidemic in the early 1980s. We shall refine our estimate as we go, extend it to other age-groups, and attribute it to specific causes, but it will serve for now as a ballpark estimate of what is involved, and to establish that what we are dealing with is indeed a major catastrophe.

Another measure of importance is to look at what has been happening to life expectancy at birth. Because life expectancy is more sensitive to deaths at younger ages, only large changes in midlife mortality can affect it. For whites, life expectancy at birth fell by one-tenth of a year between 2013 and 2014.

In the next three years, between 2014 and 2015, 2015 and 2016, and again between 2016 and 2017, life expectancy fell for the US population as a whole. These declines reflect mortality at all ages, not just in midlife, but are, in fact, heavily influenced by what has been happening to whites in midlife. Any decline in life expectancy is extremely uncommon. With a three-year decline, we are in unfamiliar territory;

American life expectancy has never fallen for three years in a row since states’ vital registration coverage was completed in 1933.5 For the subset of states that had registration of deaths before then, the only precedent is a century ago, from 1915 through 1918, during the First World War and the influenza epidemic that followed it. Catastrophes indeed.

The Geography of Mortality If we are to begin to understand why these deaths are happening, we can first

look for clues on where the deaths are happening. If we look across states at the changes in mortality rates for whites aged forty-five to fifty-four from 1999 to 2017, we find the increases in all but six states, with the largest increases in death rates in West Virginia, Kentucky, Arkansas, and Mississippi, all states with education levels lower than the national average.

The only states where midlife white mortality fell by a noticeable amount were California, New York, New Jersey, and Illinois, all states with high levels of education.

A more detailed geography is shown in figure 2.2, where mortality rates for midlife whites are presented for about a thousand small areas across the United States in 2000 on the left and in 2016 on the right. These small areas are counties or, if the population of a county is small, a collection of adjacent counties.

Darker areas indicate higher mortality, so the maps show high mortality in the West (except California), Appalachia, and the South in 2000, intensifying and spreading by 2016 into new areas, such as Maine, upper Michigan, and parts of Texas.

We will refer back to the patterns in these maps throughout the book.

FIGURE 2.2. All-cause mortality rates, white non-Hispanics ages 45–54, by small area. Authors’ calculations using CDC data.

Carrying Their Troubles with Them: Age versus Cohort Effects Figure 2.1 compares death rates across countries for one specific age-group, those aged forty-five to fifty-four, but our concerns do not end there.

White mortality progress has reversed throughout adulthood, in contrast to what is happening in the rest of the rich world. We highlight the midlife group, those aged forty-five to fifty-four. But, as we will see, rising mortality is not simply a

baby-boomer phenomenon. For US whites, the hurdles at younger ages have also been raised.

The future of today’s midlife adults is also in question. Will those in midlife “age out” of the mortality crisis if they survive? Or will they carry their troubles with them as they age, so that tomorrow’s elderly will suffer like today’s middle aged?

Elderly Americans receive benefits, such as healthcare from Medicare and pensions from Social Security, that are not available to those in middle age so that, if these benefits are good for health, there is an argument for the positive alternative.

But if the midlife deaths are happening to people born around 1950 because of the conditions under which they have lived their lives, or because of the way they have chosen to live their lives, there can be no expectation that they will do better as they age.

Unfortunately, recent data are more consistent with the second, more negative outcome. The midlife increase in mortality has now begun to affect the elderly, as the birth cohorts born after the Second World War begin to move into old age.

The all-cause mortality rate for whites ages sixty- five to seventy-four fell on average 2 percent per year between the early 1990s and 2012; since 2012, their mortality has stopped falling.

Social scientists often try to isolate two different phenomena.

On the one hand, there may be “age” effects, when an outcome is tied to age, and on the other hand, “cohort” effects, when outcomes are attached to people born around the same time and are carried with them as they age.

Cohort and age effects are not, of course, mutually exclusive, nor do they exhaust all of the possibilities. We will argue for (a version of) the cohort interpretation, which is, unfortunately, the more pessimistic of the two accounts. There is something about these people that makes them susceptible and that they carry with them through life.

Discovering the nature of that something is our task in the rest of this book.

There are two stories, often seen as competing, though they need not be. One, the “external” or circumstantial account, emphasizes what happened to people, the opportunities that they had, the kind of education, occupation, or social environment that was available to them.

The alternative, “internal” account emphasizes what people did to themselves, not their opportunities but their choices among those opportunities, or their own preferences. It is a debate between worsening opportunities, on the one hand, and worsening preferences, or declining values or even virtues, on the other.

Before we can take the story further, we have to return to our midlife Americans in the early twenty-first century and find out more about the causes of their deaths. Not surprisingly, suicide, opioids, and alcoholism feature in the story, but they are by no means the only players.