Triumphs of Experience: The Men of the Harvard Grant Study (44 page)

—WILLIAM SHAKESPEARE

AS I’VE TRIED TO CONVEY
throughout this book, longitudinal studies are constructions of intrinsic contradiction and paradox. They require investment in massive information collection before there’s any way to know for certain whether the information being compiled is the kind that can answer the questions the Study is posing. Much of this vast accumulation will almost assuredly never answer any questions at all. Yet in the huge heaps of data, little glints may sometimes be perceived. Some will turn out to be fool’s gold. But some, suddenly—with a different cast of light or a sudden shift in context or a new analytical technology—identify themselves as twenty-four karat. There are unexpected, unexplainable, curious, and just plain odd findings in every large longitudinal study, and these are the very ones that beg for review every so often, just in case.

Here are some of the tantalizing glints from the Harvard Study of Adult Development. In time they may, as my final lesson from
Chapter 2
suggests, help to elucidate one or two of life’s enduring mysteries.

I. WHY DO THE RICH LIVE LONGER THAN THE POOR?

The last fifty years of epidemiology are making it ever clearer that the human lifespan is finite; after a point, the greatest riches in the world can’t help you live longer. Nonetheless, the poor die sooner. In the
United
States it is socioeconomic status (income, education, occupation, access to health care) that is thought to account for this disparity, not the malnutrition and infection that shorten lifespans in third world countries.

Some people blame society for discriminatory access to health care and for toxic neighborhoods, poor nutrition, poor schools, and high unemployment; some blame the victim for dropping out of school, delinquency, bad habits, and poor self-care. While there is danger that the current emphasis on individual health promotion can be used in the service of victim-blaming, I’ve shown in
Chapter 7
that the role of health-related behaviors cannot be carelessly dismissed out of political correctness.
1

Nevertheless, as Marcia Angell, former editor-in-chief of the
New England Journal of Medicine,
has pointed out, “Despite the importance of socioeconomic status to health, no one quite knows how it operates. It is, perhaps, the most mysterious of the determinants of health.”
2

Nowhere is the Harvard Study of Adult Development as powerful as when it addresses this mysterious relationship between health and social class. The College and Inner City groups were matched for several important confounders: gender, race, geography, absence of delinquency, and a 1920–1930 birth cohort. But the two samples were clearly dichotomized by social class and intelligence (at least according to standard IQ tests), since the Inner City men were matched to a low-scoring group of delinquent youths. As
Figure 10.1
illustrates, the Inner City cohort has been becoming disabled ten years earlier on average than the College sample, and dying ten years sooner; the estimated average longevity of the Inner City men is seventy years, of the College men, seventy-nine years.

The morbidity of the two samples is similar with regard to illnesses that are independent of self-care: that is, cancer (excluding
lung),
arthritis, heart disease, and brain disease. But there was twice as much lung cancer, emphysema, and cirrhosis, and three times as much Type II diabetes, among the Inner City men as in the College sample. The Inner City men were also more than three times as likely to be overweight.
All these differences, however, disappeared for Inner City men who graduated from college.

Figure 10.1 Death after fifty for College men, Inner City men, and Inner City college graduates.

The average Inner City man was much less educated than the Harvard graduates, and he also led a far less healthy lifestyle. But the more education an Inner City man obtained, the more likely he was to stop smoking, avoid obesity, and be circumspect in his use of alcohol. The estimated average age of death for the Grant Study men who did not go to graduate school and the college-educated Inner City
men
was identical—seventy-nine years, if the World War II deaths are excluded.

So we have to ask: Does education really predict healthy aging independent of social class and intelligence? The college-educated Inner City men were neither more intelligent nor more privileged socially than their peers who did not attend college, so those two factors do not explain the nine-year difference in their estimated lifespans. The college-educated Inner City men had carefully tested IQs that were on average 30 points lower than their Harvard counterparts’, and they attended lesser colleges. They were a full inch shorter, suggesting inferior childhood nutrition, and none of them had the upper-middle- or upper-class advantages typical of two-thirds of the Harvard men. In middle age, only half as many of the Inner City men as College men had made it into the ranks of the upper class, and they made only half as much money. Thus, neither intelligence nor status nor wealth can account for the disappearance of the nine-year shortfall in lifespan once an Inner City man had graduated from college. Parity of education alone was enough to produce parity in physical health.

Well-trained medical sociologists will scoff at this assertion. Do not the Whitehall Studies by Sir Michael Marmot in England appear to show that social class is one of the leading causes, if not the leading cause, of premature death in England? Did not the health of Whitehall civil servants improve with every step in their pay grade?
3
Yes. But Marmot’s early Whitehall studies did not control for education or alcoholism. Our study of the Inner City men showed that education is very significantly associated with income and job promotion, and alcoholism is very significantly deleterious to both.
4
That is, with every step in pay grade, the chances rose of a man being nonalcoholic and better educated. It is likely that it is these factors, not the job or the pay grade, that facilitate better health. This is yet another instance
where the choices we make appear to influence how long we will live.

Most social science studies, including Marmot’s own early ones, control only for self-reported alcohol consumption, which, as I have been at pains to point out, correlates very poorly with objective alcohol abuse. In our samples, however, the difference in health outcomes among occupational levels diminished sharply once we controlled for alcohol abuse, which tends to depress occupational level. Alcoholism is bad for career advancement as well as for health.

The question then becomes: If education so powerfully affects self-care (as well as the more obvious job status), what affects whether or not people stay in school? In general, pursuit of education is most successful in coherent communities that invest heavily in their families and their school systems in an atmosphere of gender and racial tolerance. People with no hope for the future don’t pursue education effectively. Providing that hope is the responsibility of the community, not the individual.

That said, however, the pursuit of education also reflects individual personality traits of
perseverance
and
planfulness
—traits that Friedman and Martin have shown to be important to longevity.
5
An important postscript to these findings is a recent paper by David Baber and colleagues, who find that it is not only increased education per se that reduces mortality; of equal importance was what Baber et al. called
health reading fluency
—the capacity to read prescription bottles, understand preventive services, and so on.
6

II. IS PTSD DUE TO COMBAT OR TO PERSONALITY DISORDER?

The frequency of posttraumatic stress disorder (PTSD) among returning Vietnam veterans has led many to wonder whether the principal
causal
factor is not severe combat stress after all, but pre-existing personality disorder. To address this question, the Grant Study took advantage of the fact that most of its members were World War II veterans. Not only had they all been extensively studied before the war, but they were all extensively debriefed after the war on their combat experiences, their physical symptoms during combat, and their persisting stress-related symptoms. John Monks, an internist particularly interested in combat experience, carried out these debriefings.
7
Forty years later, sociologist Glen Elder and I asked all the surviving College veterans (excluding the early Study dropouts) to fill out questionnaires regarding persisting symptoms of posttraumatic stress. (PTSD had not yet been “invented” in 1946, but its principal symptoms had been anticipated by the prescient Monks.) One hundred and seven men returned questionnaires. These men had also completed the NEO, which, as I’ve described in
Chapter 4
, is an extensive and popular multiple-choice scale that includes a scale for the trait
Neuroticism.
8
We were particularly interested to learn if the men who developed symptoms of posttraumatic stress had already looked vulnerable before the war, or if combat really was the primary factor in symptom development. (Please note here, however, that since the Study men had been protected to some degree by education and rank from the circumstances in which frank PTSD develops, we were studying
symptoms
of posttraumatic stress, not the disorder per se.)

First, the men who experienced the most intense combat did not appear to have been more vulnerable by nature; in fact, they manifested superior psychosocial health in adolescence and also at age sixty-five. Second, it was only the men with high combat exposure who continued to report symptoms compatible with PTSD after forty years. Third, we found that symptoms of posttraumatic stress both in 1946 and in 1988 were predicted independently by two factors: combat
exposure and number of physiological symptoms during combat stress (but not during civilian stress). As I’ve already noted, severe combat exposure also predicted early death. It was noteworthy that the symptoms of posttraumatic stress reported in 1946 were not correlated with evidence of subsequent major depressive disorder, alcohol abuse, or poor psychosocial adjustment. Only combat exposure made a significant statistical contribution to posttraumatic stress symptoms, and that contribution was very significant. And while the NEO trait
Neuroticism
was associated with bleak childhood, psychiatrist utilization, poor psychosocial outcome at age forty-seven, and physiological symptoms during civilian stress, it was
not
associated with PTSD.
9

Sixteen men who endured severe combat reported no posttraumatic stress symptoms in 1946, and in 1988 still could not recall ever having had such symptoms. When we compared these sixteen resilient men to the eighteen with high combat exposure who did experience symptoms, their
Neuroticism
scores were the same. However, their defensive styles were not. The high combat veterans who manifested less mature defenses as young adults had very significantly more symptoms than those with more mature defenses (
Chapter 8
). Equally important, seven (39 percent) of the men with high combat experience and less mature defenses were dead by age sixty-five, while no man with mature defenses was. In the Grant Study, high combat exposure per se was not associated with a higher incidence of postwar alcoholism, but as I’ve said, the Grant Study men also did not develop full-blown PTSD.

The take-home messages here are: first, that combat exposure predicted symptoms of posttraumatic stress, but pre-existing psychopathology did not; and second, premorbid emotional vulnerability predicted subsequent psychopathology, but not symptoms of post-traumatic stress. Without the kind of prospective data and prolonged
follow-up
that the Grant Study made available, this finding would not have been possible.

III. POLITICS, MENTAL HEALTH, AND . . . SEX

As an optimist, a psychoanalyst, and a Democrat, I brought many prejudices to the Grant Study. I’ve always been delighted to have them confirmed, and this actually has happened occasionally. There
is
such a thing as a happy marriage, for example. And although colleagues derided for years my conviction that alcoholics could not return safely to controlled drinking, patience (as I detailed in
Chapter 9
) has seen me vindicated.
10

Still, the greatest value of longitudinal study is its way of shattering prejudices and superstitions, and the Grant Study has done a real job on some of mine. I held a deep belief, for example, that Republicans are neither as loving nor as altruistic as Democrats. But after it occurred to me that Gandhi was a very bad father and John D. Rockefeller rather a good one, I thought that I had better test that notion out.

My first step was to identify political preferences from 1950 to 1999. Every biennial questionnaire had asked about politics, and we knew a lot about the men’s political views. We knew, for instance, that in 1954 only 16 percent had sanctioned the McCarthy hearings. We knew that like many liberal members of their generation, the Grant men were “for” equal rights, at least after the fact. They applauded the Supreme Court decisions and civil rights legislation after they happened, but only very few took an active role in trying to bring about racial or gender equality. In 1967, 91 percent were for de-escalating our involvement in Vietnam; this was true of only 80 percent of their classmates. Had the Grant Study subjects had their way, Eugene McCarthy
and Nelson Rockefeller would have been nominated in 1968 rather than Humphrey and Nixon; Gore would have easily beaten Bush in 2000, and the U.S. would never have invaded Iraq without further consultation or U.N. permission.