Personality and Social Factors 497suspiciousness as measured by the Cattell 16-PF also pre-
dicted all cause mortality for older men and women while
P. Martin et al. (1992) reported survival for centenarians was
enhanced for those high on suspiciousness„a measure of
hostility and a reversal similar to the Type A “ ndings reported
by Williams et al. (1988) earlier in the lifecycle.
Exciting new “ndings from the Nun Study (Danner,
Snowdon, & Friesen, 2001; Snowdon, 2001) suggest that
positive emotions measured in late adolescence are related to
long-term survival.
Does Disease Cause Personality Change in Adulthood?
It seems a truism to say it depends on the disease, but that
appears to be the case. Siegler (2000) reviewed the evidence
for rated personality change in AD patients. In studies in the
United States and the United Kingdom, AD patients are rated
as more vulnerable (N6) and less conscientious (C) as would
be expected. When the same technique was used to describe
patients with other brain disorders, the pattern of change was
speci“c to diagnosis, thus, it would appear that where behav-
ior change is a symptom of the disorder, it is speci“c to the
disorder.
Siegler and colleagues (1991, 1994) looked at the role of
premorbid personality in the facets of the NEO-PI. It was
clear that rank order stability is maintained on the facets of
personality that are not the hallmark of the disease. The
Spearman correlations are highly signi“cant (r•s range
from.397,p.02 to.776,p.0001) for all domains
and facets of the NEO-PI except for N6„vulnerability
(r.046,p.80), and C„conscientiousness (r.001,
p.999), which are hallmarks of a dementing disorder.
A study of sources of personality change in the University
of North Carolina (UNC) Alumni Heart Study Cohort found
that divorce versus remarriage and not the development of
a disease during midlife(from ages 42 to 50) was associ-
ated with change in personality (Costa, Herbst, McCrae, &
Siegler, 2000). Thus, disease is not an engine of personality
change unless there are major structural alternations in brain
functioning during midlife. Studies that depend on self-
reports of disease or self-rated health tend to report personal-
ity change, but these “ndings may be related to the appraisal
process.
The overall general stability of personality is also impor-
tant in the study of personality disorders (Costa, McCrae, &
Siegler, 1999). This is an area assumed to have no relevance
to later life, as personality disorders were thought to age out.
Research in this area (Costa & Widiger, 1994) indicates that
this is not true, and indeed, the persistence of disordered per-
sonality patterns can often be mistaken for the disruption of
Alzheimer•s disease (Costa, 2000; Siegler, Bastian, Steffens,
Bosworth, & Costa, in press).Role of Behavioral Risk FactorsCHD, cancer, disability, and total mortality are all increased
by •bad habitsŽ and hostility is related to these bad habits.
Some data from our ongoing UNC Alumni Heart Study help
tell this story. Siegler et al. (1992) reported on the ability of
hostility to predict risk. This was measured at college enroll-
ment, and the study continued as the subjects aged into their
early forties. Hostility was signi“cantly associated with caf-
feine use, ratio of total/HDL cholesterol, amount of alcohol,
and body mass index in the predicted direction„higher
hostility, higher risk„and to exercise in the opposite
direction„higher hostility , more exercise„controlling for
age and gender. As a follow-up to that paper, Siegler and
colleagues examined the association between the same col-
lege hostility measure and risky behaviors measured during
the following 10 years of data collection from 1989 to 1999.
In this analysis, high-risk behavior is set at a level generally
accepted in the literature. In this college-educated cohort,
high hostility in college still predicts excess risk during
midlife for both men and women if they are current smokers,
have high depressive and alcoholism symptoms, and ap-
praisals of their current life situation are negative. Those not
still predicted by college hostility were all signi“cantly pre-
dicted by hostility when remeasured at age 42 by and change
in hostility from age 19 to age 42 (Siegler et al., 2002).
Furthermore, this paper indicated that college hostility, adult
hostility, raw change in hostility, and residualized change in
hostility were all differentially associated with risk, indicat-
ing the complexity of developmental patterns relating
psychosocial factors to disease risk indicators. This is partic-
ularly important from an aging point of view because these
same risk indicators (smoking, obesity, sedentary behavior)
predict Medicare costs (Daviglus et al., 1989) and disability
(Vita, Terry, Hubert, & Fries, 1998). Early predictors of
these risk indicators can help ensure a healthy life span, and
interventions aimed at hostility modi“cation might have
signi“cant dividends and a role to play in primary preven-
tion of CHD in both men and women earlier in the life
span (Stampfer, Hu, Manson, Rimm, & Willett, 2000).
Understanding the role of genetic polymorphisms related to
neurotransmitter function is another way to understand the
potential associations of personality, risky behavior, and
disease risk (Williams et al., 2001).
There is increasing interest in health promotion and dis-
ease prevention in older persons, and interest in successful
aging shifts the focus to those older persons who are doing