Contents
Introduction
xi
Part I Behaviour, stress and health
1
1 Psychosocial correlates of health
3
Individual risk factors and disease 4
Personality and disease 11
Social and environmental influences on health 14
Summary and conclusions 20
Further reading 21
2 Stress and health
22
Towards a definition of stress 22
Psychophysiological substrates of stress 28
Investigating the stress/disease relationship 35
From stress to disease: psychophysiological processes 40
Summary and conclusions 42
Further reading 42
Part II Understanding health-related behaviour
43
3 Health-related decision making
45
Expectancy value models 46
Attitudinal models 52
Stage theories of behavioural change 55
Social cognition models: some dilemmas and questions 58
Alternative approaches to the study of behavioural
decision making 59
Summary and conclusions 62
Further reading 62
4 Health- and illness-related cognitions
63
Children’s understandings of health and illness 64
Adults’ representations of health and disease 66
Coping with illness 74
A cognitive model of pain 76
Summary and conclusions 79
Further reading 79
Part III Applied health psychology
81
5 Hospital issues
83
The experience of hospitalization 83
Coping with a diagnosis 85
Coping with mortality 87
Patient and hospital staff interactions 89
Adherence 92
Medical decision making 96
Stress and the hospital system 97
Summary and conclusions 101
Further reading 101
6 Working in the hospital system
102
Patient-based interventions 103
Increasing adherence 107
Working with the dying 112
Working as a health psychologist 115
Summary and conclusions 120
Further reading 121
7 Health promotion
122
Applying psychological theory to health promotion 122
Community programmes 127
Worksite health promotion 129
Towards a new agenda for health promotion 132
Summary and conclusions 139
Further reading 140
Part IV Clinical interventions
141
8 Psychological interventions
143
Matching interventions to individual characteristics 143
Client-centred counselling 145
Motivational interviewing 146
viii
Contents
Problem-focused counselling 147
Stress management training 151
Emotional disclosure 157
Self-management training 158
Operant conditioning 159
Informal interventions 160
Summary and conclusions 161
Further reading 162
9 Assessment issues
163
Measuring quality of life 164
Disease-specific measures 169
Measuring pain 171
Measures of affect 172
Coping with illness 175
Individual differences 178
Health-related behaviours 181
Summary and conclusions 183
Further reading 183
10 Improving quality of life
184
Information provision 185
Stress management and other cognitive behavioural
interventions 188
Coping effectiveness training 194
Problem-solving counselling 194
Operant approaches 195
Self-management 196
Emotional expression 200
Summary and conclusions 201
Further reading 201
11 Risk behaviour change
202
Screening programmes 203
Educational programmes 206
Self-help approaches 208
Behavioural programmes 209
Relaxation and stress management training 214
Motivational interviewing 217
Other interventions 218
Summary and conclusions 221
Further reading
status is not trivial. It indicates that the status of health psychology
has shifted from one of a shared scientific interest to that of an applied
profession. Health psychologists, with appropriate training, can now
achieve chartered status and work autonomously within the health service
and other settings. Health psychologists are already working in areas
such as health promotion, health-related research, and as consultants to
a variety of organizations. An increasing number are also more directly
involved in the provision of health care.
The delineation of the roles of the two professional groups in this
setting has already resulted in forests of paper being consumed in consul-
tation documents and lively debate. At present the boundaries between
the two professions are blurred. A crude position statement would indi-
cate that clinical psychologists will maintain the patient as their primary
focus, while health psychologists work at other levels: teaching, training,
working at an organizational level, and so on. Those who have both
clinical skills and the knowledge and practice base of health psychology,
and who may truly be called clinical health psychologists, may adopt
any or all these roles. However, this simple categorization fails to take
account of the training in non-patient issues that clinical psychologists
receive, and will surely be challenged by clinicians who are skilled in
working at an organizational level, by health psychologists who prove to
have excellent therapeutic skills, and so on.
Whatever the final, probably overlapping roles, adopted by each pro-
fession, what is clear is that psychologists working in medical settings
need to be aware of both health and clinical psychology theory and how
it can be applied to maximize the effectiveness of health care delivery.
This book provides an introduction to the knowledge base, theory, and
the practice of both health and clinical psychology as applied to health,
and is relevant to professionals, trainees or students wishing to gain an
understanding of health and clinical psychology as applied to the care of
the physically ill. It is divided into four parts:
u
Part I: Behaviour, stress and health
u
Part II: Understanding health-related behaviour
u
Part III: Applied health psychology
u
Part IV: Clinical interventions
Part I: Behaviour, stress and health
The two chapters in Part I consider the relationship between behaviour,
stress and health. Chapter 1 firstly considers the risk for disease associated
with a number of individual behaviours such as smoking or poor diet,
before moving to consider more ‘social’ causes of ill health such as low
socio-economic status or poor working conditions. It also considers
how the effects of gender on health may be behaviourally mediated and
xii
Introduction
not simply a function of biology. Each section considers some of the
controversies that have been associated with the relevant area of research
as well as what we know about the associations between each risk factor
and disease. Chapter 2 focuses on social, psychological and physiological
theories of stress. It considers the relationship between theories that con-
sider stress to be ‘in the eye of the beholder’ (for example Lazarus and
Folkman 1984) and those that consider stress to be a more direct function
of environmental demands and resources (for example Hobfoll (1989) ).
It then describes the physiological processes that underlie the stress
response, considering the cortical, sympathetic and immune systems.
Part II: Understanding health-related behaviour
The second part also comprises two chapters. The first provides a critical
overview of some of the most influential theories of behavioural decision
making used and developed by health psychologists. These social cogni-
tion theories attempt to identify key variables (attitudes, social norms,
cost/benefit analyses and so on) that underpin choices related to health
behaviours. They have not been without their critics, however, and
the chapter explores some of the strengths of other approaches to the
study of health-related behavioural choices. The second chapter in Part II
examines children’s and adults’ concepts of both health and illness and
how these concepts influence their response to illness. It also considers
how cognitions form an important part of our response to one particular
symptom: pain.
Part III: Applied health psychology
This part has three chapters. The first identifies a number of elements of
the care system that impact on how people cope with illness and react to
hospitalization. The second considers how these may be influenced by
psychologists and other health professionals to maximize the effective-
ness of the health care system and minimize its adverse psychological
consequences on the individuals who enter it. The chapters cover issues
such as the experience of hospitalization, how people cope with a dia-
gnosis of severe or chronic illness, staff/patient interactions, adherence to
medication and behavioural programmes, factors that influence medical
decision making, and stress and the hospital system. The second chapter
includes examples of the work currently conducted by health psychologists
within the health care system.
The third chapter in Part III adopts a critical stance in relation to
health promotion. It argues that while educational and community-wide
programmes based on psychological principles have proven effective in
changing behaviour in the past, future initiatives should focus on changing
Introduction
xiii
new risk factors and social and structural moderators of disease, including
socio-economic inequalities and work factors, rather than focusing
exclusively on changing individual behaviour. In doing so, the chapter
refers back to some of the risk factors discussed in the Chapter 1 and
social cognition theories discussed in Chapter 3.
Part IV: Clinical interventions
The final part has four chapters. The first provides a brief description of
the clinical interventions that are conducted with patients who are physi-
cally ill. The types of intervention selected for inclusion in this chapter are
those most commonly used with patients in acute medical settings. They
include interventions that have been developed specifically for use with
patients with chronic health problems, and others that are in more general
use. The effectiveness of these approaches in the management of disease
states, reducing risk for disease progression, and helping people to cope
with the emotional sequelae to their illness is considered in the final two
chapters. Sitting between these chapters is one that focuses on the assess-
ment of health and psychological status in physically ill patients. This
interrupts the flow between the chapters describing therapeutic approaches
and their application in health care settings. This was done for two
reasons. First, to emphasize the importance of assessment and not to
make it the final, ‘add-on’, chapter. Secondly, some of the assessment
instruments described in this chapter contribute to the evaluative research
reported in the following two chapters.
And finally . . .
Writing this text has encouraged (nay, forced!) me to read in areas of
health and clinical psychology about which I previously knew very little.
It has instilled an interest in areas previously hidden from me, some of
which I am now actively involved in researching. I hope the book is able
to provoke such an interest in you, and that you enjoy reading it.
Paul Bennett
xiv
Introduction
Psychosocial correlates of health
1
Part I
Behaviour, stress and health
Psychosocial correlates of health
3
Psychosocial correlates of health
The assumption that behaviour or personality is linked to health is not
new. The Ancient Greeks and Romans described associations between
personality and health, and this presumptive link has since continued
through medical folklore and even psychoanalytic theory. However, the
scientific exploration of links between behaviour and health is a relatively
young venture. Even ‘classic’ studies identifying a link between smoking
and disease were conducted only in the 1950s and much relevant research
is more recent. This research has had three primary foci. The first has
been the link between behaviours, such as smoking or eating habits, that
confer risk of disease indirectly. Smoking, for example, may cause disease
as a result of the carcinogens inhaled. The behavioural repertoire
associated with smoking is not in itself harmful. A second set of research
has focused on behaviours that
directly
moderate risk of disease. This
includes exercise, but perhaps more excitingly from a psychological
perspective has also included individual differences, including Type A and
C behaviour. A third strand of research has focused on elements of the
social or psychological environment that influence disease rates, including
social support and socio-economic status. The latter, in particular, has
recently emerged as an extremely important area of research.
Rather than simply report the relationship between these variables and
disease rates, this chapter not only reports such data but also looks at
some of the issues or controversies raised by each set of research. The
chapter considers:
u
Individual risk factors and disease
u
Personality and disease
u
Social and environmental influences on health
1
4
Behaviour, stress and health
Individual risk factors and disease
Investigating the behaviour/disease relationship
Two differing methodologies have been used to measure the association
between psychosocial factors and health. The simplest is known as a case-
control design, and involves comparing individuals who have a disease
with controls matched on important variables such as socio-economic status
(SES) and age. Any between-group differences found on other variables of
interest are thought to imply causality. This type of approach is fairly cost-
effective but has a number of weaknesses. First, it considers only differences
between the survivors of illness and controls. Those who die of their disease
are excluded from such an analysis, potentially weakening the magnitude
of any observed behaviour/disease relationship. Secondly, the method
allows associations between variables to be identified but the directions
of such relationships have to be assumed. In some cases the direction of
any relationship may be obvious: people with lung cancer, for example,
are unlikely to start smoking as a consequence of their disease. However,
causality can be more difficult to disentangle in other cases. Findings of
high levels of stress in individuals with a debilitating disease, for example,
may raise the question as to whether the disease or the stress came first.
An alternative approach, known as a
longitudinal design
, involves meas-
urement of behaviour in a cohort of (typically) healthy individuals prior
to disease onset. As the study progresses, those that develop disease are
identified and comparisons are made between the baseline characteristics
of these individuals and those who do not develop the disease. Again,
any differences on such measures are thought to imply causality. This
method has the benefit that the behavioural variables are measured prior
to disease onset, so causality can be assured. However, there are a number
of problems inherent in this approach. First, the cohort of individuals
has to be sufficiently large to ensure that a statistically significant number
of individuals will develop the diseases under investigation. Secondly,
the method is based on the assumption that baseline levels of behaviour
will remain constant over the period of the study. The long duration of
such studies makes this assumption questionable. It is possible that many
individuals within the cohort will make considerable lifestyle changes
over follow-up periods that may last up to twenty years or even more.
Any such changes may attenuate the relationship between behaviour and
the initiation of disease. That such studies still find relationships between
behaviour and disease attests to the strength of such relationships.
Behaviour and health
Some of the first evidence to substantiate a link between behaviour and
health came from the Alameda County Study (Berkman and Syme 1979).
Psychosocial correlates of health
5
This longitudinal study has followed nearly seven thousand initially healthy
individuals for a period of more than 20 years and identified which factors
measured at baseline were associated with health or ill health over this
period. One of their earlier reports was the first to highlight an association
between ‘lifestyle factors’ and increased longevity. The behaviours, now
known as the Alameda Seven, were: sleeping 7–8 hours a day, having break-
fast every day, not smoking, rarely eating between meals, being near or at
prescribed weight, moderate consumption of alcohol, and regular exercise.
Cross-cultural comparisons have also shown an association between
behaviour and health. The longevity of the Abraskians, a people who live
in a remote part of Russia and who reputedly live to extreme ages, for
example, has been attributed to genetics and a variety of behavioural factors,
including a low animal fat and high vegetable diet, high levels of social
support, no consumption of alcohol or nicotine and vigorous work activity.
Evidence linking behaviour and health is now overwhelming. Peto and
Lopez (1990), for example, estimated that 75 per cent of all cancer-related
deaths are attributable, at least in part, to behaviour. Others, including
the World Bank (1993), have stated that a significant number of chronic
diseases and up to half of all premature deaths can be attributed to beha-
vioural factors. Five behaviours in particular are associated with risk for
disease: smoking, alcohol misuse, poor nutrition, low levels of exercise and
unprotected sexual intercourse.
Smoking
Smoking doubles the risk of premature death. Approximately 3 million
people die of tobacco use each year across the world (Peto and Lopez
1990). It is responsible for approximately 30 per cent of cases of coronary
heart disease (CHD), 75 per cent of cases of cancer, 80 per cent of cases
of chronic obstructive airways disease, and 90 per cent of deaths associ-
ated with lung cancer. The risks attributable to passive smoking are also
substantial. It is estimated that about 25 per cent of lung cancers that
occur in non-smokers are attributable to passive smoking. In Greece, the
risk for cancer attributable to passive smoking is considered comparable
to that of smoking itself.
Present morbidity levels associated with smoking reflect the cumulative
risk of smoking over many years, and historical processes of some decades
ago. Increased lung cancer rates in women over the past two decades are
thought to be the result of a rapid increase in the numbers of women
smoking during and after the Second World War. Reductions in lung
cancer rates amongst men may reflect the introduction of cigarette filters
at about the same time. In contrast, childhood illnesses represent the more
immediate impact of smoking. In the USA, an estimated half million cases
of childhood pneumonia and bronchitis are attributed to parental smoking.
Smoking cessation decreases risk for all smoking related diseases: former
smokers live longer than persistent smokers do.
6
Behaviour, stress and health
In the West, smokers are now a minority in every age and social group.
In the UK, for example, adult smoking rates between 1974 and 1996 fell
by 26 per cent among men and 22 per cent in women, to 29 and 28 per
cent of the population, respectively (see Figure 1.1). Smoking rates among
young people fell consistently between 1974 and 1992. More recently
this decline has slowed and there is evidence of increasing smoking rates
in this group, particularly among young women, who may soon prove
the majority of smokers in a number of countries.
Excessive alcohol consumption
Excessive alcohol consumption may impact adversely on both short- and
long-term health. It is thought to contribute to 3 per cent of all cancers.
Alcohol also contributes to conditions such as cirrhosis of the liver and
hypertension. However, the most damaging effect of alcohol may be
behavioural. It is estimated that 20 per cent of psychiatric admissions,
60 per cent of suicide attempts, 30 per cent of divorces, and 40 per cent
of incidences of domestic violence are associated to some degree with
alcohol misuse.
In general, there is a linear relationship between level of alcohol con-
sumption and disease rates. The one exception to this is that relating
to CHD. A number of cross-sectional studies in the 1970s reported a
J-shaped relationship between consumption and disease rates. This sug-
gested that moderate consumption of alcohol is associated with lower
risk for CHD than total abstinence, while higher levels of consumption
increase risk. This unexpected relationship was met initially with some
caution, with some suggesting that it may have been an artefact of the
populations studied. It was suggested that the higher than expected rates
% Smokers
60
40
20
0
1996
Men
Women
1992
1988
1984
1980
1976
1972
Figure 1.1
Changes in the prevalence of male and female smokers in the
UK
(
Source
: Of
fi
ce of National Statistics 1999)
Psychosocial correlates of health
7
of CHD among the non-drinkers may have reflected the inclusion of
individuals who had stopped drinking as a consequence of drink-related
health problems. However, a number of longitudinal studies (see, for example,
Shaper
et al
. 1994) have found the J-shaped relationship between alcohol
and CHD after following cohorts of individuals free from disease at
baseline. A mechanism through which cholesterol reduces risk of CHD has
now also been found: moderate consumption appears to reduce harmful
cholesterol levels.
Defining what is meant by excess alcohol consumption has proven far
from simple. This confusion is illustrated by changes to health advice
made by the UK government in 1995. Between 1986 and 1995 the recom-
mended limits for weekly consumption were 21 units of alcohol or less
for men, and 14 units or less for women. In 1995, a government commit-
tee established to review these guidelines recommended they be increased
to 28 and 21 units per week, respectively. These changes caused a furore
and much criticism among alcohol experts, particularly as they were not
based on any new evidence (see, for example,
British Medical Journal
,
vol. 293). Consequently, a number of health promotion and alcohol
agencies have been reluctant to adopt these guidelines and there is a lack
of clear advice concerning the recommended limits to consumption.
The percentage of the population to exceed the 21/14 unit limits has
remained quite stable throughout the past decade, although there has
been a slight increase among women: 27 per cent of men and 12 per cent
of women exceeded these limits in 1996 (see Figure 1.2). Consumption
declines with age: 40 per cent of men aged 18–24 and 18 per cent of those
over 64 years report drinking over the recommended limits. The same
pattern is found among women, although consumption is lower, with
rates of 24 and 7 per cent respectively. Those in the lower socio-economic
groups tend to drink more than the more socially advantaged.
All males aged
16 and over
Recommended
weekly number
of units
40
40
Males
30
20
10 0
21
16
–
24
Females
14
010
20
30
All females aged
16 and over
25
–
44
45
–
64
65 and
over
Figure 1.2
Average levels of alcohol consumption in British men and
women according to age in 1998
(
Source
: Of
fi
ce of National Statistics 1999)
8
Behaviour, stress and health
Cholesterol
Raised serum cholesterol levels increase risk for CHD. The Multiple Risk
Factor Intervention Trial (MRFIT), for example, followed over 350,000
adults for six years and found a linear relationship between baseline
cholesterol level and the incidence of CHD or stroke (Neaton
et al
. 1992).
Individuals within the top third of cholesterol levels were three and a
half times more likely to develop cardiovascular disease than those in the
lowest third. While there is no threshold level below which there is no
risk for CHD, risk is significantly increased by cholesterol levels above
5.2 mmol/litre for those aged over 30 years, and above 4.7 mmol/litre for
younger people. These margins place about two-thirds of the UK popu-
lation at some risk for CHD as a consequence of their serum cholesterol
levels (Lewis
et al
. 1986).
Cholesterol is essential to life. It is a constituent of every cell in the body,
and is implicated in a variety of bodily functions, including the production
of sex hormones and the bile necessary for digestion. A significant per-
centage of our cholesterol is synthesized by the liver; the rest is absorbed
from food. Circulating levels of cholesterol are also mediated by stress
(see Chapter 2) and exercise levels. Nevertheless, the most frequent method
by which public health authorities have tried to control cholesterol is
through dietary means. Recommended levels of intake are frequently
substantially lower than actual levels. In the USA, for example, approxim-
ately 44 per cent of calories are consumed as fat, contrasting with the
recommended level of 30 per cent. Despite these figures, there is some
evidence that the British diet is becoming more healthy (see Figure 1.3).
The unexpected twist in the cholesterol story is that low cholesterol
levels also confers risk of premature mortality. The MRFIT study found
that individuals with low cholesterol levels carried a risk of suicide or
trauma-related death 1.4 times greater than that of men in the mid-range.
Even more dramatically, a longitudinal study of 52,000 Swedish adults
reported that participants in the low cholesterol range evidenced a rate
of non-illness-related mortality 2.8 times higher than those in the mid-
range: risk for suicide was 4.2 times greater (Lindberg
et al
. 1992).
The link between low levels of cholesterol, suicide and accident rates
may appear, at first consideration, somewhat surprising. However, there
is considerable evidence from forensic studies that low levels of choles-
terol are associated with aggression, personality disorder and low mood.
With these findings in mind, the excess mortality following cholesterol
reduction is perhaps not so surprising. More problematic is finding
an explanation. Current explanations are focusing on a link between
cholesterol and serotonin levels. Low levels of cholesterol are associated
with low serotonergic activity, which, in turn, is linked to aggression and
disinhibition of behaviour. While there is reasonable evidence to support
both these links, the mechanisms through which the effects are mediated
are unclear. One possibility is that low cholesterol levels mediate changes
Psychosocial correlates of health
9
Figure 1.3
Changes in the British diet over the past two decades
(
Source
: Of
fi
ce of National Statistics 1999)
in cell membrane function resulting in alterations to the serotonergic
neurotransmission processes.
Exercise
Those who are physically active throughout their adult life live longer
than those who are sedentary. One of the earliest studies to report such
a relationship compared CHD rates among bus drivers and bus conductors
who shared a similar working environment, but engaged in significantly
different levels of exercise. Bus conductors evidenced significantly lower
rates of CHD. In retrospect, some of these differences may be attributable
to differences in stress levels associated with the differing job types (see
below). However, these findings have been supported by a number of
longitudinal studies. Paffenbarger
et al
. (1986), for example, monitored
leisuretime activity in a cohort of Harvard graduates for a period of
16 years. Those who expended more than 2000 kcal of energy in active
leisure activities per week lived, on average, two and a half years longer
than those who expended less than 500 calories in exercise. How this
protection is achieved, whether through short, intense periods of exercise
or longer, less intense periods, appears unimportant and no additional
health gain is achieved by exceeding these limits. Uptake of exercise is
protective against CHD, leading to reductions in resting blood pressure,
cholesterol and triglyceride levels. Exercise is also an important aspect of
weight control and, particularly in women, is protective against osteo-
porosis. As a function of the relatively low levels of fat in those who
exercise regularly, it may also protect against some cancers.
Grams per person per week
300
200
100
0
Low and
reduced
fat spreads
Poultry
Beef and
veal
Mutton
and lamb
Fish
1997
1991
1986
1981
1976
1971
Meat and
fi
sh
Grams per person per week
300
200
100
0
Butter
Margarine
Lard
1997
1991
1986
1981
1976
1971
Fats
10
Behaviour, stress and health
Forty-five per cent of the UK population report engaging in some form
of leisure exercise at least once a month, with the figure rising to 64 per
cent if walking is included in these activities. A lesser number of indi-
viduals achieve the levels of exercise considered necessary to protect against
CHD: here, the figure is nearer to 25 per cent of the adult population
(Norman
et al
. 1998). These figures, however, represent a significant rise
in exercise participation: in 1985, only 20 per cent of British men and
2 per cent of women engaged in such levels of exercise. Those who engage
in exercise are more likely to be young, male, and members of higher
socio-economic groups. Participation in leisure exercise among profes-
sionals, for example, is virtually double that among unskilled manual
workers (80 versus 45 per cent), although the latter may engage in more
physically demanding work activity. Those who participated in some
form of sport in their youth are almost three times more likely to exercise
in adulthood.
Unsafe sex
Estimates of the prevalence of HIV in 1997 suggested that a total of
30.6 million people were infected with HIV worldwide. In the UK, the
primary route of infection has been through sex between men, account-
ing for 72 per cent of all AIDS cases reported by 1998. Eighteen per cent
of cases resulted from heterosexual sex. However, the incidence of new
cases among gay men is falling slowly while the incidence of HIV infec-
tion within the heterosexual community is rising. In addition, young
people are at increasing risk of infection: adolescent heterosexuals
account for about 20 per cent of all newly reported cases in the USA
(Stiff
et al
. 1990).
Nearly half of British adolescents aged between 16 and 17 report
having had at least one sexual partner during the previous year. However,
they are unlikely to plan intercourse and those using a condom are in
the minority. The findings of a large-scale British survey conducted by
Wellings
et al
. (1994) indicated that only about half of those whose first
sexual experience occurred between the ages of 16 and 24 years used a
condom at first intercourse: 31 per cent of men and 24 per cent of
women report using no form of contraception. Younger people were less
likely to take precautions: over 60 per cent of respondents who had their
first sexual experience at the age of 13 reported not using any form of
contraception.
Heterosexuals in the general adult population use condoms consist-
ently only about 10–15 per cent of the time with primary partners and
15 per cent of the time with secondary partners (Dolcini
et al
. 1995).
Even more concerning from a disease prevention perspective is the choice
made by those who are known to be HIV positive not to use a condom.
Sobo (1993), for example, was able to identify a group of HIV positive
women who did not use a condom with their lover or husband, but did
Psychosocial correlates of health
11
Table 1.1
The percentage of respondents engaging in one or more key
health-related behaviours
No. of health behaviours Percentage of sample
0 6.2
1 20.6
2 34.3
3 30.9
4 8.1
Source
: Norman
et al
. 1998
so with more casual sexual partners (see Chapter 3). The likelihood of
an increase in the rate of non-protected intercourse appears to be rising,
perhaps because of the wide prevalence of beliefs that new treatments
for HIV will be curative (Kalichman
et al
. 1998).
Is there a healthy lifestyle?
In an attempt to answer this question, Norman
et al
. (1998) categorized
over 13,000 people in a British survey as either engaging or not engaging
in up to four health-related behaviours: not smoking, moderate alcohol
use, exercising three or more times a week, and eating fruit or vegetables
regularly. While 93 per cent of the sample reported engaging in at least
one of these behaviours, only 8 per cent reported that they did them all
(Table 1.1).
Personality and disease
Type A behaviour and hostility
First identified by two cardiologists, Rosenman and Friedman, Type A
behaviour (TAB) was defined as an excess of competitiveness, time urgency
and easily aroused hostility. The absence of such characteristics is referred
to as Type B behaviour. Early case-control and longitudinal studies
showed significant associations between TAB and CHD. Such was the
research consensus achieved that, by the mid-1980s, TAB was con-
sidered to confer the same degree of risk for CHD as the more tradi-
tional risk factors of high blood pressure and cholesterol. Subsequent,
and frequently methodologically flawed, studies failed to find a relation-
ship between TAB and disease progression in men either at high risk of
CHD or who had already experienced a myocardial infarction (MI). As
a consequence, this degree of consensus is no longer evident. However,
one component of TAB, hostility, appears to convey the degree of risk
12
Behaviour, stress and health
previously ascribed to TAB and this has formed the primary focus of
more recent research.
A number of case-control studies have reported significant associations
between hostility and CHD, with the strongest relationship being found
among those under 50 years of age. Such a finding is not unique: the
predictive power of other biological and behavioural risk factors is also
strongest among younger people. However, the most convincing evidence
linking hostility to CHD has stemmed from longitudinal studies. In one
of the earliest such studies, Barefoot
et al
. (1983) followed 255 physicians
for 25 years and found that those who scored above the median on
measures of hostility taken while in training were nearly five times more
likely to experience an MI over this period than those who scored below
the median. These data have been supported by findings in older popu-
lations. European studies (for example Everson
et al
. 1997) following
middle-aged men for up to 25 years have found an association between
hostility and coronary events. Hostility may also contribute to the develop-
ment of CHD in women (Lahad
et al
. 1997). Data from some of these
and other studies combined into meta-analysis suggested a significant
association between hostility and CHD (Miller
et al
. 1996).
A second set of studies has measured the association between hostil-
ity and the degree of atheroma of the cardiac arteries, measured using
angiography. Although there have been a number of positive findings,
several studies have failed to find any relationship between hostility and
atheroma. One explanation for these disparate findings may be biased
sampling. Angiograms are invasive procedures usually conducted only on
patients who report some degree of symptomatology. Accordingly, the
samples used are likely to be non-representative, and the results of most
of these studies should be considered with caution. Two types of study
are the exception to this rule: those that track the progression of atheroma
over time, and those involving angiograms in representative populations.
In the first type, Julkunen
et al
. (1994) measured the relationship between
hostility and the progression of atherosclerosis over a two-year period in
a sample of 119 middle-aged men. The progression of atheroma was
almost twice as fast in highly hostile participants as in those who were
low in hostility. Utilizing the second methodology, Barefoot
et al
. (1994)
took advantage of the need for airline pilots to have routine angiograms
to ensure their fitness to fly. They compared the hostility scores of pilots
with angiographic evidence of CHD with those having none. Among
non-smokers only, pilots found to have evidence of CHD had higher
hostility scores than did those in the comparison group.
Type C behaviour
Type C behaviour has been independently described by Greer and Morris
(1975) and Temoshok (1987). It is defined as an aggregate of several
Psychosocial correlates of health
13
coping styles, in particular being stoic, cooperative, appeasing, unassertive
and inexpressive of negative emotions, particularly anger. It is thought
to be linked to the development of cancer.
A number of studies have shown an association between Type C per-
sonality and the incidence or progression of cancer. Using a case-control
design, Kune
et al
. (1991), for example, compared patients newly diag-
nosed with colorectal cancer and community controls matched for age
and gender. Patients with cancer were significantly more likely to report
histories of unhappiness in childhood and recent adult life and to have
strong feelings of discomfort when experiencing feelings of anger. These
differences may have been influenced by knowledge of a diagnosis of
cancer. Accordingly, more weight should be ascribed to the findings of
studies where a diagnosis was not known at the time of psychological
assessment. A number of such studies (for example Greer and Morris
1975) have found that women who suppressed their anger or had a
conforming personality were more likely to have malignant changes than
those without these characteristics. Longitudinal studies have also shown
Type C characteristics to predict cancer. Shaffer
et al
. (1987), for exam-
ple, followed 972 physicians for a 30-year period and found that partici-
pants characterized by high levels of ‘acting out’ and emotional expression
had a less than 1 per cent risk of developing cancer. Participants charac-
terized as ‘loners’ and thought to inhibit emotional expression were 16
times more likely to develop cancer than those in this group.
Concordant with the Type C hypothesis is work examining the impact
of ‘fighting spirit’ on cancer progression. This constellation of behavi-
ours is considered to be the opposite of Type C behaviour, and has been
associated with longer survival following diagnosis. In the first study to
identify this characteristic as a prognostic factor, Derogatis
et al
. (1979)
found that women who showed ‘fighting spirit’ during treatment for meta-
static breast cancer lived significantly longer than those who did not.
However, many of these women also received less chemotherapy, sug-
gesting that disease severity may have been worse in the poor survival
group or that their higher levels of fighting spirit could have been a
consequence of being subjected to lower levels of a physically and men-
tally debilitating treatment. Greer (1991) also found fighting spirit to be
an important determinant of survival. In a longitudinal study following
62 women with early non-metastatic breast cancer, he identified five
reaction types: denial, fighting spirit, stoic acceptance, helplessness and
hopelessness, and anxious preoccupation. Breast cancer recurrences and
mortality were recorded 5, 10 and 15 years later. By 15-year follow-up,
45 per cent of the women who were categorized in the fighting spirit
group were alive without recurrences; this compared with 17 per cent in
all the other groups. Similar evidence has been published in a variety of
cancers, including melanoma and lung cancer. However, there have also
been some negative findings. In a larger replication study of the original
Greer study, for example, Dean and Surtees (1989) found denial measured
14
Behaviour, stress and health
after surgery was associated with a
favourable
outcome in a group of 125
women with non-metastatic breast cancer followed for up to eight years.
Such contradictory findings have meant that the relationship between
Type C behaviour, ‘fighting spirit’, and cancer remains controversial and
far from proven. A related strand of research has focused on the health
outcomes following attempts to facilitate active coping with disease and
the expression of emotions. These studies are considered in Chapters 7
and 8.
Social and environmental influences on health
Socio-economic status and health
There is strong historical evidence that the more affluent members of
society have lived longer than the less well off. More recent evidence
of this health gradient can be found in a study of nineteenth century
obelisks in Glasgow graveyards. In an imaginative study, Davey Smith
et al
. (1992) measured the height of obelisks in Glasgow graveyards as
a proxy for the wealth of the individual buried beneath them. They
compared these with the ages of the first generation buried below and
found a strong linear relationship between the height of the obelisks and
the age of their first occupant, suggesting that the more wealthy lived
longer. What is important is not just that this relationship existed, but
that the families buried in the graves represented a small and wealthy
fraction of the Glasgow population. This is not evidence that the very
poor and immiserated did not live as long as the rich did: rather, that
the relatively rich did not live as long as the very rich. Such a gradient
still exists. It is progressive, and throughout the social classes. It holds
for women as well as men and is characteristic of all western countries
(Wilkinson, 1992).
A number of explanations have been proposed to account for these
differences. People in lower socio-economic groups may be exposed to
more environmental insults, low quality and damp accommodation,
and air pollution. An alternative explanation may be that less well off
individuals engage in more health-damaging behaviours, such as smok-
ing or excessive alcohol consumption. While both these explanations
may account for some of the differences in some of the studies, they
cannot explain them all. Marmot
et al
. (1984), for example, explored
the impact of a number of these variables on the health of British civil
servants working in London over a period of ten years. Their findings
indicated that while those in the more deprived social groups did engage
in more health-damaging behaviours, these did not fully explain the
health/SES relationship. When variations in smoking, obesity, plasma
cholesterol and blood pressure were statistically partialled out of the risk
Psychosocial correlates of health
15
equation, occupational-status-related differentials in health still remained.
Mortality was three times higher among men in the lowest grade than
those in the highest.
While people who occupy the lower socio-economic groups may engage
in more health-damaging behaviours, the adverse health effects of these
behaviours may be overwhelmed by factors associated with their economic
position. Hein
et al
. (1992), for example, reported data from a 17-year
prospective study of CHD in Danish men. Adjusting for a variety of
confounding factors, they found that men who smoked were three and a
half times more likely to develop CHD than non-smokers. However,
when these data were analysed according to SES, white-collar smokers
were six and a half times more likely to experience a cardiac event than
the equivalent non-smokers. Amongst blue-collar workers, smoking status
conferred no additional risk for CHD. Fewer middle-class people may
smoke, but those that do may be particularly vulnerable to its health-
damaging effects. Conversely, the impact of smoking on the health of the
less well off is seemingly overwhelmed by social factors.
Comparisons of life expectancy across different western countries sug-
gest some intriguing explanations of the relationship between social class
and health. Wilkinson (1992) provided powerful evidence that it is not
absolute wealth that determines health. He drew on evidence that showed
only a weak relationship between the absolute wealth of the society and
overall life expectancy. More predictive is the distribution of wealth
within a society. The narrower the distribution, whatever its absolute level,
the better the overall health of the nation. Accordingly, although Japan
and Cuba differ substantially on measures of economic wealth, both have
relatively equitable distributions of income and long life expectancies
throughout their populations. Of particular interest is evidence from
Scotland, which tracked average age of mortality and income distribution
over the life of the Thatcher government. As earning differentials rose,
so premature mortality among the less well off increased despite their
access to material goods, food, clothing and so on remaining relatively
constant. These data led Wilkinson to suggest that, for the majority of
people in western countries, health hinges on relatively more than absolute
living standards.
Wilkinson’s explanation of health differentials suggests that we engage
in some form of comparison of our living conditions with others in society,
and that knowledge of a relative deprivation in some way increases risk
of disease. Three different psychological processes may also be implicated
in the health gradient. Individuals in the lower socio-economic groups
report more stressors than those in higher groups and that these stressors
are frequently linked directly to their material conditions. In addition, the
less well off have less control over their environment and fewer personal
resources to moderate the impact of such stressors than the better off.
Finally, social support, a powerful mediator of health status, is less avail-
able to those in lower socio-economic groups (Adler
et al
. 1994).
16
Behaviour, stress and health
Social isolation and health
There is substantial evidence that both men and women who have a
small number of social contacts are more likely to die earlier than those
who have more extended networks. Data from the Alameda County
Study (Berkman and Syme 1979), for example, showed increased lon-
gevity to be associated with relatively high numbers of social ties as a
consequence of marriage, contacts with close friends and relatives, church
membership and membership of other organizations. The most isolated
were the most prone to premature death even after controlling for factors
such as smoking, alcohol use and levels of physical activity. In a later
study, Reynolds and Kaplan (1990) found that women who had few
social contacts and were socially isolated were at double the risk of
developing hormone-related cancers and evidenced an almost fivefold
increase in risk of dying from them than less isolated women.
Similar results have been reported in European samples. Orth-Gomer
and Johnsson (1987), for example, followed a cohort of 17,400 men and
women for a period of six years and found that both men and women
who had a restricted social network evidenced a 50 per cent greater risk
of CHD than those who were socially embedded. However, the relation-
ship between social contact and health was not always linear: those with
many social contacts did not always benefit in terms of health. Older
women, for example, who had many social contacts evidenced higher
mortality than those with medium-sized networks. To explain these
apparently anomalous findings, Orth-Gomer and Johnsson analysed their
data not simply according to the absolute number of contacts, but taking
into account the nature of the contacts. When they did this, they found
the strongest predictor of mortality was a lack of social integration,
which the authors considered to provide guidance, practical help and
a feeling of belonging. They found only a low association between the
provision of emotional support and CHD, although a later study by
the same group found this to be an important protective factor. In an
alternative interpretation of the Type A hypothesis, Orth-Gomer and
Unden (1990) suggested that TAB or hostility might confer risk of CHD
as a consequence of associated social isolation. In a longitudinal study,
following a cohort of men for ten years, they found no differences in
mortality between Type A and Type B men. However, over this period,
69 per cent of the socially isolated Type A men in the cohort had experi-
enced an MI, in contrast to the 17 per cent incidence among those who
were socially integrated.
Further evidence of the impact of social isolation can be found in
studies of populations already experiencing disease. Williams
et al
. (1992),
for example, found that patients with CHD who were unmarried and
without a confidant experienced a threefold higher risk of mortality over
a five-year follow-up period than those who were. Reflecting the subject-
ive nature of social support, some studies that have failed to find a
Introduction
xi
Part I Behaviour, stress and health
1
1 Psychosocial correlates of health
3
Individual risk factors and disease 4
Personality and disease 11
Social and environmental influences on health 14
Summary and conclusions 20
Further reading 21
2 Stress and health
22
Towards a definition of stress 22
Psychophysiological substrates of stress 28
Investigating the stress/disease relationship 35
From stress to disease: psychophysiological processes 40
Summary and conclusions 42
Further reading 42
Part II Understanding health-related behaviour
43
3 Health-related decision making
45
Expectancy value models 46
Attitudinal models 52
Stage theories of behavioural change 55
Social cognition models: some dilemmas and questions 58
Alternative approaches to the study of behavioural
decision making 59
Summary and conclusions 62
Further reading 62
4 Health- and illness-related cognitions
63
Children’s understandings of health and illness 64
Adults’ representations of health and disease 66
Coping with illness 74
A cognitive model of pain 76
Summary and conclusions 79
Further reading 79
Part III Applied health psychology
81
5 Hospital issues
83
The experience of hospitalization 83
Coping with a diagnosis 85
Coping with mortality 87
Patient and hospital staff interactions 89
Adherence 92
Medical decision making 96
Stress and the hospital system 97
Summary and conclusions 101
Further reading 101
6 Working in the hospital system
102
Patient-based interventions 103
Increasing adherence 107
Working with the dying 112
Working as a health psychologist 115
Summary and conclusions 120
Further reading 121
7 Health promotion
122
Applying psychological theory to health promotion 122
Community programmes 127
Worksite health promotion 129
Towards a new agenda for health promotion 132
Summary and conclusions 139
Further reading 140
Part IV Clinical interventions
141
8 Psychological interventions
143
Matching interventions to individual characteristics 143
Client-centred counselling 145
Motivational interviewing 146
viii
Contents
Problem-focused counselling 147
Stress management training 151
Emotional disclosure 157
Self-management training 158
Operant conditioning 159
Informal interventions 160
Summary and conclusions 161
Further reading 162
9 Assessment issues
163
Measuring quality of life 164
Disease-specific measures 169
Measuring pain 171
Measures of affect 172
Coping with illness 175
Individual differences 178
Health-related behaviours 181
Summary and conclusions 183
Further reading 183
10 Improving quality of life
184
Information provision 185
Stress management and other cognitive behavioural
interventions 188
Coping effectiveness training 194
Problem-solving counselling 194
Operant approaches 195
Self-management 196
Emotional expression 200
Summary and conclusions 201
Further reading 201
11 Risk behaviour change
202
Screening programmes 203
Educational programmes 206
Self-help approaches 208
Behavioural programmes 209
Relaxation and stress management training 214
Motivational interviewing 217
Other interventions 218
Summary and conclusions 221
Further reading
status is not trivial. It indicates that the status of health psychology
has shifted from one of a shared scientific interest to that of an applied
profession. Health psychologists, with appropriate training, can now
achieve chartered status and work autonomously within the health service
and other settings. Health psychologists are already working in areas
such as health promotion, health-related research, and as consultants to
a variety of organizations. An increasing number are also more directly
involved in the provision of health care.
The delineation of the roles of the two professional groups in this
setting has already resulted in forests of paper being consumed in consul-
tation documents and lively debate. At present the boundaries between
the two professions are blurred. A crude position statement would indi-
cate that clinical psychologists will maintain the patient as their primary
focus, while health psychologists work at other levels: teaching, training,
working at an organizational level, and so on. Those who have both
clinical skills and the knowledge and practice base of health psychology,
and who may truly be called clinical health psychologists, may adopt
any or all these roles. However, this simple categorization fails to take
account of the training in non-patient issues that clinical psychologists
receive, and will surely be challenged by clinicians who are skilled in
working at an organizational level, by health psychologists who prove to
have excellent therapeutic skills, and so on.
Whatever the final, probably overlapping roles, adopted by each pro-
fession, what is clear is that psychologists working in medical settings
need to be aware of both health and clinical psychology theory and how
it can be applied to maximize the effectiveness of health care delivery.
This book provides an introduction to the knowledge base, theory, and
the practice of both health and clinical psychology as applied to health,
and is relevant to professionals, trainees or students wishing to gain an
understanding of health and clinical psychology as applied to the care of
the physically ill. It is divided into four parts:
u
Part I: Behaviour, stress and health
u
Part II: Understanding health-related behaviour
u
Part III: Applied health psychology
u
Part IV: Clinical interventions
Part I: Behaviour, stress and health
The two chapters in Part I consider the relationship between behaviour,
stress and health. Chapter 1 firstly considers the risk for disease associated
with a number of individual behaviours such as smoking or poor diet,
before moving to consider more ‘social’ causes of ill health such as low
socio-economic status or poor working conditions. It also considers
how the effects of gender on health may be behaviourally mediated and
xii
Introduction
not simply a function of biology. Each section considers some of the
controversies that have been associated with the relevant area of research
as well as what we know about the associations between each risk factor
and disease. Chapter 2 focuses on social, psychological and physiological
theories of stress. It considers the relationship between theories that con-
sider stress to be ‘in the eye of the beholder’ (for example Lazarus and
Folkman 1984) and those that consider stress to be a more direct function
of environmental demands and resources (for example Hobfoll (1989) ).
It then describes the physiological processes that underlie the stress
response, considering the cortical, sympathetic and immune systems.
Part II: Understanding health-related behaviour
The second part also comprises two chapters. The first provides a critical
overview of some of the most influential theories of behavioural decision
making used and developed by health psychologists. These social cogni-
tion theories attempt to identify key variables (attitudes, social norms,
cost/benefit analyses and so on) that underpin choices related to health
behaviours. They have not been without their critics, however, and
the chapter explores some of the strengths of other approaches to the
study of health-related behavioural choices. The second chapter in Part II
examines children’s and adults’ concepts of both health and illness and
how these concepts influence their response to illness. It also considers
how cognitions form an important part of our response to one particular
symptom: pain.
Part III: Applied health psychology
This part has three chapters. The first identifies a number of elements of
the care system that impact on how people cope with illness and react to
hospitalization. The second considers how these may be influenced by
psychologists and other health professionals to maximize the effective-
ness of the health care system and minimize its adverse psychological
consequences on the individuals who enter it. The chapters cover issues
such as the experience of hospitalization, how people cope with a dia-
gnosis of severe or chronic illness, staff/patient interactions, adherence to
medication and behavioural programmes, factors that influence medical
decision making, and stress and the hospital system. The second chapter
includes examples of the work currently conducted by health psychologists
within the health care system.
The third chapter in Part III adopts a critical stance in relation to
health promotion. It argues that while educational and community-wide
programmes based on psychological principles have proven effective in
changing behaviour in the past, future initiatives should focus on changing
Introduction
xiii
new risk factors and social and structural moderators of disease, including
socio-economic inequalities and work factors, rather than focusing
exclusively on changing individual behaviour. In doing so, the chapter
refers back to some of the risk factors discussed in the Chapter 1 and
social cognition theories discussed in Chapter 3.
Part IV: Clinical interventions
The final part has four chapters. The first provides a brief description of
the clinical interventions that are conducted with patients who are physi-
cally ill. The types of intervention selected for inclusion in this chapter are
those most commonly used with patients in acute medical settings. They
include interventions that have been developed specifically for use with
patients with chronic health problems, and others that are in more general
use. The effectiveness of these approaches in the management of disease
states, reducing risk for disease progression, and helping people to cope
with the emotional sequelae to their illness is considered in the final two
chapters. Sitting between these chapters is one that focuses on the assess-
ment of health and psychological status in physically ill patients. This
interrupts the flow between the chapters describing therapeutic approaches
and their application in health care settings. This was done for two
reasons. First, to emphasize the importance of assessment and not to
make it the final, ‘add-on’, chapter. Secondly, some of the assessment
instruments described in this chapter contribute to the evaluative research
reported in the following two chapters.
And finally . . .
Writing this text has encouraged (nay, forced!) me to read in areas of
health and clinical psychology about which I previously knew very little.
It has instilled an interest in areas previously hidden from me, some of
which I am now actively involved in researching. I hope the book is able
to provoke such an interest in you, and that you enjoy reading it.
Paul Bennett
xiv
Introduction
Psychosocial correlates of health
1
Part I
Behaviour, stress and health
Psychosocial correlates of health
3
Psychosocial correlates of health
The assumption that behaviour or personality is linked to health is not
new. The Ancient Greeks and Romans described associations between
personality and health, and this presumptive link has since continued
through medical folklore and even psychoanalytic theory. However, the
scientific exploration of links between behaviour and health is a relatively
young venture. Even ‘classic’ studies identifying a link between smoking
and disease were conducted only in the 1950s and much relevant research
is more recent. This research has had three primary foci. The first has
been the link between behaviours, such as smoking or eating habits, that
confer risk of disease indirectly. Smoking, for example, may cause disease
as a result of the carcinogens inhaled. The behavioural repertoire
associated with smoking is not in itself harmful. A second set of research
has focused on behaviours that
directly
moderate risk of disease. This
includes exercise, but perhaps more excitingly from a psychological
perspective has also included individual differences, including Type A and
C behaviour. A third strand of research has focused on elements of the
social or psychological environment that influence disease rates, including
social support and socio-economic status. The latter, in particular, has
recently emerged as an extremely important area of research.
Rather than simply report the relationship between these variables and
disease rates, this chapter not only reports such data but also looks at
some of the issues or controversies raised by each set of research. The
chapter considers:
u
Individual risk factors and disease
u
Personality and disease
u
Social and environmental influences on health
1
4
Behaviour, stress and health
Individual risk factors and disease
Investigating the behaviour/disease relationship
Two differing methodologies have been used to measure the association
between psychosocial factors and health. The simplest is known as a case-
control design, and involves comparing individuals who have a disease
with controls matched on important variables such as socio-economic status
(SES) and age. Any between-group differences found on other variables of
interest are thought to imply causality. This type of approach is fairly cost-
effective but has a number of weaknesses. First, it considers only differences
between the survivors of illness and controls. Those who die of their disease
are excluded from such an analysis, potentially weakening the magnitude
of any observed behaviour/disease relationship. Secondly, the method
allows associations between variables to be identified but the directions
of such relationships have to be assumed. In some cases the direction of
any relationship may be obvious: people with lung cancer, for example,
are unlikely to start smoking as a consequence of their disease. However,
causality can be more difficult to disentangle in other cases. Findings of
high levels of stress in individuals with a debilitating disease, for example,
may raise the question as to whether the disease or the stress came first.
An alternative approach, known as a
longitudinal design
, involves meas-
urement of behaviour in a cohort of (typically) healthy individuals prior
to disease onset. As the study progresses, those that develop disease are
identified and comparisons are made between the baseline characteristics
of these individuals and those who do not develop the disease. Again,
any differences on such measures are thought to imply causality. This
method has the benefit that the behavioural variables are measured prior
to disease onset, so causality can be assured. However, there are a number
of problems inherent in this approach. First, the cohort of individuals
has to be sufficiently large to ensure that a statistically significant number
of individuals will develop the diseases under investigation. Secondly,
the method is based on the assumption that baseline levels of behaviour
will remain constant over the period of the study. The long duration of
such studies makes this assumption questionable. It is possible that many
individuals within the cohort will make considerable lifestyle changes
over follow-up periods that may last up to twenty years or even more.
Any such changes may attenuate the relationship between behaviour and
the initiation of disease. That such studies still find relationships between
behaviour and disease attests to the strength of such relationships.
Behaviour and health
Some of the first evidence to substantiate a link between behaviour and
health came from the Alameda County Study (Berkman and Syme 1979).
Psychosocial correlates of health
5
This longitudinal study has followed nearly seven thousand initially healthy
individuals for a period of more than 20 years and identified which factors
measured at baseline were associated with health or ill health over this
period. One of their earlier reports was the first to highlight an association
between ‘lifestyle factors’ and increased longevity. The behaviours, now
known as the Alameda Seven, were: sleeping 7–8 hours a day, having break-
fast every day, not smoking, rarely eating between meals, being near or at
prescribed weight, moderate consumption of alcohol, and regular exercise.
Cross-cultural comparisons have also shown an association between
behaviour and health. The longevity of the Abraskians, a people who live
in a remote part of Russia and who reputedly live to extreme ages, for
example, has been attributed to genetics and a variety of behavioural factors,
including a low animal fat and high vegetable diet, high levels of social
support, no consumption of alcohol or nicotine and vigorous work activity.
Evidence linking behaviour and health is now overwhelming. Peto and
Lopez (1990), for example, estimated that 75 per cent of all cancer-related
deaths are attributable, at least in part, to behaviour. Others, including
the World Bank (1993), have stated that a significant number of chronic
diseases and up to half of all premature deaths can be attributed to beha-
vioural factors. Five behaviours in particular are associated with risk for
disease: smoking, alcohol misuse, poor nutrition, low levels of exercise and
unprotected sexual intercourse.
Smoking
Smoking doubles the risk of premature death. Approximately 3 million
people die of tobacco use each year across the world (Peto and Lopez
1990). It is responsible for approximately 30 per cent of cases of coronary
heart disease (CHD), 75 per cent of cases of cancer, 80 per cent of cases
of chronic obstructive airways disease, and 90 per cent of deaths associ-
ated with lung cancer. The risks attributable to passive smoking are also
substantial. It is estimated that about 25 per cent of lung cancers that
occur in non-smokers are attributable to passive smoking. In Greece, the
risk for cancer attributable to passive smoking is considered comparable
to that of smoking itself.
Present morbidity levels associated with smoking reflect the cumulative
risk of smoking over many years, and historical processes of some decades
ago. Increased lung cancer rates in women over the past two decades are
thought to be the result of a rapid increase in the numbers of women
smoking during and after the Second World War. Reductions in lung
cancer rates amongst men may reflect the introduction of cigarette filters
at about the same time. In contrast, childhood illnesses represent the more
immediate impact of smoking. In the USA, an estimated half million cases
of childhood pneumonia and bronchitis are attributed to parental smoking.
Smoking cessation decreases risk for all smoking related diseases: former
smokers live longer than persistent smokers do.
6
Behaviour, stress and health
In the West, smokers are now a minority in every age and social group.
In the UK, for example, adult smoking rates between 1974 and 1996 fell
by 26 per cent among men and 22 per cent in women, to 29 and 28 per
cent of the population, respectively (see Figure 1.1). Smoking rates among
young people fell consistently between 1974 and 1992. More recently
this decline has slowed and there is evidence of increasing smoking rates
in this group, particularly among young women, who may soon prove
the majority of smokers in a number of countries.
Excessive alcohol consumption
Excessive alcohol consumption may impact adversely on both short- and
long-term health. It is thought to contribute to 3 per cent of all cancers.
Alcohol also contributes to conditions such as cirrhosis of the liver and
hypertension. However, the most damaging effect of alcohol may be
behavioural. It is estimated that 20 per cent of psychiatric admissions,
60 per cent of suicide attempts, 30 per cent of divorces, and 40 per cent
of incidences of domestic violence are associated to some degree with
alcohol misuse.
In general, there is a linear relationship between level of alcohol con-
sumption and disease rates. The one exception to this is that relating
to CHD. A number of cross-sectional studies in the 1970s reported a
J-shaped relationship between consumption and disease rates. This sug-
gested that moderate consumption of alcohol is associated with lower
risk for CHD than total abstinence, while higher levels of consumption
increase risk. This unexpected relationship was met initially with some
caution, with some suggesting that it may have been an artefact of the
populations studied. It was suggested that the higher than expected rates
% Smokers
60
40
20
0
1996
Men
Women
1992
1988
1984
1980
1976
1972
Figure 1.1
Changes in the prevalence of male and female smokers in the
UK
(
Source
: Of
fi
ce of National Statistics 1999)
Psychosocial correlates of health
7
of CHD among the non-drinkers may have reflected the inclusion of
individuals who had stopped drinking as a consequence of drink-related
health problems. However, a number of longitudinal studies (see, for example,
Shaper
et al
. 1994) have found the J-shaped relationship between alcohol
and CHD after following cohorts of individuals free from disease at
baseline. A mechanism through which cholesterol reduces risk of CHD has
now also been found: moderate consumption appears to reduce harmful
cholesterol levels.
Defining what is meant by excess alcohol consumption has proven far
from simple. This confusion is illustrated by changes to health advice
made by the UK government in 1995. Between 1986 and 1995 the recom-
mended limits for weekly consumption were 21 units of alcohol or less
for men, and 14 units or less for women. In 1995, a government commit-
tee established to review these guidelines recommended they be increased
to 28 and 21 units per week, respectively. These changes caused a furore
and much criticism among alcohol experts, particularly as they were not
based on any new evidence (see, for example,
British Medical Journal
,
vol. 293). Consequently, a number of health promotion and alcohol
agencies have been reluctant to adopt these guidelines and there is a lack
of clear advice concerning the recommended limits to consumption.
The percentage of the population to exceed the 21/14 unit limits has
remained quite stable throughout the past decade, although there has
been a slight increase among women: 27 per cent of men and 12 per cent
of women exceeded these limits in 1996 (see Figure 1.2). Consumption
declines with age: 40 per cent of men aged 18–24 and 18 per cent of those
over 64 years report drinking over the recommended limits. The same
pattern is found among women, although consumption is lower, with
rates of 24 and 7 per cent respectively. Those in the lower socio-economic
groups tend to drink more than the more socially advantaged.
All males aged
16 and over
Recommended
weekly number
of units
40
40
Males
30
20
10 0
21
16
–
24
Females
14
010
20
30
All females aged
16 and over
25
–
44
45
–
64
65 and
over
Figure 1.2
Average levels of alcohol consumption in British men and
women according to age in 1998
(
Source
: Of
fi
ce of National Statistics 1999)
8
Behaviour, stress and health
Cholesterol
Raised serum cholesterol levels increase risk for CHD. The Multiple Risk
Factor Intervention Trial (MRFIT), for example, followed over 350,000
adults for six years and found a linear relationship between baseline
cholesterol level and the incidence of CHD or stroke (Neaton
et al
. 1992).
Individuals within the top third of cholesterol levels were three and a
half times more likely to develop cardiovascular disease than those in the
lowest third. While there is no threshold level below which there is no
risk for CHD, risk is significantly increased by cholesterol levels above
5.2 mmol/litre for those aged over 30 years, and above 4.7 mmol/litre for
younger people. These margins place about two-thirds of the UK popu-
lation at some risk for CHD as a consequence of their serum cholesterol
levels (Lewis
et al
. 1986).
Cholesterol is essential to life. It is a constituent of every cell in the body,
and is implicated in a variety of bodily functions, including the production
of sex hormones and the bile necessary for digestion. A significant per-
centage of our cholesterol is synthesized by the liver; the rest is absorbed
from food. Circulating levels of cholesterol are also mediated by stress
(see Chapter 2) and exercise levels. Nevertheless, the most frequent method
by which public health authorities have tried to control cholesterol is
through dietary means. Recommended levels of intake are frequently
substantially lower than actual levels. In the USA, for example, approxim-
ately 44 per cent of calories are consumed as fat, contrasting with the
recommended level of 30 per cent. Despite these figures, there is some
evidence that the British diet is becoming more healthy (see Figure 1.3).
The unexpected twist in the cholesterol story is that low cholesterol
levels also confers risk of premature mortality. The MRFIT study found
that individuals with low cholesterol levels carried a risk of suicide or
trauma-related death 1.4 times greater than that of men in the mid-range.
Even more dramatically, a longitudinal study of 52,000 Swedish adults
reported that participants in the low cholesterol range evidenced a rate
of non-illness-related mortality 2.8 times higher than those in the mid-
range: risk for suicide was 4.2 times greater (Lindberg
et al
. 1992).
The link between low levels of cholesterol, suicide and accident rates
may appear, at first consideration, somewhat surprising. However, there
is considerable evidence from forensic studies that low levels of choles-
terol are associated with aggression, personality disorder and low mood.
With these findings in mind, the excess mortality following cholesterol
reduction is perhaps not so surprising. More problematic is finding
an explanation. Current explanations are focusing on a link between
cholesterol and serotonin levels. Low levels of cholesterol are associated
with low serotonergic activity, which, in turn, is linked to aggression and
disinhibition of behaviour. While there is reasonable evidence to support
both these links, the mechanisms through which the effects are mediated
are unclear. One possibility is that low cholesterol levels mediate changes
Psychosocial correlates of health
9
Figure 1.3
Changes in the British diet over the past two decades
(
Source
: Of
fi
ce of National Statistics 1999)
in cell membrane function resulting in alterations to the serotonergic
neurotransmission processes.
Exercise
Those who are physically active throughout their adult life live longer
than those who are sedentary. One of the earliest studies to report such
a relationship compared CHD rates among bus drivers and bus conductors
who shared a similar working environment, but engaged in significantly
different levels of exercise. Bus conductors evidenced significantly lower
rates of CHD. In retrospect, some of these differences may be attributable
to differences in stress levels associated with the differing job types (see
below). However, these findings have been supported by a number of
longitudinal studies. Paffenbarger
et al
. (1986), for example, monitored
leisuretime activity in a cohort of Harvard graduates for a period of
16 years. Those who expended more than 2000 kcal of energy in active
leisure activities per week lived, on average, two and a half years longer
than those who expended less than 500 calories in exercise. How this
protection is achieved, whether through short, intense periods of exercise
or longer, less intense periods, appears unimportant and no additional
health gain is achieved by exceeding these limits. Uptake of exercise is
protective against CHD, leading to reductions in resting blood pressure,
cholesterol and triglyceride levels. Exercise is also an important aspect of
weight control and, particularly in women, is protective against osteo-
porosis. As a function of the relatively low levels of fat in those who
exercise regularly, it may also protect against some cancers.
Grams per person per week
300
200
100
0
Low and
reduced
fat spreads
Poultry
Beef and
veal
Mutton
and lamb
Fish
1997
1991
1986
1981
1976
1971
Meat and
fi
sh
Grams per person per week
300
200
100
0
Butter
Margarine
Lard
1997
1991
1986
1981
1976
1971
Fats
10
Behaviour, stress and health
Forty-five per cent of the UK population report engaging in some form
of leisure exercise at least once a month, with the figure rising to 64 per
cent if walking is included in these activities. A lesser number of indi-
viduals achieve the levels of exercise considered necessary to protect against
CHD: here, the figure is nearer to 25 per cent of the adult population
(Norman
et al
. 1998). These figures, however, represent a significant rise
in exercise participation: in 1985, only 20 per cent of British men and
2 per cent of women engaged in such levels of exercise. Those who engage
in exercise are more likely to be young, male, and members of higher
socio-economic groups. Participation in leisure exercise among profes-
sionals, for example, is virtually double that among unskilled manual
workers (80 versus 45 per cent), although the latter may engage in more
physically demanding work activity. Those who participated in some
form of sport in their youth are almost three times more likely to exercise
in adulthood.
Unsafe sex
Estimates of the prevalence of HIV in 1997 suggested that a total of
30.6 million people were infected with HIV worldwide. In the UK, the
primary route of infection has been through sex between men, account-
ing for 72 per cent of all AIDS cases reported by 1998. Eighteen per cent
of cases resulted from heterosexual sex. However, the incidence of new
cases among gay men is falling slowly while the incidence of HIV infec-
tion within the heterosexual community is rising. In addition, young
people are at increasing risk of infection: adolescent heterosexuals
account for about 20 per cent of all newly reported cases in the USA
(Stiff
et al
. 1990).
Nearly half of British adolescents aged between 16 and 17 report
having had at least one sexual partner during the previous year. However,
they are unlikely to plan intercourse and those using a condom are in
the minority. The findings of a large-scale British survey conducted by
Wellings
et al
. (1994) indicated that only about half of those whose first
sexual experience occurred between the ages of 16 and 24 years used a
condom at first intercourse: 31 per cent of men and 24 per cent of
women report using no form of contraception. Younger people were less
likely to take precautions: over 60 per cent of respondents who had their
first sexual experience at the age of 13 reported not using any form of
contraception.
Heterosexuals in the general adult population use condoms consist-
ently only about 10–15 per cent of the time with primary partners and
15 per cent of the time with secondary partners (Dolcini
et al
. 1995).
Even more concerning from a disease prevention perspective is the choice
made by those who are known to be HIV positive not to use a condom.
Sobo (1993), for example, was able to identify a group of HIV positive
women who did not use a condom with their lover or husband, but did
Psychosocial correlates of health
11
Table 1.1
The percentage of respondents engaging in one or more key
health-related behaviours
No. of health behaviours Percentage of sample
0 6.2
1 20.6
2 34.3
3 30.9
4 8.1
Source
: Norman
et al
. 1998
so with more casual sexual partners (see Chapter 3). The likelihood of
an increase in the rate of non-protected intercourse appears to be rising,
perhaps because of the wide prevalence of beliefs that new treatments
for HIV will be curative (Kalichman
et al
. 1998).
Is there a healthy lifestyle?
In an attempt to answer this question, Norman
et al
. (1998) categorized
over 13,000 people in a British survey as either engaging or not engaging
in up to four health-related behaviours: not smoking, moderate alcohol
use, exercising three or more times a week, and eating fruit or vegetables
regularly. While 93 per cent of the sample reported engaging in at least
one of these behaviours, only 8 per cent reported that they did them all
(Table 1.1).
Personality and disease
Type A behaviour and hostility
First identified by two cardiologists, Rosenman and Friedman, Type A
behaviour (TAB) was defined as an excess of competitiveness, time urgency
and easily aroused hostility. The absence of such characteristics is referred
to as Type B behaviour. Early case-control and longitudinal studies
showed significant associations between TAB and CHD. Such was the
research consensus achieved that, by the mid-1980s, TAB was con-
sidered to confer the same degree of risk for CHD as the more tradi-
tional risk factors of high blood pressure and cholesterol. Subsequent,
and frequently methodologically flawed, studies failed to find a relation-
ship between TAB and disease progression in men either at high risk of
CHD or who had already experienced a myocardial infarction (MI). As
a consequence, this degree of consensus is no longer evident. However,
one component of TAB, hostility, appears to convey the degree of risk
12
Behaviour, stress and health
previously ascribed to TAB and this has formed the primary focus of
more recent research.
A number of case-control studies have reported significant associations
between hostility and CHD, with the strongest relationship being found
among those under 50 years of age. Such a finding is not unique: the
predictive power of other biological and behavioural risk factors is also
strongest among younger people. However, the most convincing evidence
linking hostility to CHD has stemmed from longitudinal studies. In one
of the earliest such studies, Barefoot
et al
. (1983) followed 255 physicians
for 25 years and found that those who scored above the median on
measures of hostility taken while in training were nearly five times more
likely to experience an MI over this period than those who scored below
the median. These data have been supported by findings in older popu-
lations. European studies (for example Everson
et al
. 1997) following
middle-aged men for up to 25 years have found an association between
hostility and coronary events. Hostility may also contribute to the develop-
ment of CHD in women (Lahad
et al
. 1997). Data from some of these
and other studies combined into meta-analysis suggested a significant
association between hostility and CHD (Miller
et al
. 1996).
A second set of studies has measured the association between hostil-
ity and the degree of atheroma of the cardiac arteries, measured using
angiography. Although there have been a number of positive findings,
several studies have failed to find any relationship between hostility and
atheroma. One explanation for these disparate findings may be biased
sampling. Angiograms are invasive procedures usually conducted only on
patients who report some degree of symptomatology. Accordingly, the
samples used are likely to be non-representative, and the results of most
of these studies should be considered with caution. Two types of study
are the exception to this rule: those that track the progression of atheroma
over time, and those involving angiograms in representative populations.
In the first type, Julkunen
et al
. (1994) measured the relationship between
hostility and the progression of atherosclerosis over a two-year period in
a sample of 119 middle-aged men. The progression of atheroma was
almost twice as fast in highly hostile participants as in those who were
low in hostility. Utilizing the second methodology, Barefoot
et al
. (1994)
took advantage of the need for airline pilots to have routine angiograms
to ensure their fitness to fly. They compared the hostility scores of pilots
with angiographic evidence of CHD with those having none. Among
non-smokers only, pilots found to have evidence of CHD had higher
hostility scores than did those in the comparison group.
Type C behaviour
Type C behaviour has been independently described by Greer and Morris
(1975) and Temoshok (1987). It is defined as an aggregate of several
Psychosocial correlates of health
13
coping styles, in particular being stoic, cooperative, appeasing, unassertive
and inexpressive of negative emotions, particularly anger. It is thought
to be linked to the development of cancer.
A number of studies have shown an association between Type C per-
sonality and the incidence or progression of cancer. Using a case-control
design, Kune
et al
. (1991), for example, compared patients newly diag-
nosed with colorectal cancer and community controls matched for age
and gender. Patients with cancer were significantly more likely to report
histories of unhappiness in childhood and recent adult life and to have
strong feelings of discomfort when experiencing feelings of anger. These
differences may have been influenced by knowledge of a diagnosis of
cancer. Accordingly, more weight should be ascribed to the findings of
studies where a diagnosis was not known at the time of psychological
assessment. A number of such studies (for example Greer and Morris
1975) have found that women who suppressed their anger or had a
conforming personality were more likely to have malignant changes than
those without these characteristics. Longitudinal studies have also shown
Type C characteristics to predict cancer. Shaffer
et al
. (1987), for exam-
ple, followed 972 physicians for a 30-year period and found that partici-
pants characterized by high levels of ‘acting out’ and emotional expression
had a less than 1 per cent risk of developing cancer. Participants charac-
terized as ‘loners’ and thought to inhibit emotional expression were 16
times more likely to develop cancer than those in this group.
Concordant with the Type C hypothesis is work examining the impact
of ‘fighting spirit’ on cancer progression. This constellation of behavi-
ours is considered to be the opposite of Type C behaviour, and has been
associated with longer survival following diagnosis. In the first study to
identify this characteristic as a prognostic factor, Derogatis
et al
. (1979)
found that women who showed ‘fighting spirit’ during treatment for meta-
static breast cancer lived significantly longer than those who did not.
However, many of these women also received less chemotherapy, sug-
gesting that disease severity may have been worse in the poor survival
group or that their higher levels of fighting spirit could have been a
consequence of being subjected to lower levels of a physically and men-
tally debilitating treatment. Greer (1991) also found fighting spirit to be
an important determinant of survival. In a longitudinal study following
62 women with early non-metastatic breast cancer, he identified five
reaction types: denial, fighting spirit, stoic acceptance, helplessness and
hopelessness, and anxious preoccupation. Breast cancer recurrences and
mortality were recorded 5, 10 and 15 years later. By 15-year follow-up,
45 per cent of the women who were categorized in the fighting spirit
group were alive without recurrences; this compared with 17 per cent in
all the other groups. Similar evidence has been published in a variety of
cancers, including melanoma and lung cancer. However, there have also
been some negative findings. In a larger replication study of the original
Greer study, for example, Dean and Surtees (1989) found denial measured
14
Behaviour, stress and health
after surgery was associated with a
favourable
outcome in a group of 125
women with non-metastatic breast cancer followed for up to eight years.
Such contradictory findings have meant that the relationship between
Type C behaviour, ‘fighting spirit’, and cancer remains controversial and
far from proven. A related strand of research has focused on the health
outcomes following attempts to facilitate active coping with disease and
the expression of emotions. These studies are considered in Chapters 7
and 8.
Social and environmental influences on health
Socio-economic status and health
There is strong historical evidence that the more affluent members of
society have lived longer than the less well off. More recent evidence
of this health gradient can be found in a study of nineteenth century
obelisks in Glasgow graveyards. In an imaginative study, Davey Smith
et al
. (1992) measured the height of obelisks in Glasgow graveyards as
a proxy for the wealth of the individual buried beneath them. They
compared these with the ages of the first generation buried below and
found a strong linear relationship between the height of the obelisks and
the age of their first occupant, suggesting that the more wealthy lived
longer. What is important is not just that this relationship existed, but
that the families buried in the graves represented a small and wealthy
fraction of the Glasgow population. This is not evidence that the very
poor and immiserated did not live as long as the rich did: rather, that
the relatively rich did not live as long as the very rich. Such a gradient
still exists. It is progressive, and throughout the social classes. It holds
for women as well as men and is characteristic of all western countries
(Wilkinson, 1992).
A number of explanations have been proposed to account for these
differences. People in lower socio-economic groups may be exposed to
more environmental insults, low quality and damp accommodation,
and air pollution. An alternative explanation may be that less well off
individuals engage in more health-damaging behaviours, such as smok-
ing or excessive alcohol consumption. While both these explanations
may account for some of the differences in some of the studies, they
cannot explain them all. Marmot
et al
. (1984), for example, explored
the impact of a number of these variables on the health of British civil
servants working in London over a period of ten years. Their findings
indicated that while those in the more deprived social groups did engage
in more health-damaging behaviours, these did not fully explain the
health/SES relationship. When variations in smoking, obesity, plasma
cholesterol and blood pressure were statistically partialled out of the risk
Psychosocial correlates of health
15
equation, occupational-status-related differentials in health still remained.
Mortality was three times higher among men in the lowest grade than
those in the highest.
While people who occupy the lower socio-economic groups may engage
in more health-damaging behaviours, the adverse health effects of these
behaviours may be overwhelmed by factors associated with their economic
position. Hein
et al
. (1992), for example, reported data from a 17-year
prospective study of CHD in Danish men. Adjusting for a variety of
confounding factors, they found that men who smoked were three and a
half times more likely to develop CHD than non-smokers. However,
when these data were analysed according to SES, white-collar smokers
were six and a half times more likely to experience a cardiac event than
the equivalent non-smokers. Amongst blue-collar workers, smoking status
conferred no additional risk for CHD. Fewer middle-class people may
smoke, but those that do may be particularly vulnerable to its health-
damaging effects. Conversely, the impact of smoking on the health of the
less well off is seemingly overwhelmed by social factors.
Comparisons of life expectancy across different western countries sug-
gest some intriguing explanations of the relationship between social class
and health. Wilkinson (1992) provided powerful evidence that it is not
absolute wealth that determines health. He drew on evidence that showed
only a weak relationship between the absolute wealth of the society and
overall life expectancy. More predictive is the distribution of wealth
within a society. The narrower the distribution, whatever its absolute level,
the better the overall health of the nation. Accordingly, although Japan
and Cuba differ substantially on measures of economic wealth, both have
relatively equitable distributions of income and long life expectancies
throughout their populations. Of particular interest is evidence from
Scotland, which tracked average age of mortality and income distribution
over the life of the Thatcher government. As earning differentials rose,
so premature mortality among the less well off increased despite their
access to material goods, food, clothing and so on remaining relatively
constant. These data led Wilkinson to suggest that, for the majority of
people in western countries, health hinges on relatively more than absolute
living standards.
Wilkinson’s explanation of health differentials suggests that we engage
in some form of comparison of our living conditions with others in society,
and that knowledge of a relative deprivation in some way increases risk
of disease. Three different psychological processes may also be implicated
in the health gradient. Individuals in the lower socio-economic groups
report more stressors than those in higher groups and that these stressors
are frequently linked directly to their material conditions. In addition, the
less well off have less control over their environment and fewer personal
resources to moderate the impact of such stressors than the better off.
Finally, social support, a powerful mediator of health status, is less avail-
able to those in lower socio-economic groups (Adler
et al
. 1994).
16
Behaviour, stress and health
Social isolation and health
There is substantial evidence that both men and women who have a
small number of social contacts are more likely to die earlier than those
who have more extended networks. Data from the Alameda County
Study (Berkman and Syme 1979), for example, showed increased lon-
gevity to be associated with relatively high numbers of social ties as a
consequence of marriage, contacts with close friends and relatives, church
membership and membership of other organizations. The most isolated
were the most prone to premature death even after controlling for factors
such as smoking, alcohol use and levels of physical activity. In a later
study, Reynolds and Kaplan (1990) found that women who had few
social contacts and were socially isolated were at double the risk of
developing hormone-related cancers and evidenced an almost fivefold
increase in risk of dying from them than less isolated women.
Similar results have been reported in European samples. Orth-Gomer
and Johnsson (1987), for example, followed a cohort of 17,400 men and
women for a period of six years and found that both men and women
who had a restricted social network evidenced a 50 per cent greater risk
of CHD than those who were socially embedded. However, the relation-
ship between social contact and health was not always linear: those with
many social contacts did not always benefit in terms of health. Older
women, for example, who had many social contacts evidenced higher
mortality than those with medium-sized networks. To explain these
apparently anomalous findings, Orth-Gomer and Johnsson analysed their
data not simply according to the absolute number of contacts, but taking
into account the nature of the contacts. When they did this, they found
the strongest predictor of mortality was a lack of social integration,
which the authors considered to provide guidance, practical help and
a feeling of belonging. They found only a low association between the
provision of emotional support and CHD, although a later study by
the same group found this to be an important protective factor. In an
alternative interpretation of the Type A hypothesis, Orth-Gomer and
Unden (1990) suggested that TAB or hostility might confer risk of CHD
as a consequence of associated social isolation. In a longitudinal study,
following a cohort of men for ten years, they found no differences in
mortality between Type A and Type B men. However, over this period,
69 per cent of the socially isolated Type A men in the cohort had experi-
enced an MI, in contrast to the 17 per cent incidence among those who
were socially integrated.
Further evidence of the impact of social isolation can be found in
studies of populations already experiencing disease. Williams
et al
. (1992),
for example, found that patients with CHD who were unmarried and
without a confidant experienced a threefold higher risk of mortality over
a five-year follow-up period than those who were. Reflecting the subject-
ive nature of social support, some studies that have failed to find a