Journal of Public Health Medicine
492±499
The health of students in institutes of higher
education: an important and neglected public
health problem?
Sarah Stewart-Brown, Julie Evans, Jacoby Patterson, Sophie Petersen,
Helen Doll, John Balding and David Regis
Abstract
Background A survey of students in three UK higher
education establishments was undertaken to obtain informa-
tion about students' physical and emotional well-being, their
attitudes to, and beliefs about health, and the prevalence of
risk factors for future ill health.
Methods Health was measured by the prevalence of long-
standing illness and by the SF-36 health status measurement
tool. Survey results were compared with equivalent data for
18- to 34-year-olds in the local population. The prevalence of
long-standing illness was also compared with two national
surveys.
Results The survey achieved a 49 per cent response rate.
More than one-third of respondents reported a long-standing
illness, a higher prevalence than in all comparison surveys.
Students scored signi®cantly worse than their peers in the
local population on all eight SF-36 dimensions. The greatest
difference was for role limitations as a result of emotional
problems. The main sources of emotional distress were
study or work problems and money.
Conclusion The poor response rate in this survey dictates the
need for caution in interpretation of the results. However,
they suggest that the health of students is poor relative to
that of their peers, and that their emotional health is more of
a problem than their physical health. Public health practi-
tioners might want to pay more attention to the health of this
important and relatively neglected group. Worries about
studies and money appear to be affecting students' academic
work, and this should be of concern to higher education
establishments.
Keywords: student, health, survey
Introduction
In studies of health and health-related lifestyles, young people
are an under-researched group and there are few surveys of the
health of students at universities and other higher education
institutions. Most surveys achieve poor response rates from
younger age groups, and surveys of students present even more
of a challenge for a number of reasons: they often have more
than one address; their term-time addresses may be temporary;
they have many distractions, from academic pressure to social
Vol. 22, No. 4, pp.
Printed in Great Britain
and sporting activities. As a result, information about students'
health is scarce.
The health of students is nevertheless important. As potential
policy makers, professionals and senior managers of the future,
their health-related lifestyles, and their attitudes and beliefs
about health, are likely to have a disproportionate in¯uence on
the population's health. Many students leave home for the ®rst
time to study for a degree and create a lifestyle free from
parental in¯uence. Health-related habits formed during this
period may be dif®cult to change later in life. The system of
®nancing higher education has changed over the last 20 years
and ®nancial hardship may be having a detrimental impact on
students' health.
In 1995 a survey of students' health and lifestyles was under-
taken ± a collaboration between four district health authorities
and the universities of Oxford and Exeter. The aim was to obtain
information related to students' physical and emotional well-
being, their attitudes to, and beliefs about health, and the
prevalence of risk factors for future ill health. We surveyed
students from an established university (institution A), a new
university (B) and a college of higher education (C).
A self-completion questionnaire was designed, based on a
selection of well-validated questions, including some previously
Health Services Research Unit, Department of Public Health, University of
Oxford, Institute of Health Sciences, Old Road, Headington, Oxford OX3 7LF.
Sarah Stewart-Brown, Director
Julie Evans, Research Of®cer
Jacoby Patterson, Clinical Lecturer in Public Health Medicine.
Sophie Petersen,* Research Of®cer
Helen Doll, Statistician
Schools Health Education Unit, University of Exeter, Heavitree Road, Exeter
EX1 2LU.
John Balding, Director
David Regis, Research Manager
Address correspondence to Dr S. Stewart-Brown.
E-mail: sarah.stewart-brown@dphpc.ox.ac.uk
*Present address: British Heart Foundation Health Promotion Research Group,
Department of Public Health, University of Oxford, Institute of Health
Sciences, Old Road, Headington, Oxford OX3 7LF.
q Faculty of Public Health Medicine 2000
 HEAL TH OF STUDENTS
used in schools by the Schools Health Education Unit, together
with a number of experimental questions, some with open-
ended responses. The questionnaire included the Of®ce of
Population Censuses and Surveys (now Of®ce for National
Statistics) General Household Survey question on long-
standing illness,1 and the anglicized version of the Medical
Outcomes Trust Short Form 36 health status measurement tool
(SF-36).2 The SF-36 measures health in eight multi-item
dimensions: physical functioning; social functioning; role
limitations due to physical problems; role limitations due to
emotional problems; mental health; energy and vitality; pain;
and general perception of health. For each variable item scores
are coded, summed, and transformed onto a scale from zero
(worst possible health state) to 100 (best possible health state).
The SF-36 includes two very similar questions that ask about
the impact of physical and emotional health problems on
respondents' work or other regular daily activity over the past 4
weeks. The word `studies' was added to both questions for the
purposes of this survey, to increase the relevance to students.
The SF-36 is a well-validated instrument and population norms
are available for the area comprising the counties of Oxford-
shire, Berkshire, Buckinghamshire and Northamptonshire in
which our surveyed institutions are located.3 Students were
asked how often they worried about a series of potential
problems, including their studies, money, their health, the
environment and uncertainty about the future. They were asked
to indicate their response, on a ®ve-point scale ranging from
`never' to `most days'.
In addition, an extensive range of socio-economic and
health-related lifestyle questions were included, the results of
which are not reported here. The questionnaire was substantial,
taking around 40 min to complete.
Methods
Because of problems anticipated in surveying this group, the
approach was tailored to each institution using local knowledge
about how students obtain information and mail. To raise
awareness of the survey, assistance was sought from the student
health and registrar's of®ces and chairs of student groups.
Advertisements were placed on notice boards and articles
written for student newsletters.
The names and addresses of 1000 full-time students were
randomly selected from the register at each of institutions A and
C, and 500 from B. The different sample sizes were a result of
varying resources at the different institutions. The question-
naires were accompanied by a letter explaining the study's
objectives, guaranteeing anonymity and con®dentiality, and
offering a prize draw of £250 per institution. The letter also
provided contact details of the institution's counselling service
in case any student needed support as a result of any issues
raised by completing the questionnaire.
At institutions A and C, all the questionnaires were posted to
students' term-time addresses during the spring term 1996.
493
Reminder cards were sent to non-responding students 2 weeks
after the initial mailing, again to their term-time addresses. In
this mailing identi®cation numbers were omitted from ques-
tionnaires to encourage response. At institution B, question-
naires were posted initially to students' internal mail addresses
during the autumn term 1996. However, very few of the
questionnaires were picked up, so the remainder were retrieved
and mailed to students' term-time addresses. A reminder card
was sent 2 weeks after the initial mailing.
At institution C, a second reminder letter and questionnaire
were sent to students' term-time addresses 2 weeks after the
reminder card. At institution A, a second reminder letter and
another questionnaire were sent to student's permanent
addresses (usually their parental address) during the Easter
vacation. This included a pre-paid reply envelope addressed to
the research unit.
Students at institutions A and B were originally requested to
return their completed questionnaires to the Student Advice
Centre. Students at institution C were requested to return
completed questionnaires to the nearest campus reception, from
where they were passed to the registrar's of®ce. Students
completing questionnaires after receiving a ®nal reminder were
requested to return them direct to the research unit, in a further
attempt to encourage response by ensuring anonymity.
Data were coded and entered by staff of the School Health
Education Unit at the University of Exeter. Analysis of the data
was undertaken at the Health Services Research Unit using
SPSS for Windows versions 7.5 and 9.0. Statistical signi®cance
was de®ned as p < 0.01.
The age of respondents ranged from 18 to 64, but 90 per cent
were aged 33 or lower and the average age was 23 (median 21).
Results of the student survey were compared with equivalent
data for 18- to 34-year-olds living in the four counties of
Oxfordshire, Berkshire, Buckinghamshire and Northampton-
shire obtained from a series of local population health surveys
undertaken in 1986±1987, 1991±1992 and 1997,4±6 and from
published data derived from national surveys.1,7±12 SF-36
scores were compared with the results of the 1991±1992
survey because the question format on two of the dimensions of
the SF-36 was slightly improved in the 1997 survey. Questions
about worries were compared with the results of the 1997
survey because they were not included in previous surveys.
Long-standing illness prevalence was compared with the results
of all three local population surveys plus the equivalent age
group in national surveys conducted at several points in time:
the General Household Survey1,7 and the Health Survey for
England.8±12
To investigate the possibility that the results we obtained
were attributable to differences in the age, sex or social class
distribution of the student population, we undertook logistic
regression analyses on the categorical and binary dependent
variables (long-standing illness and causes of worries) and
analysis of variance on the continuously distributed dependent
variables (SF-36 scores) in the local surveys. In the combined
494 JOURNAL OF PUBLIC HEALTH MEDICINE
Table 1 Health status of students and their 18- to 34-year-old peers
18- to 34-year-olds in 1991±1992 popln
survey
..........................................................
SF-36 dimension Mean score
General health 75.95
Energy/vitality 61.62
Mental health 72.89
Pain 84.09
Physical function 92.50
Role mental 82.19
Role physical 89.33
Social function 88.33
datasets, which for the local population surveys covered the age
range 18±64 years, age, sex and social class were entered as
independent variables alongside student status.
Results
Response rates
Forty-nine per cent of the student sample responded to the
questionnaire (1208/2457). Men were less likely to respond
than women (36 per cent of respondents were male compared
with 45 per cent of the student population), younger students
were slightly less likely to respond than older ones (25 per cent
of respondents were aged under 20 compared with 27 per cent
of the student population), and postgraduate students were
slightly under-represented (11 per cent of respondents were
postgraduate students compared with 17 per cent of the student
body). In other ways, for example, in terms of the courses they
were undertaking, the respondents were representative of the
student body.
The response rate varied between institutions, with 54 per
cent responding from institution A, 49 per cent from C and 41
per cent from B. Comparing respondents from the three
institutions: A (traditional university) had the highest propor-
tion of men (42 per cent); C (higher education college) had the
highest proportion of women (71 per cent) and a social class
distribution closest to that of the general population in the four
counties (39 per cent in classes I and II compared with 33 per
cent of the population); B (new university) had the highest
proportion of students in social classes I and II (51 per cent),
and also the highest proportion of ethnic minority groups (17
per cent). One-®fth of respondents in all three institutions
were aged over 25. These results mirror socio-demographic
differences in the student bodies of the three institutions.
In all analyses, we compared response between institutions.
If there was an important bias attributable to varying response
rates (for example, if those who were in poor health were more
likely to respond), we would have expected to ®nd a difference
between sickness rates at institution A, which had the highest
Students
...........................................................
95% CI Mean score 95% CI
(75.33±76.56) 66.01 (64.89±67.12)
(61.00±62.24) 53.00 (51.90±54.11)
(72.34±73.44) 65.57 (64.60±66.55)
(83.44±84.75) 79.57 (78.34±80.80)
(92.00±92.99) 87.86 (86.95±88.77)
(81.14±83.24) 58.00 (55.64±60.36)
(88.48±90.19) 76.78 (74.82±78.74)
(87.74±88.93) 80.75 (79.57±81.93)
response rate, and at B, where fewest students responded.
However, this was not the case. We also compared the
responses of those students who responded promptly and
those responding after receipt of the second reminder letter. We
found very few signi®cant differences and none that suggested
an important response bias. Estimated response rates for the
18±24 age group in the three local population surveys were 80
per cent (1986±1987); 64 per cent (1991±1992) and 42 per cent
(1997); and 70, 69 and 56 per cent, respectively, for the 25±34
age group.
Health status (SF-36)
Students scored signi®cantly lower on all eight dimensions of
the SF-36 than the equivalent age group in the 1991±1992 local
population survey (p < 0.01) (Table 1, Figure 1). Despite the
varying response rates between institutions, SF-36 scores were
remarkably similar in all three. The only signi®cant differences
between institutions were attributable to sex. Amongst both the
student and the population samples, scores for women were
between three and seven points lower than those for men on
seven of the eight dimensions (p < 0.01); sex differences in
general health perception scores were smaller and did not
achieve statistical signi®cance at the 1 per cent level.
The greatest difference between the student sample and the
general population was for role limitation due to emotional
problems, with students scoring more than 20 points lower than
the general population. This score is derived from three items,
with a binary response (yes or no), that elicit the extent to which
emotional problems interfered with studies, work or other
regular daily activities. Thirty-eight per cent of students said
they had `cut down on the amount of time spent on studies,
work or other regular activities' because of emotional health
problems, 49 per cent had `accomplished less than they would
like', and 39 per cent `didn't do studies, work or other activities
as carefully as possible' (Table 2). Equivalent ®gures for 18- to
34-year-olds in the general population were 13 per cent, 21 per
cent and 19 per cent, respectively. Physical health problems had
a smaller impact on daily activities as measured by four items.
 HEAL TH OF STUDENTS
Figure 1 Health status of students and their 18- to 34-year-old peers. Black columns, students; white columns, 1991±1992
population.
One-quarter of students (25 per cent) had `cut down on the
amount of time spent on studies, work or other regular
activities' because of physical problems, 30 per cent had
`accomplished less than they would like', 13 per cent were
`limited in the kind of studies, work or other activities' they
could do, and 25 per cent had `dif®culty performing their
studies, work or other activities' (Table 3). Equivalent ®gures
for the local population were 9, 13, 9 and 12 per cent,
respectively.
Worries
The most common causes of worry were `study or work
problems' and `money problems'. Sixty-two per cent of
students said they worried about study or work problems
`often' or `most days'. The corresponding ®gure for money
worries was 52 per cent (Table 4). Money and pressure at work
were also the two most common sources of worry amongst 18-
to 34-year-olds in the 1997 local population survey, but the
frequency of these worries was considerably lower; only 36 per
Table 2 Role limitations due to emotional problems
Role limitation
Cut down on the amount of time you spent on work or other activities
Accomplished less than you would like
Didn't do work or other activities as carefully as possible
495
cent of 18- to 34-year-olds worried about money, and 25 per
cent about pressure at work `a lot' or `all of the time'.
Long-standing illness
Four hundred and three students (33.5 per cent) reported long-
standing illness. Female students were more likely to report
illness than men (37 per cent compared with 28 per cent,
p < 0.01). There was no signi®cant difference in the prevalence
of long-standing illness between the three institutions. The most
common illnesses were asthma and musculo-skeletal problems,
each reported by more than 8 per cent of students.
The prevalence of long-standing illness was higher amongst
the students in our survey than among the equivalent age group
in all the comparison studies (Figure 2). In comparing the
results from these different studies it is important to note the
differences in technique used, which could affect response to
this question (Table 5). Both the General Household Survey
and the Health Survey for England were conducted by
interview using a household sampling frame, whereas the
18- to 34-year-olds in Students
1991±1992 popln .......................................
survey (%) (%) p value
13.0 37.9 < 0.001
21.4 48.9 < 0.001
19.1 39.2 < 0.001
496 JOURNAL OF PUBLIC HEALTH MEDICINE

Table 3 Role limitations due to physical problems

18- to 34-year-olds in Students

1991±1992 popln .......................................

Role limitation survey (%) (%) p value

Cut down on the amount of time you spent on work or other activities 8.9 24.7 < 0.001
Accomplished less than you would like 13.1 30.2 < 0.001
Were limited in the kind of work or other activities 9.0 13.3 < 0.001
Had dif®culty in performing the work or other activities 11.9 24.7 < 0.001

local population surveys were administered by post using
general practice registers as the sampling frame. They show a
gradual upward trend over time. The steep rise in reported long-
standing illness over time in the three local surveys is likely to
be attributable in part to changes in the format of the long-
standing illness question. The wording was adapted between
surveys to more clearly de®ne what constituted a long-standing
illness. In the 1997 survey, a checklist of conditions was
provided instead of inviting a free-text response. The checklist
had been developed from the illnesses reported in the earlier
two surveys. The format of the question used in the student
survey was similar to that used in the two national interview
surveys, which indicate a prevalence of between 13 and 26 per
cent, and to that in the 1991±1992 local survey, which reported
a prevalence of 18 per cent in this age group. The prevalence of
long-standing illness in the student survey was signi®cantly
greater than that in the 1986±1987 and 1991±1992 local
surveys (p < 0.01), but not from the 1997 local population
surveys.
Adjustment for differences in age, sex and social class
SF-36 scores in the two populations showed minor variations by
age, sex and social class. Analysis of variance, however,
demonstrated that the differences in scores between the students
and the local population were not attributable to differences in
age, sex or social class. After adjustment, the differences
changed by less that one point for all dimensions except mental
health, where the difference in scores increased by one point, and
general health, where the difference decreased by one point.
Similarly, the proportion of respondents in the three surveys
(student, 1991±1992 population and 1997 population) who
reported worries about money or work, or who reported that
their emotional and physical health affected their work, varied
somewhat according to age, sex and social class. Logistic
regression analyses with these outcomes as dependent vari-
ables, however, also showed that the differences between
students and the local population were not attributable to
differences in these independent variables.
Discussion
The results of surveys based on a response rate below 70 per
cent can be misleading. Those who respond to surveys tend to
differ from those who do not in ways that can be dif®cult to
predict. Although the students responding to this survey were
similar in terms of age, sex and type of course to the student
body in general they may have differed in other ways. In
particular, students who were prone to feel ill may have been
more likely to respond to this survey than those who did not.
However, the fact that so few signi®cant differences were found
between the three institutions with different response rates, and
the fact that response rates were also low in local comparison
surveys suggests that differences between students and the
general population are likely to be real.
Previously published surveys of student health have achieved
response rates ranging between 18 and 100 per cent. However,
those that achieved higher response rates than ours used different
techniques. For example, personally administering questionnaires

Table 4 Frequency of worry about money and work

Worry about money Worry about work

................................................................................... ..................................................................................

18- to 34-year-olds 18- to 34-year-olds

1997 popln survey (%) Students (%) 1997 popln survey (%) Students (%)

Never 7.3 3.7 19.4 0.9

Rarely/a little 25.5 13.3 25.5 8.5
Some 31.4 30.7 30.1 29.2
Often/a lot 24.8 33.7 18.5 38.0
Most/all of the time 11.0 18.7 6.6 23.5
p value < 0.001 < 0.001
 HEAL TH OF STUDENTS 497

Figure 2 Long-standing illness, disability or in®rmity. V, General Household Survey, all persons aged 16±44 (n < 9000)
(interview). Source: Refs 1 and 7. W, Health Survey for England, all persons aged 16±34 (n < 5000) (interview). Source: Refs
8±12. X, Local population survey, all persons aged 18±34 (n < 4187, 3685 and 2982 in 1986±1987, 1991±1992 and 1997,
respectively) (postal). O, Student health and lifestyle survey, all persons (n ˆ 1208) (postal).

Table 5 Methods of surveys using the long-standing illness question

Survey Method Question Response format

GHS & HS England Interview Do you have any long-standing illness, disability or in®rmity
(anything that has troubled you over a period of time or is likely
to affect you over a period of time)?
If yes, what is the matter with you? Open
Local popln 1986±1987 Postal Do you have any long-standing illness or disability?
If yes please give brief details Open
Local popln 1991±1992 Postal Do you have any long-standing illness, disability or in®rmity
(that has troubled you over time or is likely to affect you over a
period of time)?
If yes, please specify Open
Local popln 1997 Postal Do you have any long-standing illness, disability or in®rmity?
Anything that has affected your work or other regular daily
activity over a period of time or is likely to affect you over a
period in the future.
Please circle the number that best describes your illness(es) Checklist
Student Postal Do you have any long-standing illness, disability or in®rmity?
By long-standing we mean anything that has troubled you over
a period of time or is likely to affect you over a period of time
in the future.
If yes, please describe Open
498 JOURNAL OF PUBLIC HEALTH MEDICINE
to students during classes resulted in an almost 100 per cent
response rate from those who attended the class,13,14 but when
the number of students not attending the class was taken into
account the response rate was much lower.15 A US survey of the
health and lifestyles of medical students demonstrated how
different approaches can affect response, with rates ranging
from 93 per cent of those whose questionnaires were
administered in class to 48, 30 and 18 per cent in three
groups whose questionnaires were administered via internal
mail.16
Other postal surveys of student health that obtained response
rates ranging between 65 and 76 per cent focused on a single
aspect of health behaviour, such as alcohol or substance abuse.
In those surveys the questionnaire was relatively quick to
complete, and did not contain the personal questions that our
survey did, for example, about sexual health and suicidal
feelings.17±21
The response rate to this survey is suf®ciently low to
question whether publication of results will add knowledge of
value to public health. However, response rates to local
population surveys have been falling recently and in our 1997
survey were similar for this age group to those achieved in the
student population. The results are presented because, in spite
of the low response rate, this is the most comprehensive survey
of health and health-related lifestyles of British students to have
been published to date. In the absence of other studies of student
health our ®ndings are important, if only to draw attention to the
likelihood that this group has poor health, and to encourage
more robust studies. In setting up such studies the dif®culties
that we report, and that are likely to be encountered in future
studies, need to be taken into account and allowed for in the
study method.
The results we present suggest that, contrary to what might
be expected of this high achieving group of young people, the
health of students is poor relative to that of their peers. The most
common causes of physical ill health were asthma and
musculo-skeletal problems, but students' emotional health
was more of a problem than their physical health. The fact
that between one- and two-thirds (allowing for the low response
rate) of students reported that anxiety about their studies was
limiting their capacity to work must be of some concern to
higher education establishments, as must the fact that ®nancial
concerns were affecting the academic work of between 25 and
50 per cent of students. Public health practitioners might want
to tackle the health problems amongst students by supporting
the concept of health-promoting universities, both at national
level and in local health improvement plans.22
Our results are consistent with those of other recent UK
surveys,13,14 which, using the Hospital Anxiety and Depression
scale, have shown high rates of anxiety amongst students. Both
the implications and the possible cause of these results are
worthy of consideration. Anxiety is part of normal experience
and the occasional episode is unlikely to have a health impact.
However, both observational and experimental animal studies
reviewed by Wilkinson23 and Brunner24 provide clear evidence
that prolonged or repeated stress has a detrimental impact on
the immune and cardiovascular systems. The impact of stress
on human health has been demonstrated in a range of different
observational studies also reviewed by Wilkinson.23 High
levels of anxiety amongst students are therefore potentially the
cause of their high levels of long-standing illness. Perhaps more
importantly, if high levels of emotional distress come to be
regarded as normal or unavoidable during early adulthood,
students will not aim to seek out low-stress occupations in
future life. Their experience as students may therefore set them
up for a lifetime of stress with important implications for their
long-term physical health. This group is one that includes the
policy makers, managers, doctors and teachers of future
generations, and their belief that high levels of stress and
anxiety are normal and unavoidable is likely, through social
policy, workplace practices, medical opinion and approaches to
teaching, to have a disproportionate impact on the health and
well-being of the public.
Although the illness prevalence rates in our survey need
validating in larger, well-funded studies, there are good reasons
why public health practitioners might want to take these
®ndings seriously and to consider the causes of these high levels
of anxiety. Students clearly attribute much of their anxiety to
their studies or to ®nancial concerns. This is perhaps not
surprising given that studies are their main occupation and that
their disposable income is likely to be lower than that of their
peers in employment. It is, however, also possible that an
ever-present pressure to compete and succeed is in itself
detrimental to human health. Such a possibility has far-reaching
implications for educational and social policy.
Acknowledgements
The student health and lifestyle survey was conducted by the
Oxford University Health Services Research Unit in conjunc-
tion with colleagues at the Schools Health Education Unit at
Exeter. It was supported ®nancially by the National Adolescent
and Student Health Unit and the Directors of Public Health in
the four counties of Oxfordshire, Berkshire, Buckinghamshire
and Northamptonshire. We are grateful to Eileen Goddard,
Anne Simms, Phil Robson and Anna Hinton for reviewing
drafts of the student questionnaire and for their advice and
support in the development of the project. We are also indebted
to the staff and students of the three institutions for agreeing to
participate in the study, and to Pamela Wyville-Staples for
preliminary data analysis.
References
1 Of®ce for Population Censuses and Surveys. General household survey
1993. London: HMSO, 1995.
 HEAL TH OF STUDENTS
2 Jenkinson C, Layte R, Wright L, Coulter A. The U.K. SF-36: an analysis
and interpretation manual. Oxford: Health Services Research Unit, 1996.
3 Jenkinson C, Coulter A, Wright L. Short form 36 (SF 36) health survey
questionnaire: normative data for adults of working age. Br Med J 1993;
306: 1437±1440.
4 Jamieson M, Grif®ths J. Health for all in the Oxford region. Oxford:
Oxford Regional Health Authority, 1987.
5 Wright L, Harwood D, Coulter A. Health and lifestyles in the Oxford
region. Oxford: Health Services Research Unit, 1992.
6 Petersen S, Stewart-Brown S, Peto V. Health and lifestyles in four
counties. Oxford: Health Services Research Unit, 1998.
7 Of®ce for National Statistics. Living in Britain: results from the 1995
general household survey. London: The Stationery Of®ce, 1997.
8 Of®ce for Population Censuses and Surveys. Health survey for England
1991. London: HMSO, 1993.
9 Of®ce for Population Censuses and Surveys. Health survey for England
1993. London: HMSO, 1995.
10 Joint Health Surveys Unit. Health survey for England 1994. London:
HMSO, 1996.
11 Joint Health Surveys Unit. Health survey for England 1995. London:
The Stationery Of®ce, 1997.
12 Joint Health Surveys Unit. Health survey for England 1996. London:
The Stationery Of®ce, 1998.
13 Webb E, Ashton CH, Kelly P, Kamali F. Alcohol and drug use in UK
university students. Lancet 1996; 348: 922±925.
14 Ashton CH, Kamali F. Personality, lifestyles, alcohol and drug
consumption in a sample of British medical students. Med Educ 1995;
29: 187±192.
499
15 Richmond RL, Kehoe L. Smoking behaviour and attitudes among
Australian medical students. Med Educ 1997; 31: 169±176.
16 Delnevo CD, Abatemarco DJ, Gotsch AR. Health behaviours and health
promotion/disease prevention perceptions of medical students. Am J
Prev Med 1996; 12(1): 38±43.
17 Delk EW, Meilman PW. Alcohol use among college students in
Scotland compared with norms from the United States. J Am Coll Hlth
1996; 44(6): 274±281.
18 Anderson P. Alcohol consumption of undergraduates at Oxford
University. Alcohol Alcoholism 1984; 19(1): 77±84.
19 West R, Drummond C, Eames K. Alcohol consumption, problem
drinking and anti-social behaviour in a sample of college students. Br J
Addiction 1990; 85: 479±486.
20 Whitty C, Jones R. A comparison of prospective and retrospective diary
methods of assessing alcohol use among university undergraduates.
J Publ Hlth Med 1992; 14(3): 264±270.
21 Ghodse AH, Howse K. Substance use of medical students: a nationwide
survey. Hlth Trends 1994; 26(3): 85±88.
22 Tsouros AD, Dowding JT, Dooris M. Health promoting universities:
concept, experience and framework for action. Copenhagen: WHO
(EURO), 1998.
23 Wilkinson RG. Unhealthy societies: the af¯ictions of inequality.
London: Routledge, 1996.
24 Brunner E. Stress and the biology of inequality. Br Med J 1997; 314:
1472±1475.
Accepted on 4 May 2000