-
Addiction (1993) 88, 655-663
RESEARCH REPORT
Doctors and substance misuse: types ofdoctors, types of
problems
DEBORAH BROOKE, GRIFFITH EDWARDS & TOBY ANDREWSAddiction
Research Unit, Institute of Psychiatry, National Addiction Centre,
4 Windsor Walk,London SE5 8AF, UK
AbstractThe casenotes of 144 doctors who had received
treatmentfor substance misuse were analysed. There were
nodifferences between general practitioners (n = 61) and hospital
doctors (n - 58) in terms of their substancemisuse histories or the
problems they incurred. Differences emerged between the consultant
(n = 24) and thenon-consultant (n = 34) grades of hospital doctor.
The consultants were older at onset of problematic use(42.6 8.6 vs.
29.9 9.8 years); they suffered fewer career problems and misused
fewer substances. The mostfrequent pathways into substance use were
personality difficulties (76 subjects, 52.8%) and anxiety
ordepression (46 subjects, 31.9%). A history of depression (n = 36)
was associated with perceived stress at work(p = 0.014), and at
home (p - 0.06). Past neurotic disturbances (n = 20) were
associated with personalitydifficulties (p - 0.035), anxiety or
depression (p = 0.004), and with an earlier onset of problematic
substanceuse (30.2 8.3 vs. 36.5 9.8 years, p 0.014). Principal
components of possible antecedents yielded onemajor component on
which all elements loaded; this was labelled the 'disturbance
score'. This score showeda reduction with increasing age of onset
of problematic substance use.
IntroductionIn a previous report,' we described the
character-istics of 144 doctors with drug or alcoholproblems who
were seen at the Bethlem andMaudsley hospitals between 1969 and
1988.The study was based on retrospective analysis ofcasenotes.
These problems affected everyspeciality and all degrees of
seniority. The meanage of presentation was 43 years; subjects
hadexperienced problems with their substance mis-use on average for
over 6 years. Alcohol was thecurrent problem for 42% and drug
misuse for26%; 31% were misusing both alcohol and drugs
Correspondence to; Deborah Brooke, Depanment ofPsychiatry, Epsom
General Hospital, Dorking Road, Epsom,Surrey KT18 7EG, UK.
at presentation. In the present study we take theanalysis
further, looking at associations betweenvariables.
MethodThe sample comprised all doctors attending theMaudsley or
Bethlem hospitals between 1969and 1988 who had received an ICD
diagnosis ofalcohol dependence or drug dependence, orboth. Data
were abstracted by D.B. and G.E.,using a structured schedule. The
data dealt within the earlier paper comprised such variables asage,
marital status, ethnic origin, place ofqualification, speciality
and employment dura-tion, plus clinical details of drugs misused
and
655
-
656 Deborah Brooke et al.
their sources, and details of routes into treat-ment. In this
paper, we present the findings forcertain additional areas:
Characteristics of the subject at key contact. Relat-ing to
medical career and type of employment,and details of previous
medical and psychiatricillness. (Within the UK, National Health
Servicedoctors work in primary care as 'family doctors',i.e.
general practitioners (36%) and general prac-titioner trainees
(2%); as hospital doctorseither fully trained consultants (21%) or
as non-consultants in training grades (33%) and inPublic and
Community Health (5%). Percent-ages are based on British Medical
Associationfigures for 1990; n = 88 191. Small categorieshave been
excluded for clarity.)
Substance use characteristics at key contact andsubstance use
history. Here such variables werecovered as age when drug and/or
alcohol use firstbecame problematic and, where relevant, age
atwhich drugs were first injected, and movementfrom alcohol to
drugs or vice versa. Any treat-ments received and any self-help
groupsattended were noted.
Pathways into substance misuse. This section ofthe schedule
required the abstractor to make ajudgement as to the likely major
aetiologicalfactors relating to the subject's misuse of alcoholor
drugs. Given the nature of the case materialfirom which the
information was being taken, itappeared better to code in terms of
certainbroadly defined 'pathways', rather than employan extended
check list of individual items wherethere might be missing
information. One ormore of the following altemative pathways
couldbe coded: personality problems; non-specificdrift into
drinking; anxiety or depression; pain,injury or accident; stress at
work; family stress;bereavement. To score positively on
'personalitydifficulties', evidence of long-standing relation-ship
or occupational difficulties from earliestadulthood was needed, in
the absence of con-comitant psychiatric problems and distinct
fromconsequences of substance misuse.
Measures of impairment. Three scores werederived for each
subject. These were (i) a scorefor number of substances ever
misused, with onepoint given for misuse of a compound in each ofsix
classes (alcohol; opiates; benzodiazepines and
barbiturates; cannabis, hallucinogens and sol-vents; stimulants;
antidepressants and others),(ii) Score for substance-related
problems in-curred at any time in each of seven defined
areas(family relationships and financial difficulties;patient care;
forensic involvement; drink-drivingcharges scored separately firom
other forensiccontact; impaired personal health; suicide at-tempt;
General Medical Council involvement),(iii) Score for detrimental
effects of substancemisuse on career progression, with three
gradesof severity: 'impaired', (for example, longperiods of
unemployment or sickness absence);'chequered', (forced periods
abroad or workingoutside own speciality); and 'blocked', (failure
toprogress in preferred speciality).
ResultsThe frequency distribution for age of problemonset
against type of substance misused is shownin Fig. 1. Alcohol
problems developed in thisgroup across all age bands, but drug
problemsdeclined in incidence with increasing age.
Exploratory correlation analysisInitially, we carried out an
exploratory correla-tion analysis putting sex, ethnicity,
generalpractice vs. hospital employment, and consultantvs. other
hospital grades, against substance usecharacteristics, measures of
impairment, andtreatments received. No significant differenceswere
found between sex (124 men, 20 women),or between ethnic group
(Caucasian 130, other14). Fifteen of 61 general practitioners as
op-posed to 5 of 58 hospital doctors had attendedself-help groups
(p = 0.02), but otherwise gen-eral practitioners and hospital staff
showed nodifferences. When, however, a comparison wasmade between
doctors of consultant and non-consultant grades, a number of
findings emerged(Table 1).
Non-consultants had an earlier age of problemonset than
consultants (non-consultants, n = 34,mean age of problem onset =
29.9 9.8 years,consultants, n = 24, mean age of problem on-set =
42.6 8.6, p = 0.000). With respect tochoice of substance, alcohol
was the dominantsubstance of misuse among the consultants,while
drugs and alcohol were equally representedamong the more junior
doctors (p = 0.008).Junior staff had a higher mean drug score,
that
-
Doctors and substance misuse 657
00
^ S
13 o
o - o+1 -H 9^O VO C>" : 9 II
+1+1in t-r *
- - CM O+) +1 9
enCO00
iri
CMrrt
a
"e
,*(oodII
VO 00 i00 o; jj+1 +1 qvO Ov oCM Ov II
CM 3
" e
2 o3" ouM Ic co o
-
658 Deborah Brooke et al.
16-14-12-
S2 10-1ooi 8-
o 6-
2-
0 I I 0-23 24-29 30-35 36-41
Age of problem onset42-47 48+
Figure 1. Age of problem onset by type of substance first
misused (\3 % alcohol, 0 % drug(s); n = 133).
is, they had misused more substances (p = 0.04).They also showed
a higher mean career handicapscore (p = 0,004). On the other hand,
the con-sultants scored more highly on the problem score(p =
0.002).
The importance of past psychiatric problems. Thenumbers of
subjects in the diagnostic categoriesof schizophrenia (n - 2),
hypomania (n = 8) andbrain damage (w 3) were too small for
statisti-cal analysis and these subjects were omitted.Data for
other psychiatric problems are pre-sented in Table 2, The analysis
for subjects witha past history of depression shows that they hada
tendency to present with drug misuse ratherthan alcohol misuse,
although this finding doesnot reach conventional levels of
significance(p = 0,052). There were 20 subjects with a pasthistory
of varieties of neurotic disturbance (ex-cluding depression). They
developed problemswith substance misuse at a younger age(30.2 8.3
vs. 36.5 9.8 years, p - 0.014). Boththe group with past depression
and the groupwith other neurotic disturbances hadsignificantly
smaller problem scores (p = 0.05 ineach instance).
Continuity and change in choice of substance. Aseparate analysis
(see Table 3) was carried outon the frequency with which subjects
shiftedbetween a drug or alcohol problem.
Overalli there was continuity, but with sometendency to
addition. Eighty-two subjects(56.9%) presented with problems
related to their
originally problematic substance, while 43(29.9%) had added a
drug to a pre-existingalcohol problem, or vice versa. This tendency
toaddition was more marked within the group whostarted by misusing
drugs. Some 45% of thesesubjects added an alcohol problem, while
only25% of those whose initial difficulties were withalcohol went
on to develop an additional drugproblem. Eight subjects (5.6%)
substituted onecategory for the other.
Pathways leading into substance misuse. Themost frequently
recorded pathways were person-ality difficulties (76 subjects,
52.8%) and anxietyor depression (46, 31.9%). Non-specific driftand
family stress were each coded for 38 subjects(26.4%), and stress at
work was coded for 33(22.9%). Pain and bereavement occurred in
14(9.7%) and 13 (9.0%) respectively. No pathwaywas cited for seven
subjects (4.9%). More thanone pathway was cited for 72 subjects
(50.0%).
As regards cross-tabulation of pathways withsubject
characteristics, women (n = 20) weremore likely than men (n = 124)
to be coded for'family stress' (45% vs. 23%, p = 0.04).
Othersignificant findings emerged in relation to previ-ous
experience of depression or other psychiatricillness, and here the
relevant data are given inTable 4.
Those with a past history of depression(n = 36) were more likely
than those withoutsuch a diagnosis (n = 93) to be coded for
theanxiety and depression pathway (81% vs. 12%,p< 0.0001), for a
pathway via bereavement
-
Doctors and substance misuse 659
I
on u
Ii
- - s -^ ' -^ +1 +1 ^ +1 +1 ^
O\ O rt Ji^ t"^ 00 GO
"Id "od^ o^ 3 ^ d^ o^
-
660 Deborah Brooke et al.
Table 3 . Continuities in substances used between problem onset
and index contact, numbers withpercentages of whole sample given in
brackets (n = 144)
First problem Stayed same Added drugs Added alcohol
SubstitutedAlcoholDrugsTotals
49 (34.0)33 (22.9)82 (56.9)
16 (11,1)27 (18.8)
43 (29.9)Information was incomplete on 11 subjects (7,6),
4 (2,8)4 (2,8)8 (5.6)
(19% VS. 3%, p - 0.005), for stress at work (39%vs. 18%, p =
0.014), or, at a marginal level ofsignificance, for stress in the
family (36% vs.20%, p = 0.06). Conversely, they were less
oftendiagnosed as having had a drift pathway (8% vs.35%, p =
0.002).
The subjects with a diagnosis of psychiatricillness
(predominantly of a neurotic nature)other than depression (n = 20;
n for those with-out this past diagnosis = 121), were also
morelikely to have an anxiety and depression pathway(60% vs. 27%, p
= 0.004) and this group con-tained more subjects coded for
personalitydifficulties (75% vs. 50%, p = 0.04). In general,these
findings appear to suggest that subjectswith a demonstrated
psychiatric vulnerabilitymay be prone to finding pathways into
substancemisuse which are characterised by mood dis-turbance or
failure to find ways of coping withvarieties of stress.
Multivariate analysisPrincipal components analysis. Here we
sought
to examine the postulate that two distinct typesof doctors with
substance problems might beidentifiable. The two types were
hypothesized interms of (i) younger doctors with greater evi-dence
of psychiatric disturbance other thandepression, and with coding
for personality oranxiety and depression pathways; (ii) older
doc-tors, with contrasting characteristics on thesedimensions.
Rather than using multiple vari-ables, we used a multivariate
technique to detectunderlying trends: a principal components
analy-sis of two pathways (anxiety/depression andpersonality
difficulties) and a past history of neu-rotic disturbance
(excluding depression) yieldedone major component on which all
elementsloaded, and which accounted for 48% of thevariance, rather
than there being two distinctcomponents. This component is a new
variable,composed of the two pathways and the past
history, as above. We interpreted this variable asa 'disturbance
score'. We concluded that thehypothesis that two distinct types of
troubleddoctors could be identified was, in these
terms,disconfirmed.
Analysis of variance. The disturbance scoreswere then entered
into a univariate analysis ofvariance as dependent variables.
Additional de-pendent variables were personality difficulties,
apast history of neurotic disturbance and thehandicap score. The
explanatory variables weresex, age of problem onset and type of
problem(alcohol or drugs). The following conclusionscan be
drawn:
(i) The personality difficulties pathway exertsa main effect
upon age of problem onset: thiscoding was significantly more
frequent atyounger ages (df = 5, F-3.8,p = 0.003).
(ii) There is an interaction between sex andage of problem onset
for the disturbance score(df = 4, F = 2.85, p = 0.027). Thus
disturbancescores showed a reduction with increasing age ofproblem
onset, except for an increase amongmen in the 42-48 year group. The
maximumdisturbance score among men was found inthose who presented
at age 24 or younger, andthe maximum disturbance score among
womenwas found in those who presented between theages of 24-30.
(iii) An interaction effect was demonstratedbetween age of
problem onset, type of substanceproblem and handicap score (df=5, F
= 2.56,p = 0.03). Those who present with a problemover the age of
48 have a greatly increased careerhandicap score. This is due to
the small sub-group in this age bracket with drug problems(w = 4),
rather than those with alcohol problems.
DiscussionWe wish to consider three issues. Firstly,
-
Doctors and substance misuse 661
Si
S
I
inOo
L C30 o in m o
>,? ^ o NO ^ ino^ tf^ O o^ o^ O00 in ~; o o j ;tn o m o
II
8-18 u oS oOZ
-
662 Deborah Brooke et al.
whether sub-groups of doctors with substanceproblems can be
identified; secondly, what mightbe the nature of factors that
predispose to sub-stance misuse among doctors; and thirdly,
thepolicy implications of our findings. There aremethodological
limitations to our study. Thesubjects were limited to the sample of
substance-misusing doctors who presented to apost-graduate teaching
hospital over a 20-yearperiod. The sample is large in UK terms but
thenumber is still small. There is no comparisongroup drawn from
other occupations, or fromdoctors who do not misuse substances. The
datawas collected retrospectively. Nonetheless, wehave explored
some hypotheses, and this is thefirst time that multivariate
techniques have beenused in this field. The results point to a
series ofsignificant questions, which deserve further
con-sideration.
The identification of sub-groupsAmong alcoholics in the general
population,several authors have described a group with ayounger age
of onset and a family history ofalcoholism, a more severe course
and associatedpsychiatric problems or personality difficulties.^
'^A sub-group may be identified within the hospi-tal doctors,
consisting of those below consultantgrade. This group is
characterised by a youngerage at onset of substance problems, large
num-bers of substances misused, a greater proportionmisusing drugs
and experiencing markedly dis-advantageous effects on their
careers. Those atconsultant grade have a later onset of
problem-atic use, showing that they are not the cohort ofyoung,
troubled doctors marched forward; theyaccumulate more
substance-related problems,but their careers suffer less, except
for those whodevelop a drug problem in later life. The highcareer
handicap scores suffered by such individu-als in this study were
due to taking earlyretirement, and it may have been that this
coursewas infiuenced by the perceived illegality of illicitdrug
use, as much as the health needs of thepractitioner. We
hypothesized that these twoclusters might emerge on principal
componentanalysis. However, principal component analysisof our data
did not show two separate dimen-sions of older and younger
subjects. It yielded aderived score of 'disturbance' for each
subject.This derived score decreased with age, except foran
increase in men in their forties. This finding
suggests a continuum of vulnerability to sub-stance misuse
problems across all age groups; thecontribution made by personality
difficulties inyounger substance misusers may inflate the
dis-turbance score in the youngest age groups, butnot sufficiently
to demonstrate a specific type.Doctors in all age groups continue
to sufferdisturbance, and this data suggests that emo-tional
problems are of major importance.
Pathways and past history: the nature of vulner-abilityThe
genesis of an individual's substance misuseproblem cannot be
reduced to a single factor andour concept of 'pathways' serves to
draw atten-tion to premorbid individual susceptibilities. So,for
example, over half of the subjects were codedpositively for
personality difficulties. This is ahigh proportion with poor adult
adjustment andlimited coping strategies. Furthermore, anxietyor
depression had contributed to the develop-ment of substance misuse
in about one-third.Whether anxiety and depression were experi-enced
as a past event, or whether they were, inretrospect, a pathway into
substance misuse,they emerge as frequent antecedents to sub-stance
misuse at all ages.
It appears paradoxical, but both a past historyof depression and
of other neurotic conditionswere associated with a lower problem
score, pos-sibly because these subjects came to theattention of
helping agents before they had timeto accumulate substance-related
problems. Thegroup with other neurotic conditions was com-posed of
a variety of diagnoses: seven hadsuffered from anxiety, five had
sought help withrelationship difficulties and the remainder
com-prised anorexia nervosa, conduct disorder,morbid jealousy or
episodes of self-harm. Thisgroup had a younger age of onset of
problematicsubstance use and some of these episodes ofpsychiatric
illness may have been secondary tothe developing substance problem.
Not all of oursubjects were considered to have a
definablevulnerability; about one-quarter were thought tohave
drifted into substance misuse.
Implications for policyThe detection and prevention of drug and
alco-hol misuse at work is within the remit ofoccupational health
services. Substance misuse
-
Doctors and substance misuse 663
by any member of an organization is damagingand wasteful of
their training; it is entirely appro-priate that occupational
health services prioritisethis issue. In addition to the increasing
emphasison counselling and health promotion, occupa-tional health
has an expanding role in theimplementation of UK and EEC
regulationsgoveming conditions in the workplace. All ofthese
demands have resource implications, butshort cuts in occupational
health services arecostly in the long term. The hospital doctors
intraining grades show a predilection to misuse avariety of drugs,
in addition to alcohol. It is clearthat this is partly in response
to personalitydifficulties, but also this group are known to
besubject to a collection of stressors unique to thefirst years of
qualification. The connections be-tween demoralisation and
self-medication areunclear but failing to address misery due
toorganizational deficiencies sets the scene for re-duced
expectations and lack of motivation in thefuture. Strategies have
been suggested both inthe UK^'' and the US^ to improve their
experi-ences. As a preventive measure, the issue ofself-prescribing
could be addressed via post-graduate education.
About 36% of Britain's doctors are self-employed general
practitioners. They consulttheir own general practitioners (often a
partnerin the practice) at about one-tenth the rate ofnon-general
practitioner patients.' They fre-quently diagnose, treat and refer
themselves,sometimes inappropriately. They have access
tomood-altering drugs as a matter of routine, andmany of our
subjects had attempted to alleviatedistress by misuse of
self-prescribed medication.This is facilitated by a culture within
the profes-sion that regards self-treatment as an
appropriateinitial response to illness.*'' It has been
suggestedthat a preferential health care system for doctorswould
facilitate access to independent medicalcare.'"
Many subjects developed a substance misuseproblem in middle age,
including one-quarter ofour number who were considered to have
driftedinto substance misuse. This illustrates the needfor
continuing interest by the profession in thewelfare of all its
members. Concentrating on thetraining grades and the
psychologically vulner-
able would fail to reach this group. They con-tribute to the
numbers of undiagnosed addictdoctors. Not only missed, they are
also misman-aged because of lack of awareness thatconfidential,
expert and effective help is avail-able.""" Wider publicity for
this help wouldencourage earlier entry into treatment and dispelthe
aura of gloom that has imprisoned addicteddoctors by paralysing
those around them.
References1, BROOKE, D, , EDWARDS, G, & TAYLOR, C,
(1991)
Addiction as an occupational hazard: 144 doctorswith drug and
alcohol problems, British Joumal ofAddiction, 86, pp.
1011-1016.
2, BUYDENS-BRANCHEY, L,, BRANCHEY, M . & Nou-MAIR, D, (1989)
Age of alcoholism onset:relationship to psychopathology, Archives
ofGeneral Psychiatry, 46, pp, 225-230,
3, CLONINGER, C, R, (1987) Neurogenetic adaptivemechanisms in
alcoholism, Science, 236, pp, 410-416,
4, DowLiNG, S, & BARRETT, S, (1991) Doctors in theMaking.
The Experience of the Pre-registration Year(Bristol, SAUS
Publications, Bristol University),
5, GARRUD, P, (1990) Counselling needs and experi-ence of junior
hospital doctors, British MedicalJoumal, 300, pp, 445-447,
6, RESIDENT SERVICES COMMITTEE, ASSOCIATION OFPROGRAM DIRECTORS
IN INTERNAL MEDICINE(1988) Stress and impairment during
residencytraining: strategies for reduction, identification
andmanagement, Annals of Intemal Medicine, 109, pp,154-161,
7, CHAMBERS, R, & BELCHER, J, (1992) Self-reportedhealth
care over the past ten years: a survey ofgeneral practitioners,
British Joumal of GeneralPractice, 42, pp, 153-156,
8, AixiBONE, A,, OAKES, D . & SHANNON, H, (1981)The health
and health care of doctors, Joumal ofthe Royal College of General
Practitioners, 31, pp,728-734,
9, RICHARDS, R, (1989) The Health of Doctors, ProjectPaper No,
78 (London, King's Fund PublishingOffice).
10, CHAMBERS, R, (1989) The health of general practi-tioners: a
cause for concem? (Editorial), Joumal ofthe Royal College of
General Practitioners, 39, pp,179-181,
11, LLOYD, G, (1990) Alcoholic doctors can recover,British
Medical Joumal, 300, pp, 728-730,
12, RAWNSLEY, K, (1985) Helping the sick doctor: anew service,
British Medical Joumal, 291, p, 922,
13, SMITH, R, (1989) Dealing with sickness and in-competence:
success and failure, British MedicalJoumal, 298, pp, 1695-1698,