- Email: [email protected]
Is drug use a response to stress?
Drug and Alcohol Dependence,
Elsevier Scientific Publishers
29 (1991) 97 - 106 Ireland Ltd.
97
Is drug use a response to stress? Fiona O’Doherty Department
of Psychology,
University sf Strathclyde,
George St., Glasgow CU.&
(Received July lOth, 1991) The perceived impact of life events on drug use was tested and specifically the hypothesis that drug use is a response to stress. Three groups of drug users were examined: heroin users, heavy drinkers and smokers, These were compared to non-drug using controls and the frequency of life events in the lives of each group was recorded. Whereas both the heroin and alcohol groups reported more stressful life events in their lives, no such difference was found for the smokers when compared to their control group. However, the differences between the heroin users and their control group on the one hand, and between the drinkers and their control group on the other, were found to consist largely of events which were consequences of the drug use itself. Furthermore, these two drug user groups (alcohol and heroin respectively) apparently experienced less stressful events than their control groups when events unconnected with drug use were examined alone. Nonetheless, this is interpreted as due to the way these groups perceive events rather than as an accurate record of their experiences. The conclusion is that heroin and alcohol use act as stress ‘buffers’, in that they reduce awareness of ongoing stressful life events. But in so doing they introduce a lot of new stresses which possibly maintain the drug use and therefore ultimately increase stress. Key words: heroin use; drinking; smoking; stress
Introduction
The theory that life events can have an impact on drug use arises from two assumptions. The first of these is that drug use is a response to stress, and the second is that drug use is functional in reducing stress. In support of the first idea, high scores in the life event profiles of drug users have been found in some studies (Duncan 1977; Kosten et al., 1986; Dudley et al., 1974). The stress reducing role of drug use has also been examined, for instance, by Neff and Hussaini (1982) in a group of drinkers. Their hypothesis was that, given the same amount of life stress, heavier drinkers would experience less depression than lighter drinkers. Their conclusions, however, were equivocal. Although they Correspondence to: Fiona O’Doherty, Department of Business Studies, Edinburgh University, George Square, Edinburgh. U.K. 0376-8716/91/$03.50 0 1991 Elsevier Scientific Publishers Printed and Published in Ireland
found a negative relationship between life stress and depression amongst the heavy drinkers and in the light drinking group, this relationship did not occur in the moderate drinking group. Life stress has also been identified as a possible factor leading to relapse in drug use after a period of remission (Marlatt and Gordon, 1980). Consequently the identification of relapse precipitants, such as stressful situations and events, has been the focus of some studies (e.g., Litman et al., 1977) and relapse prevention programmes. The assumption is that avoidance of the stressful situation would result in avoidance of the relapse. In a review, O’Doherty and Davies (1987) examined the evidence that life events influence drug use and concluded that despite the abundance of research on the topic, no clear understanding of the relationship between stressful events and drug use has yet emerged. One possibility is that whereas life events may Ireland Ltd.
98
play an important role in initiating drug use, subsequent life events may be of less importance once a pattern of drug use has been established. The failure of research to identify which aspects of stress are important and how it impacts on drug use, is due to a number of methodological problems common to studies on life events. Such methodological problems lead to difficulty with interpretation and typically these consist of: (I) confounding the independent and dependent variables; (2) an absence of control groups; and atheoretical approach to (3) a general research in this field. These problems have been dealt with adequately by a number of authors (Schroeder and Costa, 1984; Paykel, 1983; O’Doherty and Davies, 1987) and will not therefore be considered in detail here. Nonetheless, these difficulties indicate that although, for example, Duncan (1977) has concluded his results support ‘the theory that drug use is a stress induced disorder’ and although Dudley et al. (1976), in a study of both heroin and alcohol abusers, felt that ‘high life change could act as a causal factor in perpetuating chronicity and complicating treatment’, such conclusions are still unproven hypotheses. The present authors are unaware of any controlled studies on life events and drug use which adequately take account of the methodological problems referred to above. Consequently a number of basic questions remain to be answered about the role that life events play in changing drug consumption. The study presented in this paper aims to answer such questions by examining the frequency of life events in the lives of drug users compared to the lives of non-users. The simple and preliminary hypothesis is that drug users will report more life stresses than non-users. Method
Three groups of drug users were examined. These consisted of 31 heroin users recruited from a community based drug clinic, 33 heavy
drinkers (i.e., they consumed on average more than 50 standard units of alcohol per week,, measured by a l-week retrospective drinking diary) and also recruited through alcohol problems clinics, and 39 self-confessed cigarette addicts. A further 63 people formed a control pool. Control subjects were all non-smokers, non-heroin users and drank less than 8 standard units of alcohol per week. From this control pool three groups were selected which matched the sociodemographic characteristics of the substance groups. These three groups consisted of 36 heroin controls, 27 alcohol controls and 63 tobacco controls. As there was obviously some overlap of control subjects, no inter-drug comparisons were made. A total of 166 subjects participated in the study. Subjects were interviewed at 3-month intervals over a period of 18 months. A maximum of six interviews was administered to each subject. A life events schedule designed for the study was administered at each interview which recorded information about events which had occurred since the last interview (or in the case of interview one, events which had occurred during the previous 3 months). The schedule is reproduced in Table I. Furthermore, at each interview a l-week retrospective diary of drug consumption was collected and used as a sample of drug use for that 3-month period around the interview. The interview schedule will be described fully here, despite the fact that only part of it will be used in the analysis in this paper. For the most part the items are similar to those of Saracon et al. (1978) Life Experiences Survey (LES) on which the present list was based. Some addiction specific items have been added. The schedule differs, however, from the LES in the following important ways. Firstly, although information was recorded on the standard form shown in Table I, this information was collected by means of an interview as opposed to a self-completed checklist. Trained interviewers were used to elicit information about the events and to discuss the events with the subjects before recording that information. This was done to avoid the problems which result from use of the self-completed checklist approach (see
-
I
_
.~
The life events
c
-
.^
.~
of peace
_
(4) jail for disturbance
driving
theft, etc.
(3) jail for drunk
petty
(2) jail for (drug related)
(1) any imprisonment
LEGAL
117) overdosed on drugs (18) any other illness
plcasc)
to GP
(specify
(16) visits
(specify please)
(15) any hospitalirutions (other than for above1
(13) anxiety (14) loss of appetite
(11) sexual difficulties (121 sleep disturbance
_
profile.
(IO) recent conccntratmn and memory problem?,
(9) depression (diagnosed)
(7) liver damage (8) stomach ulcer
(-5) abscess (6) serious tooth decay
(3) bronchitis (4) hepatitis
(e.g. strokes)
HEALTH (I) heart problems (2) circulation problrma
EVENT
CODE NC).
Table I.
_
\.
_
DATE
_.
_~
,
,
_
1
,_
_
_
_
_
I! = unrelated
^
_*
_
,,,
_
_
_
--
N = no effect
r_
_ _ _
I = increase
~
in
but not
leaving
conflict
(9) divorce/separation (10) at centre of family
(7) retirement (3) son/daughter home
(6) finish formal schooling/university
(4) death of spouse (5) death of relative/ friend
(3) split with girlfriend/ boyfriend
-
-
-
-
-
-
-
-
SOCIAL (1) new circle of friends (2) meet new girlfriend/ boyfriend
-
-
-
-
-
-
(4) increase in financial resources
of
-
(3) reduced income (for any reason)
(2) pressure to pay
(1) increase debt (borrowed more)
FINANCIAL
(9) arrested, charged
appearance
(8) any other court
above, but not found guilty (7) jail for possession illegal substances
connection with the
(6) court appearance
(5) loss of licencelfine
Table I. (cod.)
-
I
_
family
(15) other
please)
conflict
tried)
with
pressure
with boss/
working
.~
c
-
.^
.~
health;
_
_
_
\.
M, Mood; L, Legal;
life recently
your
*P, Physical
have affected
that
(12)any other events
hours
(11)changed
(10)failed exams
(9)sat exams
teacher
(8)problems
(7)promotion
(6)added responsibility
at work/study
(5)increased
(4)redundancy
dismissal
(3)threatened
(specifically
(1) new job (2) failure to find work
EMPLOYMENT
house
(specify
(14) Moved
(13) pregnancy
(12) marriage
(11) other
_
_.
_~
S. Social;
,
_
,_
F, Financial;
_
_
_
_
E, Employment.
^
_*
_
,,,
_
_
_
--
r_
_ _ _
~
6
102
Paykel, 1983; Brown and Harris, 1982). Subjects therefore indicated which problems had occurred in any of the 3 previous months (locating the event in one of the 3 months), and said whether they were directly involved (‘self events) or whether the main participant was a significant ‘other’ in their lives (‘others’ events). Examples of ‘others’ events were when a close friend became ill, or a parent became unemployed. The second way in which this schedule differs is that in the LES only one overall rating of each event, as being negative, positive or neutral, is obtained. Events, however, commonly have both negative and positive aspects, which are not necessarily exclusive categories. For this reason the schedule presented here collects information on various aspects of an event including whether the effect on each of the following was positive, negative or neutral: (1) the effect of the event on Physical Health; (2) on Mood; (3) on Legal; (4) Social; (5) Financial; and (6) Employment aspects of life. These ratings are not, however, used in the current paper which examines only the frequency of events in conjunction with the following ratings. These ratings correspond to the third way in which the schedule differs from the LES, in that it obtains also information about: (1) the subjects beliefs as to the cause of each event; and (2) subjects beliefs about the effect each event has had on their subsequent drug consumption. In the first of these, subjects were asked whether they believed an event was caused by, or, was unrelated to, their drug use. In the second, subjects had to choose whether the event had lead to increased consumption, decreased consumption, or had no effect on consumption. There is some similarity between the classification of caused/unrelated events in this study and Brown and Harris’ (1978) distinction between dependent and independent events. However, an important difference does exist between the two. In Brown and Harris’ categorization, events were labelled ‘dependent’ if they had been within the control of the subject, regardless of whether or not they were actually caused by depression (the dependent variable in their study). For example, events such as mar-
riage or leaving home would be considered dependent. Any event outside the subjects control was classified by the authors as ‘independent’. Physical illness would be an example of an independent event. Independent events were therefore those considered to be independent of the subject and not, as in this study, independent of the dependent variable (drug use). In the present study, events were classified as ‘caused’ when the subjects believed them to have been caused by the drug use (and therefore were within the control of the subject, theoretically at least). Unrelated events, on the other hand, were events unrelated to drug use and could either be within or outwith the subjects control. To be specific, marriage would be classified as an ‘unrelated event’ in this study , but as a ‘dependent’ event in Brown and Harris’ classification. Whereas physical illness could be either a ‘caused’ or ‘unrelated’ event, depending on the subjects perception and the type of illness. All of the ratings in the current study were given by the subjects, whereas in Brown and Harris’ study the ratings were applied by the interviewer after the interview had been terminated. The weighting of events according to normative means is common in life event questionnaires. For instance the frequently used Social Readjustment Rating Scale (SRRS), by Holmes and Rahe (1967), provides a score, (a weight), for each event indicating the relative severity of the event. The total sum of the weights checked by a subject indicates the level of stress reported by that subject. However, Rahe and Arthur (1978) have found that the total of these weights correlated significantly with the simple number of events reported and was just as good a predictor of illness. This finding has also been confirmed by more recent work, for example, Skinner and Lei (1980), or Marziali and Pilkonis (1986). In the following analysis therefore, no attempt is made to weight the events and only the total numbers of events reported per interview are used. By means of this simple preliminary analysis, devoid of contamination by weights, it was hoped to establish whether or not there was a prima face case to say that drug users experience more stress than non users.
103
Table
II.
group
Heroin control
N
1v = 131
Heroin = 94
f-Value
“-Tail prohahility
X
X
TE, Total number of events SE, Number of self events
7.64 4.9
4.77 2.0
5.44 8.05
O.Ooo* 0.000’
OE, Number of others events CSE, Number of caused self events USE, Number of unrelated self events COE, Number caused others events
2.7 3.6 1.2 1.2
UOE,
1.5
2.7 0.06 1.9 0.4 2.3
0.08 11.6 -3.41 5.3 -2.9
0.93 o.ooo* 0.001* 0.000” O.OW’
Number
of unrelated
others
events
_ ‘P
< 0.01.
Results
Table II shows the results of t-tests comparing the frequency of events reported by heroin users with the frequency of events reported by the matched control group. Tables III and IV show the same comparisons for the alcohol and tobacco groups respectively. Figures I- 3 show these results graphically. Overall the heroin users reported significantly more life events than tl12 control group (see ‘Total number of events’ in Table I ). However, a further examination of the table shows that an excess of events was not reported by the heroin users for all comparisons. Firstly, significantly more events about ‘self’
n q
6
a Z
9 >
+ A Eel ;m -: 09
Heroin Group Heroin Control Group
6
4
2
c”: 0 l-E
sz
CE
Fig. 1. Frequency of events heroin control groups.
CsEEEmuE reported
by the heroin
and
were reported by the heroin users than by the controls, but there was no difference in the number of events reported about ‘significant others’. Secondly, amongst those events occurring to the drug users themselves, significantly more were caused by the drug use amongst the heroin users than in the control group. But as such events could rarely happen to the controls, this finding is hardly surprising. What is more difficult to explain, however, is the finding that the heroin users reported experiencing significantly fewer events than did the control group when events unrelated to heroin were examined alone. A similar pattern emerged when events about significant others were broken down. The heroin users reported significantly more of these to have been caused by heroin than did the control group, and they reported significantly fewer events unrelated to heroin use than the control group. Table III shows the same comparisons for the alcohol group and their matched control group. A comparable pattern emerged. The drinkers reported significantly more events overall than did the control group, and also more events about themselves than did the control group. Both groups reported a similar number of events about significant others. Amongst the ‘self’ events significantly more were caused by drinking and significantly less were unrelated to drinking in the alcohol group than in the control group. Amongst the events occurring to signifi-
104 TabIe III.
TE, Total number of events SE, Number of self events OE, Number of other events CSE, Number of caused self events USE, Number of unrelated self events COE, Number caused other events UOE, Number of unrelated others events
Alcohol group X N = 122
Alcohol control group X N = 106
t-Value
2-Tail probability
6.86 3.6 3.2 2.3 1.3 0.6 2.5
4.6 1.9 2.6 0.01 1.9 0.16 2.5
4.8 5.26 1.78 9.8 - 2.94 3.48 0.33
o.ooo* 0.001” 0.076 o.ooo* 0.004* 0.001* 0.73
Tobacco group x N = 146
Tobacco control
t-Value
2-Tail probability
0.38 0.43 0.91 2.73 - 1.0 -0.0 1.0
0.70 0.67 0.36 0.007* 0.29 0.99 -3.0
‘P < 0.01. Table IV.
TE, Total number of events SE, Number of self events OE, Number of other events CSE, Number of caused self events USE, Number of unrelated self events COE, Number caused other events UOE, Number of unrelated others events
group N = 236
4.8 1.9 2.9 0.16 1.7 0.29 2.6
4.7 2.0 2.6 0.04 1.9 0.29 2.3
‘P < 0.01.
6
:
‘;
5
n n
6
Alcoholgroup Akohol control
“$3
$5
4
0
“8 g:
Tobacco group Tobacco control group
SS
=$I
:,
n n
t4
group
EZ t
8*2 ‘0:
J 2
P:
lE
3E
CE
c9uQoIlE
Fig. 2. Frequency of events reported by alcohol and alcoho1 control grc ups.
1
lE
SE
CE
CsElsimuE
Fig. 3. Frequency of events tobacco control groups.
reported
by tobacco and
105
cant others, more events caused by drinking were reported by the alcohol group than by the control group. Unlike the heroin group, however, there was no significant difference between the drinkers and the controls in the number of events reported about other people which were unrelated to drinking. Table IV shows the results for the tobacco group. The smokers can be seen to be almost identical to their control group in their reports of life events. In fact the only significant difference between the smokers and their controls was, not surprisingly, that the smokers reported significantly more events occurring to themselves which were caused by tobacco than did the non-smoking control group. Discussion
The results can be interpreted in a number of ways. Firstly, both the heroin and alcohol groups reported experiencing more stressful events than their respective control groups, but on the face of it this excess of stressful events consisted of consequences of drug use itself. In other words, it appears that increased stress in the lives of drug users could be the result of the drug use rather than a cause of it. If anything, it appears that these two groups, i.e. the alcohol and the heroin users, experienced less life events than did the control groups, when stressful events uncontaminated by drug use are examined alone. By contrast the smokers showed an almost identical pattern of event reporting to that of their control group. It is almost certain that heroin users and drinkers experience more life events caused by their drug use than do non-drug using controls. Both heroin and alcohol taken frequently in large doses can result in negative consequences for health, not always caused directly by the substance itself. People who drink heavily are statistically more likely to become involved in accidents and regular use of large doses of heroin runs the risk of developing abcesses or contamination from AIDS or hepatitis. The possession of heroin increases ones legal events, through arrest or imnrisonment and the same
applies to alcohol use in certain circumstances such as drunk driving. Financial and employment problems can also be shown to associate with heroin and alcohol abuse. The negative consequences of smoking in terms of the type of events just outlined are probably less common. Tobacco is not illegal (except the sale of cigarettes to those under 16 years of age). Although attitudes may be changing, it is also more socially acceptable than either heroin or heavy alcohol use, and is therefore less likely to lead to social or family problems. Furthermore, even for a heavy smoker, the financial outlay to maintain the habit is less than that required for an alcohol or heroin habit. Smoking therefore is less likely to lead to negative financial consequences than use of either of these other drugs. Negative health consequences for this group may also be less likely in the short term, and less likely to be reported as life events, although long term prolonged use of tobacco may lead to a worse outcome than either alcohol or heroin. Increased life stress in the lives of drug users is, therefore, at least in part a consequence of the drug use rather than a cause of it. Nonetheless, this increased life stress is possibly a factor in maintaining drug use once a habit has been established. This cannot, however, be the full explanation for the above findings. When ‘self’ events unconnected with drug use (the unrelated events) were examined alone, they were found to be reported with less frequency by the drug using groups than by the controls. Furthermore, the friends and families of the heroin users reportedly experienced less stressful events unrelated to drug use than did the friends and families of the control group. It seems unlikely that this could actually be the case. Is it not more likely that both the alcohol and heroin users have become over-involved with their drug use resulting in less awareness of ongoing unrelated events and heightened awareness of drug related events? (A further possibility is that events are attributed to drug use when this may not actually be the case, although from a detailed examination of the actual raw events reported, this seems to be unlikely).
106
The suggestion is that the alcohol and heroin users experience a narrowing of the perceptual repertoire, where their awareness or perception of non-drug experiences is narrowed and they become more focussed on drug related events. This interpretation of the findings is considered more plausible than the first because it is unlikely to be the case that the people around heroin users actually experience less life stress, and in particular less events unrelated to drug use, than do the friends and families of the control group. It is also hard to explain why heroin users and drinkers would genuinely experience less life stress unrelated to their drug use than would matched controls. Perhaps, as the user becomes more and more involved in the drug habit it is necessary to adopt a lifestyle in which drug use is a dominant and central component. This may cause them to become less aware of what is happening to other people in their lives and become more focussed on events which are related to drug use in their own lives. Tobacco use on the other hand has fewer immediate consequences for one’s life. Maintaining a smoking habit does not involve major changes in lifestyle and it is less disruptive of lifestyle than heavy use of either alcohol or heroin. Furthermore, procurement of the drug tobacco presents no particular problems and does not preoccupy the smoker unless he/she runs short, when it will soon dominate the consciousness. No narrowing of the smoking repertoire, either in practice or perception would therefore normally be expected. Conclusion
It is not possible to determine from this set of life event data exactly how drug use relates to stress. Undoubtedly heroin and alcohol users report more stressful events than do matched control groups. It seems that many of the events reported are those which are actually caused by the drug use, although it is not clear whether this reporting of events is an accurate reflection of actual experience. The possibility that drug use is functional in ‘stress-buffering’, which was raised in the in-
troduction, is given some support by the data, in that both heroin and alcohol users reported less self events which were unrelated to their drug use. However, it would appear that if drug use is a means of stress reduction, it is not a particularly efficient method. Although it may reduce awareness of one type of life stress, it introduces many new stresses into one’s life which possibly become functional in maintaining the drug use and thereby creating further stress. References
Brown, G.W. and Harris, T. (1978) The Social Origins pression. Tavistock London.
of De-
Brown, G.W. and Harris, T. (1982) Fall-off in the reporting of Life Events. Social Psychiatry 17, 23-28. Dudley, D.L., Mules, J.E., Roszell, D.K., Glickfield, G. and Hague, W.H. (1976) Frequency and magnitude distrihution of life change in heroin and alcohol addicts. Int. .I. Addict. I 1, 977 - 987. Duncan, D.F. (1977) Life stress as a precurser to adolescent drug dependence. Int. J. Addict. 12, 1047 - 1056. Holmes, T. and Rahe, R. (1967) The Social Readjustment rating scale. J. Pschosom. Res. 11, 2133218. Kosten, T.R., Rounsaville. M.D., Herbert, D. and Kleber, M.D. (1986) A 2.5 year follow-up of depression, life crises and treatment effects on abstinence among opioid addicts. Arch. Gen. Psychiatry 43, 733-738. Litman, G.K., Eiser, J.R. and Rawson. N. (1977) Towards a typology of relapse. Drug Alcohol Depend. 2, 157- 162. Marlatt, G.A and Gordon, J.R. (1980). Determinants of relapse: Implications for the maintainance of behaviour change. In: Behavioural Medicine: Changing Health Life Styles (Davidson, P.O. and Davidson, S.M., eds), Brunner/Mazel, Marziali, E.A Subjective Stress 12,
New York. and Pilkonis, P.A. (1986) The Measurement of Response to Stressful Life Events. J. Hum. 5- 12.
Neff, J.A. and Husaini, B.A. (1982) Life Events, Drinking Patterns and Depressive Symptomatology. .J. Stud. Alcohol. 43, 301-318. O’Doherty, F. and Davies, J. (1987) Life events and Addiction A critical Review. Br. J. Addict. 82, 127- 137. Paykel, E.S. (1983) Methodological aspects of life events research. J. Psychosom. Res. 27. 341-352. Rahe, R.H. and Arthur, R.H. (1978) Life changes and illness studies. J. Hum. Stress 4, 3- 15. Saracon, I.G., Johnson, J.H. and Siegal, J.M. (1978) Assessing the impact of life changes development of the life experiences survey. J. Consult. Clin. Psychol. 46, 932-946. Schroeder, D.H. and Costa, P.T. (1984) The influence of life stress on physical illness: Substantive effects of methodological flaws? .J. Pers. Sot. Psychol. 46, 853 - 863.
Elsevier Scientific Publishers
29 (1991) 97 - 106 Ireland Ltd.
97
Is drug use a response to stress? Fiona O’Doherty Department
of Psychology,
University sf Strathclyde,
George St., Glasgow CU.&
(Received July lOth, 1991) The perceived impact of life events on drug use was tested and specifically the hypothesis that drug use is a response to stress. Three groups of drug users were examined: heroin users, heavy drinkers and smokers, These were compared to non-drug using controls and the frequency of life events in the lives of each group was recorded. Whereas both the heroin and alcohol groups reported more stressful life events in their lives, no such difference was found for the smokers when compared to their control group. However, the differences between the heroin users and their control group on the one hand, and between the drinkers and their control group on the other, were found to consist largely of events which were consequences of the drug use itself. Furthermore, these two drug user groups (alcohol and heroin respectively) apparently experienced less stressful events than their control groups when events unconnected with drug use were examined alone. Nonetheless, this is interpreted as due to the way these groups perceive events rather than as an accurate record of their experiences. The conclusion is that heroin and alcohol use act as stress ‘buffers’, in that they reduce awareness of ongoing stressful life events. But in so doing they introduce a lot of new stresses which possibly maintain the drug use and therefore ultimately increase stress. Key words: heroin use; drinking; smoking; stress
Introduction
The theory that life events can have an impact on drug use arises from two assumptions. The first of these is that drug use is a response to stress, and the second is that drug use is functional in reducing stress. In support of the first idea, high scores in the life event profiles of drug users have been found in some studies (Duncan 1977; Kosten et al., 1986; Dudley et al., 1974). The stress reducing role of drug use has also been examined, for instance, by Neff and Hussaini (1982) in a group of drinkers. Their hypothesis was that, given the same amount of life stress, heavier drinkers would experience less depression than lighter drinkers. Their conclusions, however, were equivocal. Although they Correspondence to: Fiona O’Doherty, Department of Business Studies, Edinburgh University, George Square, Edinburgh. U.K. 0376-8716/91/$03.50 0 1991 Elsevier Scientific Publishers Printed and Published in Ireland
found a negative relationship between life stress and depression amongst the heavy drinkers and in the light drinking group, this relationship did not occur in the moderate drinking group. Life stress has also been identified as a possible factor leading to relapse in drug use after a period of remission (Marlatt and Gordon, 1980). Consequently the identification of relapse precipitants, such as stressful situations and events, has been the focus of some studies (e.g., Litman et al., 1977) and relapse prevention programmes. The assumption is that avoidance of the stressful situation would result in avoidance of the relapse. In a review, O’Doherty and Davies (1987) examined the evidence that life events influence drug use and concluded that despite the abundance of research on the topic, no clear understanding of the relationship between stressful events and drug use has yet emerged. One possibility is that whereas life events may Ireland Ltd.
98
play an important role in initiating drug use, subsequent life events may be of less importance once a pattern of drug use has been established. The failure of research to identify which aspects of stress are important and how it impacts on drug use, is due to a number of methodological problems common to studies on life events. Such methodological problems lead to difficulty with interpretation and typically these consist of: (I) confounding the independent and dependent variables; (2) an absence of control groups; and atheoretical approach to (3) a general research in this field. These problems have been dealt with adequately by a number of authors (Schroeder and Costa, 1984; Paykel, 1983; O’Doherty and Davies, 1987) and will not therefore be considered in detail here. Nonetheless, these difficulties indicate that although, for example, Duncan (1977) has concluded his results support ‘the theory that drug use is a stress induced disorder’ and although Dudley et al. (1976), in a study of both heroin and alcohol abusers, felt that ‘high life change could act as a causal factor in perpetuating chronicity and complicating treatment’, such conclusions are still unproven hypotheses. The present authors are unaware of any controlled studies on life events and drug use which adequately take account of the methodological problems referred to above. Consequently a number of basic questions remain to be answered about the role that life events play in changing drug consumption. The study presented in this paper aims to answer such questions by examining the frequency of life events in the lives of drug users compared to the lives of non-users. The simple and preliminary hypothesis is that drug users will report more life stresses than non-users. Method
Three groups of drug users were examined. These consisted of 31 heroin users recruited from a community based drug clinic, 33 heavy
drinkers (i.e., they consumed on average more than 50 standard units of alcohol per week,, measured by a l-week retrospective drinking diary) and also recruited through alcohol problems clinics, and 39 self-confessed cigarette addicts. A further 63 people formed a control pool. Control subjects were all non-smokers, non-heroin users and drank less than 8 standard units of alcohol per week. From this control pool three groups were selected which matched the sociodemographic characteristics of the substance groups. These three groups consisted of 36 heroin controls, 27 alcohol controls and 63 tobacco controls. As there was obviously some overlap of control subjects, no inter-drug comparisons were made. A total of 166 subjects participated in the study. Subjects were interviewed at 3-month intervals over a period of 18 months. A maximum of six interviews was administered to each subject. A life events schedule designed for the study was administered at each interview which recorded information about events which had occurred since the last interview (or in the case of interview one, events which had occurred during the previous 3 months). The schedule is reproduced in Table I. Furthermore, at each interview a l-week retrospective diary of drug consumption was collected and used as a sample of drug use for that 3-month period around the interview. The interview schedule will be described fully here, despite the fact that only part of it will be used in the analysis in this paper. For the most part the items are similar to those of Saracon et al. (1978) Life Experiences Survey (LES) on which the present list was based. Some addiction specific items have been added. The schedule differs, however, from the LES in the following important ways. Firstly, although information was recorded on the standard form shown in Table I, this information was collected by means of an interview as opposed to a self-completed checklist. Trained interviewers were used to elicit information about the events and to discuss the events with the subjects before recording that information. This was done to avoid the problems which result from use of the self-completed checklist approach (see
-
I
_
.~
The life events
c
-
.^
.~
of peace
_
(4) jail for disturbance
driving
theft, etc.
(3) jail for drunk
petty
(2) jail for (drug related)
(1) any imprisonment
LEGAL
117) overdosed on drugs (18) any other illness
plcasc)
to GP
(specify
(16) visits
(specify please)
(15) any hospitalirutions (other than for above1
(13) anxiety (14) loss of appetite
(11) sexual difficulties (121 sleep disturbance
_
profile.
(IO) recent conccntratmn and memory problem?,
(9) depression (diagnosed)
(7) liver damage (8) stomach ulcer
(-5) abscess (6) serious tooth decay
(3) bronchitis (4) hepatitis
(e.g. strokes)
HEALTH (I) heart problems (2) circulation problrma
EVENT
CODE NC).
Table I.
_
\.
_
DATE
_.
_~
,
,
_
1
,_
_
_
_
_
I! = unrelated
^
_*
_
,,,
_
_
_
--
N = no effect
r_
_ _ _
I = increase
~
in
but not
leaving
conflict
(9) divorce/separation (10) at centre of family
(7) retirement (3) son/daughter home
(6) finish formal schooling/university
(4) death of spouse (5) death of relative/ friend
(3) split with girlfriend/ boyfriend
-
-
-
-
-
-
-
-
SOCIAL (1) new circle of friends (2) meet new girlfriend/ boyfriend
-
-
-
-
-
-
(4) increase in financial resources
of
-
(3) reduced income (for any reason)
(2) pressure to pay
(1) increase debt (borrowed more)
FINANCIAL
(9) arrested, charged
appearance
(8) any other court
above, but not found guilty (7) jail for possession illegal substances
connection with the
(6) court appearance
(5) loss of licencelfine
Table I. (cod.)
-
I
_
family
(15) other
please)
conflict
tried)
with
pressure
with boss/
working
.~
c
-
.^
.~
health;
_
_
_
\.
M, Mood; L, Legal;
life recently
your
*P, Physical
have affected
that
(12)any other events
hours
(11)changed
(10)failed exams
(9)sat exams
teacher
(8)problems
(7)promotion
(6)added responsibility
at work/study
(5)increased
(4)redundancy
dismissal
(3)threatened
(specifically
(1) new job (2) failure to find work
EMPLOYMENT
house
(specify
(14) Moved
(13) pregnancy
(12) marriage
(11) other
_
_.
_~
S. Social;
,
_
,_
F, Financial;
_
_
_
_
E, Employment.
^
_*
_
,,,
_
_
_
--
r_
_ _ _
~
6
102
Paykel, 1983; Brown and Harris, 1982). Subjects therefore indicated which problems had occurred in any of the 3 previous months (locating the event in one of the 3 months), and said whether they were directly involved (‘self events) or whether the main participant was a significant ‘other’ in their lives (‘others’ events). Examples of ‘others’ events were when a close friend became ill, or a parent became unemployed. The second way in which this schedule differs is that in the LES only one overall rating of each event, as being negative, positive or neutral, is obtained. Events, however, commonly have both negative and positive aspects, which are not necessarily exclusive categories. For this reason the schedule presented here collects information on various aspects of an event including whether the effect on each of the following was positive, negative or neutral: (1) the effect of the event on Physical Health; (2) on Mood; (3) on Legal; (4) Social; (5) Financial; and (6) Employment aspects of life. These ratings are not, however, used in the current paper which examines only the frequency of events in conjunction with the following ratings. These ratings correspond to the third way in which the schedule differs from the LES, in that it obtains also information about: (1) the subjects beliefs as to the cause of each event; and (2) subjects beliefs about the effect each event has had on their subsequent drug consumption. In the first of these, subjects were asked whether they believed an event was caused by, or, was unrelated to, their drug use. In the second, subjects had to choose whether the event had lead to increased consumption, decreased consumption, or had no effect on consumption. There is some similarity between the classification of caused/unrelated events in this study and Brown and Harris’ (1978) distinction between dependent and independent events. However, an important difference does exist between the two. In Brown and Harris’ categorization, events were labelled ‘dependent’ if they had been within the control of the subject, regardless of whether or not they were actually caused by depression (the dependent variable in their study). For example, events such as mar-
riage or leaving home would be considered dependent. Any event outside the subjects control was classified by the authors as ‘independent’. Physical illness would be an example of an independent event. Independent events were therefore those considered to be independent of the subject and not, as in this study, independent of the dependent variable (drug use). In the present study, events were classified as ‘caused’ when the subjects believed them to have been caused by the drug use (and therefore were within the control of the subject, theoretically at least). Unrelated events, on the other hand, were events unrelated to drug use and could either be within or outwith the subjects control. To be specific, marriage would be classified as an ‘unrelated event’ in this study , but as a ‘dependent’ event in Brown and Harris’ classification. Whereas physical illness could be either a ‘caused’ or ‘unrelated’ event, depending on the subjects perception and the type of illness. All of the ratings in the current study were given by the subjects, whereas in Brown and Harris’ study the ratings were applied by the interviewer after the interview had been terminated. The weighting of events according to normative means is common in life event questionnaires. For instance the frequently used Social Readjustment Rating Scale (SRRS), by Holmes and Rahe (1967), provides a score, (a weight), for each event indicating the relative severity of the event. The total sum of the weights checked by a subject indicates the level of stress reported by that subject. However, Rahe and Arthur (1978) have found that the total of these weights correlated significantly with the simple number of events reported and was just as good a predictor of illness. This finding has also been confirmed by more recent work, for example, Skinner and Lei (1980), or Marziali and Pilkonis (1986). In the following analysis therefore, no attempt is made to weight the events and only the total numbers of events reported per interview are used. By means of this simple preliminary analysis, devoid of contamination by weights, it was hoped to establish whether or not there was a prima face case to say that drug users experience more stress than non users.
103
Table
II.
group
Heroin control
N
1v = 131
Heroin = 94
f-Value
“-Tail prohahility
X
X
TE, Total number of events SE, Number of self events
7.64 4.9
4.77 2.0
5.44 8.05
O.Ooo* 0.000’
OE, Number of others events CSE, Number of caused self events USE, Number of unrelated self events COE, Number caused others events
2.7 3.6 1.2 1.2
UOE,
1.5
2.7 0.06 1.9 0.4 2.3
0.08 11.6 -3.41 5.3 -2.9
0.93 o.ooo* 0.001* 0.000” O.OW’
Number
of unrelated
others
events
_ ‘P
< 0.01.
Results
Table II shows the results of t-tests comparing the frequency of events reported by heroin users with the frequency of events reported by the matched control group. Tables III and IV show the same comparisons for the alcohol and tobacco groups respectively. Figures I- 3 show these results graphically. Overall the heroin users reported significantly more life events than tl12 control group (see ‘Total number of events’ in Table I ). However, a further examination of the table shows that an excess of events was not reported by the heroin users for all comparisons. Firstly, significantly more events about ‘self’
n q
6
a Z
9 >
+ A Eel ;m -: 09
Heroin Group Heroin Control Group
6
4
2
c”: 0 l-E
sz
CE
Fig. 1. Frequency of events heroin control groups.
CsEEEmuE reported
by the heroin
and
were reported by the heroin users than by the controls, but there was no difference in the number of events reported about ‘significant others’. Secondly, amongst those events occurring to the drug users themselves, significantly more were caused by the drug use amongst the heroin users than in the control group. But as such events could rarely happen to the controls, this finding is hardly surprising. What is more difficult to explain, however, is the finding that the heroin users reported experiencing significantly fewer events than did the control group when events unrelated to heroin were examined alone. A similar pattern emerged when events about significant others were broken down. The heroin users reported significantly more of these to have been caused by heroin than did the control group, and they reported significantly fewer events unrelated to heroin use than the control group. Table III shows the same comparisons for the alcohol group and their matched control group. A comparable pattern emerged. The drinkers reported significantly more events overall than did the control group, and also more events about themselves than did the control group. Both groups reported a similar number of events about significant others. Amongst the ‘self’ events significantly more were caused by drinking and significantly less were unrelated to drinking in the alcohol group than in the control group. Amongst the events occurring to signifi-
104 TabIe III.
TE, Total number of events SE, Number of self events OE, Number of other events CSE, Number of caused self events USE, Number of unrelated self events COE, Number caused other events UOE, Number of unrelated others events
Alcohol group X N = 122
Alcohol control group X N = 106
t-Value
2-Tail probability
6.86 3.6 3.2 2.3 1.3 0.6 2.5
4.6 1.9 2.6 0.01 1.9 0.16 2.5
4.8 5.26 1.78 9.8 - 2.94 3.48 0.33
o.ooo* 0.001” 0.076 o.ooo* 0.004* 0.001* 0.73
Tobacco group x N = 146
Tobacco control
t-Value
2-Tail probability
0.38 0.43 0.91 2.73 - 1.0 -0.0 1.0
0.70 0.67 0.36 0.007* 0.29 0.99 -3.0
‘P < 0.01. Table IV.
TE, Total number of events SE, Number of self events OE, Number of other events CSE, Number of caused self events USE, Number of unrelated self events COE, Number caused other events UOE, Number of unrelated others events
group N = 236
4.8 1.9 2.9 0.16 1.7 0.29 2.6
4.7 2.0 2.6 0.04 1.9 0.29 2.3
‘P < 0.01.
6
:
‘;
5
n n
6
Alcoholgroup Akohol control
“$3
$5
4
0
“8 g:
Tobacco group Tobacco control group
SS
=$I
:,
n n
t4
group
EZ t
8*2 ‘0:
J 2
P:
lE
3E
CE
c9uQoIlE
Fig. 2. Frequency of events reported by alcohol and alcoho1 control grc ups.
1
lE
SE
CE
CsElsimuE
Fig. 3. Frequency of events tobacco control groups.
reported
by tobacco and
105
cant others, more events caused by drinking were reported by the alcohol group than by the control group. Unlike the heroin group, however, there was no significant difference between the drinkers and the controls in the number of events reported about other people which were unrelated to drinking. Table IV shows the results for the tobacco group. The smokers can be seen to be almost identical to their control group in their reports of life events. In fact the only significant difference between the smokers and their controls was, not surprisingly, that the smokers reported significantly more events occurring to themselves which were caused by tobacco than did the non-smoking control group. Discussion
The results can be interpreted in a number of ways. Firstly, both the heroin and alcohol groups reported experiencing more stressful events than their respective control groups, but on the face of it this excess of stressful events consisted of consequences of drug use itself. In other words, it appears that increased stress in the lives of drug users could be the result of the drug use rather than a cause of it. If anything, it appears that these two groups, i.e. the alcohol and the heroin users, experienced less life events than did the control groups, when stressful events uncontaminated by drug use are examined alone. By contrast the smokers showed an almost identical pattern of event reporting to that of their control group. It is almost certain that heroin users and drinkers experience more life events caused by their drug use than do non-drug using controls. Both heroin and alcohol taken frequently in large doses can result in negative consequences for health, not always caused directly by the substance itself. People who drink heavily are statistically more likely to become involved in accidents and regular use of large doses of heroin runs the risk of developing abcesses or contamination from AIDS or hepatitis. The possession of heroin increases ones legal events, through arrest or imnrisonment and the same
applies to alcohol use in certain circumstances such as drunk driving. Financial and employment problems can also be shown to associate with heroin and alcohol abuse. The negative consequences of smoking in terms of the type of events just outlined are probably less common. Tobacco is not illegal (except the sale of cigarettes to those under 16 years of age). Although attitudes may be changing, it is also more socially acceptable than either heroin or heavy alcohol use, and is therefore less likely to lead to social or family problems. Furthermore, even for a heavy smoker, the financial outlay to maintain the habit is less than that required for an alcohol or heroin habit. Smoking therefore is less likely to lead to negative financial consequences than use of either of these other drugs. Negative health consequences for this group may also be less likely in the short term, and less likely to be reported as life events, although long term prolonged use of tobacco may lead to a worse outcome than either alcohol or heroin. Increased life stress in the lives of drug users is, therefore, at least in part a consequence of the drug use rather than a cause of it. Nonetheless, this increased life stress is possibly a factor in maintaining drug use once a habit has been established. This cannot, however, be the full explanation for the above findings. When ‘self’ events unconnected with drug use (the unrelated events) were examined alone, they were found to be reported with less frequency by the drug using groups than by the controls. Furthermore, the friends and families of the heroin users reportedly experienced less stressful events unrelated to drug use than did the friends and families of the control group. It seems unlikely that this could actually be the case. Is it not more likely that both the alcohol and heroin users have become over-involved with their drug use resulting in less awareness of ongoing unrelated events and heightened awareness of drug related events? (A further possibility is that events are attributed to drug use when this may not actually be the case, although from a detailed examination of the actual raw events reported, this seems to be unlikely).
106
The suggestion is that the alcohol and heroin users experience a narrowing of the perceptual repertoire, where their awareness or perception of non-drug experiences is narrowed and they become more focussed on drug related events. This interpretation of the findings is considered more plausible than the first because it is unlikely to be the case that the people around heroin users actually experience less life stress, and in particular less events unrelated to drug use, than do the friends and families of the control group. It is also hard to explain why heroin users and drinkers would genuinely experience less life stress unrelated to their drug use than would matched controls. Perhaps, as the user becomes more and more involved in the drug habit it is necessary to adopt a lifestyle in which drug use is a dominant and central component. This may cause them to become less aware of what is happening to other people in their lives and become more focussed on events which are related to drug use in their own lives. Tobacco use on the other hand has fewer immediate consequences for one’s life. Maintaining a smoking habit does not involve major changes in lifestyle and it is less disruptive of lifestyle than heavy use of either alcohol or heroin. Furthermore, procurement of the drug tobacco presents no particular problems and does not preoccupy the smoker unless he/she runs short, when it will soon dominate the consciousness. No narrowing of the smoking repertoire, either in practice or perception would therefore normally be expected. Conclusion
It is not possible to determine from this set of life event data exactly how drug use relates to stress. Undoubtedly heroin and alcohol users report more stressful events than do matched control groups. It seems that many of the events reported are those which are actually caused by the drug use, although it is not clear whether this reporting of events is an accurate reflection of actual experience. The possibility that drug use is functional in ‘stress-buffering’, which was raised in the in-
troduction, is given some support by the data, in that both heroin and alcohol users reported less self events which were unrelated to their drug use. However, it would appear that if drug use is a means of stress reduction, it is not a particularly efficient method. Although it may reduce awareness of one type of life stress, it introduces many new stresses into one’s life which possibly become functional in maintaining the drug use and thereby creating further stress. References
Brown, G.W. and Harris, T. (1978) The Social Origins pression. Tavistock London.
of De-
Brown, G.W. and Harris, T. (1982) Fall-off in the reporting of Life Events. Social Psychiatry 17, 23-28. Dudley, D.L., Mules, J.E., Roszell, D.K., Glickfield, G. and Hague, W.H. (1976) Frequency and magnitude distrihution of life change in heroin and alcohol addicts. Int. .I. Addict. I 1, 977 - 987. Duncan, D.F. (1977) Life stress as a precurser to adolescent drug dependence. Int. J. Addict. 12, 1047 - 1056. Holmes, T. and Rahe, R. (1967) The Social Readjustment rating scale. J. Pschosom. Res. 11, 2133218. Kosten, T.R., Rounsaville. M.D., Herbert, D. and Kleber, M.D. (1986) A 2.5 year follow-up of depression, life crises and treatment effects on abstinence among opioid addicts. Arch. Gen. Psychiatry 43, 733-738. Litman, G.K., Eiser, J.R. and Rawson. N. (1977) Towards a typology of relapse. Drug Alcohol Depend. 2, 157- 162. Marlatt, G.A and Gordon, J.R. (1980). Determinants of relapse: Implications for the maintainance of behaviour change. In: Behavioural Medicine: Changing Health Life Styles (Davidson, P.O. and Davidson, S.M., eds), Brunner/Mazel, Marziali, E.A Subjective Stress 12,
New York. and Pilkonis, P.A. (1986) The Measurement of Response to Stressful Life Events. J. Hum. 5- 12.
Neff, J.A. and Husaini, B.A. (1982) Life Events, Drinking Patterns and Depressive Symptomatology. .J. Stud. Alcohol. 43, 301-318. O’Doherty, F. and Davies, J. (1987) Life events and Addiction A critical Review. Br. J. Addict. 82, 127- 137. Paykel, E.S. (1983) Methodological aspects of life events research. J. Psychosom. Res. 27. 341-352. Rahe, R.H. and Arthur, R.H. (1978) Life changes and illness studies. J. Hum. Stress 4, 3- 15. Saracon, I.G., Johnson, J.H. and Siegal, J.M. (1978) Assessing the impact of life changes development of the life experiences survey. J. Consult. Clin. Psychol. 46, 932-946. Schroeder, D.H. and Costa, P.T. (1984) The influence of life stress on physical illness: Substantive effects of methodological flaws? .J. Pers. Sot. Psychol. 46, 853 - 863.