Abstract

Background. Contrary to short-term use, long-term benzodiazepine use is undesirable. Nevertheless, its prevalence is high. To prevent long-term use, it is important to know which short-term users are at risk of becoming long-term users.

Objectives. The purpose of the present study was to identify patient-related factors of long-term versus short-term use of benzodiazepines.

Methods. A cross-sectional study was carried out in family practices among users of benzodiazepines with regard to DSM-IV diagnosis, coping and psychosocial characteristics,. In a multivariate logistic regression analysis, long-term use of benzodiazepines was the dependent variable.

Results. A total of 164 short-term and 158 long-term benzodiazepine users participated in the study. Having a DSM-IV disorder and psychiatric co-morbidity, being older, less educated, lonely and using more avoidance coping behaviour was associated with long-term use of benzodiazepines compared with short-term use.

Conclusion. The associations found point to possibilities to reduce long-term benzodiazepine use, for example if patients with these characteristics are treated with the alternatives to benzodiazepines or are monitored closely for a short period after being prescribing benzodiazepines.

Zandstra SM, van Rijswijk E, Rijnders CATh, van de Lisdonk EH, Bor JHJ, van Weel C and Zitman FG. Long-term benzodiazepine users in family practice: differences from short-term users in mental health, coping behaviour and psychological characteristics. Family Practice 2004; 21: 266–269.

Introduction

Guidelines advise benzodiazepines for short periods of time, on strict indications, but discourage their longer use because of side effects (dependency, cognitive impairments, falls and traffic accidents).1 Yet daily practice, with long-term use by 1–8% of the population, is quite different.2 The reasons for the discrepancy are not clear, but are probably related to sub-DSM-IV mental health problems and psychological distress that is not classified as formal ‘psychiatric’morbidity. This is at the core of the dispute on the long-term efficacy of benzodiazepines.

As not all those prescribed benzodiazepine will become long-term users, it is worthwhile to investigate which factors are associated with long-term use. For instance, comparing psychiatric morbidity, psychosocial circumstances and coping in short- and long-term users might identify such risk factors. A literature search in Medline, Psyclit and Pubmed on psychiatric diagnosis (DSM-IV or ICD-10) of long-term benzodiazepine users in family practices yielded only a few publications. Ohayon described DSM-IV mental disorders, yet did not specify the duration of benzodiazepine use.3 Vissers concluded in a comparison of long- versus short-term users that coping with life's problems, but not the experienced life problems themselves, were a determinant,4 but did not analyse psychiatric morbidity. Most studies compare long-term users with non-users,5,6 which is less relevant for the prediction of long-term use.

The aim of our study is to identify patient-related factors of long-term benzodiazepine use in family practice.

Material and methods

This cross-sectional study compared short- and long-term benzodiazepine users in family practices with regards to DSM-IV diagnosis, coping and psychosocial characteristics. A survey on psychopathology in the practice population of 32 family practices provided the study data.2 In the Dutch health care system, every citizen is registered with a family physician, and practice populations are therefore equivalent to the general population. Participating family physicians had to have a computerized patient and medication registration system, which currently is the case for >80% of practices. The 64 practices in the region fulfilling these criteria were approached, of which 32 participated. Details of the approach and representativeness of the study population have been described.2 Participating practices comprised more training practices (chi-square = 5.6; P = 0.02) and had fewer patients aged 45–74 but more aged 25–44 in their practice list compared with the Dutch average. The practice populations contained 4% short-term and 2% long-term benzodiazepine users, which is similar to other Dutch studies.7,8 To control for family physicians' work style, equal numbers of short- and long-term users were selected randomly from every practice.

Benzodiazepines were defined as the ATC-coded groups N05BA, CD, CF and CG. Benzodiazepine use was defined as one or more prescriptions recorded in the computerized family practice prescription files. For every registered patient, the prescription was translated into use/non-use for each day, using the issued daily dosage and number of tablets, for the previous 12 months.

Use was defined according to WHO criteria9,10 as short term (prescribed benzodiazepine for ≤90 days) and long term (prescribed for ≥180 days). Those patients who used benzodiazepines for 91–179 days were excluded.

The psychosocial characteristics of every patient were determined using the following (i) Social Support List for Interactions, 12-item version (SSL 12-I);11 (ii) Loneliness questionnaire;12 (iii) Brugha questionnaire;13 (iv) socio-economic status (SES), income and education level; and (v) three questions from the Ontario Health Survey14 on traumatic youth experiences: “were you placed in a children's home?”; “were you placed in a youth detention centre?”; and “were you raised in foster homes before the age of 16?”

The short version ‘Coping Inventory for Stressful Situations’ (CISS)15 was administerd to determine how the patients were coping and any psychiatric disorders were determined by the Schedules for Clinical Assessment in Neuropsychiatry version 2.1 (SCAN-2.1).16 A clinical psychologist, purpose-trained for this study, conducted all the interviews.

For all SCAN data, DSM-IV diagnoses were automatically computed by means of the SCAN-2.1 algorithm (Table 1).16

Table 1

Univariate results of psychosocial circumstances and DSM-IV diagnosis for short-term versus long-term users


 
Short term n = 122 (%) or mean
 
Long term n = 128 (%) or mean
 
t-testa, chi-squareb or Fisher'sc exact test
 
DSM-IV caseness   0.010b,* 
    No case 82 (67.2%) 66 (51.6%)  
    Case: one diagnostic category 30 (24.6%) 37 (28.9%)  
    Case: 2–8 diagnostic categories 10 (8.2%) 25 (19.5%)  
DSM-IV diagnostic category    
    Psycho-organic disorder 0 (0%) 1 (0.8%) 1.0c 
    Substance-related disorder 6 (4.9%) 14 (10.9%) 0.103c 
    Psychotic disorder 1 (0.8%) 2 (1.6%) 1.0c 
    Mood disorder 13 (10.7%) 23 (18.0%) 0.108c 
    Anxiety disorder 13 (10.7%) 24 (18.8) 0.077c 
    Eating disorder 1 (0.8%) 1 (0.8%) 0.273c 
    Somatoform disorder 7 (5.7%) 10 (7.8%) 0.618c 
    Sleeping disorder 13 (10.7%) 22 (16.4%) 0.149c 
    Dissociative disorder 2 (1.6%) 6 (4.7%) 0.282c 
    Other DSM-IV disorder 0 (0%) 2 (1.6%) 1.0c 
Gender    
    Female 79 (64.8%) 88 (68.8%) 0.6b 
Age 48.1 57.4 0.0000a,* 
SES/nett income/month   0.6b 
    < fl 1800 (< € 817) 23 (19.7%) 24 (20.2%)  
    fl 1800–fl 2500 (€ 817–€ 1134) 27 (23.1%) 36 (30.3%)  
    fl 2500–fl 4500 (€ 1134–€ 2042) 50 (42.7%) 42 (35.3%)  
    > fl 4500 (> € 2042) 17 (14.5%) 17 (14.3%)  
SES/education   .001b,* 
    Elementary school (lower level) 14 (11.5%) 39 (30.5%)  
    Secondary education (intermediate level) 60 (49.2%) 59 (46.1%)  
    Secondary education (higher level) 26 (21.3%) 17 (13.3%)  
    Higher education (university) 22 (18.0%) 13 (10.2%)  
Youth experience    
    Before the age of 16; placed in a children's home or youth detention centre or raised in foster homes 1 (0.8%) 6 (4.7%) 0.06b 
Brugha number of life events    
Total score range 0–12 0.53 0.55 0.9a 
Coping    
    Avoidance coping range 7–35 17.9 18.2 0.7a 
    Task-oriented range 7–35 23.8 22.5 0.1a 
    Emotion-oriented range 7–35 18.7 19.0 0.7a 
SSL-12-I    
    Total score of supportive interactions range 12–48 31.4 30.3 0.1a 
    Everyday support range 4–16 11.3 10.8 0.08a 
    Social support in problems range 4–16 9.5 9.4 0.7a 
    Esteem support range 4–16 10.7 10.2 0.1a 
Loneliness    
    Total score range 0–11 2.6 3.5 0.03a,* 

 
Short term n = 122 (%) or mean
 
Long term n = 128 (%) or mean
 
t-testa, chi-squareb or Fisher'sc exact test
 
DSM-IV caseness   0.010b,* 
    No case 82 (67.2%) 66 (51.6%)  
    Case: one diagnostic category 30 (24.6%) 37 (28.9%)  
    Case: 2–8 diagnostic categories 10 (8.2%) 25 (19.5%)  
DSM-IV diagnostic category    
    Psycho-organic disorder 0 (0%) 1 (0.8%) 1.0c 
    Substance-related disorder 6 (4.9%) 14 (10.9%) 0.103c 
    Psychotic disorder 1 (0.8%) 2 (1.6%) 1.0c 
    Mood disorder 13 (10.7%) 23 (18.0%) 0.108c 
    Anxiety disorder 13 (10.7%) 24 (18.8) 0.077c 
    Eating disorder 1 (0.8%) 1 (0.8%) 0.273c 
    Somatoform disorder 7 (5.7%) 10 (7.8%) 0.618c 
    Sleeping disorder 13 (10.7%) 22 (16.4%) 0.149c 
    Dissociative disorder 2 (1.6%) 6 (4.7%) 0.282c 
    Other DSM-IV disorder 0 (0%) 2 (1.6%) 1.0c 
Gender    
    Female 79 (64.8%) 88 (68.8%) 0.6b 
Age 48.1 57.4 0.0000a,* 
SES/nett income/month   0.6b 
    < fl 1800 (< € 817) 23 (19.7%) 24 (20.2%)  
    fl 1800–fl 2500 (€ 817–€ 1134) 27 (23.1%) 36 (30.3%)  
    fl 2500–fl 4500 (€ 1134–€ 2042) 50 (42.7%) 42 (35.3%)  
    > fl 4500 (> € 2042) 17 (14.5%) 17 (14.3%)  
SES/education   .001b,* 
    Elementary school (lower level) 14 (11.5%) 39 (30.5%)  
    Secondary education (intermediate level) 60 (49.2%) 59 (46.1%)  
    Secondary education (higher level) 26 (21.3%) 17 (13.3%)  
    Higher education (university) 22 (18.0%) 13 (10.2%)  
Youth experience    
    Before the age of 16; placed in a children's home or youth detention centre or raised in foster homes 1 (0.8%) 6 (4.7%) 0.06b 
Brugha number of life events    
Total score range 0–12 0.53 0.55 0.9a 
Coping    
    Avoidance coping range 7–35 17.9 18.2 0.7a 
    Task-oriented range 7–35 23.8 22.5 0.1a 
    Emotion-oriented range 7–35 18.7 19.0 0.7a 
SSL-12-I    
    Total score of supportive interactions range 12–48 31.4 30.3 0.1a 
    Everyday support range 4–16 11.3 10.8 0.08a 
    Social support in problems range 4–16 9.5 9.4 0.7a 
    Esteem support range 4–16 10.7 10.2 0.1a 
Loneliness    
    Total score range 0–11 2.6 3.5 0.03a,* 
*

Statistically significant level 0.05.

First, univariate analyses of the caseness (presence of at least one specific diagnosis), as well as the total scores of the questionnaires were carried out with long-term benzodiazepine use as dependent variable. Secondly, all independent variables were introduced into a multivariate logistic regression analysis, followed by a manual backward procedure. All analyses were performed with the SAS statistical software package.

Results

The study recruited 164 short-term and 158 long-term benzodiazepine users. In 72 cases, no SCAN interview could be scheduled in time due to logistical reasons, leaving 122 short-term and 128 long-term users with complete data to study.

Comparing study patients with all short- and long-term benzodiazepine users in the practices, short-term users were older (participants > 45 years 63.4% versus 53.8%), but otherwise no differences were found.

Analysis (Table 1)

Caseness, in particular with morbidity diagnosed in two or more diagnostic categories, was statistically significantly associated with long-term benzodiazepine use. No statistically significant association was found for specific diagnoses. Of the psychosocial characteristics, a lower level of education, older age and loneliness were statistically significantly associated with long-term benzodiazepine use. No statistically significant differences were found for the other independent variables.

The multivariate logistic regression showed that the one diagnostic category [odds ratio (OR) 2.38; confidence interval (CI) 1.21–4.69], 2–8 diagnostic categories (OR 3.61; CI 1.40–9.26), avoidance coping style (OR per unit 1.05; CI 1.00–1.10), loneliness (OR per unit 1.09; CI 0.99–1.19), lower level of education (OR 2.00; CI 0.90–4.43) and older age (OR per unit 1.07; CI 1.04–1.09) were related to long-term use. The goodness of fit of the model was satisfying (Nagelkerke R2 = 0.2750). Interactions and confounders were not found between the different variables.

Discussion

Psychiatric (co-) morbidity, an avoidance coping behaviour, loneliness, older age and lower education were associated with long-term benzodiazepine use compared with short-term use. The cross-sectional design implies that we cannot analyse cause–consequence relationships. However, comparing long-term with short-term users rather than with non-users provides clear suggestions of factors related to long-term use. As we controlled for the influence of the prescribing family physician on long-term use, our findings represent true patient factors.

Although psychiatric morbidity was more strongly associated with long-term than with short-term use (in about half versus a third of the cases), the presence of a ‘formal’ DSM-IV diagnosis could not in itself completely explain long-term use.

Psychological characteristics and coping were inter-related with psychiatric morbidity, which has been described.17,18 Nevertheless, loneliness, education level and avoidance coping were related in their own right to long-term benzodiazepine use.

Training practices were over-represented, which may have resulted in a more restricted benzodiazepine prescribing routine in the study sample.

Usually, psychological characteristics, coping and co-morbidity are largely ignored in guidelines and recommendations of benzodiazepine use, but our findings point out that this characterizes long-term users. There is a lack of evidence of the (long-term) effectiveness of benzodiazepines for these features that may underline the complexity and vulnerability of these patients. In the absence of further evidence of benzodiazepine effectiveness, we recommend caution in starting prescribing benzodiazepines, and, when prescribing, regular evaluation of the effects, in particular when prescribed in the absence of a psychiatric diagnosis.

We thank the participating practices and patients for their co-operation during this study. This work was funded by a grant from the Prevention Fund [now the Council for Medical and Health Research (ZonMW)].

References

1
Leipzig RM, Cumming RG, Tinetti ME. Drugs and falls in older people: a systematic review and meta-analysis: I. Psychotropic drugs.
J Am Geriatr Soc
 
1999
;
47
:
30
–39.
2
Zandstra S, Furer J, van de Lisdonk E, van 'tHof M, Bor J, van Weel C, Zitman F. Different study criteria affect the prevalence of benzodiazepine use.
Soc Psychiatry Psychiatr Epidemiol
 
2002
;
37
:
139
–144.
3
Ohayon MM, Caulet M, Priest RG, Guilleminault C. Psychotropic medication consumption patterns in the UK general population.
J Clin Epidemiol
 
1998
;
51
:
273
–283.
4
Vissers FHJA. Gebruik van slaap- en kalmeringsmiddelen in het dagelijks leven. UMC Maastricht, The Netherlands;
1998
.
5
Salinsky JV, Dore CJ. Characteristics of long term benzodiazepine users in general practice.
J R Coll Gen Pract
 
1987
;
37
:
202
–204.
6
Simpson RJ, Power KG, Wallace LA, Butcher MH, Swanson V, Simpson EC. Controlled comparison of the characteristics of long-term benzodiazepine users in general practice.
Br J Gen Pract
 
1990
;
40
:
22
–26.
7
Van der Waals F. Sex Differences Among Recipients of Benzodiazepines in Dutch General Practice. Amsterdam: Academic Medical Centre, University of Amsterdam, Department of Women's Health;
1995
.
8
Van Hulten R. Blue Boy–Why Not? Studies of Benzodiazepine Use in a Dutch Community. PhD Thesis, University of Utrecht;
1998
.
9
Committee on the Safety of Medicines (CSM). Benzodiazepines, dependence and withdrawal symptoms. Currents Problems
1988
; 21.
10
Salzman C. The APA Task Force report on benzodiazepine dependence, toxicity, and abuse.
Am J Psychiatry
 
1991
;
148
:
151
–152.
11
Van Eijk LM, Kempen GI, Van Sonderen FL. [A short scale for measuring social support in the elderly: the SSL12-I].
Tijdschr Gerontol Geriatr
 
1994
;
25
:
192
–196.
12
Van Tilburg TG, De Jong GJ. [Reference standards for the loneliness scale].
Tijdschr Gerontol Geriatr
 
1999
;
30
:
158
–163.
13
Brugha TS, Cragg D. The list of threatening experiences: the reliability and validity of a brief life events questionnaire.
Acta Psychiatr Scand
 
1990
;
82
:
77
–81.
14
Ontario Ministry of Health. Microdata Manual: User's Guide. Toronto, Ontario: Queen's Printer for Ontario;
1992
.
15
Endler NS, Parker JDA. Assessment of multidimensional coping: task, emotion, and avoidance strategies.
Psychol Assess
 
1994
;
6
:
50
–60.
16
Rijnders CA, Van den Berg JF, Hodiamont PP, Nienhuis FJ, Furer JW, Mulder J, Giel R. Psychometric properties of the schedules for clinical assessment in neuropsychiatry (SCAN-2.1).
Soc Psychiatry Psychiatr Epidemiol
 
2000
;
35
:
348
–352.
17
Endler NS, Parker JDA, Summerfeldt LJ. Coping with health problems: developing a reliable and valid multidimensional measure.
Psychol Assess
 
1998
;
10
:
195
–205.
18
Endler NS, Parker JDA. The multidimensional assessment of coping: concepts, issues, and measurement. In: Deary I, Van Heck G, Nowack W (eds). Personality psychology in Europe, Tilburg: Tilburg University Press;
1993
: 309–319.

Author notes

Department of Family Practice and aDepartment of Social Medicine, University Medical Centre St Radboud Nijmegen and bDepartment of Psychiatry, University Medical Center Leiden, The Netherlands