- Split View
-
Views
-
Cite
Cite
WJMJ Gorgels, RC Oude Voshaar, AJJ Mol, EH van de Lisdonk, AJLM van Balkom, MHM Breteler, HJM van den Hoogen, J Mulder, FG Zitman, Predictors of discontinuation of benzodiazepine prescription after sending a letter to long-term benzodiazepine users in family practice, Family Practice, Volume 23, Issue 1, February 2006, Pages 65–72, https://doi.org/10.1093/fampra/cmi065
- Share Icon Share
Abstract
Background. Predictors of benzodiazepine discontinuation after sending a discontinuation letter by the family practitioner have not been established sufficiently.
Objective. To identify predictors of short- and long-term discontinuation of benzodiazepine use and relapse in use after a minimal intervention with a discontinuation letter followed by an offer for an evaluation consultation.
Methods. Predictors of benzodiazepine discontinuation and relapse in use were studied by logistic regression analysis and survival analysis within a family practice population of long-term benzodiazepine users (n = 1707) addressed by a discontinuation letter and followed for 21 months.
Results. A lower baseline prescription, a shorter duration of use, male gender and use of an agent with a half-life time <24 hours were predictive of complete discontinuation in the short (6 months) and long term (21 months). Multiple agent use at baseline, use of antidepressants at 6 months and benzodiazepine type (anxiolytic/hypnotic) at baseline predicted relapse. Attendance at an evaluation consultation 3 months after the letter was sent was not predictive of discontinuation or relapse.
Conclusions. Amount of baseline use and duration of use are the main determinative characteristics of successful discontinuation. The discontinuation letter intervention is suitable for use with a broad group of long-term benzodiazepine users in family practice and can be used as a first step within a stepped care approach to decrease long-term benzodiazepine use.
Gorgels WJMJ, Oude Voshaar RC, Mol AJJ, van de Lisdonk EH, van Balkom AJLM, Breteler MHM, van den Hoogen HJM, Mulder J and Zitman FG. Predictors of discontinuation of benzodiazepine prescription after sending a letter to long-term benzodiazepine users in family practice. Family Practice 2006; 23: 65–72.
Introduction
National and international guidelines advise prescription of benzodiazepines only for short periods. Nevertheless, long-term benzodiazepine use is highly prevalent in western communities. A strategy, advised by the British Mental Health Foundation, to bring down the number of existing long-term users is a stepped care model in which a minimal intervention (the first step) is followed by a more intensive approach with a taper scheme with or without psychotherapeutical support (second step).1 Minimal interventions as well as intensive interventions were effective.2–7
Knowledge of factors that predict successful quitting of benzodiazepines or relapse after minimal interventions may contribute to understanding the mechanisms of benzodiazepine continuation and may enhance the efficacy of the stepped care strategy, for instance by identifying subgroups that do not respond to the minimal intervention.
A lower baseline benzodiazepine use predicted benzodiazepine reduction and endured abstinence following a minimal intervention with a discontinuation letter sent by the family practitioner to long-term users.3,8,9 Furthermore, subjects who discussed the letter with their family practitioner and those with shorter durations of use were more likely to reduce their use.10 No clinically significant association was observed for age, gender, number of different benzodiazepines at baseline, duration of benzodiazepine use, and benzodiazepine type (anxiolytic or hypnotic use).3,8 The latter studies, however, lacked power.
More intensive treatments of insomnia-related, long-term benzodiazepine use by gradual tapering and cognitive behavioural therapy also showed a lower initial dosage to predict quitting of benzodiazepines.5,6 Predictor studies of successful taper outcome in secondary care populations, however, were inconclusive. Associations with quitting were reported for a lower initial dose, males and females, younger and older subjects and for the presence of psychopathology and personality disturbances.11–16
As the major goals of benzodiazepine discontinuation interventions are lasting changes it is important to establish predictors of long-term abstinence. Within a large family practice-based stepped care intervention study (the Benzoredux study) we studied the predictors of short-term and long-term discontinuation of benzodiazepine use following the minimal intervention of a discontinuation letter and an evaluation consultation offer.4
Methods
This prospective intervention study in family practice was carried out between August 1998 and December 2001.4
Study population
Long-term benzodiazepine users were identified by performing a computerised search for prescriptions of benzodiazepines. We searched for the ATC (Anatomical Therapeutical Classification) codes of benzodiazepines (anxiolytics (N05BA**), hypnotics (N05CD**) and the anti-epileptic clonazepam (N03AE01) in the prescription administration of 27 family practices participating in a benzodiazepine reduction program (the Benzoredux study), covering the primary care of approximately 110 000 people.4,17 The practices were chosen to have a large variety in location [11 urban (Amsterdam, Nijmegen and Almere) and 16 rural (villages nearby Nijmegen)] and organisation type (4 health centres, 10 group practices and 13 solo practices). To fulfil the long-term user definition, subjects had to meet the following criteria: (i) having received benzodiazepine prescriptions for more than 3 months, and (ii) within the last 3 months before inclusion in the study having received benzodiazepine prescription in an amount sufficient for 60 days of use according to the family practitioner prescription rules. Application of these criteria resulted in 2425 long-term benzodiazepine users available for the analysis.
Intervention
The intervention consisted of a discontinuation letter sent by the family practitioner followed by an offer to arrange an evaluation consultation with the family practitioner after 12 weeks (evaluation consultation). In the letter, which was constructed after a letter used by Cormack et al.3, the subjects were advised to gradually discontinue benzodiazepine use by themselves. Of the 2425 long-term benzodiazepine users, 1707 subjects were addressed by the discontinuation letter and 718 were excluded (Fig. 1). Of the subjects addressed by the discontinuation letter 1152 (67%) accepted the invitation and arranged an evaluation consultation with the family practitioner at a median of 14 weeks from the date of the sending of the discontinuation letter.
Assessments
Benzodiazepine prescription data (recorded prescription) were extracted from the electronic medical dossier (EMD) as used by most family practitioners in The Netherlands. For each benzodiazepine prescription the number of prescribed daily dosages (PDDs) was calculated using benzodiazepine equivalence rules by Zitman and Couvee.18 Subsequently, for each subject all PDDs were summed within 3-month periods (each 3 month period consisted of 91 days exactly). Baseline recorded prescription was defined as the sum of all PDDs of benzodiazepine prescriptions issued within the 3 months immediately before the mailing of the discontinuation letter. The discontinuation letter sending date itself was moment zero and the follow up was until 21 months after the discontinuation letter was sent (divided in 7 periods of 3 months).
Short-term quitters were defined as subjects with no benzodiazepine prescription in the 3-month period between 4–6 months after the discontinuation letter was sent. Relapse was defined as any new issue of a benzodiazepine prescription until the end of follow up (21 months) in short-term quitters. Long-term quitters were defined as those who received no benzodiazepine prescription in the full period of 4–21 months after the discontinuation letter was sent.
Other patient characteristics extracted from the family practitioner EMD were: gender, age, and health insurance status (private or National Health Service). Antidepressant use at baseline was defined as any prescription of an antidepressant (ATC code N06A) in the baseline three-month period. Antidepressant use at 6 months was defined as any prescription of an antidepressant in the three-month period of 4–6 months after the letter. Self-reported duration of use was assessed at the evaluation consultation. As a consequence it was available only for those subjects actually appearing at this evaluation consultation (evaluation visit: n = 1152; duration of use data available: n = 1105; duration of use data missing: n = 47).
Analysis
The analysis was performed on all subjects with complete follow up for 6 months (short-term analysis, n = 1678) and 21 months (long-term analysis, n = 1601). Logistic regression analysis was applied to identify predictors of quitting. The dependent variables were short-term quitter (yes/no) and long-term quitter (yes/no). With respect to relapse our scope was different as we were also interested in the time to relapse. Therefore a Cox-regression analysis was used to analyse predictors of relapse within the group of short-term quitters, with time to relapse as the dependent variable.
The following independent variables were entered in the regression procedures:
age;
gender;
health insurance type;
amount of baseline benzodiazepine prescription;
duration of use;
multiple agent use [subjects were classified as single users (using only a single type of benzodiazepine in the baseline period) and multiple agent users (using more than one type of benzodiazepine within the baseline period). Multiple agent use may either result from using two or more different agents at the same time, or subsequently within the baseline period];
ATC category N05CD (hypnotic) or N05BA (anxiolytic) [N.B. six subjects with use of N03AE01 (clonazepam) were included in the anxiolytics group];
half-life time of the agent (or its main metabolites) among single users (t1/2 <24 hours or ≥24 hours) (A half life of ≥24 hours was considered for chloordiazepoxide, clobazam, clonazepam, clorazepinic acid, diazepam, flurazepam, ketazolam, medazepam, nordazepam and prazepam. A half-life <24 hours was considered for alprazolam, bromazepam, brotizolam, flunitrazepam, loprazolam, lorazepam, lormetazepam, midazolam, nitrazepam, oxazepam and temazepam);19
evaluation consultation attendance;
baseline use of antidepressants (in logistic regression); and
use of antidepressants at 6 months (in Cox-regression).
We choose to categorize age, duration of use and baseline benzodiazepine prescription in three categories as we could not assume linearity between the score on the variable and the chance to quit.
Variables that had a P-value of ≤0.15 in univariate analysis were entered in the multivariate model. Using the SAS package, both a stepwise forward procedure and a backward procedure were carried out, resulting in similar explanatory models.20 Two sided P-values (α = 0.05) and 95% confidence intervals (CIs) were used in all analysis.
Some restrictions had to be dealt with in the analysis. First, as duration of use was only available for a subgroup of subjects (n = 1105), its relationship with quitting and reduction was evaluated within this subgroup. Furthermore, we choose to perform separate analysis in single users to estimate the odds ratios of ‘ATC category’ and ‘half-life time’. This resulted in three regression models: for those with full follow-up, for those with duration of use data, and for those with single use. To check possible confounding by practice, we entered all practices as dummy variables in the terminal models. As this did not change any of the predictor estimates this extensive model will not be reported.
Results
Baseline characteristics (Table 1)
The discontinuation letter group had a mean age of 63 years and a high percentage of females and National Health insured subjects (Table 1). Mean baseline prescription was 71 PDD in 3 months, which corresponds to an average defined daily dose (DDD) of 0.8, with a median of 0.5 DDD. The most prescribed benzodiazepines were oxazepam, temazepam and diazepam. Sixteen percent of the users had prescriptions of more than one benzodiazepine at baseline (multiple agent users). The subgroup of subjects for which duration of use data were available was not different in terms of baseline characteristics from the intervention group as a whole. Overall there was a long reported duration of use: 4% had used benzodiazepines for less than 12 months and the median duration of use was 10 years.
. | Discontinuation letter intervention sample (n = 1707) . | |
---|---|---|
Demography | ||
Gender, % female | 73 | |
Age (yrs), mean (SD) | 63 (14) | |
Health insurance, % NHS | 78 | |
Benzodiazepine usage characteristics | ||
Baseline benzodiazepine PDD, mean (SD) | 71 (78) | |
Subjects in subgroups of baseline PDD: | ||
≤65 PDD, n (%) | 1096 (64%) | |
>65–≤130 PDD, n (%) | 413 (24) | |
>130 PDD, n (%) | 198 (12) | |
≥2 benzodiazepines at baseline, n (%) | 280 (16) | |
Single benzodiazepine at baseline, n (%) | 1427 (84) | |
Benzodiazepine type at baseline | ||
Oxazepam, % | 32 | |
Temazepam, % | 16 | |
Diazepam, % | 11 | |
Other, % | 25 | |
ATC group | ||
N05BAb (anxiolytic), % | 65 | |
N05CD (hypnotic), % | 35 | |
Half life time | ||
<24 hours, % | 73 | |
≥24 hours, % | 27 | |
Duration of benzodiazepine use (n = 1105) in months, mean (SD) | 149 (113) | |
Subjects in subgroups of duration of use categories | ||
3–≤24 months, n (%) | 100 (9) | |
>24–≤120 months, n (%) | 520 (47) | |
>120 months, n (%) | 485 (44) | |
Baseline use of antidepressants, n (%) | 251 (15) |
. | Discontinuation letter intervention sample (n = 1707) . | |
---|---|---|
Demography | ||
Gender, % female | 73 | |
Age (yrs), mean (SD) | 63 (14) | |
Health insurance, % NHS | 78 | |
Benzodiazepine usage characteristics | ||
Baseline benzodiazepine PDD, mean (SD) | 71 (78) | |
Subjects in subgroups of baseline PDD: | ||
≤65 PDD, n (%) | 1096 (64%) | |
>65–≤130 PDD, n (%) | 413 (24) | |
>130 PDD, n (%) | 198 (12) | |
≥2 benzodiazepines at baseline, n (%) | 280 (16) | |
Single benzodiazepine at baseline, n (%) | 1427 (84) | |
Benzodiazepine type at baseline | ||
Oxazepam, % | 32 | |
Temazepam, % | 16 | |
Diazepam, % | 11 | |
Other, % | 25 | |
ATC group | ||
N05BAb (anxiolytic), % | 65 | |
N05CD (hypnotic), % | 35 | |
Half life time | ||
<24 hours, % | 73 | |
≥24 hours, % | 27 | |
Duration of benzodiazepine use (n = 1105) in months, mean (SD) | 149 (113) | |
Subjects in subgroups of duration of use categories | ||
3–≤24 months, n (%) | 100 (9) | |
>24–≤120 months, n (%) | 520 (47) | |
>120 months, n (%) | 485 (44) | |
Baseline use of antidepressants, n (%) | 251 (15) |
The baseline period is defined as the last 3 months before start of the intervention (sending the letter).
6 subjects with N03AE01 (clonazepam) were included in the N05BA group (anxiolytics).
. | Discontinuation letter intervention sample (n = 1707) . | |
---|---|---|
Demography | ||
Gender, % female | 73 | |
Age (yrs), mean (SD) | 63 (14) | |
Health insurance, % NHS | 78 | |
Benzodiazepine usage characteristics | ||
Baseline benzodiazepine PDD, mean (SD) | 71 (78) | |
Subjects in subgroups of baseline PDD: | ||
≤65 PDD, n (%) | 1096 (64%) | |
>65–≤130 PDD, n (%) | 413 (24) | |
>130 PDD, n (%) | 198 (12) | |
≥2 benzodiazepines at baseline, n (%) | 280 (16) | |
Single benzodiazepine at baseline, n (%) | 1427 (84) | |
Benzodiazepine type at baseline | ||
Oxazepam, % | 32 | |
Temazepam, % | 16 | |
Diazepam, % | 11 | |
Other, % | 25 | |
ATC group | ||
N05BAb (anxiolytic), % | 65 | |
N05CD (hypnotic), % | 35 | |
Half life time | ||
<24 hours, % | 73 | |
≥24 hours, % | 27 | |
Duration of benzodiazepine use (n = 1105) in months, mean (SD) | 149 (113) | |
Subjects in subgroups of duration of use categories | ||
3–≤24 months, n (%) | 100 (9) | |
>24–≤120 months, n (%) | 520 (47) | |
>120 months, n (%) | 485 (44) | |
Baseline use of antidepressants, n (%) | 251 (15) |
. | Discontinuation letter intervention sample (n = 1707) . | |
---|---|---|
Demography | ||
Gender, % female | 73 | |
Age (yrs), mean (SD) | 63 (14) | |
Health insurance, % NHS | 78 | |
Benzodiazepine usage characteristics | ||
Baseline benzodiazepine PDD, mean (SD) | 71 (78) | |
Subjects in subgroups of baseline PDD: | ||
≤65 PDD, n (%) | 1096 (64%) | |
>65–≤130 PDD, n (%) | 413 (24) | |
>130 PDD, n (%) | 198 (12) | |
≥2 benzodiazepines at baseline, n (%) | 280 (16) | |
Single benzodiazepine at baseline, n (%) | 1427 (84) | |
Benzodiazepine type at baseline | ||
Oxazepam, % | 32 | |
Temazepam, % | 16 | |
Diazepam, % | 11 | |
Other, % | 25 | |
ATC group | ||
N05BAb (anxiolytic), % | 65 | |
N05CD (hypnotic), % | 35 | |
Half life time | ||
<24 hours, % | 73 | |
≥24 hours, % | 27 | |
Duration of benzodiazepine use (n = 1105) in months, mean (SD) | 149 (113) | |
Subjects in subgroups of duration of use categories | ||
3–≤24 months, n (%) | 100 (9) | |
>24–≤120 months, n (%) | 520 (47) | |
>120 months, n (%) | 485 (44) | |
Baseline use of antidepressants, n (%) | 251 (15) |
The baseline period is defined as the last 3 months before start of the intervention (sending the letter).
6 subjects with N03AE01 (clonazepam) were included in the N05BA group (anxiolytics).
Predictors of discontinuation (Table 2)
In the whole group, 468 (28%, n = 1678) were short-term quitters and 233 (15%, n = 1601) were long-term quitters. Among the subjects for which duration of use data were available the numbers were 311 (28%) short-term quitters and 157 (15%) long-term quitters.
Female gender, a higher baseline prescription and a longer duration of use were associated with a lower probability of quitting benzodiazepines in both the short- and long-term (Table 2). In univariate analysis, multiple agent users had a lower chance to quit. When adjustment was made for baseline prescription the odds ratio of multiple agent use was no longer significant in short-term evaluation. Concerning long-term quitting, however, multiple agent use remained an independent predictor in the multivariate model. Categorisation by ATC category as anxiolytics and hypnotics was not predictive. Half-life time, however, was a statistically significant independent predictor in multivariate analysis in the short- as well as long-term: subjects using agents with a longer half-life time were less likely to quit their benzodiazepine use. Table 3 shows the relationship of the two strongest predictors concerning short-term discontinuation. It shows that of the subjects with a short duration of use, 86% were classified in the low benzodiazepine prescription category. Only 1 subject from the highest category of baseline use and duration of use was able to quit. This person relapsed during follow up.
Variables . | Short-term (ST) evaluation (6 months) n = 1678 ST quitters n = 468 . | . | . | Long-term (LT) evaluation (21 months) n = 1601 LT quitters n = 233 . | . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Univariate . | . | Multivariate . | . | Multivariate . | . | ||||||
. | OR (95% CI) . | P-value . | Adj. OR (95% CI) . | P-value . | Adj. OR (95% CI) . | P-value . | ||||||
Intercept | ||||||||||||
Gender (F = 1, M = 0) | 0.7 (0.6–0.9) | 0.004 | 0.6 (0.5–0.8) | <0.001 | 0.6 (0.4–0.8) | <0.001 | ||||||
Age (yr) | ||||||||||||
(reference) <50 | – | – | – | |||||||||
50–<75 | 1.2 (0.9–1.6) | 0.16 | – | – | ||||||||
≥75 | 1.0 (0.7–1.4) | 0.90 | – | – | ||||||||
Health insurance (private = 1, NHS = 0) | 1.2 (1.0–1.6) | 0.12 | – | – | ||||||||
Baseline BZ prescription | ||||||||||||
(reference) ≤65 PDD | – | – | ||||||||||
>65–≤130 PDD | 0.3 (0.2–0.4) | <0.001 | 0.3 (0.2–0.4) | <0.001 | 0.3 (0.2–0.4) | <0.001 | ||||||
>130 PDD | 0.2 (0.1–0.3) | <0.001 | 0.2 (0.1–0.3) | <0.001 | 0.2 (0.1–0.4) | <0.001 | ||||||
≥2 benzodiazepines at baseline | 0.5 (0.3–0.7) | <0.001 | – | 0.5 (0.3–0.9) | −0.03 | |||||||
Evaluation consultation (yes/no) | 1.1 (0.9–1.4) | 0.49 | – | |||||||||
Baseline use of antidepressants (yes/no) | 1.0 (0.7–1.3) | 0.77 | – | – | ||||||||
Duration of usea | ||||||||||||
(reference) 3–≤24 months | – | – | – | |||||||||
>24–≤120 months | 0.4 (0.3–0.6) | <0.001 | 0.5 (0.3–0.8) | 0.001 | 0.5 (0.3–0.8) | 0.003 | ||||||
>120 months | 0.2 (0.1–0.3) | <0.001 | 0.2 (0.1–0.4) | <0.001 | 0.2 (0.1–0.4) | <0.001 | ||||||
ATCb | ||||||||||||
(reference, anxiolytics) N05BA | – | – | – | |||||||||
(hypnotics) N05CD | 0.8 (0.6–1.0) | 0.10 | – | – | ||||||||
Half-life timeb | ||||||||||||
(reference) <24 hours | – | – | – | |||||||||
≥24 hours | 0.7 (0.5–0.9) | 0.003 | 0.6 (0.5–0.8) | 0.001 | 0.6 (0.4–0.9) | 0.01 |
Variables . | Short-term (ST) evaluation (6 months) n = 1678 ST quitters n = 468 . | . | . | Long-term (LT) evaluation (21 months) n = 1601 LT quitters n = 233 . | . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Univariate . | . | Multivariate . | . | Multivariate . | . | ||||||
. | OR (95% CI) . | P-value . | Adj. OR (95% CI) . | P-value . | Adj. OR (95% CI) . | P-value . | ||||||
Intercept | ||||||||||||
Gender (F = 1, M = 0) | 0.7 (0.6–0.9) | 0.004 | 0.6 (0.5–0.8) | <0.001 | 0.6 (0.4–0.8) | <0.001 | ||||||
Age (yr) | ||||||||||||
(reference) <50 | – | – | – | |||||||||
50–<75 | 1.2 (0.9–1.6) | 0.16 | – | – | ||||||||
≥75 | 1.0 (0.7–1.4) | 0.90 | – | – | ||||||||
Health insurance (private = 1, NHS = 0) | 1.2 (1.0–1.6) | 0.12 | – | – | ||||||||
Baseline BZ prescription | ||||||||||||
(reference) ≤65 PDD | – | – | ||||||||||
>65–≤130 PDD | 0.3 (0.2–0.4) | <0.001 | 0.3 (0.2–0.4) | <0.001 | 0.3 (0.2–0.4) | <0.001 | ||||||
>130 PDD | 0.2 (0.1–0.3) | <0.001 | 0.2 (0.1–0.3) | <0.001 | 0.2 (0.1–0.4) | <0.001 | ||||||
≥2 benzodiazepines at baseline | 0.5 (0.3–0.7) | <0.001 | – | 0.5 (0.3–0.9) | −0.03 | |||||||
Evaluation consultation (yes/no) | 1.1 (0.9–1.4) | 0.49 | – | |||||||||
Baseline use of antidepressants (yes/no) | 1.0 (0.7–1.3) | 0.77 | – | – | ||||||||
Duration of usea | ||||||||||||
(reference) 3–≤24 months | – | – | – | |||||||||
>24–≤120 months | 0.4 (0.3–0.6) | <0.001 | 0.5 (0.3–0.8) | 0.001 | 0.5 (0.3–0.8) | 0.003 | ||||||
>120 months | 0.2 (0.1–0.3) | <0.001 | 0.2 (0.1–0.4) | <0.001 | 0.2 (0.1–0.4) | <0.001 | ||||||
ATCb | ||||||||||||
(reference, anxiolytics) N05BA | – | – | – | |||||||||
(hypnotics) N05CD | 0.8 (0.6–1.0) | 0.10 | – | – | ||||||||
Half-life timeb | ||||||||||||
(reference) <24 hours | – | – | – | |||||||||
≥24 hours | 0.7 (0.5–0.9) | 0.003 | 0.6 (0.5–0.8) | 0.001 | 0.6 (0.4–0.9) | 0.01 |
Analysis in the subgroup of subjects of with duration of use was available (n = 1098 (short-term analysis, ST quitters = 311) and n = 1050 (Long-term analysis, LT quitters = 157), respectively). In this subgroup analysis, in the multivariate model, the estimates of duration of use were adjusted for gender, age, health insurance, baseline BZ prescription and ≥2 BZ at baseline. There were no clinically important changes in the estimates of gender, baseline prescription and ‘≥2 benzodiazepines at baseline’ compared with the estimates from the overall model.
Analysis in the subgroup of subjects with baseline prescription of a single benzodiazepine agent only [n = 1404 (short-term analysis, ST quitters = 421) and n = 1343 (Long-term analysis, LT quitters = 218)].
In the multivariate analysis ATC and half-life time were adjusted for gender, age, health insurance, baseline BZ prescription and duration of use. There were no clinically important changes in the estimates of gender and baseline prescription compared with the estimates from the overall model.
Variables . | Short-term (ST) evaluation (6 months) n = 1678 ST quitters n = 468 . | . | . | Long-term (LT) evaluation (21 months) n = 1601 LT quitters n = 233 . | . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Univariate . | . | Multivariate . | . | Multivariate . | . | ||||||
. | OR (95% CI) . | P-value . | Adj. OR (95% CI) . | P-value . | Adj. OR (95% CI) . | P-value . | ||||||
Intercept | ||||||||||||
Gender (F = 1, M = 0) | 0.7 (0.6–0.9) | 0.004 | 0.6 (0.5–0.8) | <0.001 | 0.6 (0.4–0.8) | <0.001 | ||||||
Age (yr) | ||||||||||||
(reference) <50 | – | – | – | |||||||||
50–<75 | 1.2 (0.9–1.6) | 0.16 | – | – | ||||||||
≥75 | 1.0 (0.7–1.4) | 0.90 | – | – | ||||||||
Health insurance (private = 1, NHS = 0) | 1.2 (1.0–1.6) | 0.12 | – | – | ||||||||
Baseline BZ prescription | ||||||||||||
(reference) ≤65 PDD | – | – | ||||||||||
>65–≤130 PDD | 0.3 (0.2–0.4) | <0.001 | 0.3 (0.2–0.4) | <0.001 | 0.3 (0.2–0.4) | <0.001 | ||||||
>130 PDD | 0.2 (0.1–0.3) | <0.001 | 0.2 (0.1–0.3) | <0.001 | 0.2 (0.1–0.4) | <0.001 | ||||||
≥2 benzodiazepines at baseline | 0.5 (0.3–0.7) | <0.001 | – | 0.5 (0.3–0.9) | −0.03 | |||||||
Evaluation consultation (yes/no) | 1.1 (0.9–1.4) | 0.49 | – | |||||||||
Baseline use of antidepressants (yes/no) | 1.0 (0.7–1.3) | 0.77 | – | – | ||||||||
Duration of usea | ||||||||||||
(reference) 3–≤24 months | – | – | – | |||||||||
>24–≤120 months | 0.4 (0.3–0.6) | <0.001 | 0.5 (0.3–0.8) | 0.001 | 0.5 (0.3–0.8) | 0.003 | ||||||
>120 months | 0.2 (0.1–0.3) | <0.001 | 0.2 (0.1–0.4) | <0.001 | 0.2 (0.1–0.4) | <0.001 | ||||||
ATCb | ||||||||||||
(reference, anxiolytics) N05BA | – | – | – | |||||||||
(hypnotics) N05CD | 0.8 (0.6–1.0) | 0.10 | – | – | ||||||||
Half-life timeb | ||||||||||||
(reference) <24 hours | – | – | – | |||||||||
≥24 hours | 0.7 (0.5–0.9) | 0.003 | 0.6 (0.5–0.8) | 0.001 | 0.6 (0.4–0.9) | 0.01 |
Variables . | Short-term (ST) evaluation (6 months) n = 1678 ST quitters n = 468 . | . | . | Long-term (LT) evaluation (21 months) n = 1601 LT quitters n = 233 . | . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Univariate . | . | Multivariate . | . | Multivariate . | . | ||||||
. | OR (95% CI) . | P-value . | Adj. OR (95% CI) . | P-value . | Adj. OR (95% CI) . | P-value . | ||||||
Intercept | ||||||||||||
Gender (F = 1, M = 0) | 0.7 (0.6–0.9) | 0.004 | 0.6 (0.5–0.8) | <0.001 | 0.6 (0.4–0.8) | <0.001 | ||||||
Age (yr) | ||||||||||||
(reference) <50 | – | – | – | |||||||||
50–<75 | 1.2 (0.9–1.6) | 0.16 | – | – | ||||||||
≥75 | 1.0 (0.7–1.4) | 0.90 | – | – | ||||||||
Health insurance (private = 1, NHS = 0) | 1.2 (1.0–1.6) | 0.12 | – | – | ||||||||
Baseline BZ prescription | ||||||||||||
(reference) ≤65 PDD | – | – | ||||||||||
>65–≤130 PDD | 0.3 (0.2–0.4) | <0.001 | 0.3 (0.2–0.4) | <0.001 | 0.3 (0.2–0.4) | <0.001 | ||||||
>130 PDD | 0.2 (0.1–0.3) | <0.001 | 0.2 (0.1–0.3) | <0.001 | 0.2 (0.1–0.4) | <0.001 | ||||||
≥2 benzodiazepines at baseline | 0.5 (0.3–0.7) | <0.001 | – | 0.5 (0.3–0.9) | −0.03 | |||||||
Evaluation consultation (yes/no) | 1.1 (0.9–1.4) | 0.49 | – | |||||||||
Baseline use of antidepressants (yes/no) | 1.0 (0.7–1.3) | 0.77 | – | – | ||||||||
Duration of usea | ||||||||||||
(reference) 3–≤24 months | – | – | – | |||||||||
>24–≤120 months | 0.4 (0.3–0.6) | <0.001 | 0.5 (0.3–0.8) | 0.001 | 0.5 (0.3–0.8) | 0.003 | ||||||
>120 months | 0.2 (0.1–0.3) | <0.001 | 0.2 (0.1–0.4) | <0.001 | 0.2 (0.1–0.4) | <0.001 | ||||||
ATCb | ||||||||||||
(reference, anxiolytics) N05BA | – | – | – | |||||||||
(hypnotics) N05CD | 0.8 (0.6–1.0) | 0.10 | – | – | ||||||||
Half-life timeb | ||||||||||||
(reference) <24 hours | – | – | – | |||||||||
≥24 hours | 0.7 (0.5–0.9) | 0.003 | 0.6 (0.5–0.8) | 0.001 | 0.6 (0.4–0.9) | 0.01 |
Analysis in the subgroup of subjects of with duration of use was available (n = 1098 (short-term analysis, ST quitters = 311) and n = 1050 (Long-term analysis, LT quitters = 157), respectively). In this subgroup analysis, in the multivariate model, the estimates of duration of use were adjusted for gender, age, health insurance, baseline BZ prescription and ≥2 BZ at baseline. There were no clinically important changes in the estimates of gender, baseline prescription and ‘≥2 benzodiazepines at baseline’ compared with the estimates from the overall model.
Analysis in the subgroup of subjects with baseline prescription of a single benzodiazepine agent only [n = 1404 (short-term analysis, ST quitters = 421) and n = 1343 (Long-term analysis, LT quitters = 218)].
In the multivariate analysis ATC and half-life time were adjusted for gender, age, health insurance, baseline BZ prescription and duration of use. There were no clinically important changes in the estimates of gender and baseline prescription compared with the estimates from the overall model.
. | Baseline benzodiazepine prescription . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|
. | ≤65 PDD . | >65–≤130 PDD . | >130 PDD . | Total . | ||||
Duration of use | ||||||||
3–≤24 months | 0.62 (53/86) | 0.10 (1/10) | 0 (0/3) | 0.55 (54/99) | ||||
>24–≤120 months | 0.38 (135/357) | 0.22 (26/116) | 0.23 (10/44) | 0.33 (171/517) | ||||
>120 months | 0.27 (74/277) | 0.08 (11/139) | 0.02 (1/66) | 0.18 (86/482) | ||||
Total | 0.36 (262/720) | 0.14 (38/265) | 0.10 (11/113) | 0.28 (311/1098) |
. | Baseline benzodiazepine prescription . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|
. | ≤65 PDD . | >65–≤130 PDD . | >130 PDD . | Total . | ||||
Duration of use | ||||||||
3–≤24 months | 0.62 (53/86) | 0.10 (1/10) | 0 (0/3) | 0.55 (54/99) | ||||
>24–≤120 months | 0.38 (135/357) | 0.22 (26/116) | 0.23 (10/44) | 0.33 (171/517) | ||||
>120 months | 0.27 (74/277) | 0.08 (11/139) | 0.02 (1/66) | 0.18 (86/482) | ||||
Total | 0.36 (262/720) | 0.14 (38/265) | 0.10 (11/113) | 0.28 (311/1098) |
Figures are probability to have discontinued benzodiazepine prescription at short-term assessment and (n quitters/n of subjects).
. | Baseline benzodiazepine prescription . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|
. | ≤65 PDD . | >65–≤130 PDD . | >130 PDD . | Total . | ||||
Duration of use | ||||||||
3–≤24 months | 0.62 (53/86) | 0.10 (1/10) | 0 (0/3) | 0.55 (54/99) | ||||
>24–≤120 months | 0.38 (135/357) | 0.22 (26/116) | 0.23 (10/44) | 0.33 (171/517) | ||||
>120 months | 0.27 (74/277) | 0.08 (11/139) | 0.02 (1/66) | 0.18 (86/482) | ||||
Total | 0.36 (262/720) | 0.14 (38/265) | 0.10 (11/113) | 0.28 (311/1098) |
. | Baseline benzodiazepine prescription . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|
. | ≤65 PDD . | >65–≤130 PDD . | >130 PDD . | Total . | ||||
Duration of use | ||||||||
3–≤24 months | 0.62 (53/86) | 0.10 (1/10) | 0 (0/3) | 0.55 (54/99) | ||||
>24–≤120 months | 0.38 (135/357) | 0.22 (26/116) | 0.23 (10/44) | 0.33 (171/517) | ||||
>120 months | 0.27 (74/277) | 0.08 (11/139) | 0.02 (1/66) | 0.18 (86/482) | ||||
Total | 0.36 (262/720) | 0.14 (38/265) | 0.10 (11/113) | 0.28 (311/1098) |
Figures are probability to have discontinued benzodiazepine prescription at short-term assessment and (n quitters/n of subjects).
Analysis of relapse (Table 4)
A Cox-regression analysis of relapse among the short-term quitters showed three significant variables of relapse: multiple agent use, use of antidepressants at 6 months, and ATC classification as anxiolytics or hypnotics. Results of the regression model are shown in Table 4. Antidepressant use at baseline was similarly related to relapse as antidepressant use at 6 months. Half of the subjects who relapsed did so within the first 3 months after quitting.
Variables . | n short-term quitters = 468 Relapse: 205 Censored: 263 (abstinent at end of follow up n = 233; lost to follow up n = 30) . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|
. | Univariate . | . | Multivariate . | . | ||||
. | Hazard ratio (95% CI) . | P-value . | Adj. HR (95% CI) . | P-value . | ||||
Intercept | – | |||||||
Gender (F = 1, M = 0) | 1.3 (0.9–1.7) | 0.12 | ||||||
Age (yrs) | ||||||||
(reference) <50 | – | |||||||
50–<75 | 1.2 (0.8–1.8) | 0.29 | ||||||
≥75 | 1.4 (0.9–2.2) | 0.16 | ||||||
Health insurance (private = 1, NHS = 0) | 1.3 (1.0–1.8) | 0.06 | NS | |||||
Baseline BZ prescription | ||||||||
(reference) ≤65 PDD | – | |||||||
>65–≤130 PDD | 1.3 (0.9–1.9) | 0.15 | ||||||
>130 PDD | 1.5 (0.8–2.8) | 0.23 | ||||||
≥2 BZ at baseline | 1.8 (1.2–2.6) | 0.004 | 1.9 (1.3–2.8) | 0.001 | ||||
Evaluation consultation (yes = 1, no = 2) | 0.8 (0.6–1.1) | 0.14 | ||||||
Use of antidepressants at 6 months (yes/no) | 0.6 (0.4–1.0) | 0.04 | 0.6 (0.3–0.9) | 0.02 | ||||
Duration of usea | ||||||||
(reference) 3–≤24 months | - | |||||||
>24–≤120 months | 1.3 (0.8–2.1) | 0.38 | ||||||
>120 months | 1.7 (1.0–3.0) | 0.06 | NS | |||||
ATCb | ||||||||
(reference, anxiolytics) N05BA | – | – | ||||||
(hypnotics) N05CD | 0.7 (0.5–1.0) | 0.08 | 0.7 (0.5–1.0) | 0.03 | ||||
Half-life timeb | ||||||||
(reference) <24 hours | – | |||||||
≥24 hours | 1.0 (0.7–1.5) | 0.87 |
Variables . | n short-term quitters = 468 Relapse: 205 Censored: 263 (abstinent at end of follow up n = 233; lost to follow up n = 30) . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|
. | Univariate . | . | Multivariate . | . | ||||
. | Hazard ratio (95% CI) . | P-value . | Adj. HR (95% CI) . | P-value . | ||||
Intercept | – | |||||||
Gender (F = 1, M = 0) | 1.3 (0.9–1.7) | 0.12 | ||||||
Age (yrs) | ||||||||
(reference) <50 | – | |||||||
50–<75 | 1.2 (0.8–1.8) | 0.29 | ||||||
≥75 | 1.4 (0.9–2.2) | 0.16 | ||||||
Health insurance (private = 1, NHS = 0) | 1.3 (1.0–1.8) | 0.06 | NS | |||||
Baseline BZ prescription | ||||||||
(reference) ≤65 PDD | – | |||||||
>65–≤130 PDD | 1.3 (0.9–1.9) | 0.15 | ||||||
>130 PDD | 1.5 (0.8–2.8) | 0.23 | ||||||
≥2 BZ at baseline | 1.8 (1.2–2.6) | 0.004 | 1.9 (1.3–2.8) | 0.001 | ||||
Evaluation consultation (yes = 1, no = 2) | 0.8 (0.6–1.1) | 0.14 | ||||||
Use of antidepressants at 6 months (yes/no) | 0.6 (0.4–1.0) | 0.04 | 0.6 (0.3–0.9) | 0.02 | ||||
Duration of usea | ||||||||
(reference) 3–≤24 months | - | |||||||
>24–≤120 months | 1.3 (0.8–2.1) | 0.38 | ||||||
>120 months | 1.7 (1.0–3.0) | 0.06 | NS | |||||
ATCb | ||||||||
(reference, anxiolytics) N05BA | – | – | ||||||
(hypnotics) N05CD | 0.7 (0.5–1.0) | 0.08 | 0.7 (0.5–1.0) | 0.03 | ||||
Half-life timeb | ||||||||
(reference) <24 hours | – | |||||||
≥24 hours | 1.0 (0.7–1.5) | 0.87 |
NS = not significant.
Analysis in the subgroup of subjects with duration of use data available (n = 311, relapse = 129).
Analysis in the subgroup of subjects with baseline prescription of a single benzodiazepine agent only (n = 421, relapse = 176). In the multivariate model of this subgroup, the estimate of ‘Use of antidepressants at 6 months’ was 0.5 (95% CI 0.3 − 0.9; P-value 0.01).
Variables . | n short-term quitters = 468 Relapse: 205 Censored: 263 (abstinent at end of follow up n = 233; lost to follow up n = 30) . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|
. | Univariate . | . | Multivariate . | . | ||||
. | Hazard ratio (95% CI) . | P-value . | Adj. HR (95% CI) . | P-value . | ||||
Intercept | – | |||||||
Gender (F = 1, M = 0) | 1.3 (0.9–1.7) | 0.12 | ||||||
Age (yrs) | ||||||||
(reference) <50 | – | |||||||
50–<75 | 1.2 (0.8–1.8) | 0.29 | ||||||
≥75 | 1.4 (0.9–2.2) | 0.16 | ||||||
Health insurance (private = 1, NHS = 0) | 1.3 (1.0–1.8) | 0.06 | NS | |||||
Baseline BZ prescription | ||||||||
(reference) ≤65 PDD | – | |||||||
>65–≤130 PDD | 1.3 (0.9–1.9) | 0.15 | ||||||
>130 PDD | 1.5 (0.8–2.8) | 0.23 | ||||||
≥2 BZ at baseline | 1.8 (1.2–2.6) | 0.004 | 1.9 (1.3–2.8) | 0.001 | ||||
Evaluation consultation (yes = 1, no = 2) | 0.8 (0.6–1.1) | 0.14 | ||||||
Use of antidepressants at 6 months (yes/no) | 0.6 (0.4–1.0) | 0.04 | 0.6 (0.3–0.9) | 0.02 | ||||
Duration of usea | ||||||||
(reference) 3–≤24 months | - | |||||||
>24–≤120 months | 1.3 (0.8–2.1) | 0.38 | ||||||
>120 months | 1.7 (1.0–3.0) | 0.06 | NS | |||||
ATCb | ||||||||
(reference, anxiolytics) N05BA | – | – | ||||||
(hypnotics) N05CD | 0.7 (0.5–1.0) | 0.08 | 0.7 (0.5–1.0) | 0.03 | ||||
Half-life timeb | ||||||||
(reference) <24 hours | – | |||||||
≥24 hours | 1.0 (0.7–1.5) | 0.87 |
Variables . | n short-term quitters = 468 Relapse: 205 Censored: 263 (abstinent at end of follow up n = 233; lost to follow up n = 30) . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|
. | Univariate . | . | Multivariate . | . | ||||
. | Hazard ratio (95% CI) . | P-value . | Adj. HR (95% CI) . | P-value . | ||||
Intercept | – | |||||||
Gender (F = 1, M = 0) | 1.3 (0.9–1.7) | 0.12 | ||||||
Age (yrs) | ||||||||
(reference) <50 | – | |||||||
50–<75 | 1.2 (0.8–1.8) | 0.29 | ||||||
≥75 | 1.4 (0.9–2.2) | 0.16 | ||||||
Health insurance (private = 1, NHS = 0) | 1.3 (1.0–1.8) | 0.06 | NS | |||||
Baseline BZ prescription | ||||||||
(reference) ≤65 PDD | – | |||||||
>65–≤130 PDD | 1.3 (0.9–1.9) | 0.15 | ||||||
>130 PDD | 1.5 (0.8–2.8) | 0.23 | ||||||
≥2 BZ at baseline | 1.8 (1.2–2.6) | 0.004 | 1.9 (1.3–2.8) | 0.001 | ||||
Evaluation consultation (yes = 1, no = 2) | 0.8 (0.6–1.1) | 0.14 | ||||||
Use of antidepressants at 6 months (yes/no) | 0.6 (0.4–1.0) | 0.04 | 0.6 (0.3–0.9) | 0.02 | ||||
Duration of usea | ||||||||
(reference) 3–≤24 months | - | |||||||
>24–≤120 months | 1.3 (0.8–2.1) | 0.38 | ||||||
>120 months | 1.7 (1.0–3.0) | 0.06 | NS | |||||
ATCb | ||||||||
(reference, anxiolytics) N05BA | – | – | ||||||
(hypnotics) N05CD | 0.7 (0.5–1.0) | 0.08 | 0.7 (0.5–1.0) | 0.03 | ||||
Half-life timeb | ||||||||
(reference) <24 hours | – | |||||||
≥24 hours | 1.0 (0.7–1.5) | 0.87 |
NS = not significant.
Analysis in the subgroup of subjects with duration of use data available (n = 311, relapse = 129).
Analysis in the subgroup of subjects with baseline prescription of a single benzodiazepine agent only (n = 421, relapse = 176). In the multivariate model of this subgroup, the estimate of ‘Use of antidepressants at 6 months’ was 0.5 (95% CI 0.3 − 0.9; P-value 0.01).
Discussion
This study focussed on predictors of complete benzodiazepine discontinuation after a minimal intervention with a discontinuation letter from the family practitioner. Our main finding was that the amount of benzodiazepines used before the intervention and the duration of use were the two most important predictors of quitting. In the subgroup of subjects with a duration of use of ≤2 years and an average baseline prescription of ≤0.7 DDD, 62% quitted benzodiazepine use in the short-term. On the contrary, in the subgroup with both the highest baseline use (>1.4 DDD) and longest duration of use (>10 years), only 2% of the subjects discontinued in the short-term. The observation that quitting was more apparent among males was recently supported by findings from another Dutch benzodiazepine discontinuation study, but not from other discontinuation letter studies.3,8,10,21 Furthermore it has not been previously reported that short-term quitters using antidepressants relapsed less frequently, just like baseline single agent users when using a hypnotic agent. The finding that users of short-acting benzodiazepines quitted more frequently is in line with results of another minimal intervention study.2
Some remarks can be made concerning the generalizability of our findings. First, the intervention we studied actually consisted of a discontinuation letter and an evaluation consultation offer. This consultation however was not structured and the majority of the quitters had already quit before the consultation.4 Furthermore, as these long-term benzodiazepine users have a higher than average consultation rate we assume the letter would have been discussed whether or not a consultation was scheduled officially. We entered attendance at the evaluation consultation as a separate variable in the regression models and the consultation itself appeared to be no predictor of quitting or relapse. Therefore, a comparison with studies with no officially scheduled evaluation consultation is valid.3
Second, due to exclusion criteria and the family practitioners ‘freedom’ to additionally exclude long-term users on their own insight, the study population may have been selected towards subjects with a higher a priori chance to quit. This ‘healthy subject’ effect may have affected, for instance, a possible relationship of age with quitting if family practitioners selectively excluded subjects of older age. We however anticipated this and encouraged the family practitioners at the start of the study to actively include subjects, and especially the elderly, as much as possible. There were no differences in mean age between the group of long-term users as a whole and those included to receive the letter.4 So, our results meet the findings of Cormack and do not support exclusion for this intervention based on an age criterion.3 This is an important message as family practitioners may be more reluctant to offer this intervention to the elderly while it is the elderly who experience the drawbacks of long-term benzodiazepine use the most.
A third issue concerns the measurement of benzodiazepine use, as we measured prescription by family practitioners and not actual intake. Subjects could have received benzodiazepines from other sources, for instance psychiatrists or other specialists. In the Dutch Health Care system, the family physician functions as a gatekeeper for secondary medical specialist care and all patients (70% in National Health Plan, 30% private insurance) are linked to one family physician, often for many years. In The Netherlands long-term prescription of benzodiazepines is considered a family practice issue: 89% of all prescriptions are given by the family practitioner, 8% are prescribed by a specialist and 3% are prescribed in institutions.22 This makes it likely that those receiving prescriptions from other physicians are not in our study due to our exclusion criteria.
Fourth, duration of use was only available for those subjects attending an evaluation visit 3 months after the letter was sent. Our results do not indicate selection bias for this subgroup. There were no differences between the total group and this subgroup as far as our baseline characteristics are concerned. Furthermore, attendance at an evaluation visit itself was neither a predictor of quitting nor relapse.
The two strongest predictors of quitting in this study were amount of use and duration of use. This is in line with other reports of minimal interventions and secondary care interventions.2,3,10,11,13 The half-life time of the agent and its main metabolites was related to quitting, with subjects using short-acting agents having a better chance to quit. This relationship was independent of anxiolytic/hypnotic classification on the basis of ATC codes, and does not support the hypothesis that users of long half-life benzodiazepines may quit more easily because they experience less severe withdrawal symptoms due to a more gradual decrease of benzodiazepine plasma levels.23
Multiple agent use predicted short-term quitting in the unadjusted univariate analysis but not in the analysis adjusted for the amount of baseline benzodiazepine prescription suggesting that the univariate relationship reflected a dosage issue. Multiple agent use also predicted long-term quitting in the multivariate model. In line with this, multiple agent use was a strong independent predictor of relapse. Multiple agent use may possibly reflect a characteristic of patients in which the threshold to use or ask for drugs is low.
A primary care benzodiazepine taper-off study in patients with depression showed that, concerning the chance to become long-term benzodiazepine abstinent, a pre-treatment with a selective serotonin re-uptake inhibitor during 6 weeks was of minimal benefit.18 We also observed that baseline use of antidepressants was not related to quitting benzodiazepines. However, antidepressant use was a relapse preventive factor. Antidepressant use may decrease depression- or anxiety-related use of benzodiazepines.
The discontinuation letter proved to be an excellent first step within our stepped care model. As this intervention has substantial effects and demands only minor effort and costs it can be applied within a broad group of benzodiazepine users. A second step can be a more intensive strategy, like a taper scheme. Taper programmes were very successful among motivated subjects in secondary as well as in primary care.5–7,13,18,23 There is no reason to exclude subjects from the minimal intervention on the basis of gender or age. We, however, found evidence that the efficiency of the stepped care model can be enhanced when subjects with baseline use of >2 DDD or with a duration of use of more than 10 years are left out of the first step with the discontinuation letter. Instead, they could be offered a taper programme directly.
Future research should be undertaken to establish the value of relapse preventive measures, for instance additional consultations during follow-up. These may be useful especially for single users of anxiolytic agents and multiple-agent users as these subgroups appeared to be more vulnerable to relapse.
Declaration
Funding: the Dutch Health Care Insurance Council funded the study.
Ethical approval: the study was approved by the Committee on Research involving Human Subjects (currently known as CMO Arnhem/Nijmegen).
Conflicts of interest: none.
References
Russel J, Lader MH. Guidelines for the prevention and treatment of benzodiazepine dependence: on behalf of the Substance Abuse Committee of the Mental Health Foundation. The Mental Health Foundation;
Bashir K, King M, Ashworth M. Controlled evaluation of brief intervention by general practitioners to reduce chronic use of benzodiazepines.
Cormack MA, Sweeney KG, Hughes-Jones H, Foot GA. Evaluation of an easy, cost-effective strategy for cutting benzodiazepine use in general practice.
Gorgels WJMJ, Oude Voshaar RC, Mol AJJ, van de Lisdonk EH, van Balkom AJLM, van den Hoogen HJM et al. Discontinuation of long-term benzodiazepine use by sending a letter to users in family practice: a prospective controlled intervention study.
Baillargeon L, Landreville P, Verreault R, Beauchemin JP, Gregoire JP, Morin CM. Discontinuation of benzodiazepines among older insomniac adults treated with cognitive-behavioural therapy combined with gradual tapering: a randomized trial.
Morin CM, Bastien C, Guay B, Radouco-Thomas M, Leblanc J, Vallieres A. Randomized clinical trial of supervised tapering and cognitive behavior therapy to facilitate benzodiazepine discontinuation in older adults with chronic insomnia.
Oude Voshaar RC, Gorgels WJMJ, Mol AJJ, van Balkom AJLM, van de Lisdonk EH, Breteler MH et al. Tapering-off long-term benzodiazepine use with or without simultaneous group cognitive behavioural therapy: a three-condition randomised controlled trial.
Cormack MA, Owens RG, Dewey ME. The effect of minimal interventions by general practitioners on long-term benzodiazepine use.
Voshaar RO, Gorgels W, Mol A, van Balkom A, Breteler M, van de Lisdonk E et al. Predictors of relapse after discontinuation of long-term benzodiazepine use by minimal intervention: a 2-year follow-up study.
Morgan JD, Wright DJ, Chrystyn H. Pharmacoeconomic evaluation of a patient education letter aimed at reducing long-term prescribing of benzodiazepines.
Rickels K, Schweizer E, Case WG, Greenblatt DJ. Long-term therapeutic use of benzodiazepines. I. Effects of abrupt discontinuation.
Lader M, Farr I, Morton S. A comparison of alpidem and placebo in relieving benzodiazepine withdrawal symptoms.
Rickels K, Case WG, Schweizer E, Garcia-Espana F, Fridman R. Long-term benzodiazepine users 3 years after participation in a discontinuation program.
Schweizer E, Rickels K, De Martinis N et al. The effect of personality on withdrawal severity and taper outcome in benzodiazepine dependent patients.
Holton A, Riley P, Tyrer P. Factors predicting long-term outcome after chronic benzodiazepine therapy.
Murphy SM, Tyrer P. A Double-Blind Comparison of the Effects of Gradual Withdrawal of Lorazepam, Diazepam and Bromazepam in Benzodiazepine Dependence.
WHO Collaborating Centre for Drugs Statistics Methodology. Guidelines ATC classification and DDD assignment. (First edition). Oslo: WHO/NCM;
Zitman FG, Couvee JE. Chronic benzodiazepine use in general practice patients with depression: an evaluation of controlled treatment and taper-off: report on behalf of the Dutch Chronic Benzodiazepine Working Group.
Niessen WJM, Stewart RE, Broer J, Haaijer-Ruskamp FM. Vermindering van gebruik van benzodiazepinen door een brief van de eigen huisarts aan chronische gebruikers.
Geneesmiddelen Informatie Project (GIP). GIPsignaal–Gebruik van benzodiazepinen 1993–1998. Amstelveen, The Netherlands: College voor Zorgverzekeringen;
Author notes
aDepartment of General Practice and Family Medicine, Radboud University Nijmegen Medical Centre, bDepartment of Psychiatry, Radboud University Nijmegen Medical Centre, cDepartment of Psychiatry and Institute for Research in Extramural Medicine, VU University Medical Center Amsterdam, dDepartment of Clinical Psychology, Radboud University Nijmegen and eDepartment of Psychiatry, Leiden University Medical Center, The Netherlands.