Reducing benzodiazepine prescribing in primary care


Benzodiazepines (BZD) are a group of medicines that act as sedatives and are very effective at reducing anxiety and helping reduce insomnia. They work in a similar way to alcohol. However, with regular long term use, tolerance develops (this means you need more of the drug to have the same effect) and withdrawal symptoms will occur if ceased abruptly.

Many people become trapped in their dependence on benzodiazepines and are not able to function without them. The irony is that the withdrawal symptoms are often far worse than the original problem they were taken for, and people experience horrendous anxiety, panic, sweating, shakes, nausea and insomnia. In severe cases, people can have seizures as a result of withdrawal.

Therefore because of the severe and unpleasant withdrawals, it is very difficult to reduce and stop taking them, and often people end up taking them for years, with a significant negative impact on their lives. As a result, there is a need to develop effective ways to help empower people to move on with their lives, free from the reliance on these pills.

The withdrawal symptoms of benzodiazepines are often far worse that the original condition they were taken to treat.

The withdrawal symptoms of benzodiazepines are often far worse that the original condition they were intended to treat.

Four out of 5 benzodiazepine prescriptions are made by GPs

This blog will be considering evidence for brief interventions for reducing benzodiazepines in primary care. Primary care is an obvious setting for such interventions as 4 out of 5 BZD prescriptions are made by GPs, and they are in a prime position to intervene. However, GP time is limited and so brief and structured interventions may well be the most pragmatic and feasible method of delivery.

One of the papers (Gould et al, 2014)  reports on a systematic review and meta-analysis of interventions to reduce BZD in older adults, and the second paper (Vicens, et al 2014) reports on a study that compared two interventions to reduce BZD in primary care in a 3-arm cluster randomised trial in Spain.

Evidence from the systematic review (Gould et al, 2014)

The first step in understanding the outcomes of a systematic review is to consider the quality of the review process in order to ensure that its conclusions are not limited by potential bias in searching, selecting and reporting of studies (see PRISMA for more information).

This review has a clear aim, it described the search strategy, along with search terms and inclusion/exclusion criteria. The papers identified from the search were screened independently against selection criteria and included papers were assessed for risk of bias (using Cochrane Risk of Bias tool). It reported the numbers of papers included and excluded at each step using a flow chart. However, it made no reference to an a priori protocol for the search, which would help assess if they undertook the search exactly as they had planned it.


  • To review evidence for interventions aimed at reducing benzodiazepine use in older people

Inclusion criteria

  • Peer reviewed, randomised controlled trials of interventions that aimed to reduce BZD use (either by withdrawal regimes or changing prescribing)


  • They identified 10 withdrawal studies, and 8 changing prescribing studies from originally 208 potentially eligible studies
  • Excluded studies where there was insufficient data to calculate effect size


  • Supervised withdrawal with psychotherapy
    • The odds of not using BZD at follow-up was 5.06 times higher than for control interventions at post-intervention
    • As time went on past the intervention, the effect lessened but was still statistically significant at 0.5 – 3 months
    • In addition, the odds of not using BZD with withdrawal and prescribing interventions was 1.43 compared with control interventions including treatment as usual, education placebo, or psychotherapy alone


The authors rightly point out that their review contains so few studies that it limits the strength and reliability of the conclusions. It also limits the ability to perform further analyses on any possible effects on outcome from variations in setting, participant characteristics, physician effects and so on. All of these things could have significant effects on the success (or not) of an intervention. Like all multi-faceted complex interventions it is hard to determine what is the “active ingredient” and so it is hard to make concrete recommendations based on what works with this current evidence.

Authors conclusions:

Despite these limitations, the authors make cautious recommendations for clinical practice including:

  • The importance of medication review together with a withdrawal schedule and education about BZD use
  • Provision of supervised withdrawal alongside psychotherapy (mainly addressed at underlying mental health issues).  This produced a better effect on being free of BZD, but the authors rightly point out that this approach may not be feasible for those who can’t or won’t access the therapy sessions
Supervised withdrawal alongside psychotherapy was a component in the cautious recommendations from this review.

Supervised withdrawal alongside psychotherapy was a component in the cautious recommendations from this review.

Evidence from the Single study (Vicens et al, 2014)

This was an RCT comparing two types of structured interventions which aimed to reduce BZD use in primary care. This was a three arm multi-centre cluster trial recruiting from 21 primary care centres in Spain. Participants were eligible for the trial if they were 18-60 and had been taking BZD for at least 6 months.

Interventions were:

  1. A structured educational intervention with gradual tapering, then fortnightly visits (SIF)
  2. The same structured educational intervention, only with written instructions rather than follow-up visits (SIW)
  3. GP usual care

Care was taken to minimize bias by independent randomisation, and the allocation was masked from those involved in data collection and entry. The sample tended to be older people with an age range of 50-72 years old.


  • The main result was that benzodiazepine reduction was 3 times more likely than in treatment as usual arm for both the SIF and SIW.

Authors conclusions:

As both the SIF and the SIW produced similar results, the recommendation by the authors is to suggest the cheapest intervention which is the written individualised stepped dose reduction with written instructions, as opposed to the more time consuming intervention which involved follow-up visits.


This trial concluded that a structured educational intervention with written instructions was the most effective option for reducing benzodiazepine use.

Overall conclusions

It is clear that there is a need to offer effective interventions that support people in reducing BZD in primary care. The review and RCT both offer some support to a more structured approach than is typically offered in routine care.

There are important issues around what is the active ingredient in interventions that are complex (i.e. multiple components). The Vicens et al RCT offers some clues to this, as they found that irrespective of the follow-up support, a structured withdrawal with information is three times more effective than treatment as usual. So, the required solution may not necessarily need complex psychosocial input.

However, qualitative data on the patient perceptions of such interventions and issues of acceptability and feasibility would be useful. It may be that the Vicens interventions were effective for those who were already very motivated to change, and there is a question about whether this would work for people with more complex co-morbidities and BZD use.

Both kinds of interventions investigated by these studies were associated with reduction in BZD use, but the psychotherapy intervention had a larger effect.

Both kinds of interventions investigated by these studies were associated with reduction in BZD use, but the psychotherapy intervention had a larger effect.


Gould RL. Interventions for reducing benzodiazepine use in older people: meta-analysis of randomised controlled trials. Br J Psychiatry. 2014 Feb;204(2):98-107. doi: 10.1192/bjp.bp.113.126003. [PubMed abstract]

Vicens C. Comparative efficacy of two interventions to discontinue long-term benzodiazepine use: cluster randomised controlled trial in primary care. Br J Psychiatry. 2014 Feb 13. [Epub ahead of print] [PubMed abstract]

Share on Facebook Tweet this on Twitter Share on LinkedIn Share on Google+
Mark as read
Create a personal elf note about this blog
Liz Hughes

Liz Hughes

Liz is a mental health nurse by clinical background, and is Professor of Applied Mental Health Research at the University of Huddersfield, which is a joint appointment with South West Yorkshire Mental Health Partnership NHS Foundation Trust. She is also a visiting senior fellow in the Mental Health and Addictions Research Group led by Professor Simon Gilbody at the University of York. Her role is to develop a programme of funded health research (related to multi-morbidities in mental health) as well as build research capacity. Her clinical experience spans acute psychiatric inpatient settings as well as in inpatient and community addictions treatment services in London and the south of England. Her main research (and teaching) interests include dual diagnosis of mental health and substance use; physical and sexual health and relationships in people with serious mental health problems; and workforce development in mental health. Liz has published and presented widely related to her interests and has worked on a range of workforce development products including strategy documents, training resources and E-learning for dual diagnosis for the Department of Health which is cited as a resource to support NICE guidance for Psychosis and Substance Misuse. She is particularly passionate about improving the quality of care that marginalised groups of mental health service users receive, and developing the capabilities of the workforce to address this.

More posts - Website

Follow me here –