Why do GPs over prescribe benzodiazepines? Synthesis of qualitative studies

May 15 2014

Benzodiazepines are used to treat insomnia, anxiety and chronic back pain due to their sedative and muscle relaxing effects. They’ve got a sting in the tail though and can cause memory disruption, loss of coordination and dependence if used long term. It’s therefore recommended that other treatments, such as psychological interventions, are tried first and that the drugs are prescribed for short term use only.

However, research has shown that clinicians are over prescribing benzodiazepines and giving it to patients for long periods. The aim of this study was to review qualitative studies of clinicians’ decisions to prescribe or withdraw benzodiazepines and blend those studies into an explanatory model, to inform the development of interventions that could encourage clinicians to follow the prescribing guidelines.

Methods

The authors searched multiple databases for papers. Papers were included if they were qualitative, in a primary care setting and published between 1990-2011.

Two pairs of researchers independently extracted data from the included papers. The authors include a PRISMA diagram of the process which shows clearly the stages of the search and the review.

The authors state they used thematic analysis to synthesise the results. Thematic analysis is a stalwart of qualitative research and has been recommended for synthesising qualitative reviews (see the paper by Thomas & Harden, 2008). However, it is arguably not as rigorous as meta-ethnographic approaches which call on reviewers to specify how they integrated the multiple data sources, for example by using a ‘line of argument’ approach which aims to discover a ‘whole’ among multiple ‘parts’.

The reviewers wanted to find out why GPs over prescribe or withdraw benzodiazepines

Results

Eight papers were included in the review from seven countries, and the authors report seven analytic themes. They encompassed the changing context of prescribing (for example, changing guidance and also societal attitudes), GPs perceptions of themselves and of the patients and a perceived lack of alternative treatments. Overall, the authors report that GPs are ambivalent about prescribing and experienced conflicting pressures.

The authors present an explanatory model that aims to synthesise the multiple themes. I can firmly tell you that it has arrows and boxes in it.

Conclusions

The authors’ state:

Making decisions on whether or not to prescribe was often uncomfortable, demanding and complex within the time and pressure constraints of daily practice.

The GPs often made their decisions on a case-by-case basis rather than having consistent practices. For example, one theme was of ‘the deserving patient’ and GPs decided whether the patient could ‘legitimately’ use benzodiazepines. The authors suggest that these decisions might be understood as examples of cognitive heuristics, and I’d certainly be interested to see a study of heuristic decision making in this context (any health psychologists at the back?).

The review paints a picture of the pressures and stresses inherent in general practice

Limitations

Databases were searched in October 2011. It’s unusual that the review hasn’t been updated, particularly as the authors note the importance of changing contexts in primary care.
Personally I found the ‘explanatory model’ quite confusing – quite how the themes interrelate and compare wasn’t very clear. Possibly adopting a more specific analytical process, such as line-of-argument, might have been helpful to translate the multiple themes into a more streamlined model. As it is, the authors present a good summary of the themes from the multiple papers, but there is perhaps a missed opportunity here to rigorously employ synthesis techniques and clearly bring together those themes into an overarching model. Of course, the proof of the pudding is in the eating – the test will be whether the model is useful for informing future work.
Again perhaps because they employed a broader thematic approach rather than meta-ethnographic approach, the authors didn’t separate the data into ‘first order’ and ‘second order’ constructs. First order refers to quotations lifted from the paper, second order to comments or observations by the original study authors. It can be helpful to distinguish these to get a sense of how much of the original data (participant quotations) are being synthesised and for example to explore whether different studies put forward consistent or different interpretations.
The authors conducted a quality analysis using the CASP qualitative research appraisal checklist (PDF) but it was notable that all the studies scored very highly and none were considered for exclusion. It’s possible that researchers publishing in this field are just all really good at qualitative research, but it might also indicate the authors were quite generous in their assessments or that the CASP checklist isn’t very helpful for making these kind of decisions – there is a debate on whether it’s even appropriate to assess quality in this way or whether it’s an inaccurate attempt to replicate the methods used in quantitative reviews.
The authors suggest that training is needed to address GPs ‘knowledge deficits’ and educate and raise awareness and particularly target high over prescribers. However they themselves note that GPs had mixed feelings about engaging in further training and considering that those who flaunt the guidelines most are perhaps least likely to access such training, it’s not clear how helpful this would be. Would interventions aimed at decision-making be more helpful or employing external motivators such as QOF checks?
None of the studies included nurse prescribers. Further work may be needed to explore other clinicians’ views, and also the views of patients themselves.