Article Text

Download PDFPDF

Counselling may be more expensive than usual care for people with mental health problems, but may not be more effective for improving Beck depression score in the long term
  1. Jonathan R Price, MSc, MRCPsych, Dr
  1. Clinical Tutor
    University of Oxford Department of Psychiatry, UK

Statistics from

OpenUrlAbstract/FREE Full Text

QUESTION: Is counselling people with mental health problems more cost-effective than usual care? Can meta-analysis of cost data from primary studies increase the reliability of economic analyses?


Meta-analysis of individual patient data.

Data sources

Studies were identified from a Cochrane systematic review.

Study selection

Eligible studies compared counselling in primary care versus usual general practitioner care in people with mental health problems. Studies with insufficient detail on healthcare utilisation, or inadequate randomisation methods were excluded.

Data extraction

Data on type of counselling, trial inclusion criteria, length of follow up, primary clinical outcomes, health service use and sample size were extracted. Standardised national unit costs in pounds sterling for the 1999–2000 financial year were applied. The British National Formulary was used to calculate drug costs. Counselling and usual care costs were based on salary information from relevant national organisations. Costs incurred by employers’ and overheads were also included. Outcomes were cost, effectiveness of treatment, and cost-effectiveness.

Main results

Cost: Counselling cost significantly more than usual general practitioner care (see table).

Weighted mean difference in costs (95% CI) between counselling and usual general practitioner care, fixed effects model.

Effectiveness: Counselling led to improved scores on the Beck depression inventory score compared with usual care in the short term, but not in the long term.

Cost effectiveness: For a 1 point improvement in the Beck depression inventory score, counselling cost £50 more than usual care in the short term, and £196 more than usual care in the long term.

Feasibility of meta-analysis: many clinical trials are not powered to examine economic effects. Meta-analysis increased the power to detect differences in costs However, the results were sensitive to changes in the assumptions underlying the cost calculations (the cost of counsellor sessions and the duration of the general practitioner consultations).


It is likely that counselling is more expensive than usual general practitioner care. Meta-analysis can increase the power of the study to detect economic differencesbetween interventions above the power of the primary studies, but is sensitive to variability between studies and the underlying assumptions made.


Counselling is a common treatment for a range of emotional problems presenting to primary care. It is popular with patients, and its high acceptability may have contributed to its remarkable recent expansion. Probably about one half of UK general practices now have a counsellor employed by or attached to their practice. Counselling must therefore have a major impact on stretched budgets and contribute to inequity of health care provision.

The rapid rise of counselling must be seen alongside continued uncertainty about its comparative clinical effectiveness and cost. This uncertainty reflects a lack of good quality randomised evidence. A recently updated Cochrane review by the same team found only seven relevant randomised controlled trials, despite energetic searching.1 They concluded that counselling is moderately more effective than usual primary care in the short term (up to 6 months). However, even this modest conclusion was tempered by concern about the quantity of evidence - their main conclusion was based on data from just 108 participants - and its quality. Regarding costs, trial data - when it is available - gives mixed messages. This probably reflects inadequate sample sizes, which give power appropriate to avoid type 2 errors on clinical outcomes but not on cost. Size is particularly important for economic analysis. 2

This meta-analysis of cost data for counselling is therefore welcome. It is of high quality, using individual patient data. However, it includes data from only 4 trials – those of the 7 in the original review that report data on healthcare use. This is a major weakness. No amount of analysis and re-analysis of the available data will overcome fundamental shortcomings in the evidence base.

We desperately need further randomised controlled trials, of high quality, of counselling in primary care. In the meantime, the reviewers and methodologists will remain innovative and busy, but primary care practitioners and service providers will remain uncertain of the role of counselling.



  • Source of funding: Funded by the National Primary Care Research and Development Centre at the University of Manchester, through the Department of Health.

  • For correspondence: P Bower. National Primary Care Research and Development Centre, University of Manchester, Manchester M13 9PL. E-mail: peter.bower{at}

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.