Counselling in primary care

Review
Results
Comment

The place, if any, of counselling in primary care is one of those topics that continues to attract attention. There is a Cochrane review, and some countries, including the UK, have produced guidelines based on evidence. What is the evidence, and how good is it? An updated review [1] by the authors of the Cochrane review is helpful.

Review

The search strategy for the original Cochrane review was heroic, and included 10 electronic databases, handsearching specialist journals, as well as consulting experts. The updated review used six databases, including specialist trials registers. Randomised and controlled trials were eligible that tested the hypothesis that counsellors treating patients in primary care are more effective than usual care provided by the GP or alternative mental health treatments.

Participants were patients consulting a GP with psychological or psychosocial problems. Intervention by a counsellor was undefined other than counsellors had to have been trained to the British standard for accreditation. Outcomes were clinical effectiveness of psychological outcomes like depressive symptoms and measures of social function.

Results

Seven trials were included from 12 publications, with 444 patients given counselling and 297 given usual care. On a quality scoring system all scored reasonably well, but how many of these trials were randomised or used blinded outcome assessments is not mentioned in this review.

Six trials had short-term outcomes (six weeks to six months). Three, including two with shorter observations, showed significant benefit of counselling over usual care, and three, all with longer observations, did not. The overall outcome was statistically significant, but of a modest size. Results in terms of outcomes for patients are best described by Figure 1. With counselling there was a small increase in the percentage of patients having a reliable and clinically meaningful change.

Figure 1: Outcomes in counselling and usual care in a mixed population of patients with different problems in primary care

Four trials had longer-term outcomes (9-12 months). There was no significant difference between counselling and usual care.

Comment

This is an interesting and thoughtful review, well worth reading if primary care counselling is of interest. There are sensitivity analyses, and useful discussion. The authors do not seek to disguise the basic problem, that of insufficient information. Some of the issues are worth looking at, because they arise time and again in meta-analysis:

Who were the patients? Some were depressed, some anxious, some had relationship or family problems, some were bereaved, some had sexual difficulties, or substance misuse problems. Some trials used a mix of patients, others used just patients with depression, or anxiety, or emotional problems.
What was the counselling intervention? It was not always described, and rarely standardised. The number of sessions varied, and not all patients attended.
What was usual care? Usually it was not described in the papers to any satisfactory degree, other than specifying that counsellors were trained to a British standard.
What was the outcome? Good question, this, as Bandolier has no idea. The best came from descriptions of change (Figure 1). But describing this is very, very, difficult.
How good were the trials? Pretty good using a specialist scoring system, but we do not know whether they were all randomised, or whether a blinded assessor made assessment of outcome. Without this knowledge we have no idea about potential sources of bias. A question that needs answering was the gross imbalance in numbers in some trials.
How much information do we have? At best on 444 patients given counselling in comparisons with usual care. The important result, that of a difference in the proportion of patients with reliable and statistically significant change, depended on just 108 patients.

And yet this is the only information we have that can inform the question of counselling in primary care. Who among us would conclude either that it works, or does not work? The best response is that we cannot possibly know, but that large advantages of counselling are unlikely. Saying any more is to make far too much from far too little.

It is relevant to compare the weight and quality of evidence we have here with the weight and quality of evidence we expect from a newly introduced pharmacological therapy. There is little comparison. It is not even possible to say that counselling is better than usual care, and the trials say nothing about possible harms. For instance, might there be rare but serious harm from counselling that outweighs any possible small benefit?

Again, it is not possible to say anything about any cost consequences, because without knowing anything about effectiveness, we can say nothing about costs. On the basis of the evidence we have from this review, would it be a sensible decision to begin a widespread use of counselling in primary care?

References:

P Bower et al. The clinical effectiveness of counselling in primary care: a systematic review and meta-analysis. Psychological Medicine 2003 33: 203-215.

previous or next story in this issue