CBT cost-effective for treatment resistant depression

The cost of depression to the individual and to society as a whole is well documented, and has become something of a talking point lately; thanks in part to Richard Layard and David Clark’s new book Thrive. Health service use associated with depression can be expensive, though these sums are often dwarfed by the costs of lost productivity. Not to mention the huge impact on quality of life.

Cognitive-behavioural therapy (CBT) has been shown to be cost-effective for various conditions, including depression. Generally, CBT hasn’t been found to be all that costly an intervention, with estimates ranging from a few hundred to a few thousand pounds depending on delivery method (Curtis, 2013). The CoBalT trial, which we covered in a previous blog, demonstrated that CBT was beneficial to patients with treatment-resistant depression (Wiles et al, 2013). An economic evaluation was carried out alongside the trial and was recently reported in The British Journal of Psychiatry (Hollinghurst et al, 2014).

It’s possible that treatments for depression are cost-saving for society, but proving this would require a lot more research.

Methods

The intervention consisted of between 12 and 18 CBT sessions of about an hour delivered at a GP surgery over the course of 12 months; full details are outlined in the original paper (Wiles et al, 2013) and corresponding elf blog post. Outcomes included BDI-II score, SF-12 and the EQ-5D-3L.

234 participants were randomised to CBT plus usual care, while 235 received usual care.

The economic evaluation took two approaches:

Firstly, the authors carried out a cost-consequences analysis (CCA), which tallies up the various costs and outcomes associated with a strategy but does not combine them into a single metric. This is useful when we’re interested in lots of costs and outcomes, but aren’t sure about the relative importance of each.
Secondly, the authors also used a cost-utility analysis (CUA) to estimate the cost-per-QALY (quality-adjusted life year) of CBT. The CUA adopted a NHS and personal social services perspective, as advised by NICE.

Some resource use (such as out of pocket expenditures, time off work and hospital visits) was self-reported by patients at 6 month intervals. Other information, including use of primary care and prescribed medication, was extracted from GP records. Therapists recorded information about the number and length of CBT sessions in order to estimate a cost for the intervention itself.

The authors carried out a number of sensitivity analyses and used bootstrapping to estimate the level of uncertainty associated with the cost-effectiveness estimates.

Results

The average direct cost of the CBT intervention was £910 per person.
There were no clear differences between the groups in terms of other NHS costs, personal expenditure or lost productivity. However, 27% of participants were unemployed and it’s possible that 12 months isn’t enough time to get back into work for the long-term unemployed.
Improvements were demonstrated in BDI-II score, SF-12 mental sub-scale and QALYs.
Under the CUA framework, the incremental cost of CBT was £850 and the incremental benefit was 0.06 QALYs. The cost per QALY was therefore just under £15,000.
The probability that the intervention would be cost-effective at a willingness to pay of £20,000 per QALY was 0.74, rising to 0.91 at a willingness to pay of £30,000.
The results are not clearly undermined by any of the authors’ sensitivity analyses.

The study didn’t find much difference in lost productivity, though presenteeism (which may be a bigger problem than absenteeism) was not measured.

Conclusion

The authors concluded that:

CBT when added to usual care is a cost-effective treatment for patients with treatment-resistant depression.

The confidence in this conclusion is, I think, justified. The cost of the intervention is relatively low and cannot rise that high in the extreme because of the nature of CBT.

As with many economic evaluations, the study suffered from missing data. However, the authors addressed this well by carrying out multiple imputation.

Capturing the health benefits associated with treatment for depression under a general QALY framework is tricky. In one of the authors’ sensitivity analyses they used QALYs estimated using the SF-6D rather than the EQ-5D, and found the intervention to be less effective and therefore less cost-effective. Measures like the SF-6D and EQ-5D have been accused of insensitivity to mental health problems, and it’s possible that they underestimate the impact of treatments for depression. That they might overestimate the benefit is highly improbable; current methods are likely to give a conservative estimate of the benefits.

Research is ongoing to better reflect mental health problems in QALYs. Meanwhile we can expect the evidence base for the cost-effectiveness of CBT for depression to get stronger.

QALYs are a general measure of health outcome, but piecing together exactly what constitutes good health (including mental and physical illness) is no mean feat.