CBT is more cost-effective than SSRI alone as treatment for panic disorder

In addition to its impact on quality of life, panic disorder can have a number of costly consequences such as lost productivity – particularly if also associated with agoraphobia. Cost-effectiveness is therefore an important consideration in choosing the optimal treatment for panic disorder, which might improve value via the cost side of the equation.

A recent study by van Apeldoorn et al estimates the cost-effectiveness of three treatment modalities, all of which have been found to be effective treatments for panic disorder: CBT, pharmacotherapy or the combination of the two. The study is the first to compare the cost-effectiveness of the 3 strategies while accounting for wider societal costs.

Methods

Based at 11 treatment facilities in the Netherlands, 150 patients were randomised to the three trial arms. The study involved 12 months of active treatment and a 12 month follow-up period, with patients assessed at months 0, 9, 12, 18 and 24. Treatments received were:

CBT – up to 21 treatment sessions of 50 minutes, consisting of interoceptive exposure, cognitive interventions and exposure in vivo.
SSRI – visited a therapist up to 12 times; each treatment lasting around 20 minutes. Prescribing clinicians were free to choose between 5 SSRIs: fluoxetine, paroxetine, sertraline, citalopram and fluvoxamine. Treatment tapering began in week 40.
CBT+SSRI – as above, with treatments delivered in parallel by 2 different therapists.

The primary cost-effectiveness analysis adopted a societal cost perspective; accounting for direct and indirect medical costs as well as non-medical costs. Productivity losses were estimated using the friction cost method and compensation mechanisms were taken into account (which together gives a more conservative and short-run estimate of the economic impact of illness than a traditional human capital approach). Most resource use data were collected using face-to-face questionnaires. Unit costs were used and based on Dutch 2005 euro prices.

The primary outcome used was the Hamilton Anxiety Rating Scale (HAM-A), with the treatment effect based on the difference between baseline and 24 month follow-up. Incremental cost-effectiveness ratios (ICERs) were calculated and cost-effectiveness acceptability curves drawn. Additionally, 2 sensitivity analyses were carried out: i) exclusion of productivity costs and ii) imputation of missing data using the expectation maximisation algorithm.

The friction cost method only counts productivity losses until an employee is replaced

Results

Full follow-up data were available for 91 people. In the sample, 52% suffered from moderate or severe agoraphobia. All groups showed an improvement in HAM-A. Costs incurred by each person, on average, across the 3 arms were:

	CBT	SSRI	CBT+SSRI
Intervention	€690	€257	€924
Productivity	€776	€1136	€1068
Total	€2224	€3118	€3590

CBT was found to dominate SSRI as cheaper and more effective. CBT+SSRI was more effective but more costly than SSRI or CBT, in which case a decision-maker’s willingness-to-pay for the health outcome becomes important. Below approximately €700 per point improvement on the HAM-A, CBT is the preferred treatment. Above about €700, CBT+SSRI is preferred.

Discussion

The authors do not report ICERs in the paper, either for their primary analysis or the sensitivity analyses. I have taken the liberty to estimate these myself and report them here, where A, B and C are the primary analysis and the 2 sensitivity analyses respectively.

	Incremental Cost			Incremental Benefit	ICER (cost per 1-point improvement)
	A	B	C		A	B	C
CBT vs SSRI	-€894	-€179	-€320	3.9	-€229	-€46	-€82
CBT+SSRI vs SSRI	€472	€467	€568	5.6	€84	€83	€101
CBT+SSRI vs CBT	€1366	€646	€888	1.7	€804	€380	€522

How and whether to capture productivity costs is one of the most enduring debates among health economists. The ICERs in the table above show that the exclusion of productivity costs (analysis B) has a large impact on the incremental costs of the interventions.

As the authors point out, the study would have benefited from the inclusion of a more generalisable measure of health such as QALYs. This would enable comparisons of cost-effectiveness with other unrelated treatments for the purpose of resource allocation. It is not clear what a reasonable decision makers’ willingness-to-pay value might be for improvements in the HAM-A, while the value attached to a QALY has been elicited in various ways from various groups of people.

Conclusion

The authors conclude that:

the present cost-effectiveness analysis shows that the optimal treatment for [panic disorder] with or without [agoraphobia] is either CBT or CBT+SSRI

While the study is persuasive in demonstrating that SSRIs alone are the least cost-effective option, and that CBT should be considered for people with panic disorder, it is not clear whether CBT or CBT+SSRI is the optimal treatment. This is due to uncertainties in the monetary value of improvements in an individual’s HAM-A score, and health economists’ indecision about the most appropriate method for costing productivity losses. Hopefully future work will address these difficulties.