Revisiting the relative cost-effectiveness of selective serotonin re-uptake inhibitors and tricyclic antidepressants

In a previous paper (Stewart, 1994) a cost model was constructed to evaluate episodic antidepressant pharmacotherapy, using a tricyclic antidepressant or a selective serotonin reuptake inhibitor. The analysis focused on those patients who were diagnosed and initially treated in the community, by GPs. Alternative drug regimens are compared in terms of expected costs per patient, over a maximum of one year.

<p><p><p>The conclusion was that while the cost advantage of TCAs was not as great as is usually inferred from the acquisition cost, there was no clear cost argument demonstrated for switching from TCAs to SSRIs. This paper develops that analysis in the light of the effect of inflation on the unit costs of the resources used in treatment of depressions. The question of interest is whether inflation has a differential impact on the two groups of drugs.

<p><p><p>A further effect that is tested is that outlined in a recent paper (Donoghue and Tylee, 1996) evaluating the extent of subtherapeutic prescribing of drugs by GPs. The earlier analysis used clinical data derived from RCTs at therapeutic doses. These figures have been adjusted to allow for those patients receiving ineffective doses of drugs, with efficacies being adjusted to assume that those on subtherapeutic doses show only placebo level responses. All figures are based on published meta-analyses for the two groups of drugs.

<p><p><p>The basic answers to the two questions are that in the former case, tackling the question of inflation, there is a shift in favour of SSRIs as a cost-effective option. Inflation has been slightly negative in drug acquisition costs, the area where TCAs showed a significant advantage before, but has been positive in other resource areas, where TCAs result in higher resource usage. The net effect is that the expected costs of TCAs showed a higher level of inflation. The latter case, adjusting for subtherapeutic doses has a significant effect in shifting the balance in favour of using SSRIs.

<p><p><p>These two categories of change illustrate that decision analysis models represent only a snapshot of clinical knowledge and hence should be updated at intervals. They also represent a snapshot of the costs structure, providing what is possibly a more economically significant reason why they should be updated.