Anzhealthpolicy.com

Reviewer's report
Title: Impact of evergreening on patients and health insurance: a case study of
citalopram/escitalopram antidepressant.
Version: 1 Date: 20 August 2012
Reviewer: Jacqueline Birks
Reviewer's report:
Study of citalopram vs escitalopram - discretionary revisions The search strategy was thorough and systematic, and trials were identifiedthrough published reviews, contacting Lundbeck, the pharmaceutical companythat manufactures both of these drugs, and direct searches for individual trialreports. RCTs were sought for citalopram compared with escitalopram, andeither compared with placebo, assessing short term efficacy (approximately 8weeks) in patients with major depression. The dose is not mentioned, and itappears that any dose of either drug could be included.
Efficacy was assessed by the proportion of responders, defined as patients witha decrease in depression score, either the MADRS or the Hamilton scale, frombaseline to follow up of at least 50%. This outcome was routinely used byLundbeck in their clinical trials, and the majority of included trials were Lundbecksponsored trials. Treatment acceptability was assessed as the proportion ofpatients who did not drop out of allocated treatment by end point.
The treatment effect for both outcomes was measured by odds ratios.
In order to compare the two drugs, direct comparisons were sought from head tohead trials, and to compute indirect comparisons, direct comparisons of eachdrug with placebo were sought. The methodology followed best practice, as setout by Cochrane Collaboration, for the conduct of meta analyses.
My main question about the meta analyses concerns the drug doses. The usualdose seems to be 20mg/day for citalopram, and 10mg/day for escitalopram.
Treatment at any level is included in the meta-analyses. There is no explanationas to why there has been no differentiation between different doses. Most trialsare testing the usual dose, but some do not. In the head to head meta analyses,Moore 2005 tests double the dose for both drugs, Burke 2002 is a three arm trialtesting 10, and 20 escitalopram compared with 20mg citalopram, Yevtushenko2007 compares 20mg escitalopram with 2 arms of citalopram of 10 and 20 mg,and Ou compares 10mg escitalopram with 20 citalopram, although both armscould double the dose in the latter part of the study. SCT-MD-02 is anunpublished study and I do not know how many arms there were to the study, orwhat doses were tested. In these meta analyses, where there are 2 arms for onedrug at different doses, the data are pooled, to provide an overall comparison ofthe two drugs. There appears to be a difference in response between 2 differentlevels of a single drug. Does this variation in dose account for some of the heterogeneity across trials? The meta analyses of the placebo controlledcomparisons also contain pooled data from more than one arm. Is theinconsistency between direct and indirect measurements of treatment effect dueto dose differences? The acquisition and analysis of these data provide extremely interesting results.
One fact that these data seem to show is that the costs of a DDD unit of each ofthe three drugs (citalopram, generic citalopram and escitalopram) has declinedsince 2005 from, in the same order, (0.70 0.46 0.55) euros to (0.58 0.35 0.50)euros. The cost of a DDD generic is 70% of a DDD of escitalopram, and a DDDof citalopram is more expensive than escitalopram DDD. The difference is small.
Of course if it were true that the treatments were equally effective the genericwould provide the best value and by the 2010 data would save 3 million euros ayear in France. It would be interesting for the authors to comment on thesefigures.
Quality of written English: Acceptable
Statistical review: No, the manuscript does not need to be seen by a
statistician.
Declaration of competing interests:
I declare that I have no competing interests

Source: http://www.anzhealthpolicy.com/imedia/6680129479112353_comment.pdf