The aim of this study was to estimate the cost effectiveness of nefazo
done compared with imipramine or fluoxetine in treating women with maj
or depressive disorder. Clinical decision analysis and a Markov state-
transition model were used to estimate the lifetime health outcomes an
d medical costs of 3 antidepressant treatments. The model, which repre
sents ideal primary care practice, compares treatment with nefazodone
to treatment with either imipramine or fluoxetine. The economic analys
is was based on the healthcare system of the Canadian province of Onta
rio, and considered only direct medical costs, Health outcomes were ex
pressed as quality-adjusted life years (QALYs) and costs were in 1993
Canadian dollars ($Can; $Can1 = $US0.75, September 1995). Incremental
cost-utility ratios were calculated comparing the relative lifetime di
scounted medical costs and QALYs associated with nefazodone with those
of imipramine or fluoxetine. Data for constructing the model and esti
mating necessary parameters were derived from the medical literature,
clinical trial data, and physician judgement. Data included informatio
n on: Ontario primary care physicians' clinical management of major de
pression; medical resource use and costs; probabilities of recurrence
of depression; suicide rates; compliance rates; and health utilities.
Estimates of utilities for depression-related hypothetical health stat
es were obtained from patients with major depression (n = 70). Medical
costs and QALYs were discounted to present value using a 5% rate. Sen
sitivity analyses tested the assumptions of the model by varying the d
iscount rate, depression recurrence rates, compliance rates, and the d
uration of the model. The base case analysis found that nefazodone tre
atment costs $Can1447 less per patient than imipramine treatment (disc
ounted lifetime medical costs were $Can50 664 vs $Can52 111) and incre
ases the number of QALYs by 0.72 (13.90 vs 13.18). Nefazodone treatmen
t costs $Can14 less than fluoxetine treatment (estimated discounted li
fetime medical costs were $Can50 664 vs $Can50 678) and produces sligh
tly more QALYs (13.90 vs 13.79). In the sensitivity analyses, the cost
-effectiveness ratios comparing nefazodone with imipramine ranged from
cost saving to $Can17 326 per QALY gained. The cost-effectiveness rat
ios comparing nefazodone with fluoxetine ranged from cost saving to $C
an7327 per QALY gained. The model was most sensitive to assumptions ab
out treatment compliance rates and recurrence rates. The findings sugg
est that nefazodone may be a cost-effective treatment for major depres
sion compared with imipramine or fluoxetine. The basic findings and co
nclusions do not change even after modifying model parameters within r
easonable ranges.