Methodologies used in cost-effectiveness models for evaluating treatments in major depressive disorder: a systematic review

Zimovetz, Evelina A; Wolowacz, Sorrel E; Classi, Peter M; Birt, Julie

doi:10.1186/1478-7547-10-1

Cost Effectiveness and Resource Allocation

Table 1 Summary of methods and conclusions of included studies

From: Methodologies used in cost-effectiveness models for evaluating treatments in major depressive disorder: a systematic review

Author, Year, Country	Compara-tors	Analysis Type, Model Structure	Time Horizon, Perspective	Primary Out-come	Definition of Effective-ness	Source of Primary Clinical Data	Sensitivity Analyses	Main Finding
Armstrong et al., 2007 [8] US	Escitalopram vs. sertraline	CUA; decision- tree	6 months, payer perspective	QALYs	Response: ≥ 50% improve-ment in MADRS	8-week head-to-head trial	Univariate, probabilistic	Escitalopram dominated^a sertraline
Armstrong et al., 2008 [9] US	Escitalopram vs. duloxetine	CUA; Markov structure; 1-week cycle	1 year, payer perspective	QALWs	Remission: MADRS ≤ 12 or HAMD-17 ≤ 7	Pooled analysis of 10 RCTs	Univariate, probabilistic	Escitalopram dominated^a duloxetine
Aziz et al., 2005 [10] US	MPT vs. MECT^b	CUA; Markov structure; 6-month cycle	Lifetime, payer and societal perspective	QALYs	Remission: not explicitly defined	Published literature	Univariate	MECT may be more cost-effective vs. MPT
Benedict et al., 2010 [11] Scotland	Primary care: duloxetine vs. SSRIs, venlafaxine ER, mirtazapine	CUA; Markov structure; 8-week cycle	48 weeks, payer perspective	QALYs	Remission and response: HAMD-17 scale, scores not reported	Pooled analysis of 8 RCTs, a meta- analysis^c	Univariate, probabilistic	For the commonly accepted WTP thresholds, duloxetine was the preferred option
	Secondary care: duloxetine vs. venlafaxine ER, mirtazapine					Pooled analysis of 2 head-to-head trials, a meta-analysis^d		Duloxetine dominated^a venlafaxine ER and mirtazapine
Borghi and Guest, 2000 [12] UK	Mirtazapine vs. amitriptyline	CEA; decision-tree	7 months, payer perspective, societal perspective	Treatment success	Remission: HAMD-17 ≤ 7	Meta-analysis of 4 RCTs of 7-month duration	Univariate	Mirtazapine was cost-effective vs. amitriptyline and fluoxetine
	Mirtazapine vs. fluoxetine		6 months, payer perspective, societal perspective	Treatment success	Response: ≥ 50% improve-ment in 17-HAMD	6-week head-to-head trial
Brown et al., 2000 [13] France	Mirtazapine vs. fluoxetine	CEA; decision-tree	6 months, societal perspective	Treatment success	Response: ≥ 50% improve-ment in 17-HAMD, HAMD-21, or a score of 1 or 2 on CGI	6-week head-to-head trial	Univariate	Mirtazapine was cost-effective vs. fluoxetine
Casciano et al., 2001 [14] 10 countries^e	Venlafaxine ER vs. SSRIs, TCAs	CEA; decision-tree	6 months, payer perspective	Treatment success, SFDs^f	Response: 50% improve-ment in HAMD or MADRS	Meta-analysis	Univariate, probabilistic	Venlafaxine ER dominated^a SSRIs and TCAs in 9 of the 10 countries
Casciano et al., 2000 [15] US	Venlafaxine ER vs. SSRIs, TCAs	CEA; decision-tree	6 months, payer perspective	Treatment success, SFDs^f	Response: 50% improve-ment in HAMD or MADRS	Meta-analysis	Univariate, probabilistic	Venlafaxine ER dominated^a SSRIs and TCAs
Dardennes et al., 2000 [16] France	Preventative strategy^g vs. episodic strategy^h	CUA; Markov structure; 8-week cycle	12 months, payer perspective	QALYs	Remission: HDRS-21 < 8	12-month double-blind trial	Univariate	Cost of maintenance therapy was partially offset by the gain from recurrence prevention
Demyttenaere et al., 2005 [17] Belgium	Escitalopram vs. citalopram, venlafaxine	CEA; decision-tree	6 months, payer and societal perspec-tives	Treatment success	Remission: MADRS ≤ 12	Meta-analysis of three 8-week RCTs	Univariate, probabilistic	Escitalopram dominated^a citalopram and was cost-effective vs. venlafaxine
Doyle et al., 2001 [18] 10 countries^e	Venlafaxine vs. SSRIs, TCAs	CEA; decision-tree	6 months, payer perspective	Treatment success, SFDs^f	Response: 50% improve-ment in HAMD or MADRS	2 meta-analyses	Univariate, probabilistic	Venlafaxine dominated^a SSRIs and TCAs in 9 of 10 countries (inpatients); 8 of 10 countries (outpatients)
Francois et al., 2002 [19] Finland	Escitalopram vs. citalopram, fluoxetine, venlafaxine	CUA; decision-tree	6 months, societal perspective	Treatment success, QALYs	Remission: MADRS ≤ 12	8-week head-to-head trial, indirect comparison	Univariate	Escitalopram dominated^a citalopram, fluoxetine and venlafaxine
Francois et al., 2003 [20] Norway	Escitalopram vs. citalopram, fluoxetine, venlafaxine	CEA; decision-tree	6 months, societal perspective	Treatment success	Remission: MADRS ≤ 12	8-week head-to-head trial, indirect comparison	Univariate	Escitalopram dominated^a citalopram, fluoxetine and venlafaxine
Freeman et al., 2000 [21] UK	Venlafaxine vs. SSRIs, TCAs	CEA; decision-tree	6 months, payer perspective	Treatment success, SFDs^f	Response: 50% improve-ment in HAMD or MADRS	Meta-analysis	Univariate, probabilistic	Venlafaxine dominated^a SSRIs and TCAs
Haby et al., 2004 [22] Australia	CBT vs. SSRIs	CEA; decision-tree	9 months, payer perspective	DALYs	Multiple outcomes averaged for individual studies	Meta-analysis	Probabilistic	CBT provided by public psychologist was the most cost-effective option
Hemels et al., 2004 [23] Austria	Escitalopram vs. citalopram	CEA; decision-tree	6 months, payer and societal perspec-tives	Treatment success	Remission: MADRS ≤ 12	8-week head-to-head trial	Univariate, probabilistic	Escitalopram dominated^a citalopram
Howard and Knight, 2004 [24] Austria	Venlafaxine ER, venlafaxine IR, SSRIs	CEA; decision-tree	16 week, payer perspective	SFDs^f	Remission: not explicitly reported	Meta-analysis	Probabilistic	Venlafaxine ER was cost-effective vs. venlafaxine IR and SSRIs; SSRIs were least cost-effective
Kongsakon and Bunchapat-tanasakda, 2008 [25] Thailand	Escitalopram vs. fluoxetine, venlafaxine	CEA; decision-tree	6 months, payer and societal perspec-tives	Treatment success	Remission: MADRS ≤ 12	2 meta-analysesⁱ	Univariate, probabilistic	Escitalopram dominated^a fluoxetine and venlafaxine
Kulp et al., 2005 [26] Germany	Escitalopram vs. venlafaxine ER	CEA; Markov structure; 2-week cycle	70 days, payer perspective	Treatment success	Response: > 50% improve-ment in MADRS Partial response: 25-50% No response: < 25%	8 week head-to-head trial	None	Escitalopram was cost-effective vs. venlafaxine ER
Lenox-Smith et al., 2004 [27] UK	Venlafaxine vs. TCAs, SSRIs (fluoxetine, paroxetine, and fluvoxamine)	CEA; decision-tree	6 months, payer perspective	SFDs^f	Remission: 17- HAMD ≤ 7 Response: ≥ 50% improve-ment in HAMD-21	Meta-analysis and a single study	Univariate	Venlafaxine dominated^a SSRIs and TCAs
Lenox-Smith et al., 2009 [28] UK	Venlafaxine vs. fluoxetine, amitriptyline	CUA; decision-tree	6 months, payer perspective	QALYs	Remission: HAMD-17 ≤ 7 Response: ≥ 50% improve-ment in HAMD-17	Pooled data from 13 clinical trials	Univariate	Venlafaxine dominated^a fluoxetine and amitriptyline; fluoxetine dominated^a amitriptyline
Löthgren et al., 2004 [29] Sweden	Escitalopram vs. citalopram, venlafaxine	CEA; decision-tree	6 months, payer and societal perspective	Treatment success	Remission: MADRS ≤ 12	Meta-analysis	Univariate, probabilistic	Escitalopram dominated^a citalopram and venlafaxine
Machado et al., 2007 [30] Brazil	SNRIs vs. SSRIs, TCAs	CEA; decision-tree	6 months, payer perspective	Treatment success	Remission: score ≤ 7 on HAMD or ≤ 12 on MADRS	Meta-analysis	Univariate, probabilistic	SNRIs dominated^a SSRIs and TCAs
Malone, 2007 [31] US	SSRIs, escitalopram, paroxetine CR, sertraline, venlafaxine ER	CEA; structure not explicitly reported	6 months, payer perspective	Treatment success	Response: ≥ 50% improve-ment in HAMD or MADRS Remission: HAMD ≤ 7 or MADRS ≤ 10	Pooled analysis of trials	Univariate, probabilistic	Venlafaxine had the lowest ICER followed by escitalopram and sertraline^j; paroxetine was dominated^k
Nuijten, 2001 [32] The Netherlands	Prolongation of antidepressant medication vs. no prolongation	CUA; Markov structure; 8-week cycle	9 months, payer and societal perspec-tives	QALYs, TWD	Not explicitly reported	Published literature	Univariate	Continuation treatment was not cost-effective, unless extended to maintenance
Perlis et al., 2009 [33] US	Test for SSRI responsive-ness vs. no test	CUA; Markov structure; 3-month cycle	3 years, societal perspective	QALYs	Remission: Instrument not explicitly reported	STAR*D trial [34]	Univariate, two-way	The ICER for the genetic test would not be considered cost effective
Sado et al., 2009 [35] Japan	COMBI vs. AD	CUA; decision-tree	12 months, payer and societal perspec-tives	Treatment success, QALYs	No response: HRSD-17 > 6 or HRSD-24 > 8	Meta-analysis of 8 RCTs	Univariate, probabilistic	COMBI was cost-effective
Simon et al., 2006 [36] UK	COMBI vs. AD	CUA; decision-tree	15 months, payer perspective	Treatment success, QALYs	Remission: HRSD-17 ≤ 6 or HRSD-24 ≤ 8	Meta-analysis	Univariate, probabilistic	COMBI was cost-effective
Sobocki et al., 2008 [37] Sweden	Venlafaxine maintenance treatment vs. placebo	CUA; Markov structure; 1-month cycle	2 years, payer and societal perspec-tives	QALYs	Time to recurrence: 17-HAMD > 12 and ≥ 50% improve-ment in 17-HAMD	2-year trial	Univariate, probabilistic	Maintenance treatment with venlafaxine was cost-effective
Sorenson et al., 2007 [38] Denmark	Escitalopram vs. citalopram, venlafaxine ER	CEA; decision-tree	6 months, payer and societal perspec-tives	Treatment success	Remission: MADRS ≤ 12	Meta-analysis	Univariate, probabilistic	Escitalopram dominated^a citalopram; similar cost-effectiveness vs. venlafaxine ER
Sullivan et al., 2004 [39] US	Escitalopram, citalopram, fluoxetine, venlafaxine ER, sertraline, paroxetine, paroxetine CR, venlafaxine	CUA; decision-tree	6 months, payer perspective	QALYs	Response: > 50% improve-ment in MADRS (with treatment maintained for ≥ 180 days)	N/A^l	Univariate, probabilistic	Escitalopram dominated^a all treatments^m
Trivedi et al., 2004 [40] US	Venlafaxine ER vs. SSRIs (fluvoxamine, fluoxetine, paroxetine)	CUA; decision-tree	8 weeks, payer perspective	QADs and DFDs	Response: HAMD < 15 and/or ≥ 50% improve-ment in HAMD. Remission: HAMD ≤ 7 Response without remission: HAMD 8-14	Pooled analysis of 8 RCTs	Probabilistic	Venlafaxine ER was cost-effective vs. SSRIs
van Baardewijk et al., 2005 [41] Canada	Duloxetine vs. venlafaxine ER	CEA; decision-tree	6 months, payer and societal perspec-tives	Treatment success, SFDs	Remission: HAMD ≤ 7 or MADRS ≤ 10	Meta-analysis	Univariate, probabilistic	Venlafaxine ER dominated^a duloxetine
Vos et al., 2005 [42] Australia	Listed in footnoteⁿ	CEA; decision-tree	9 months and 5 years, payer perspective	DALYs	Multiple outcomes	Meta-analyses	Probabilistic	All interventions had a favourable ICER under Australian health service conditions
Wade et al., 2005 [43] UK	Escitalopram vs. citalopram, venlafaxine	CEA; decision-tree	6 months, payer and societal perspective	Treatment success	Remission: MADRS ≤ 12	Meta-analysis	Univariate, probabilistic	Escitalopram dominated^a citalopram; similar cost-effectiveness vs. venlafaxine
Wade et al., 2005 [44] UK	Escitalopram vs. citalopram	CEA; decision-tree	6 months, payer and societal perspec-tives	Treatment success	Remission: MADRS ≤ 12 Response: ≥ 50% improve-ment in MADRS	Meta-analysis	Univariate, probabilistic	Escitalopram dominated^a citalopram
Xie et al., 2009 [45] Singapore	Escitalopram vs. venlafaxine, fluvoxamine	CEA; decision-tree	6 months, societal perspective	Treatment success	Remission: MADRS score ≤ 12	Head-to-head trials^o	Univariate, probabilistic	Escitalopram dominated^a venlafaxine and fluvoxamine

AD = antidepressant therapy; CEA = cost-effectiveness analysis; CBT = cognitive behavioural therapy; CGI = Clinical Global Impression; COMBI = combination therapy; CR = controlled release; CUA = cost-utility analysis; DALY = disability-adjusted life-years; DFD = disease-free day; ER = extended release; HAMD = Hamilton Depression Rating Scale; HDRS = Hamilton Depression Rating Scale; HRSD = Hamilton Rating Scale for Depression; ICER = incremental cost-effectiveness ratio; IR = instant release; MADRS = Montgomery Åsberg Depression Rating Scale; MDD = major depressive disorder; MECT = maintenance electroconvulsive therapy; MPT = maintenance pharmacotherapy; N/A = not applicable; QAD = quality-adjusted day; QALW = quality-adjusted life week; QALY = quality-adjusted life-year; RCT = randomised controlled trial; SFD = symptom-free day; SNRI = serotonin-norepinephrine reuptake inhibitor; SSRI = selective serotonin reuptake inhibitor; STAR*D = Sequenced Treatment Alternatives to Relieve Depression trial; TCA = tricyclic antidepressants; TWD = time without depression; UK = United Kingdom; US = United States; WTP = willingness to pay.
^a A dominant therapy is less expensive and more effective than the comparator.
^b MPT is compared with MECT in elderly individuals with MDD who relapsed after responding to initial course of electroconvulsive therapy.
^c Included 6 placebo-controlled RCTs with SSRIs as active arm comparators and 2 head-to-head venlafaxine trials; a meta-analysis was used to obtain response and remission rates for mirtazapine.
^d In the absence of efficacy data for mirtazapine in patients with more severe illness, mirtazapine rates were calculated by applying the mean differences between the less severe and the more severe population to the mirtazapine rates reported in the meta-analysis for the patient population with HAMD-17 ≥ 18.
^e Germany, Italy, Netherlands, Poland, Spain, Sweden, Switzerland, UK, US, Venezuela.
^f SFDs were measured as time elapsed after the determination of success through the end of the period being analysed.
^g Continuous treatment with milnacipran.
^h Follow-up with no preventive treatment.
ⁱ Due to lack of published data on head-to-head comparison of escitalopram and fluoxetine, the clinical data comparing escitalopram to citalopram were used as a proxy, derived from a published meta-analysis.
^j ICERs were calculated using SSRIs as the reference group.
^k A therapy dominated by other comparators has higher cost and lower effectiveness.
^l Due to a lack of consistent and comprehensive studies demonstrating differences in efficacy across all 8 serotonin reuptake inhibitors, and in order to retain a specific focus on the impact of adverse drug reactions on treatment costs, it was assumed that on average 60% of patients respond to serotonin reuptake inhibitor therapy.
^m Escitalopram had lowest direct costs and the greatest effectiveness, followed by citalopram, generic fluoxetine, venlafaxine ER, sertraline, generic paroxetine, paroxetine CR, and venlafaxine IR.
ⁿ SSRIs (acute, continuation, maintenance), TCAs (acute, continuation, maintenance), Bibliotherapy (acute), Acute and maintenance: individual CBT public psychologist, individual CBT, private psychologist, individual CBT public psychiatrist, individual CBT private psychiatrist, group CBT public psychologist.
^o Due to lack of data on head-to-head comparisons of escitalopram and fluvoxamine, a head-to-head comparison of citalopram and fluvoxamine was used as a proxy.

Back to article page

ISSN: 1478-7547

Contact us

General enquiries: journalsubmissions@springernature.com