Cost – E ff ectiveness of Depressive Episode Pharmacological Treatment A

1 Department of Pharmacology, Th e Faculty of Medical Sciences, University of Kragujevac, Serbia 2 Department of Economics, Center for Children and Childhood Studies, Rutgers University, the State University of New Jersey, U.S.A. 3 Department of Epidemiology & Community Health, School of Health Sciences & Practice, New York Medical College, U.S.A. 4 Clinic for Neurology and Psychiatry for Children and Youth, Clinical Center of Serbia, Belgrade, Serbia 5 Psychiatric Clinic, University Clinical Center Kragujevac, Serbia 6 Centre for Clinical Pharmacology, Medical Faculty, Military Medical Academy University of Defence, Belgrade Serbia 7 Faculty of Medical Sciences University of Kragujevac, Serbia 8 Primary Care Facility Svilajnac, Serbia


INTRODUCTION
Major depressive disorder [MDD] belongs to the chronic mental illnesses with the most signifi cant budget impact worldwide [1].Th is is the case both due to demanding medical care and far reaching consequences of decreased working ability and premature mortality [2].Antidepressants [ADs] are considered to be the most frequently used intervention for mood disorders treatment.Accumulated evidence from systematic reviews and meta-analyses has shown that diff erent pharmacological ADs do not exhibit substantial effi cacy diff erentials among majority of patients [3].First choice of ADs in treating individual patients is largely determined by the clinical features of the depressive episode, patient preferences, and the aff ordability of specifi c drugs [4].
Th ere is a large volume of published evidence on cost-eff ectiveness [5] and costutility of diff erent depression treatment strategies [6].Most of these assessments were recently compared in a well designed systematic review [7].Multiple treatment meta-analysis reported that one of the selective serotonin reuptake inhibitors [SSRIs] escitalopram, was the most cost-eff ective therapy of depression in primary care settings among the other antidepressants tested [citalopram, duloxetine, fl uoxetine, fl uvoxamine mirtazapine, paroxetine, reboxetine, sertraline and venlafaxine] [8].Although very useful indeed, the international debate was raised whether the incremental cost-eff ectiveness ratio [ICER] [9] and cost per quality-adjusted life year [QALY] approaches are providing enough back up for informed decision making [10].Nevertheless cost-eff ectiveness study remains the best methodological framework presenting broad horizon and suffi cient detail to policy makers [11].However, it is important to take into account not only the drug itself, but also whether it is original or a generic copy [12].Balkan markets have been fl ooded with generic drugs manu-factured mostly by Indian companies over the course of years, substantially decreasing drug acquisition costs in the cost-eff ectiveness equation.However additional eff ort to assess actual contribution of this eff ect was rather out of scope and budget of current study.Aims of this research were to fi ll the literature gap and compare costs and clinical outcomes of pharmacological treatment protocols in major depressive disorder.Th is has been done using a prospective randomized clinical study in an upper-middle income Eastern European setting [13].

SUBJECTS AND METHODS
Setting Th e trial was conducted at a psychiatry clinic within the 1,300 bed, large university hospital of Kragujevac in Serbia.Th e hospital is the only tertiary facility in the region, providing approximately 50,000 hospital admissions and 400,000 outpatient examinations per year.Th e psychiatry clinic has a 61 bed capacity and 30 admission places in the intensive outpatient facility.Among designs considered were longitudinal cohort studies, retrospective studies and case series.Th e most proper study design to address research questions was a clinical case series.Th e previous two methods were eliminated either on grounds of being too demanding in terms or long time horizon and budget needed or lack of relevant clinical data.Th e one selected won on grounds of providing suffi cient insight into costs and outcomes while off ering best feasibility to be conducted alongside routine clinical practice and in real world setting.Participating researchers/attending physicians were not blinded for the subgroup allocation process.Patient sub-populations were compared among themselves without a placebo controlled group.Such a decision has been made due to the serious ethical issue of placebo administration in severe depression due to increased risk of suicidal tendency oc- Outpatients were 1,777 by F.32 and 844 by F.33, who were examined at the regional psychiatry clinic during the period of 2010-2012.Out of this available prevalent pool of patients, our 65 cases were randomly selected out of those fulfi lling inclusion and exclusion criteria.Th e patient recruitment was based on either experience of a fi rst depressive episode, or clinical relapse which were previously diagnosed with chronic recurrent depression.Both conditions were demanding pharmacological treatment introduction.
Principles of good clinical practice [GCP] were obeyed and patients signed an informed consent to participate in the trial.Inclusion criteria were confi rmed diagnosis of major depressive disorder-single episode or major depressive disorder by the ICD-10.Th e prescription of the selected pharmacology treatment protocols was based on attending psychiatrist's independent clinical assessment.Exclusion criteria were age less than 18, pregnancy, breast feeding, presence of other major mental illness, and presence of cognitive impairment or signifi cant physical comorbidities.
Duration of clinical follow up per patient was eight weeks from the inclusion time point.Th ree cross sections were done: A zero week inclusion time point as the baseline aft er three weeks and eight weeks from the inclusion point.Resource use patterns and costs were evidenced for up to 14 weeks from study entry.Chronological cross-sections were planned in accordance with expected occurrence of clinically signifi cant drug eff ectiveness [15].Th ere were in total 65 patients included in the trial.Five patients were lost to follow up either due to change of contact details, poor study compliance, missing physician visits, and lack of relevant data.Ultimately 60 patients providing insight into all of the necessary data were considered in further analysis.

Intervention
Pharmacology treatment protocols were prescribed according to evidence-based clinical practice guidelines [16].Each patient had one major antidepressant agent administered in a full dosing regimen.Th ese cases were treated by either one of the following medicines: escitalopram, venlafaxine, sertraline, fl uoxetine, paroxetine, maprotiline, bupropion, trazodone or mirtazapine.Based on pharmacological properties of the main antidepressant drug, selected patients belonged to either the selective serotonin reuptake inhibitors [SSRIs] group, serotonin-norepinephrine reuptake inhibitors [SNRIs] or Heterocyclic antidepressant group for the fi nal analysis.Th e vast majority of prescribed and consumed antidepressants and add-on drugs were original brand name preparations with limited generic medicines presence.

Outcomes
Patient-reported outcomes are widely regarded as a particular stronghold of treatment success estimation in mood disorder [17].An attending physician participating in our study was focused on psychometric assessment and two diff erent patient-centered qualities of life [QOL] questionnaires.Major outcomes were clinical improvement-response to treatment assessed by the Hamilton Depression Rating Scale [HDRS] [18] and life quality assessed by the Quality of Life Enjoyment and Satisfaction Questionnaire [Q-LES-Q-SF] [19].Th ese instruments used were standardized and validated in Serbian language and exploited in a previously published study [20].
Hamilton Depression Rating Scale [HDRS] was widely accepted as a "gold standard" questionnaire for clinical ratings of depressive mood disorder.Th e original 1960 scale consists of 17 items.Each item on the questionnaire is scored on a 3 or 5 point scale, depending on the item, and the total score is compared to the corresponding descriptor.Normal score range is 0-7 and indicates absence of major mood disorders.Scores ≥ 20 or higher indicate moderate or severe depression [21].
Th e Q-LES-Q-SF is a-16-item selfreported questionnaire.Th e scale is created to assess overall satisfaction with physical health, mood, job duties, household and leisure time activities, social and family relationships, daily functioning, sexual life, economic status, over-all well-being and medications.Responses are scored on a 5-point Likert type scale, where higher scores indicate better enjoyment and satisfaction with life [possible range .Fourteen summated items create the total Q-LES-Q-SF score [19].
An unit indicator to present success rate of a treatment option was calculated as a day free from depressive symptoms [DFD] [22].Th e cost per quality adjusted life year [QALY] was a proxy of a cost-utility indicator [23].In order to compare alternative treatment strategies among subgroups, this study evaluated medical care costs per cured depressive episode.Cost-eff ectiveness measure was assessed by cost/DFD unit itself.
Th e number of DFDs was calculated according to the following principles: if a patient's HDRS score was ≥ 22, it was assumed that patient had no DFD; if HDRS was ≤ 7, it was assumed to have a full DFD; if 22 ≥ HDRS ≥ 7 then DFD was weighted in a proportionate manner.Th e summary amount of DFD over eight weeks interval was obtained by adding the calculated number of DFDs for the baseline of 8 weeks HDRS score, dividing by two, and multiplying by the 56-day interval between study entry and fi nal assessment.Each whole DFD gained was multiplied by a 0.41 ratio for one additional quality adjusted day.In this way, the value of QALY gains was obtained [24][25][26][27].

Resource Utilization
Th is study adopted broad societal perspective.Th erefore, most direct medical costs and indirect productivity related costs were taken into account.Th e expenses of hospital inpatients and outpatients were extracted from electronic registry of discharge invoices provided by the institutional framework of the university clinic.Th e offi cial price list of Republican Health Insurance Fund of Serbia at the time of service provision was adopted.Th e expenses of outof-pocket drug acquisitions were evidencebased by a recommended treatment regimen at hospital discharge with patient compliance assessment.Average market prices at the time of service provision were taken into account.Indirect productivity loss attributed to patient care, absenteeism, working days lost, and premature working disability were calculated based on Grossman's human capital approach [28].Adopted time horizon was 14 weeks, which were regarded suffi cient to ob-serve most of the outcomes and resource use attributable to depressive episode and its treatment.Discounting was neglected due to rather short-time horizon not allowing for substantial currency value distortion.

Data Analytic Procedures
For continuous variables, mean [M] and standard deviation [SD] were calculated.Fisher's or χ2 test were applied to assess diff erences for categorical variables.Student t-test or Mann-Whitney U-tests were used to assess diff erences for continuous variables, depending on the distribution of data.In order to analyze statistical signifi cance of continuous variables across three chronological cross-sections within the same group of patients, unifactorial analysis of variance [ANOVA] for parameter data, and the Friedman Test for nonparameter data were employed.All p-values of 0.05 or below were considered statistically signifi cant.Th e accepted level of signifi cance for the probability of error of the fi rst order is 0.05.To process the data we will use the statistical package "SPSS" version 19.

RESULTS
All patients were between 23 and 76 years old as visible in Table 1.Th ere were 33 females or about 55% of total sample but this sex diff erence was not statistically signifi cant [χ2=0.60;р=0.439]Nevertheless, age stratifi ed analysis showed that there was signifi cant domination of 6th life decade patients [ Average indirect productivity loss related costs were also substantially higher within the SNRI group [C2=9.995;p=0.007].Although the cost of primary medical care is also highest within SNRI [10,029.08CSD], this difference was not signifi cant.Unlike these ones, costs of hospital care are highest within the SSRI group, although this diff erence was not signifi cant either.Total drug acquisition costs were insignifi cantly higher within SSRI group compared to another two.
Duration of job absenteeism among patients and/or the family member in charge of their home care was signifi cantly longer within SNRI group [49.50 days] in Table 3. Within the same group, total duration of hospital admission was signifi cantly longer.Total number of hospital admissions during follow up was insignifi cantly higher within the SNRI group.An average patient was admitted 2.44 times.Total frequency of outpatient visits was slightly higher in the SSRI group compared to others but not signifi cantly.An average patient in the total sample visited the outpatient clinic approximately 1.5 times during observation period.
In Table 4, an average duration of depressive episode intensive treatment was 50.97 days.Clear cut evidence on gradual reduction of HDRS-17 score from point zero toward the last 8th week physician visit was observed in the total sample.Th ese diff erences turned out to be statistically signifi cant.Another relevant trend was decrease of percentage of patients with most severe clinical forms of depression and increasing number of patients entering reemission stage.Th e 14-week long time horizon we selected confi rmed that clinical improvement mostly happens in an acute stage of a depressive episode lasting 6-8 weeks, conditional to the appropriate drug and dosing regimen choice.

DISCUSSION
Straight forward negative correlation be-Straight forward negative correlation between depressive symptoms of severity and tween depressive symptoms of severity and quality of life was reported in diff erent fi nd-quality of life was reported in diff erent fi ndings [29].Our results fall within the so called ings [29].Our results fall within the so called B strategy of depression treatment with cost B strategy of depression treatment with cost per QALY ≤ $ 20,000 [30 ].Depression medi-per QALY ≤ $ 20,000 [30 ].Depression medical care expenditure reported in the literature cal care expenditure reported in the literature diff ers substantially among countries.Few re-diff ers substantially among countries.Few results opposing our results were reporting the sults opposing our results were reporting the most convenient ICER values with venlafaxine most convenient ICER values with venlafaxine [SNRI] of $2,073 and highest reemission rates [SNRI] of $2,073 and highest reemission rates [1].Th e local Brazilian trial has also reported [1].Th e local Brazilian trial has also reported lowest costs of care and better treatment out-lowest costs of care and better treatment outcomes for SNRI classifi ed medicines [31].Re-comes for SNRI classifi ed medicines [31].Recently published UK data favors venlafaxine cently published UK data favors venlafaxine treatment in terms of both cost-utility [1,285£ treatment in terms of both cost-utility [1,285£ per QALY] and 21£ cost per DFD compared per QALY] and 21£ cost per DFD compared to alternatives [32].Unlike previously men-to alternatives [32].Unlike previously mentioned fi ndings, a local Columbian study re-tioned fi ndings, a local Columbian study reports lowest cost-eff ectiveness ratio for more ports lowest cost-eff ectiveness ratio for more traditional amitriptyline of $1,274 and with traditional amitriptyline of $1,274 and with cost per QALY being even $31,595 lower than cost per QALY being even $31,595 lower than the venlafaxine reported value [33].the venlafaxine reported value [33].
Observing the landscape of published Observing the landscape of published evidence on cost-eff ectiveness and cost-utility evidence on cost-eff ectiveness and cost-utility of antidepressants treatment, we may notice of antidepressants treatment, we may notice that high income clinical settings and few ma-that high income clinical settings and few ma-jor emerging ones favor SNRI [venlafaxine] jor emerging ones favor SNRI [venlafaxine] both due to sudden clinical improvement at-both due to sudden clinical improvement attributed to this medicine and most aff ordable tributed to this medicine and most aff ordable cost per depression free day achieved.Th e last cost per depression free day achieved.Th e last criterion is consistent with our fi ndings.criterion is consistent with our fi ndings.
Reported estimates of high SSRI re-Reported estimates of high SSRI related hospital resource use and costs seem to lated hospital resource use and costs seem to be consistent with our fi ndings.Among some be consistent with our fi ndings.Among some South American emerging markets, cost per South American emerging markets, cost per day of hospital treatment [$81 in Columbia and day of hospital treatment [$81 in Columbia and $293 in Brazil] remains similar to our reported $293 in Brazil] remains similar to our reported values, but in high income markets of North values, but in high income markets of North America it can be 8-15 times more expensive.America it can be 8-15 times more expensive.Th ere is evidence supporting the claim that Th ere is evidence supporting the claim that hospital admission costs amount to almost 43-hospital admission costs amount to almost 43-75% of total costs and are therefore the leading 75% of total costs and are therefore the leading contributor to the total direct medical costs of contributor to the total direct medical costs of depression care [2].Antidepressant medicine depression care [2].Antidepressant medicine costs diff er substantially among markets but costs diff er substantially among markets but usually fall within 2% -11% range [34 ]. usually fall within 2% -11% range [34 ].
Clinical assessment by means of Clinical assessment by means of HDRS-17 reported values steadily decreased HDRS-17 reported values steadily decreased in all Serbian patients during the follow up in all Serbian patients during the follow up period.Th ese diff erences among group values period.Th ese diff erences among group values of HDRS-17 and Q-LES-Q-SF detected during of HDRS-17 and Q-LES-Q-SF detected during three scheduled visits to the attending psychia-three scheduled visits to the attending psychiatrist proved to be statistically signifi cant.Th is trist proved to be statistically signifi cant.Th is evidence proves clinical improvement in terms evidence proves clinical improvement in terms of depression free days and life quality on total of depression free days and life quality on total sample.Nevertheless, this trial supports widely sample.Nevertheless, this trial supports widely adopted knowledge on modest effi ciency dif-adopted knowledge on modest effi ciency differentials among major antidepressant medi-ferentials among major antidepressant medicines.Th e dominant elderly population that cines.Th e dominant elderly population that has been observed is particularly vulnerable to has been observed is particularly vulnerable to the depressive mood disorders [35].Popula-the depressive mood disorders [35].Population of elderly has additional diffi culties with tion of elderly has additional diffi culties with healthcare access which has been proven in healthcare access which has been proven in other communities worldwide [36].other communities worldwide [36].
Many patients aft er successful out-Many patients aft er successful outcomes in psychiatric disease management en-comes in psychiatric disease management enter lasting remissions [37] which may refl ect ter lasting remissions [37] which may refl ect to life quality improvement.Real world health to life quality improvement.Real world health economic assessments alongside clinical trials economic assessments alongside clinical trials on effi ciency off er some signifi cant method-on effi ciency off er some signifi cant method- ological advantages compared to modelling ological advantages compared to modelling approach in terms of microcosting and clini-approach in terms of microcosting and clinical outcomes assessment [38].Size of mental cal outcomes assessment [38].Size of mental disorder related costs of care seems to be just disorder related costs of care seems to be just slightly lower in Serbia compared to Central slightly lower in Serbia compared to Central European values [39,40].Structure of costs in-European values [39,40].Structure of costs indicates direct costs are prevailing against indi-dicates direct costs are prevailing against indirect cost contribution.Th is is profoundly dif-rect cost contribution.Th is is profoundly different in high income settings which is caused ferent in high income settings which is caused by substantially higher labor wages and there-by substantially higher labor wages and therefore higher signifi cance of job absenteeism and fore higher signifi cance of job absenteeism and lost productivity due to lasting mood disorders lost productivity due to lasting mood disorders [41,42 ]. [41,42 ].

Study Limitations Study Limitations
Th is study had an uneasy task to address far Th is study had an uneasy task to address far reaching issues of costs and consequences of reaching issues of costs and consequences of pharmacological treatment protocols in ma-pharmacological treatment protocols in major depressive disorders.Although this chal-jor depressive disorders.Although this challenge was processed within widely adopted lenge was processed within widely adopted methodological framework, there were several methodological framework, there were several weaknesses imposing restrictions on study weaknesses imposing restrictions on study conclusions.Sample size should have been conclusions.Sample size should have been suffi cient to test the hypothesis according to suffi cient to test the hypothesis according to some published data, but nevertheless bigger some published data, but nevertheless bigger treatment groups would eliminate few study treatment groups would eliminate few study biases arising from sample heterogeneity.Age biases arising from sample heterogeneity.Age structure with tendency towards 4th, 5th and structure with tendency towards 4th, 5th and 6th life decade is limiting conclusions to these 6th life decade is limiting conclusions to these age groups.age groups.
Uncertainty impact was not assessed Uncertainty impact was not assessed and sensitivity analysis was avoided due to a and sensitivity analysis was avoided due to a rather limited sample size and short time ho-rather limited sample size and short time horizon of the study.Treatment response among rizon of the study.Treatment response among almost one half of all patients was rather poor almost one half of all patients was rather poor and failed to achieve satisfactory mood stabi-and failed to achieve satisfactory mood stabilization.Th is was probably due to insuffi cient lization.Th is was probably due to insuffi cient psychotherapy support during initial weeks of psychotherapy support during initial weeks of clinical follow up.Opposing to aforementioned clinical follow up.Opposing to aforementioned weaknesses, there are also two strengths of this weaknesses, there are also two strengths of this trial that make publishing of these data useful trial that make publishing of these data useful for policy makers.Th is was an in depth case se-for policy makers.Th is was an in depth case series analysis and one of the pioneering eff orts ries analysis and one of the pioneering eff orts on mental illness cost-eff ectiveness analysis in on mental illness cost-eff ectiveness analysis in a vast Eastern European region.Fulfi lling the a vast Eastern European region.Fulfi lling the knowledge gap in this part of middle income knowledge gap in this part of middle income Europe might be essential.Local healthcare Europe might be essential.Local healthcare settings shape cost-eff ectiveness of certain settings shape cost-eff ectiveness of certain medical technologies and are profoundly dif-medical technologies and are profoundly different when compared with the estimates re-ferent when compared with the estimates reported for high income mature economies.ported for high income mature economies.

CONCLUSIONS
Within Southeastern European healthcare set-Within Southeastern European healthcare set-tings heterocyclic antidepressant treatment of tings heterocyclic antidepressant treatment of major depressive disorder is the least costly major depressive disorder is the least costly option compared with SSRIs and SNRIs.Th e option compared with SSRIs and SNRIs.Th e main reasons for such a favorable position are main reasons for such a favorable position are low drug acquisition costs and shorter hospi-low drug acquisition costs and shorter hospitalizations with heterocyclic antidepressants.talizations with heterocyclic antidepressants.Favoring this group of antidepressants in clini-Favoring this group of antidepressants in clinical practice could bring signifi cant savings cal practice could bring signifi cant savings to health insurance systems in upper-middle to health insurance systems in upper-middle income countries of the region.Among un-income countries of the region.Among undoubtedly successful cost containment strate-doubtedly successful cost containment strategies was broad introduction of generic alterna-gies was broad introduction of generic alternatives and facilitating their market access and tives and facilitating their market access and reimbursement from administrative obstacles reimbursement from administrative obstacles [43].Similar health economic studies along- [43].Similar health economic studies alongside clinical trials on eff ectiveness in mental side clinical trials on eff ectiveness in mental illnesses should be funded more ambitiously illnesses should be funded more ambitiously in the region on grounds of rising psychiat-in the region on grounds of rising psychiatric morbidity.Such pioneering local eff orts ric morbidity.Such pioneering local eff orts of mental illness health economics could help of mental illness health economics could help better understand the healthcare market land-better understand the healthcare market landscape in Eastern Europe and the Balkans.scape in Eastern Europe and the Balkans.

ETHICAL COMMITTEE APPROVAL
Th is study adhered to the ICH-GCP and Dec-Th is study adhered to the ICH-GCP and Declaration of Helsinki principles and was previ-laration of Helsinki principles and was previously approved by Ethical Committee of the ously approved by Ethical Committee of the University Clinical Center Kragujevac, Serbia University Clinical Center Kragujevac, Serbia decision number 01-3347.Date of decision: decision number 01-3347.Date of decision: 22.04.2010.Informed consent was provided by 22.04.2010.Informed consent was provided by all patients participating in this study.all patients participating in this study.
Entry and selection criteriaTh ere were in total 1,164 patients suff ering from F.32 [Major depressive disorder-single episode] by the International Classifi cation of Diseases 10 revision-ICD-10 and 861 registered cases of F.33 [ICD-10-Major depressive disorder-recurrent] in 2010.In 2011, the total patients were 1,496 by F.32 and 768 by F.33.
Th e authors would like to express their grati-Th e authors would like to express their gratitude to the Ministry of Education, Science and tude to the Ministry of Education, Science and Technological Development of the Republic of Technological Development of the Republic of Serbia, Grant No.175014 and to the Faculty of Serbia, Grant No.175014 and to the Faculty of Medical Sciences University of Kragujevac for Medical Sciences University of Kragujevac for Junior Research Grant No 17/10 in 2010 out of Junior Research Grant No 17/10 in 2010 out of which this trial was funded.which this trial was funded.

Table 1 .
Patient demographic profi le and medical background across treatment groups www.hophonline.org

240 Volume 2 • Number 1 • January 2015 • HOPH
Th ere were no statistically signifi cant diff erences in the frequency of satisfactory treatment outcomes among groups [χ2=4.339;р=0.362].Quality of life measure-ments provided by Q-LES-Q-SF prove signs of straightforward advancement due to therapy and release of symptoms in all three groups from point zero to the 8th week.Diff erences among average group values of total Q-LES-Q-SF scores were statistically signifi cant with the best improvement detected within the heterocyclic group [means rising from 37.