An Economic Evaluation of the 2003 European Society of Hypertension–European Society of Cardiology Guidelines for the Management of Mild-to-Moderate Hypertension in Greece

Abstract

Background:

In the 2003 European Society of Hypertension–European Society of Cardiology (ESH-ESC) guidelines, it is concluded that the major classes of antihypertensive agents are suitable for the initiation and maintenance of antihypertensive therapy. The aim of this study was to compare the cost-effectiveness of each one of the major antihypertensive agents as monotherapy in the management of mild-to-moderate hypertension in Greece, when following the 2003 ESH-ESC guidelines.

Methods:

We performed a cost-effectiveness analysis based on numbers needed to treat. A decision analysis model was developed to compare chlorthalidone, propranolol, amlodipine, enalapril and losartan. Clinical inputs were derived from a meta-analysis and randomized controlled trials and cost data from public sources. The evaluation of the cost of managing hypertension includes the cost of drug therapy, monitoring, treating side effects, poor compliance and switching. All costs were calculated from a public insurance system perspective, in 2004 Euros. Future costs and clinical benefits were discounted at 5%. The time frame was 5 years. Extensive sensitivity analyses were also performed.

Results:

The cost (in Euros) of uncomplicated hypertension treatment for 5 years was 485.87, 567.66, 851.44, 607.45, and 1279.88 for chlorthalidone, propranolol, amlodipine, enalapril, and losartan, respectively. The estimated total cost (in Euros) to prevent one death was 60230.71, 70369.96, 105596.72, 75301.40, and 158659.35, respectively.

Conclusions:

In mild-to-moderate uncomplicated hypertension chlorthalidone is the most cost-effective agent. If it was the drug of choice to initiate treatment of uncomplicated hypertension, it would probably save the public insurance system organizations a great amount of expenses for benefit of the insured patients. Am J Hypertens 2005;18:1233–1240 © 2005 American Journal of Hypertension, Ltd.

Hypertension treatment, antihypertensive agents, cost effectiveness, Greece

Hypertension is a highly prevalent risk factor for cardiovascular disease (CVD), which affects approximately 1 billion individuals worldwide.^1,2 The prevalence of hypertension in Greece is more than 30% in total,^3,4 and thus it could be estimated that approximately 2.8 million Greek adults have elevated blood pressure (BP).⁴ In the Third National Health and Nutrition Examination Survey (NHANES III), almost 74% of the hypertensive subjects had mild and 21% moderate hypertension,⁵ and these data are in accordance with unpublished data of the Naoussa study.³

Hypertension predisposes to CVD, including coronary heart disease (CHD) and stroke; and this relationship has been described as “continuous, consistent and independent of other risk factors.”² The outcome data from several clinical trials and meta-analyses prove that lowering BP with the “major” classes of antihypertensive agents (diuretics, β-blockers, calcium channel blockers [CCB], angiotensin-converting enzyme [ACE] inhibitors, and angiotensin-receptor blockers [ARB])⁶ significantly reduces the risk of CVD events.^2,6–9 A large number of randomized controlled trials (RCT) based on different compounds confirm that the main benefits of antihypertensive therapy are caused by lowering of BP per se, largely independently of the drugs used to lower BP.^6,8

During the past 10 years many RCT have compared antihypertensive regimens initiated with different classes of antihypertensive agents, usually comparing older agents (diuretics, β-blockers) with newer agents (CCB, ACE inhibitors, ARB).⁶ Most of these trials have recently been reviewed, and all of them agree that newer and older classes of antihypertensive drugs provide similar reductions in cardiovascular morbidity and mortality.^7,8,10

Therefore, in the 2003 European Society of Hypertension–European Society of Cardiology guidelines for the management of hypertension (2003 ESH-ESC guidelines), it is concluded that the major classes of antihypertensive agents are suitable for the initiation and maintenance of antihypertensive therapy.⁶ However, the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC) recommendations for first-line therapy are somehow different, because in the seventh report of the JNC (JNC-7), diuretics are suggested as the initial therapy for most patients with uncomplicated hypertension.²

The newer agents are markedly more expensive than the older ones, but their use has steadily been increased in contrast to the use of diuretics.^11,12 The rising consumption of newer antihypertensive drugs may in part be explained by observational studies claiming a better adherence rate, thus implying fewer side effects, better tolerability, and fewer switches from one agent to another.^11,13

Given the prevalence of hypertension and the continuously rising expenses for its treatment, recommendations that influence practice patterns could have a tremendous economic impact on national health expenditures. Thus, cost considerations have become a crucial part of medical decisions.^14,15 Therefore, the purpose of this study was to compare the cost-effectiveness of the major antihypertensive agents for the prevention of death and major CVD endpoints in patients with uncomplicated mild-to-moderate hypertension in Greece, when following the 2003 ESH-ESC guidelines.

Methods

A decision analysis model was constructed to simulate clinical decisions and outcomes in the management of mild-to-moderate uncomplicated hypertension,^11,14,16 when following the 2003 ESH-ESC guidelines.⁶ The model compared chlorthalidone, propranolol, amlodipine, enalapril, and losartan. Effectiveness outcome was the number-needed-to-treat for 5 years (5-year NNT) to prevent one death, CVD death, CVD event, CHD, or stroke.^11,17,18 The main outcome of interest was the cost to prevent one death and secondary outcomes were the costs to prevent the other major endpoints.¹⁷ Assuming similar efficacy of the five major classes of antihypertensive agents,^6,7,8,10,17 a cost-minimization analysis was performed. The term “cost-effectiveness” was used in reference to the direct costs required to prevent one event.¹⁷ The cost-effectiveness ratio (CER) of each drug was calculated by multiplying the 5-year direct treatment costs with the specific drug by the 5-year NNT.¹⁷

Decision Analysis Model

The analysis begins at the point at which the initiation of the drug therapy has been decided.¹⁴ In accordance with the 2003 ESH-ESC guidelines, a hypothetical cohort of adult patients with uncomplicated hypertension receives a single agent at a low dose as monotherapy. Patients who do not achieve adequate control are given a higher dose of the same agent. If they still do not achieve control or experience intolerable side effects, a different single agent is prescribed to them.^6,14 The model assumes a comprehensive initial evaluation visit and follow-up visit 1 month after the drug therapy begins. In cases of change in medication, patients are re-evaluated monthly until hypertension control is achieved.^6,14 After that, patients are re-evaluated every 4 months for the remainder of the simulation.⁶

The model allowed only one opportunity to switch from one therapeutic class to another, in response to intolerable adverse events or lack of efficacy.¹¹ Patients receiving chlorthalidone switched to enalapril, and patients receiving the other agents switched to chlorthalidone, if necessary. It was also assumed that the input parameters remained constant in every year.¹¹ For simplicity, the model compared only monotherapeutic regimens and did not examine combination treatment strategies. Normally, in mild hypertension, monotherapy is most often likely to be successful.⁶ For example, in the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial, which recruited individuals with mild-to-moderate hypertension, about 60% of the patients remained on monotherapy.¹⁹

Clinical Data Inputs

Using MEDLINE searches, previous meta-analyses,^7,8,10 and journal reviews from January 1999 through December 2003, we identified studies designed to compare the first-line antihypertensive agents with no treatment on the occurrence of major CVD endpoints and all-cause mortality. To be eligible for inclusion, studies had to be RCT with a minimum of 200 person-years of observation.⁷ All eligible trials were included in a previous meta-analysis by Psaty et al.⁷ Recommended methods for study selection and data synthesis were followed.^18,20 Other clinical data, such as compliance with treatment and side effects (hypokalemia, serum potassium <3.5 mEq/L), were based on results from recently published RCT^19,21–24 (Table 1).

Table 1

Probabilities and costs used in the model

	Baseline analysis		Sensitivity analyses
Variable	Value	Reference	Value	Reference
Compliance with chlorthalidone*	0.623	21	0.164	28
Compliance with propranolol*	0.623	21	0.347	28
Compliance with amlodipine*	0.662	21	0.407	28
Compliance with enalapril*	0.613	21	0.465	28
Compliance with losartan*	0.840	24	0.509	28
Hypokalemia in chlorthalidone group†	0.080	19	0.250	17
Hypokalemia in propranolol group†	0.010	Exp	0	17
Hypokalemia in amlodipine group†	0.040	19	0	17
Hypokalemia in enalapril group†	0.020	19	0	17
Hypokalemia in losartan group†	0.010	Exp	0	17
Cost of clinical visit (plus ECG) (Euros)	7.05	MoH	40/62	PHO/CMS
Cost of routine laboratory tests‡ (Euros)	39.12	MoH	52/52	PHO/CMS
Cost of extra laboratory tests for chlorthalidone group§ (Euros)	18.90	MoH	25/27	PHO/CMS
Cost of extra laboratory tests for propranolol group∥ (Euros)	7.37	MoH	10/11	PHO/CMS
Cost of serum potassium levels (Euros)	5.22	MoH	6/5.2	PHO/CMS
Clinical visits per year for old drugs	3	6	4	Exp
Clinical visits per year for new drugs	3	6	2	Exp
Routine laboratory tests per year for old drugs	1	6	2	Exp
Routine laboratory tests per year for new drugs	1	6	1	Exp
Extra laboratory tests per year for old drugs	1	14,17	1	14,17
Extra laboratory tests per year for new drugs	0	6	0	6

CMS = Centers for Medicare & Medicaid Services; ECG = electrocardiogram; Exp = expert opinion from a panel of three clinical hypertension researchers; MoH = Fee Schedule of the Greek Ministry of Health; PHO = pricelist of a Greek private health organization.

*

Rate of continuation of initial treatment at the end of monitoring.

†

At end of monitoring.

‡

Routine laboratory tests include plasma glucose, serum total cholesterol, high-density lipoprotein, triglycerides, uric acid, creatinine, potassium, hemoglobin, haematocrit, and urinalysis.⁶

§

Extra laboratory tests for chlorthalidone group include serum potassium, creatinine, urea, total cholesterol, and triglycerides.

∥

Extra laboratory test for propranolol group includes serum total cholesterol and triglycerides.

Statistical Analysis

Analyses were performed using SPSS 10.0 (SPSS Inc., Chicago, IL), Review Manager 4.2 for Windows (Wintertree Software Inc., The Cochrane Collaboration, Oxford, England) and EasyMA 2001 Software (software for meta-analysis of clinical trials; Department of Clinical Pharmacology, Lyon, France). Overall estimates of effect were calculated with the DerSimonian and Laird method and were presented as means with 95% confidence intervals. The assumption of heterogeneity of treatment effects was tested with χ² tests, and values of P < .05 indicate heterogeneity across combined studies.^18,20

Cost Inputs

Cost analysis was performed from the perspective of the Greek social security system interested in the direct costs, including cost of acquiring drugs, clinical visits, and laboratory tests. The costs of treating adverse events and switching between therapeutic classes were also taken into account.^14,,–17 Drug prices were available from the Greek National Formulary (Table 2). The “defined daily dose” (DDD) was used as a measure for the assumed average dose for each drug. The DDD is recommended by the World Health Organization (WHO) as a standard measure for use in drug use studies.^25,26 Physician fees and charges for the tests were defined by the Fee Schedule of the Greek Ministry of Health (Table 1).

Table 2

Retail prices for selected antihypertensive agents and valuation of drug acquisition costs for 5 years of treatment

Generic name	Package* (tablets × mg)	Acquisition cost per package* (Euros)	Daily dose† (mg)	5-Years drug cost (Euros)
Baseline analysis
Chlorthalidone	20 × 50 mg	1.72	25	78.48
Propranolol	28 × 80 mg	3.06	160	418.53
Amlodipine	14 × 5 mg	7.71	5	1005.05
Enalapril	10 × 20 mg	6.01	10	548.41
Losartan	10 × 50 mg	9.08	50	1657.10
Potassium Bicarbonate	30 × 675 mg	3.52	1350	427.05
Sensitivity analysis
Indapamide	30 × 2.5 mg	3.40	2.5	206.83
Nebivolol	28 × 5 mg	16.39	5	1068.28
Nifedipine	30 × 20 mg	5.85	30	533.81
Ramipril	20 × 5 mg	10.31	2.5	470.39
Eprosartan	28 × 600 mg	23.34	600	1521.27

*

Drug prices and packages are available from the Greek National Formulary.

†

Defined daily dose (DDD) was used as a measure for the assumed average daily dose for each drug according with the World Health Organization.²⁶

To account for concerns regarding the adverse events of chlorthalidone and propranolol, the model assumed the need of extra laboratory monitoring for these agents.¹⁴ The model also assumed that the cost of treating hypokalemia included the acquisition cost of potassium supplementation, one clinical visit and two additional serum potassium level tests per year.¹⁴ It was also assumed that every switching between antihypertensive agents required two additional clinical visits and one routine laboratory test according to the estimation of three hypertension specialists.²⁷

All costs were calculated in 2004 Euros (1 Euro = $US 1.25; 2004 values) and the time horizon of the analysis was 5 years. All future costs and health outcomes were discounted at a rate of 5% per year.^11,27

Sensitivity Analyses

Sensitivity analyses tested the effect of modifying the input parameters on the economic endpoints.¹⁴ We examined the impacts on cost-effectiveness of assuming that newer agents are 30% more effective,^14,17 and that they have higher compliance (more than twofold than chlorthalidone)²⁸ and fewer side effects than the older agents,¹⁴ as shown in Table 1. To determine the sensitivity of the economic endpoints to the price of the antihypertensive agents, the older agents were substituted by the most expensive of the same therapeutic class (indapamide, nebivolol) and the newer agents by the least expensive (nifedipine, ramipril, eprosartan). The impacts of 30% reduction of the price of the newer agents and of increase in the charges for the clinical and laboratory monitoring to the real market prices were also tested.¹⁷ To examine the applicability of the findings to the United States (US), these prices and charges were substituted by the respective ones from the US setting (2004 Red Book, Centers for Medicare and Medicaid Services).²⁹ The impact of discount rates for costs and health outcomes was evaluated by varying discount rates between 0% and 10% (nondiscounted, and discounted 3%, 6%, and 10%).²⁷ One-way sensitivity analyses were performed in all inputs in favor of the agents with the worst CER. Two-way and three-way sensitivity analyses were also performed. Probabilities and costs used in the sensitivity analyses are summarized in Table 1.

Results

Effectiveness of Treatment

There were 25 eligible trials, which included 69,185 patients and had a mean duration of follow-up 4.1 years. (Details of individual study characteristics are available from the authors on request.) As shown in Table 3, compared with no treatment, any first-line antihypertensive agent was associated with significant reductions in the risk of all major outcomes. The 5-year NNT to prevent one death was 143 and just 34 patients had to receive one antihypertensive agent for 5 years to prevent one major CVD event. The validity of the results was supported by statistical evidence of homogeneity among combined trials (P > .05) with the exception of CVD events (P = .015).

Table 3

Random effects meta-analysis comparing any antihypertensive drug treatment versus no treatment and the numbers-needed-to-treat to prevent one death or other cardiovascular event

Outcome	RR (95% CI) (Treatment/no treatment*)	P value for heterogeneity† of RR	5-Year NNT (95% CI)	P value for heterogeneity† of NNT
Total mortality	0.90 (0.85–0.95)	.51	143 (97–833)	.43
CVD mortality	0.84 (0.76–0.92)	.11	134 (96–556)	.21
Major CVD‡ events	0.79 (0.74–0.85)	.008	34 (30–66)	.015
Stroke	0.68 (0.61–0.77)	.009	65 (57–123)	.25
CHD	0.87 (0.80–0.94)	.64	348 (213–NA)§	.58

CHD = coronary heart disease; CI = confidence interval; CVD = cardiovascular disease; NA = not applicable; NNT = number-needed-to-treat; RR = relative risk.

*

The no-treatment group includes placebo-treated control subjects, participants not treated in open trials, and participants receiving usual care.

†

P < .05 indicates heterogeneity of treatment effects across combined trials.

‡

CVD events include CHD, stroke, and congestive heart failure.

§

The upper 95% confidence limit cannot be estimated for NNT because the lower 95% limit of the risk difference is zero.

Cost of Antihypertensive Treatment

The drug acquisition costs for 5 years of treatment with each agent (not taking into account compliance, switching, and side effects) are shown in Table 2, in which the cost varied 20-fold. However, when the need for laboratory tests and clinical visits, compliance, side effects, and switching to another drug class (including the cost of the alternative drug) were considered, the differences between chlorthalidone and losartan were reduced to less than threefold. Chlorthalidone was the least costly alternative, with total treatment cost for 5 years 485.87 Euros. Medications accounted for 24% to 77% of the total direct costs (Fig. 1).

Direct health care costs of treating for 5 years mild-to-moderate uncomplicated hypertension in Greece.

Figure 1.

Cost-Effectiveness Ratios

Assuming equal effectiveness among first-line agents,^{6,7,8,10,11,17} the cost-effectivenss ratios (CER) depend on the cost of the treatment and the least costly alternative is the most efficient.²⁷ Therefore, chlorthalidone had the lowest CER for every outcome. The cost to prevent one death ranged from 60,230 Euros in the chlorthalidone group to 158,659 Euros in the losartan group (Table 4).

Table 4

Costs to prevent one death or other major cardiovascular events among patients with uncomplicated mild-to-moderate hypertension treated with first-line agents

Event	Chlorthalidone	Propranolol	Amlodipine	Enalapril	Losartan
Death	60230.71	70369.96	105596.72	75301.40	158659.35
CVD death	56439.97	65941.08	98950.78	70562.16	148673.79
CVD event*	14320.59	16731.32	25106.91	17903.83	37723.20
Stroke*	146575.44	171249.98	256976.64	183250.97	386108.06
CHD*	27377.60	31986.35	47998.51	34227.91	72117.89

CHD = coronary heart disease; CVD = cardiovascular disease.

All values are given in Euros (1 Euro) = $US1.25; 2004 values. All future costs and health outcomes were discounted at a rate of 5% per year.

*

Fatal and nonfatal.

Sensitivity Analyses

Results of sensitivity analyses consistently indicated the dominance of chlorthalidone. The only two variables that changed the relationship between chlorthalidone and other treatment strategies were drug efficacy and compliance (Fig. 2). All other analyses failed to change the outcome of the model with respect to cost-effectiveness. The use of different combinations of discount rates for costs and clinical outcomes had no impact on the relative outcomes, with chlorthalidone remaining the most cost-effective initial choice.

Effect of modifying clinical output parameters on cost to prevent one death.

Figure 2.

Modification of Clinical Parameters

Modification of clinical parameters is depicted in Fig. 2. Superior drug efficacy would lower the NNT, thereby improving CER.¹⁷ In the unlikely case that new antihypertensive agents are 30% more effective than chlorthalidone, enalapril would be the most cost-effective treatment. Compliance with treatment is usually less than the rates observed in clinical trials.^13,28 The assumption of poor compliance for all the evaluated regimens²⁸ altered the basic results and propranolol became marginally more cost-effective than chlorthalidone, when the former had more than twofold higher compliance than the second (34.7% v 16.4%). Although the incidence of hypokalemia with diuretics is less than 10%,²⁴ it was assumed that 25% of the patients receiving chlorthalidone would experience this side effect. The assumption did not affect the results to a significant extent.

Modification of Economic Parameters

Changes in economic parameters are depicted in Fig. 3. The results were relatively sensitive to the reduction of the price of the newer agents, the substitution of the antihypertensive agents from others of the same therapeutic class, and the changes in the daily doses (twice the DDD for the old agents, half of the DDD for the new). These modifications did not qualitatively change the conclusions of the current analysis but the differences between old and new drugs were greatly reduced. Sensitivity analyses tested the effect of switching to the other three alternative agents but the results were in every case in favor of chlorthalidone. The increase in the charges for the clinical and laboratory monitoring significantly increased the CER. In a sensitivity analysis, prices from the US setting were used in drug acquisition costs,²⁹ physician fees and charges for tests (Centers for Medicare & Medicaid Services). Chlorthalidone remained the most cost-effective agent; however, the CER were significantly increased, and the cost to prevent one CVD event ranged from 51,742 Euros in the chlorthalidone group to 88,364 Euros in the losartan group. Although patients receiving the older agents do not need extra monitoring,⁶ it was assumed that they had more clinical visits and laboratory tests than those receiving the newer agents (Table 1). Chlorthalidone remained the most cost-effective treatment, but the differences were reduced.

Effect of modifying economic input parameters on cost to prevent one death.

Figure 3.

Multi-Way Sensitivity Analyses

Results of multi-way sensitivity analyses are shown in Fig. 4. Two-way sensitivity analyses confirmed the robustness of the results. Only under the implausible assumption of simultaneously poor compliance with treatment, reduced price of the new agents, and increased charges of the monitoring (three-way sensitivity analysis) did enalapril acquire the lowest CER (131,801.50 Euros v 135,077.01 Euros to prevent one death for chlorthalidone).

Effect of modifying more than one input parameter at the same time in multi-way sensitivity analyses. ACE = angiotensin-converting enzyme; ARB = angiotensin receptor blocker; CCB = calcium channel blocker; S.A. = sensitivity analysis. †Two-way S.A. 1: poor compliance with treatment²⁸ and 30% reduced cost of new antihypertensive agents (amlodipine, enalapril, losartan); ‡Two-way S.A. 2: The older agents (chlorthalidone, propranolol) were replaced by the most expensive representatives in the same therapeutic class; the newer agents were replaced by the least expensive; and the changes in the clinical and laboratory monitoring were adjusted upward to the private sector levels. §Three-way S.A.: Poor compliance with treatment,²⁸ reduced price of the newer agents, and increased charges in clinical and laboratory monitoring.

Figure 4.

Discussion

During the first half of 2003, two of the most widely recognized guidelines for the management of hypertension, the 2003 ESH-ESC guidelines⁶ and those in the JNC-7,² have been published. These guidelines differ in their recommendations regarding the appropriate first-line antihypertensive therapy for uncomplicated hypertension.^2,6 In the present study, a decision analysis model of the 2003 ESH-ESC guidelines has been created to compare therapeutic outcomes and costs, when initial therapy is selected from each of the five recommended classes of antihypertensive agents. This analysis indicates that diuretics prevent death and major CVD events at a significantly lower cost than the other antihypertensive drugs among patients with mild-to-moderate uncomplicated hypertension (15%, 20%, 43%, and 62% lower than β-blockers, ACE inhibitors, CCB, and ARB, respectively). Economic analyses of hypertension treatment have yielded results similar to the ones presented here.^{11,14,15,17,25,30} In the US, the potential drug cost savings if diuretics were used as first-choice drugs in the treatment of uncomplicated hypertension might be more than US $1 billion.²⁵

Based on the results of recently published RCT^16,18–20 and meta-analyses,^7,8,10 the model assumed that newer and older agents did not differ in efficacy in preventing total mortality and cardiovascular morbidity and mortality. Recently, the results of the Study on Cognition and Prognosis in the Elderly (SCOPE) demonstrated a 23.6% reduction in stroke incidence in patients treated with candesartan versus that in the control group.²⁵ Sensitivity analysis showed that a risk reduction of more than 60% would be required to make ARB more cost-effective than chlorthalidone and more than 30% to make ACE inhibitors more cost-effective. Assuming equal efficacy for every alternative treatment, the cost of the treatment is the crucial part of the analysis.

The first-line agents differ significantly in the total cost of treatment. The drug acquisition costs for 5 years of treatment with losartan exceeded 1600 Euros, whereas chlorthalidone cost less than 80 Euros. It was assumed that the price of the newer agents would decline, but the price of enalapril should be 58% lower to approximate the CER of chlorthalidone.

The drug acquisition cost represents only a part of the total treatment cost. The latter is also affected by the costs of the clinical and laboratory monitoring. Although this model was less favorable to diuretics and β-blockers by assuming more side effects and the need for extra monitoring, neither a doubling of the monitoring schedule for the old agents nor an increase in the prices was sufficient to make another agent more cost-effective than chlorthalidone. Moreover, the result is also strengthened by the fact that both guidelines agree in the schedule of monitoring and recommend it should be the same for all the first-line agents.^2,6

The results of the current analysis were sensitive to the compliance with the treatment. Pharmaceutical manufacturers often claim that the increased compliance resulting from a better side-effect profile justifies the higher price of the newer agents.^13,14 Under the assumption of poor compliance according to the study of Conlin et al,²⁸ the cost of treatment with chlorthalidone was increased because of the need for extra monitoring and the switch to a more expensive agent. In contrast, the increased cost of the other agents because of the need for extra monitoring was outweighed by the reduced cost as a result of replacing the initial agent by a less expensive one (chlorthalidone). In this case, the total cost for 5 years of treatment with propranolol was 553.00 Euros, whereas the cost of treatment with chlorthalidone was 566.54 Euros. The total costs of treatment with the newer agents were lower but were still higher than the costs of the older agents.

Although economic evaluations should ideally be conducted from the societal perspective,²⁷ this analysis was conducted from the perspective of the Greek social security system, which covers almost 100% of the population.³¹ The cost calculated is a good approximation of the cost to the social security system under current reimbursement policies, and it permits the generalization and comparability of results across studies.³¹ It should be noted that the results of this study are particularly relevant to the cost-containment efforts initiated through the recent reform of the Greek National Health Service.³² If we needed to perform this study from the societal or the patient's perspective, this method would have underestimated the treatment costs; however, chlorthalidone would remain more cost-effective than the other agents.

It is well known that absolute price levels vary among countries.³³ Drug acquisition costs, physician fees, and charges for tests are lower in Greece and other countries than in the US. However, from the point of view of cost-effectiveness assessments, the critical issue is whether the relative prices of health care resources differ. The relative prices of the alternative agents, visits, and tests do not differ significantly between Greece and the US.²⁹ Thus, the relative cost-effectiveness does not differ between the two countries.

The cost of treatment to the health care provider should influence the initial choice of the antihypertensive agent, although cost considerations should not predominate over efficacy and tolerability in any individual patient.⁶ The results of this analysis support the recommendations of JNC-7,² but not the widespread use of the newer agents as first-line treatment for uncomplicated hypertension.¹¹

This study also has some limitations that have to be noted. The analysis was limited to the treatment of mild-to-moderate uncomplicated hypertension. Certain compelling indications may require the use of other antihypertensive drugs as initial therapy.^1,2,6,24 This model does not address the impact of antihypertensive therapy on health-related quality of life (QoL). The higher cost of treatment with the newer agents might be justified if they bring about improved QoL. However, RCT failed to show a difference in the QoL. In the Treatment of Mild Hypertension Study (TOMHS), all patients receiving active therapy experienced an improvement in QoL, but only the gains in patients receiving chlorthalidone or acebutolol achieved statistical significance.³⁴ Thus, the attractiveness of chlorthalidone established by this economic model appears to be reinforced by available quality of life data. Nonetheless, this study did not include ARB.³⁴ Finally, in Greece, there are no databases with patient-specific statistics on prescription drugs dispensed and dosages by diagnosis from which information on use could be extracted. To estimate resource use in the treatment of hypertension, current guidelines,^2,6 WHO methodology,²⁶ and expert opinion from a panel of three hypertension specialists were used.

In conclusion, the study showed that chlorthalidone was by far the most cost-effective agent in the treatment of mild-to-moderate uncomplicated hypertension in Greece. Prescribing new agents as first-line antihypertensive therapy in patients with uncomplicated hypertension cannot be recommended at this time unless acquisition costs of new agents become substantially lower. In a world in which economic resources are scarce and yet the demand for and cost of health care are continually rising, it is extremely important to choose therapeutic strategies that maximize the health benefits for a given cost constraint. In mild-to-moderate uncomplicated hypertension, diuretics seem to achieve this goal and should be used as the initial therapy.