cost effectiveness of ramipril in patients with non-diabetic nephropathy and hypertension

Cost Effectiveness of Ramipril inPatients with Non-DiabeticNephropathy and HypertensionEconomic Evaluation of Ramipril Efficacy In Nephropathy(REIN) Study for Germany from the Perspective of StatutoryHealth Insurance

Peter K. Schädlich,1 Josef Georg Brecht,1 Massimo Brunetti,2 Eva Pagano,3 Badrudin Rangoonwala4 and Eduard Huppertz5

1 InForMed GmbH – Outcomes Research & Health Economics, Ingolstadt, Germany2 Centro per la Valuazione della Efficacia della Assistenza Sanitaria (CEVEAS), Modena, Italy3 Progetto Sperimentazione Gestazionale, Azienda Ospedaliera San Giovanni Battista, Torino, Italy4 Hofheim/Taunus, Germany5 Aventis Pharma Deutschland GmbH, Bad Soden/Taunus, Germany

Abstract Background: In the Ramipril Efficacy In Nephropathy (REIN) trial, ramiprilsignificantly lowered the rate of reaching the combined end-point of doubling ofbaseline serum creatinine levels or end-stage renal failure (ESRF).Objective: To determine the additional cost per patient-year of chronic (longterm) dialysis avoided (PYCDA) when the ACE inhibitor, ramipril, was added toconventional treatment of patients with non-diabetic nephropathy and hypertension.Study perspective: Statutory Health Insurance (SHI) provider in Germany.Design and setting:Data from the REIN Study were used in a cost-effectivenessanalysis (CEA). Amodelling approach was used, which was based on secondaryanalysis of published data, and costs were those incurred by the SHI provider (i.e.SHI expenses). In the base-case analysis, average case-related SHI expenseswereapplied and PYCDAwere quantified using the cumulative incidence of ESRF asobserved in the REIN trial.Main outcome measures and results: The incremental cost-effectiveness ratios(ICERs) of ramipril varied between about –76 700 deutschmarks (DM) and–DM81 900 per PYCDA(DM1 ≈ 0.55 US dollars; 1999 values), according to thetreatment periods of 1 year and 3 years, respectively. In the sensitivity analysis,the robustness of the model and its results were shown when the extent of influ-ence of different model variables on the base-case results was investigated. First,probabilities of ESRF and PYCDAwere estimated according to theWeibull method.Second, the influence of the model variables on the target variable was quantifiedusing a deterministic model. Third, the dependency of the target variable (ICER)on random variables was described in a simulation. The cost for chronic dialysishad by far the greatest impact on the target variable, which was 28 times greater

ORIGINAL RESEARCH ARTICLE Pharmacoeconomics 2001; 19 (5 Pt 1): 497-5121170-7690/01/0005-0497/$22.00/0

© Adis International Limited. All rights reserved.

than the impact of clinical effectiveness of ramipril, i.e. the number of PYCDA.There were net savings per PYCDAwith ramipril treatment after 1, 2 and 3 years:95% of the 10 000 simulation steps resulted in savings of between DM69 500and DM94 600 per PYCDA after 3 years.Conclusions: Results from this evaluation show that ramipril offers enormoussavings from the perspective of the SHI provider (third-party payer) in Germanywhen added to the conventional treatment of patients with non-diabetic nephro-pathy and hypertension.

Nephropathy describes the process of disease-induced renal glomerular damage. Deterioration ofrenal function is characterised by a decline in theglomerular filtration rate (GFR) and persistent pro-teinuria. In most forms of proteinuric chronic renaldisease, there is an inexorable progression despitethe elimination of the initial cause of injury. Withthe onset of nephrotic syndrome with proteinuriaof ≥3 g/day, patients are under a high risk of end-stage renal disease (ESRD) that requires chronic(i.e. long term) dialysis or renal transplantation.[1-5]ESRD is a major healthcare problem in indus-

trialised countries as it places an enormous burdenon the individual with the disease and on society.[6]At the end of 1996, about 700 patients per millionpopulation in Germany were under chronic (i.e. longterm) renal replacement therapy (RRT) and yearlytreatment costs for chronic dialysis and transplan-tation including after-care were estimated at approxi-mately 4 billion deutschmarks (DM) or about 1.5%of total healthcare costs.[7] In light of the poor lifeexpectancy of patients with chronic RRT[8] (whichis similar to that of certain cancers), an annual growthrate of chronic RRT in Germany of about 4%[9] andthe enormous cost of RRT, the treatment strategyin patients with chronic renal disease should be todelay their progression to ESRD.[10]Hypertension plays an important role among the

various mechanisms that contribute to renal functiondeterioration,[10] and elevation of blood pressure isa strong independent risk factor for ESRD.[11] Theincidence of hypertension as a cause of ESRD isabout 13% in Europe.[12] Normalisation of syste-mic blood pressure in patientswith insulin-dependentdiabetes mellitus limits progression of renal dis-ease.[13,14] Despite similar levels of blood pressure

control, ACE inhibitors are more effective than otherantihypertensive agents in reducing the risk of ter-minal renal failure.[15-17]The Ramipril Efficacy In Nephropathy (REIN)

study investigated whether the ACE inhibitor, rami-pril, is superior to conventional antihypertensivesthat have the same level of blood pressure control,in reducing proteinuria, limiting GFR decline andpreventing ESRD in chronic non-diabetic nephro-pathy.[18] This prospective, double-blind, multicentretrial was carried out in 14 centres in Italy. Patientswere classified according to baseline proteinuria(stratum 1: 1 to 2.9 g/day; stratum 2: ≥3.0 g/day)and randomly assigned ramipril or placebo plus con-ventional therapy targeted at achieving and main-taining diastolic blood pressure under 90mm Hg.The target dose of ramipril was 5 mg/day. All thepatients were asked to limit their sodium intake andto consume 0.7 to 0.8g protein per kg bodyweightdaily. No subsequent change in diet was introducedduring the study.The primary end-point was the rate of GFR de-

cline and analysis was by intention to treat.[19] Atthe second planned interim analysis, the differencein decline in GFR between the ramipril and placebogroups in stratum 2 was highly significant (p =0.001).[19] Therefore, the final analysis was per-formed prematurely for patients in stratum 2 (pro-teinuria ≥3.0 g/day), whereas patients in stratum 1continued in the trial. Of the 352 patients randomlyassigned ramipril or placebo, 166 patients wereclassed in stratum 2. Demographic and clinical dataof these 166 patients at time of their randomisationare given in table I.The final analysis of patients in stratum 2 re-

vealed a mean follow-up of 16 months for each

498 Schädlich et al.

© Adis International Limited. All rights reserved. Pharmacoeconomics 2001; 19 (5 Pt 1)

treatment group with a range from 1 to 52 monthsfor the 78 patients in the ramipril group and from1 to 50 months for the 88 patients in the placebogroup. At end-point, the daily doses of ramipril were1.25mg in 22.7%, 2.5mg in 29.3% and 5.0mg in48% of the patients randomly assigned to the rami-pril group.[19]Themean decline in GFR per month in the rami-

pril group of 0.53 ml/min per 1.73m2 was signifi-cantly lower than that of 0.88 ml/min per 1.73m2in the placebo group (p = 0.03) for those 56 patientsassigned ramipril and 61 patients assigned placebowith at least 3 GFR measurements including base-line. In the ramipril group, the urinary protein ex-

cretion rate significantly decreased by 23% (p <0.01) by month 1 and remained lower than baselinethroughout the study period reaching a decrease of55% at 36 months. Urinary protein excretion didnot change significantly in the placebo group. Thechange in urinary protein excretion rate after ran-domisation differed significantly between the 2 treat-ment groups (p = 0.002).[19]Renal survival was significantly different be-

tween the ramipril and placebo groups. 18 of the78 patients in the ramipril group and 40 of the 88patients in the placebo group reached the combinedend-point of doubling of serum creatinine levels orend stage renal failure (ESRF) [p = 0.02]. 17 pa-

Table I. Clinical data of 166 patients classed in stratum 2 (proteinuria of ≥3.0 g/day) of the REIN study at time of their randomisation[19]

Variable Ramipril (n = 78) Placebo (n = 88) p valuesa

Demographic dataMean (SD) age (years) 48.9 (13.6) 49.7 (13.6) 0.76Male 66 (85%) 64 (73%) 0.06Female 12 (15%) 24 (27%)

Cause of renal diseaseGlomerular 50 (64%) 53 (60%) 0.93Interstitial, polycystic 4 (5%) 5 (6%)Other, unknown 24 (31%) 30 (34%)

Mean (SD) of renal function indicatorsGFR (ml/min per 1.73m2) 40.2 (19.0) 37.4 (17.5) 0.35Creatinine clearance (ml/min per 1.73m2) 47.3 (21.7) 43.7 (20.4) 0.24Serum creatinine level (μmol/L)b 212 (88) 212 (88) 0.48Urinary protein excretion (g/24h)c 5.6 (2.8) 5.1 (2.0) 0.38Urinary urea nitrogen excretion (mmol/24 h)d 357 (343) 328 (143) 0.26Urinary sodium excretion (mmol/24 h)e 211.2 (161.2) 198.9 (92.8) 0.65

Mean (SD) arterial blood pressure (mm Hg)Systolic 149.8 (17.8) 148.0 (17.3) 0.43Diastolic 92.4 (11.5) 91.3 (11.1) 0.45Mean 111.5 (12.2) 110.2 (11.6) 0.41

Patients with hypertensionAll with hypertensionf 67 (86%) 77 (88%) 0.80On hypertensive therapy 58 (74%) 70 (80%) 0.42a p values were based on Fisher’s exact test for categorical variables and on Wilcoxon test for continuous variables.[19]

b To convert to mg/dl divide by 88.4.c Mean of 2 last measurements before randomisation.[19]

d To convert to g/24h divide by 35.7.e To convert to g/24h divide by 43.5.f Patients who had sitting systolic or diastolic blood pressure above 140/90mm Hg and were receiving antihypertensive therapy were re-

garded as hypertensive.[19]

GFR = glomerular filtration rate; REIN = Ramipril Efficacy In Nephropathy; SD = standard deviation.

Cost Effectiveness of Ramipril in Nephropathy 499


tients (21.8%) in the ramipril group and 29 patients(33%) in the placebo group required dialysis or re-nal transplantation. Blood pressure control and theoverall number of cardiovascular events were sim-ilar in the ramipril and placebo groups.[19]The clinical effectiveness of ramipril in patients

with non-diabetic nephropathy and severe persist-ent proteinuria (≥3.0 g/day) has been clearly docu-mented in the REIN study.[19] This especially ap-plies to the delay in progression to ESRD and henceto avoiding chronic RRT. However, until now, aneconomic evaluation of this effect of ramipril innon-diabetic nephropathy did not exist for Germany.Economic evaluations have been published using theACE inhibitor captopril for France,[20] Germany,[21]Italy[22] and the UK[23] in patients with diabetic ne-phropathy as well as the ACE inhibitor benazeprilfor The Netherlands[24] in patients with various re-nal diseases. The purpose of our study thereforewas to estimate the cost effectiveness of ramipril inpatients with non-diabetic nephropathy and protein-uria of ≥3.0 g/day (stratum 2) from the perspectiveof the Statutory Health Insurance (SHI) provider inGermany which covers >90% of the population.Additional SHI expenses for ramipril per patient-year of chronic dialysis avoided (PYCDA) werechosen as the target variable.

Methods

Estimation of Economic Consequences

The economic evaluation of stratum2 in theREINtrial for Germany was conducted according to therecommendations of the German Society for Clin-ical Pharmacology and Therapy for performing andassessing pharmacoeconomic studies.[25] To evalu-ate the REIN study for Germany, the morbidity dataof the patients in stratum 2 from all 14 study centresin Italy were transferred. Economic consequencesof the change in morbidity, however, were estimatedusing average case-related treatment SHI expensesin Germany. In addition, a survey of specialists whoprovide chronic RRT in Germany was performed.Economic consequences were quantified in the

marginal analysis, additional cost for ramipril per

PYCDA, using a cost-effectiveness analysis (CEA)from the perspective of the SHI provider.[26] There-fore, costs are given as expenses incurred by theSHI provider. The investigation used a modellingapproach based on secondary analysis of publisheddata. The cost parameters of the model were: (i) me-dication with ramipril in outpatient care; (ii) chronicRRT given by dialysis and renal transplantation.The effectiveness parameter was PYCDA by addingramipril. The incremental cost-effectiveness ratio(ICER) of adding ramipril was estimated as:

ICER = (ΔCRx + ΔCRRT)/PYCDA

where ΔCRx is the incremental cost of adding rami-pril to conventional treatment and ΔCRRT is theincremental cost (negativemeans savings) of chronicrenal replacement therapy.

No.

of p

atie

nts

60

50

80

70

40

30

20

10

00 21 43

No. of years

Fig. 1. Number of patients in the ramipril and placebo groupsduring the REIN study. Average follow-up was 16 months;[19]

only a small number of patients were followed for >3 years. REIN= Ramipril Efficacy In Nephropathy.



The decrease in patient numbers with time overthe course of the REIN study (stratum 2) affectedthe quantification of cost and effectiveness param-eters. Less than 10 patients from each group (ra-mipril and placebo) were in the study for >3 years(fig. 1) and there were no incidental cases of ESRFin years 4 and 5 of the trial.[27] Therefore, the eval-uation was limited to 3 years and differentiated for3 patient groups, namely those who received treat-ment for 1 year, 2 years and 3 years, respectively.A detailed description of the model is given in theSensitivity Analyses section.

Materials

The following materials were used to estimatethe economic consequences of adding ramipril toconventional treatment from the perspective ofGerman SHI:• data on file of the 166 patients in stratum 2 ofthe REIN study with detailed information foreach patient on daily dose of the study medica-tion, follow-up and time to ESRF;[27]

• the price of ramipril in public pharmacies foroutpatient care, regulations governing patientprescription charges and discount received bySHI;[28-30]

• data on chronic RRT in Germany;[7,31,32]• regulations governing the reimbursement of hos-pitals for case-specific surgical procedures basedon diagnosis-related groups;[33,34]

• epidemiological information on all-cause mor-tality of patients undergoing chronic RRT;[35,36]

• expert knowledge on (i) the distribution of thedifferent dialysis procedures; (ii) the time spanbetween the onset of ESRF and renal transplan-tation; and (iii) cost for renal transplantation andafter-care.

Valuation of the Cost Parameters

The economic valuation of cost parameters wasperformed using average case-related treatmentSHI expenses. Themost recent version of each datasource was used. Therefore, expenses were basedon 1999 prices for ramipril in outpatient care andon 1996 prices for chronic dialysis, renal trans-

plantation and after-care. All expenses are given indeutschmarks (DM1 ≈ 0.55 US dollars; 1999 val-ues).

Measuring Patient-Years of Chronic Dialysis Avoided

The number of PYCDA by adding ramipril toconventional treatment was estimated from the dif-ference of the areas under the cumulative incidencecurves of ESRD for the ramipril and placebo groupsas derived from the patient data on file of stratum2 in the REIN study[27] and was corrected for all-cause mortality of patients undergoing chronic di-alysis.[35]

Discounting

The incremental CEAof ramipril covered treat-ment periods of the patients of >1 year. To deter-mine the current value of future costs and effective-ness, a common discount rate of 5% annually wasapplied to both costs and effects.[37,38]

Cum

ulat

ive

inci

denc

e of

ES

RD

requ

iring

dia

lysi

s (%

)

35

30

25

20

40

5

10

15

00 21 3

No. of years

PlaceboRamipril

Fig. 2. Cumulative incidence of ESRD requiring chronic dialysisfor patients randomised to ramipril or placebo during the REINstudy. The area between the curves corresponds to patient-years of chronic dialysis avoided per patient by adding ramiprilto conventional treatment. ESRD = end-stage renal disease;REIN = Ramipril Efficacy In Nephropathy.



Base-Case and Sensitivity Analyses

The results from the evaluation are given in 2steps. In the base-case analysis, average case-re-lated treatment expenses were applied and PYCDAwere quantified using the cumulative incidence ofESRD as observed in stratum 2 of the REIN trial.In the sensitivity analysis, the robustness of themodelwas investigated.

Results

Disease Model

Effectiveness of RamiprilThe development of ESRD requiring chronic di-

alysis as observed in stratum 2 of the REIN trial[27]is given in figure 2 in terms of the cumulative in-cidence curves in the ramipril and placebo groups.

For a hypothetical cohort of 100 patients in eachgroup at treatment start, the difference of the areasunder these curves equals 25.7 PYCDAby ramiprilafter 3 years. This value is undiscounted and notcorrected for all-cause mortality; the follow-up ofESRD patients under chronic RRT was not subjectto the REIN trial.[18,19]Therefore, annual survival of patients undergo-

ing chronic dialysis was derived from the litera-ture. For patients aged 45 to 54 years at treatmentinitiation, the 5-year all-causemortality rate is 57%in Europe.[35] This age group was selected, as themean age of the patients in the ramipril and placebogroups of the REIN trial was 48.9 and 49.7 years,respectively (table I). The influence of comorbiditywas also taken into account. Low-risk patients showa 4-year all-cause mortality rate of 5%whereas high-

No ESRD

Death fromany cause

ESRD(PYCDAi × s1)

No ESRD

Death fromany cause

ESRD(PYCDAi × s1)

Death fromany cause

ESRD(PYCDAp × s2)

No ESRD

Death fromany cause

ESRD(PYCDAi × s1)

Death fromany cause

ESRD(PYCDAp × s2)

Death fromany cause

ESRD(PYCDAp × s3)

Ramipril plusconventional

treatmentvs

placeboplus

conventionaltreatment

Patients withnondiabetic

nephropathy,proteinuria of

≥3 g/day and serumcreatinine levelof <2.5 mg/dl

Year 1 Year 2 Year 3

Fig. 3. Disease course when ramipril is added to conventional treatment. PYCDA by ramipril are differentiated into incidental PYCDA(PYCDAi) and subsequent prevalent PYCDA (PYCDAp) for each treatment interval to be corrected for all-cause mortality of patientswith ESRD undergoing chronic dialysis. This is achieved by multiplying the respective PYCDA with the interval-related survival rates,i.e. (PYCDAi × s1), (PYCDAp × s2), (PYCDAp × s3). ESRD = end-stage renal disease; PYCDA = patient-years of chronic dialysisavoided; s1 = survival rate in year 1 of chronic dialysis (incidence); s2 = survival rate in year 2 of chronic dialysis (prevalence); s3 =survival rate in year 3 of chronic dialysis (prevalence).



risk patients may experience an all-cause mortalityas high as 77% after 4 years.[35] Based on this in-formation, the interval-related survival rates of thepatients in year 1, 2 and 3 after onset of chronicdialysis were calculated as:interval-related survival rate = 1 – (interval-re-

lated all-cause mortality rate)and were applied to the incidental patient-years

of chronic dialysis avoided (PYCDAi) and subse-quent prevalent PYCDA (PYCDAp) during the 3-year time range as shown in the disease model infigure 3. The average and limits of the clinical ef-fectiveness (PYCDA) of adding ramipril to con-ventional treatment of a hypothetical cohort of 100patients are given in table II.

Renal TransplantationIn 1996, there was a total of 52 087 patients with

ESRD undergoing chronic RRT in Germany.[7,31]Of these patients, 2016[7] or 3.87% underwent re-nal transplantation. The time span between the on-set of ESRD requiring chronic RRT and renal trans-plantation in those patients with transplantation ison average 7 years with a lower limit of 2 years andan upper limit of >10 years, according to the inter-viewed specialist healthcare providers. Atime spanof 2 years (lower limit) fits into the 3-year timerange of this evaluation and was accounted for inthe sensitivity analysis.Average interval-related renal transplantation sur-

vival and the percentage of patients with loss ofrenal transplantation requiring chronic dialysis again

were estimated on the basis of information pro-vided by the EuroTransplant Foundation, Leiden,The Netherlands.[36] Assuming that renal trans-plantation takes place in the second half of year 3in our analysis, average renal transplantation sur-vival was estimated as 92.75%, and the percentageof renal transplantation loss requiring chronic dialy-sis again was calculated as 5.65%. The missing1.6% was attributed to all-cause mortality follow-ing renal transplantation.

Treatment Costs

Treatment costs in the course of this evaluationare summarised in table III. These are incrementalcosts of ramipril in outpatient care and costs forchronic RRT (e.g. dialysis, renal transplantation andafter-care following renal transplantation), each in-curred as SHI expenses.Incremental costs for ramipril (Delix® tablets of

1.25, 2.5 and 5mg) in outpatient care were calcu-lated from daily doses medicated per treatmentinterval,[27] prices according to pack size and dos-age as of 15October 1999,[28] patients’prescriptioncharges varying according to pack size,[29] and anSHI discount of 5%.[30] An average cost was de-rived from the interval-related distribution of dailydoses of 1.25, 2.5 and 5mg, respectively. The lower

Table II. Effectiveness of ramipril added to conventional treatmentof a hypothetical cohort of 100 patients at treatment initiation(corrected for all-cause mortality). Discount rate used was 5% perannum

Treatment duration PYCDALower limita Averageb Upper limitc

1 year 3.6 3.9 4.02 years 10.4 12.3 13.13 years 16.3 21.2 23.7a High-risk patients.b Age group between 45 and 54 years (mean age of the patients

in the REIN trial.)[19]

c Low-risk patients.PYCDA = patient-years of chronic dialysis avoided; REIN = Rami-pril Efficacy In Nephropathy.

Table III. Treatment costs given by SHI expenses [rounded to thenearest 5 deutschmarks (DM)]

Cost category Lower limit Average Upper limit

Ramiprila

1 year 20 615 31 480 51 9602 years 32 595 51 635 82 1653 years 38 885 62 580 98 020

Chronic RRTb

Dialysis per patient per year 51 225 84 875 134 910Tx per case 97 190 97 190 115 000After-care per Tx patient peryear

15 545 15 545 20 000

a Incremental cost for ramipril (outpatient SHI expenses) in a hy-pothetical cohort of 100 patients at treatment initiation (1999values). Discount rate used was 5% per annum.

b 1996 undiscounted values.RRT = renal replacement therapy; SHI = Statutory Health Insur-ance; Tx = renal transplantation.



limit assumption was that all patients exclusivelyreceived 1.25mg once daily whilst the upper limitassumption was that all patients exclusively recei-ved 2.5mg twice daily.The costs for chronic dialysis per patient per

year comprise dialytic procedures, medical services,erythropoeitin usage, treatment of complicationsinduced by dialysis, treatment of comorbidity, andtransportation.[32] The procedures considered were:(i) centre haemodialysis; (ii) continuous ambulatoryperitoneal dialysis; (iii) continuous cyclic perito-neal dialysis; (iv) haemodialysis at home; (v) hae-modialysis partially hospitalised; and (vi) intermit-tent peritoneal dialysis in hospital. The average costsfor chronic dialysis per patient per year were givenby the weighted mean of SHI expenses for eachdialysis procedure[32] in 1996. The frequency of thedifferent procedures was taken from the Health Re-port for Germany edited by the Federal StatisticalOffice[7] and from expert knowledge. The lowerlimit was represented by patients undergoing hae-modialysis at home, and the upper limit was givenby patients being partially hospitalised for haemo-dialysis (table III).Average and lower limits of cost for renal trans-

plantation per case were given by the case-specificsurgical procedure FP 13.01 with 97 190 points[33]and a point value of DM1.[34] The upper limit wasderived from expert knowledge. Average and lowerlimit of cost for after-care per renal transplantationpatient per year were taken from the literature[32]

whereas the upper limit was given by expert knowl-edge (table III).

Cost Effectiveness of Ramipril

Base-Case AnalysisThe base-case analysis for estimating the incre-

mental cost effectiveness of adding ramipril to con-ventional treatment of hypertension in patients withnon-diabetic nephropathy (proteinuria of ≥3.0 g/day)is given in table IV. The values are related to ahypothetical cohort of 100 patients at treatment ini-tiation and are differentiated according to treatmentduration. Negative values of incremental costs aresavings by definition (table IV).Incremental costs for ramipril were overcom-

pensated by the SHI expenses avoided for chronicdialysis (table IV). There were net SHI savings (ne-gative total incremental costs) of about DM295 900after 1 year’s treatment, and up to DM1 739 170after 3 years’ treatment (each per 100 patients attreatment initiation). Relating these savings aftereach treatment duration to the respective numberof PYCDAby adding ramipril to conventional treat-ment resulted in net SHI savings per PYCDA.On average, the ICER given by SHI savings per

PYCDA was between DM76 715 (–295 905/3.9)and DM81 930 (–1 739 170/21.2), depending ontreatment duration. SHI savings increased as treat-ment duration with ramipril increased (table IV).The ICER was most favourable for the total treat-ment period of 3 years with savings (negative in-

Table IV. Base-case analysis in a hypothetical cohort of 100 patients at treatment initiation. Incremental cost is given by SHI expenses [roundedto the nearest 5 deutschmarks (DM); 1996 values for chronic dialysis and 1999 values for ramipril]. The discount rate used was 5% per annum

Duration Incremental cost (DM) PYCDAa ICER of ramiprilb

Ramiprilc Dialysisd Totale

1 year 31 480 −327 385 −295 905 3.9 −76 7152 years 51 635 −1 039 860 −988 225 12.3 −80 6603 years 62 580 −1 801 750 −1 739 170 21.2 −81 930a Average according to table II.b Given in DM per PYCDA derived by dividing column 4 by column 5.c Average according to table III.d Avoided SHI expenses derived by multiplying the number of PYCDA (column 5) with average SHI expenses for chronic dialysis per

patient per year according to table III.e Sum of columns 2 and 3.ICER = incremental cost-effectiveness ratio; PYCDA = patient-years of chronic dialysis avoided; SHI = Statutory Health Insurance.



cremental cost) of DM1 739 170 and 21.2 years ofchronic dialysis avoided, each per 100 patients attreatment start.

Sensitivity AnalysesResults from economic analyses are more de-

pendent on temporal and local changes in basicconditions than results from controlled clinical tri-als. To estimate the applicability of an economicevaluation of a controlled clinical trial with highinternal validity, the extent to which the differentmodel variables influence the base-case result wasinvestigated.As the correlation between the target variable,

ICER, and the exogenous parameters (independentvariables) of the model can be described by anequation (see below), the variability of results couldbe investigated using different procedures. First,analysis of extremes was applied in a best-case/worst-case scenario via numerical variation of allof the exogenous parameters. Second, the cumula-tive incidence of ESRD in the ramipril and placebogroups was modelled according to theWeibull me-thod.[39] Third, the extent of impact of all exogen-ous parameters on the target variable was quantifiedby the method of the total differential (determinis-tic model).[40-43] Fourth, the dependency of the tar-get variable on random variables is described in asimulation,[41,42,44,45] the result of which showedthe variance of the target variable that was to beexpected under ‘real world’ conditions. Impact an-alysis and simulation were based on PYCDAquan-tified according to Weibull.[39]The stochastic model for quantification of the

target variable, ICER, is as follows:

ICERk =

�i=1

k

[(ICRi − PYCDAi ⋅ si ⋅ CD) ⋅ (1 − p)(i−1)]

�i=1

k

[PYCDAi ⋅ si ⋅ (1 − p)(i−1)]

where ICERk = ICER after k years of treatment;ICRi = incremental cost for ramipril in the courseof year i; PYCDAi = number of PYCDA in thecourse of year i by adding ramipril to conventional

treatment; si = survival rate under chronic dialysisin year i; CD = cost for chronic dialysis per patientper year; p = discount rate; k = duration of treat-ment (maximum of 3 years); in the sensitivity anal-ysis all exogenous variables were varied.

Scenario AnalysisTable V depicts the combination of limits of the

independent variables for use in the best-case/worst-case scenario. In the best case, the lower limit ofadditional SHI expenses for ramipril was combinedwith the upper limit of avoided costs for chronicRRT. In theworst case, the upper limit of additionalSHI expenses for ramipril was confronted with thelower limit of avoided cost for chronic RRT. In thebest case, renal transplantation was included in theanalysis. The lower limit of the time span of 2 yearsbetween the onset of ESRD and renal transplanta-tion fell into the 3-year horizon of the evaluation.Therefore, 3.87% of those patients who acquiredESRD in the first year and survived until the thirdyear, received renal transplantation in year 3 of treat-ment. Cost for after-care following renal transplan-tation in the remaining time of the third year werecorrected for mortality. The amount of PYCDAbyramipril was corrected by allowing for those pa-tientswho lost their transplant and reverted to chronicdialysis.

Table V. Scenario analysis. Combination of limits of the exogenousparameters of the model

Exogenous parameters Best-casecombination

Worst-casecombination

Cost for ramipril Lower limit Upper limitCost for dialysis Upper limit Lower limitCost for Tx Upper limit NACost for after-care following Tx Upper limit NANumber of PYCDA Average AverageSurvival during chronic dialysis Upper limit Lower limitLatent period of Tx Lower limit NASurvival after Tx 92.75%a NALoss of Txb 5.65%a NAa Estimated on the basis of information provided by EuroTranspl-

ant Foundation.[36]

b Patients who lost their Tx underwent chronic dialysis again.NA = not applicable; PYCDA = patient-years of chronic dialysisavoided; Tx = renal transplantation or transplant.



In each of the cases considered, there were netSHI savings (negative total incremental costs) ofDM133 840 after 1 year’s treatment in the worstcase up to DM3 159 780 after 3 years’ treatment inthe best case, each per 100 patients at treatmentstart. The additional SHI expenses for ramipril wereovercompensated by savings obtained through av-oided cost for chronic RRT, independent of sce-nario and treatment duration.The ICERs of adding ramipril to conventional

treatment, according to scenario and treatment du-ration, are depicted in figure 4, together with thebase-case results. The ICER of ramipril covered awide range. SHI savings (negative incremental cost)per PYCDA varied between about DM37 000 andDM130 000 after 1 year’s treatment and betweenaboutDM45 300 andDM134 800 after 3 years’treat-ment (fig. 4). This wide range indicates that out-comes may be quite different among patients. Thebest case ICER with ramipril would occur in pa-tients who undergo haemodialysis partially hospi-talised. In general, the longer that ramipril had beenadded to conventional therapy, the greater the SHIsavings, even in the worst case.

Modelling of Cumulative Incidence of ESRDThe quantification of PYCDA using the cumu-

lative incidence of ESRD (based on empirically-given data about onset of ESRD), is affected byuncertainty, especially in the third year of the REINstudy where only about 10 patients remained ineach arm of the study (fig. 1). Therefore, the cumu-lative incidence of ESRD and the effectiveness pa-rameter, PYCDA, were quantified by using the Wei-bull model,[39] a modified exponential function:

ci = 1 − e(− α t)β

where ci = cumulative incidence of ESRD; e = Eu-ler’s number; α = exponent 1; t = time in days; β =exponent 2.The exponents α and β in the Weibull distri-

bution[39] were derived by a regression from thepooled data on file of all patients in stratum 2 ofthe REIN study. They are: α = 1.384883 and β =2731.969415 for ramipril; α = 1.004808 and β =2480.641067 for placebo. The number of PYCDAper 100 patients and duration of treatment was de-rived from the difference of the areas under theWeibull curves for ramipril and placebo.In comparison to the method based on empiri-

cally given data from the REIN trial,[27] theWeibullvariant resulted in 1.1 fewer PYCDA per 100 pa-tients after the total period of 3 years, representing

SH

I sav

ings

per

PY

CD

A (D

M)

180 000

20 000

40 000

60 000

80 000

100 000

120 000

140 000

160 000

0Treatment duration (years)

Worst caseBase caseBest case

1

129 700

76 715

36 930

2

132 430

80 660

43 340

3

134 780

81 930

45 265

Fig. 4. Scenario analysis. Incremental cost-effectiveness ratiosof adding ramipril to conventional treatment of hypertension inpatients with non-diabetic nephropathy (proteinuria of ≥3.0 g/day)in Germany depending on treatment duration and on the com-bination of limits of the independent variables (rounded to thenearest 5DM). Discount rate used was 5% per annum. Thereare SHI savings (negative incremental cost) per PYCDA after1, 2 and 3 years of treatment in each of the cases considered.DM = deutschmarks; PYCDA = patient-year of chronic dialysisavoided; SHI = Statutory Health Insurance.

Table VI. Impact analysis. Extent of influence of the independentvariables of the model on the target variable, SHI savings perPYCDA by adding ramipril to conventional treatment after a periodof 3 years. Estimation of cumulative incidence of ESRD accordingto Weibull[39]

Independent variable Elasticity (%) ofthe target variablea

Cost for chronic dialysis per patient per year 1.0373Incremental cost for ramipril -0.0373Clinical effectiveness of ramipril (PYCDA) 0.0369Survival of patients undergoing chronicdialysis

0.0369

Discount rate -0.0010a Change of the target variable as a result of the change of the re-

spective independent variable by 1% when at the same timethe remaining independent variables were kept constant ataverage values (ceteris-paribus conditions).

ESRD = end-stage renal disease; PYCDA = patient-years of chronicdialysis avoided; SHI = Statutory Health Insurance.



amore conservative estimation of the effectivenessparameter.

Impact Analysis – Deterministic ModelThe total differentialmethod[40] was used to quan-

tify the impact of the independent variables of themodel on the target variable, ICER, of ramipril.This procedure has been described in detail in pre-vious publications.[41-43]The extent of influence of the independent vari-

ables of the model on SHI savings per PYCDA isgiven in table VI. The cost for chronic dialysis perpatient per year had by far the greatest impact onthe amount of SHI savings per PYCDA by addingramipril to conventional treatment. This impact ofthis variable was 28-fold that of the incrementalcost for ramipril and of the clinical effectivenessof adding ramipril. Survival of patients undergoingchronic dialysis and the discount rate had very littleinfluence on the outcome from an SHI perspective(table VI).A graphic depiction of the correlation between

the target variable, SHI savings per PYCDA andthe determinants ‘cost for chronic dialysis per pa-

tient and year’ and ‘PYCDA per patient’ is givenin figure 5. This graph shows the target variable asa grid: this is a function of alteration of the 2 in-dependent variables when at the same time theremaining independent variables of the model arekept constant at average values (ceteris-paribus con-ditions). Figure 5 enables direct estimation of theICER of ramipril in the case of future changes inbasic conditions, without the necessity to recal-culate the ICER. If the cost for chronic dialysischanges and/or the effectiveness of ramipril in day-to-day care differs from that observed in the REINtrial, the resulting ICER of ramipril can be takenfrom the graph in steps of –DM5000 (savings) perPYCDA. Because of these predictive properties,the deterministic model minimises future uncer-tainty.

Monte Carlo Simulation – Stochastic ModelTheMonte Carlo simulation is a proven method

to cope systematically with data uncertainties.[44,45]Using this method, computations were made fromcase to case with varied input data: a clinical sce-nario was calculated for each single computation

100 000-105 00095 000-100 00090 000-95 00085 000-90 00080 000-85 00075 000-80 00070 000-75 00065 000-70 00060 000-65 00055 000-60 000

SH

I sav

ings

per

PY

CD

A (D

M)

105 000

75 000

70 00065 00060 00055 000

0.2670.236

0.2050.174

0.144

90 000

85 000

80 000

100 000

95 000

106 0

95

101 8

51

97 60

7

93 36

4

89 12

0

84 87

6

80 63

2

76 38

8

72 14

5

67 90

1

63 65

7

Cost for chronic dialysis per patient per year

PYCDA per patient

Fig. 5. Impact analysis. Incremental cost effectiveness of ramipril by alteration in both cost for chronic dialysis per patient-year andnumber of PYCDA per patient by ramipril under ceteris-paribus conditions. Cost effectiveness after 3 years’ treatment is given by thegridded area. Discount rate used was 5% per annum. Estimation of cumulative incidence of ESRD according to Weibull.[39] DM =deutschmarks; ESRD = end-stage renal disease; PYCDA = patient-years of chronic dialysis avoided; SHI = Statutory Health Insurance.



with combinations of data which were selected atrandom from the empirically given data intervals.In the simulation, the discount rate was treated

as a fixed variable whereas the remaining 4 modelparameters were treated as random variables.[46]Anormal distribution was assumed in the cases ofcost for chronic dialysis, clinical effectiveness oframipril, as well as survival of patients undergoingchronic dialysis. Incremental costs for ramipril weretreated as a variable with log normal distribution.Computation of 10 000 simulation steps were madeto describe the variance of the target variable ICERof ramipril.The variance of the ICER turned out to be rela-

tively broad. Themean of –DM81 930 per PYCDAranged between about –DM53 050 and –DM104 055per PYCDA after the total treatment duration of 3years. For 95% of the 10 000 simulation steps, theICER of ramipril ranged between –DM69 460 and–DM94 595 per PYCDA.

Discussion

Adding ramipril to conventional treatment ofhypertension in patients with non-diabetic neph-ropathy (proteinuria of ≥3.0 g/day) was clearlyjudged beneficial not only from its clinical effec-tiveness, but also from the point-of-view of costsavings, which is currently an increasing concernof the SHI provider (third-party payer) in Germany.Addition of ramipril was better than addition ofplacebo from an SHI perspective because it was lesscostly and more effective. SHI savings (negativeincremental cost) occurred with significant improve-ments in patients’ health conditions. The savingsamounted to about DM17 390 per patient over the3-year period with 0.21 years of PYCDA. In otherwords, there were average savings of DM81 930per PYCDAafter 3 years with savings of DM76 715per PYCDA after the first year. In comprehensivesensitivity analyses, the robustness of the modeland its results were proven. Adding ramipril re-mained the dominant strategy.This retrospective analysis of the incremental

cost effectiveness of adding ramipril to conventionaltreatment from the German SHI perspective (third-

party payer) used a model approach. To meet theneeds of decision-makers[26,47] selection of data forcalcualtion in themodel was primarily based on theappropriateness of reflecting day-to-day practice inmedical care of the target patients covered by theSHI scheme.

Data Input

The model approach consisted of synthesisingclinical, epidemiological and economic data fromthe following sources: (i) results and patient dataon file from a controlled, multicentre clinical trial;(ii) epidemiological information on survival of pa-tients undergoing chronic RRT; (iii) structural dataon treatment costs and their distribution in chronicRRT provided by the inpatient and ambulatory sec-tor in Germany; (iv) laws and regulations govern-ing reimbursement of medical services and drugsin Germany; and (v) expert knowledge.Aclinical trial such as the REIN study[19] is well

suited as a basis for an economic evaluation of ra-mipril under conditions of daily clinical practicebecause of the following reasons: (i) it had a clas-sical randomised, double-blind, parallel-grouped,placebo-controlled study design with a maximumduration of 4 years and 4months which is long termwhen compared with many other drug trials; (ii) thisdesign was combined with naturalistic principlesas the blinded study medication was added to con-ventional antihypertensive treatment in patients withnephropathy in parallel groups targeted at achiev-ing and maintaining diastolic blood pressure under90mm Hg; and (iii) no subsequent change in dietwas introduced.As the most recent version of each data source

was used, the reference years for the estimation ofincremental costs for ramipril in outpatient care(1999) and costs for RRT (1996) differed. It maybe argued that this does not show the drug in goodlight because of higher prices in 1999 comparedwith 1996; however, the sensitivity analysis showedthat the impact of incremental drug costs was min-imal when compared to the costs for chronic dial-ysis. Choosing the same reference year (1996) wouldhave led to slightly higher savings with ramipril.



Expert knowledge was used in 3 cases: (i) toquantify the distribution of centre haemodialysisand dialysis partially hospitalised for the calcula-tion of the average costs of all 6 different proce-dures; (ii) to estimate the time gap between onsetof ESRD and renal transplantation; and (iii) to es-timate the upper limits of cost for renal transplan-tation and after-care. Though we did not undertakea representative survey on these topics, this short-coming is tolerable. First, in the case of the averagecost for chronic dialysis, the results from our an-alysis remained stable in the sensitivity analysiseven under worst-case assumptions. Second, renaltransplantation was accounted for only in the best-case scenario of the sensitivity analysis. Costs as-sociated with renal transplantation did not contrib-ute to the base-case results, as patients’ time on thewaiting list of about 5 years[7] exceeded the 3-yeartime range of our analysis. In addition, expert know-ledge with respect to cost for renal transplantationand after-care was used to estimate upper limitsof these costs, the averages of which were fixedby German healthcare regulations[33] and publish-ed data.[32,34] Therefore, we do not feel that theseestimates are prone to the shortcomings of consen-sus methods and expert panels in pharmacoecono-mic studies that have been discussed.[48]

Methodological Issues

In Germany, direct medical cost is representedby: (i) expenses of the SHI scheme which covers>90% of inhabitants and includes ambulatory andinpatient care; (ii) charges paid directly by mem-bers of the SHI plan; (iii) expenses of private in-surance funds; and (iv) expenses of pension fundsfor rehabilitation treatment.[41] This economicevaluation was undertaken from the perspective ofthe SHI provider (third-party payer) in Germanyand therefore had to concentrate on direct costsreimbursed by SHI. This means that only SHI ex-penses and not all direct costs were included in theeconomic evaluation of the REIN study for Ger-many, and all indirect costs were excluded. Fur-thermore, SHI expenses for treatment of adverseevents were excluded from the analysis, since the

number of serious non-fatal cardiovascular eventswas similar in the 2 arms of stratum 2 in the REINstudy.[19]Costs for non-ACE inhibitor antihypertensive

medication in the ramipril and placebo groups werealso excluded from the analysis. In patients alreadyreceiving antihypertensive agents (74% in the ra-mipril and 80% in the placebo groups), the study-drug dose was increased and the dose of other an-tihypertensive drugs was progressively reduced toavoid symptomatic hypotension. In each patient,the aim was to achieve and maintain a diastolicblood pressure of 90mmHg with the minimum doseof concomitant antihypertensive agents.[19] How-ever, we decided to take the conservative approachof incorporating total incremental costs for rami-pril and not to subtract the reduced costs for con-comitant antihypertensive agents (i.e as a result oframipril medication).Costs for medical services incurred by the dose

titration of ramipril were also not considered in theanalysis. For the purposes of our evaluation, it wasexpected that dose titration would occur in out-patient care during routine visits with internists/nephrologists in contrast to the inpatient initiationof adjuvant ramipril treatment that occurred in theREIN trial because of the study protocol.[19]The time range of the economic evaluation of

ramipril was restricted to 3 years because only abouthalf of the patients had a follow-up of >1 year and<21 patients were in the study for >3 years with noincidental cases of ESRF occurring in years 4 and5 of the REIN trial.[19,27] Therefore, we took a con-servative approach to avoid speculative assump-tions.

Therapeutic and Economic Implications

ACE inhibitors reduce the rate of renal functiondecline in insulin-dependent diabetes mellitus.[17,49]The benefit of the ACE inhibitor, benazepril, in non-diabetic renal disease was shown in the Angiotensin-Converting-Enzyme Inhibition in Progressive RenalInsufficiency (AIPRI) trial.[50] However, diastolicblood pressure decreased by 3.5 to 5.0mm Hg inthe benazepril group and increased by 0.2 to



1.5mm Hg in the placebo group.[50] The REIN trialclearly showed that ramipril was beneficial in non-diabetic chronic renal disease.[19] The protective ef-fect of ramipril on decline in (GFR) and risk ofESRF was accompanied by a substantial decreasein the urinary protein excretion rate, which appearedto exceed the reduction that might be expected fromblood pressure reduction alone.[19]It is generally accepted that ACE inhibitors

possess renoprotective properties and also lowerproteinuria more than other antihypertensive drugs,despite comparable blood pressure lowering prop-erties.[19,51,52] Data from clinical trials seem to in-dicate that ACE inhibitors (and possibly calciumantagonists) should be preferred in the treatment ofpatients with diabetic and non-diabetic nephropa-thies.[52] To date, other drugs have not been testedin a similar fashion to the REIN trial, where theycould have proven to have an equal effect to that oframipril.Economic evaluations of the ACE inhibitor, cap-

topril, in diabetic patients with nephropathy revealedsignificant economic savings for the healthcare sys-tems in different countries[20-23] (based on reduc-tion in incidence of ESRD).[17] Preventive treat-ment with the ACE inhibitor, benazepril, is costeffective in patients with chronic renal insuffici-ency of various causes.[24] Our analysis evaluatedramipril in non-diabetic patients with nephropathyand hypertension in Germany. The results concurwith the abovementioned evaluations. SpendingDM100 on ramipril to treat hypertension in patientswith non-diabetic nephropathy yielded net savingsof DM940 after the first year of treatment, withincreasing savings as ramipril treatment progressed.The clinical and economic importance of adding

ramipril treatment to non-diabetic patients withnephropathy and hypertension is emphasised bythe knowledge that in 69% of patients commencingchronic RRT in 1997 in Germany, diabetes was notthe primary cause of ESRD.[9] Since the number ofpatients requiring chronic RRT is expected to growin the future and the transplantation rate is far fromsufficient in Germany,[9] clinically and economi-cally beneficial treatment can be provided by addi-

ing ramipril medication, especially to those pati-ents with severe proteinuria who have the greatestrisk of developing ESRD.

Conclusions

Adding ramipril to conventional therapy is thedominant treatment strategy for patients with non-diabetic nephropathy and persistent proteinuria (≥3g/day) with more effectiveness and less cost thanconventional antihypertensive treatment. Because ofthe high burden of ESRD to the individual patientand to the healthcare system as a whole, ramiprilhas an enormous saving potential from the perspec-tive of the SHI provider (third-party payer) in Ger-many when added to conventional treatment of hy-pertension for this patient group. This more thanjustifies its additional cost in outpatient care.

Acknowledgements

This investigationwas funded byAventis PharmaDeutsch-land GmbH, D-65812 Bad Soden/Taunus, Germany.The authors would like to thank the REIN study group

(GISEN, Gruppo Italiano di Studi Epidemiologici in Nef-rologia), Ospedali Riuniti di Bergamo and Mario Negri In-stitute for Pharmacological Research, Bergamo, Italy, for thepatient data on file of stratum 2 in the REIN study and theirideational support during extensive discussions, especiallyGiuseppe Remuzzi, MD, Piero Ruggenenti, MD and AnalisaPerna, Stat. Sci. D. The authors are also indebted to thefollowing physicians in Germany who participated in thesurvey of specialists providing chronic renal replacementtherapy: Gabriele Schott, MD, internist/nephrology, Düssel-dorf, and ProfessorManfredWeber, MD, internist/nephrology,Head of Medical Clinic I, Clinics of the City of Cologne,Cologne-Merheim.At the time this researchwas conducted, PeterK. Schädlich

and Josef Georg Brecht were employees of PAREXELGmbH,Berlin, Germany. The views expressed in this article aresolely those of the authors.Meeting presentations – the information contained in this

manuscript has been presented in part as posters at the 5thAnnual InternationalMeeting of the International Society forPharmacoeconomics andOutcomesResearch,Arlington (VA),USA, May 2000 and at the Congress of Nephrology 2000,Vienna, Austria, September 2000.

References1. Remuzzi G, Ruggenenti P, Benigni A. Understanding the nature

of renal disease progression. Kidney Int 1997; 51: 2-15



2. Kobrin S, Aradhye S. Preventing progression and complica-tions of renal disease. Hosp Med 1997; 33 (11): 11-12, 17-18,20, 29-31, 35-36, 39-40

3. Brenner BM, Meyer TW, Hostetter TH. Dietary protein intakeand the progressive nature of kidney disease: the role of hae-modynamically mediated glomerular injury in the pathogen-esis of progressive gomerular sclerosis. N Engl J Med 1982;307: 652-9

4. Hostetter TH, Olson JL, Rennke HG, et al. Hyperfiltration inremnant nephrons: a potentially adverse response to renal ab-lation. Am J Physiol 1981; 241: F85-93

5. Bailey RR. Nephrotic syndrome. In: Speight T, Sutherland J,editors. New ethicals disease index. 4th ed. Auckland (NZ):Adis International Limited, 1995: 661-2

6. Songer TJ. The economic costs of NIDDM. Diabetes MetabRev 1992; 8: 389-404

7. Federal Statistical Office. Chronic renal failure. In: Health Re-port for Germany. Stuttgart: Metzler-Poeschel, 1998: 254-8

8. Pastan S, Bailey J. Dialysis therapy. N Engl J Med 1998; 338:1428-37

9. Frei U, Schober-Halstenberg HJ. Annual Report of the GermanRenal Registry 1998.QuaSi-NiereTask ForceGroup for qualityassurance in renal replacement therapy. Nephrol Dial Trans-plant 1999; 14: 1085-90

10. Klahr S. Prevention of progression of nephropathy. NephrolDial Transplant 1997; 12 Suppl. 2: 63-6

11. Klang MJ, Whelton PK, Randall BL, et al. Blood pressure andend-stage renal disease in men. N Engl JMed 1996; 334: 13-8

12. Valderrábano F, Gómez-Campderá F, Jones EH. Hypertensionas cause of end-stage renal disease: lessons learnt from inter-national registries. Kidney Int 1998; 68 Suppl.: S60-6

13. Mogensen CE. Long-term antihypertensive treatment inhibitingprogression of diabetic nephropathy. BMJ 1982; 285: 685-9

14. Parving HH, Andersen AR, Smidt UM, et al. Early aggressiveantihypertensive treatment reduces rate of decline in kidneyfunction in diabetic nephropathy. Lancet 1983; I: 1175-9

15. Parving HH, Hommel E, Smidt UM. Protection of kidney func-tion and decrease in albuminuria by captopril in insulin de-pendent diabetics with nephropathy. BMJ 1988; 297: 1086-91

16. Bjorck S, Mulec H, Johnsen SA, et al. Renal protective effectof enalapril in diabetic nephropathy. BMJ 1992; 304: 339-43

17. Lewis JB, Berl T, Bain RP, et al. Effect of intensive blood pres-sure control on the course of type 1 diabetic nephropathy.Collaborative Study Group. Am J Kidney Dis 1999; 34: 809-17

18. Gruppo Italiano di Studi Epidemiologici in Nephrologia(GISEN). A long-term, randomized clinical trial to evaluatethe effects of ramipril on the evaluation of renal function inchronic nephropathies. J Nephrol 1991; 3: 193-202

19. GISEN Group (Gruppo Italiano di Studi Epidemiologici inNefrologia). Randomised placebo-controlled trial of effect oframipril on decline in glomerular filtration rate and risk ofterminal renal failure in proteinuric, non-diabetic nephropa-thy. Lancet 1997; 349: 1857-63

20. Le Pen C, Petitjean P, Lévy P, et al. Economic evaluation of thecontribution of captopril in the treatment of diabetic nephrop-athy: a cost-benefit approach. Nephrologie 1996; 17: 321-6

21. Szucs T, Ritz E, Standl E. DieWirtschaftlichkeit von Captopril.Eine Analyse an Patienten mit diabetischer Nephropathie inDeutschland. Münch med Wschr 1994; 136: 581-5

22. Garrattini L, Brunetti M, Salvioni F, et al. Economic evaluationof ACE inhibitor treatment of nephropathy in patients withinsulin-dependent diabetes mellitus in Italy. Pharmaco-economics 1997; 12: 67-75

23. Hendry BM, Viberti GC, Hummel S, et al. Modelling and cost-ing the consequences of using an ACE inhibitor to slow theprogression of renal failure in type I diabetic patients. Q JMed1997; 90: 277-82

24. van Hout BA, Simeon GP, McDonnell J, et al. Economic eval-uation of benazepril in chronic renal insufficiency. Kidney Int1997; 63 Suppl.: S159-62

25. Brecht JG, Jenke A, Köhler ME, et al. Empfehlungen der Deu-tschen Gesellschaft für Klinische Pharmakologie und Therapiezur Durchführung und Bewertung pharmakoökonomischerStudien. Klin Pharmakol akt 1995; 6 (1): 4-11

26. Glaeske G, von Stillfried D. Pharmakoökonomie als Entschei-dungshilfe in der Arzneimittelversorgung: Perspektiven ausSicht der GKV. Ersatzkasse 1995; 75: 298-307

27. Ruggenenti P, Perna A, Mosconi L, et al. Patient data of stratum2 in the randomised placebo-controlled trial of effect of rami-pril on decline in glomerular filtration rate and risk of terminalrenal failure in proteinuric, non-diabetic nephropathy. TheGISEN Group (Gruppo Italiano di Studi Epidemiologici inNefrologia). Bergamo: Ospedali Riuniti di Bergamo and Ma-rio Negri Institute for Pharmacological Research, 1996. (Dataon file)

28. GroβeDeutsche Spezialitäten-Taxe,Lauer®-TaxemitApotheken-Ein- und Verkaufspreisen. Fürth: Pharma Daig + Lauer, 1999Okt 15

29. Gesetz zur Stärkung der Solidarität in der gesetzlichen Kran-kenversicherung, 1998

30. Sozialgesetzbuch, Fünftes Buch, 1993 mit §§ 31 und 39 überZuzahlungen von in der GKV Versicherten bei Arznei- undVerbandmitteln sowie bei Krankenhausbehandlung sowie §130 über den Rabatt in Höhe von fünf Prozent auf den Arz-neimittelabgabepreis, den die gesetzlichen Krankenkassen vonden Apotheken erhalten

31. Frei U, Schober-Halstenberg HJ. Nierenersatztherapie inDeutschland. Bericht über Dialysebehandlung und Nieren-transplantation in Deutschland 1996. Berlin: Projektgesch-äftsstelle QuaSi-Niere, Jul 1998

32. Nebel M. Behandlungskosten der verschiedenen Nierenersatz-verfahren. Präsentation beim Satelliten-Symposium ‘Wird Dia-lyse unbezahlbar – Netzwerkstrategien zur Verbesserung derFrühdiagnostik von Nierenerkrankungen’ der 105. Tagung derDeutschen Gesellschaft für Innere Medizin – Internistenkon-gress; 1999 Apr 10-14; Wiesbaden

33. Verordnung zur Regelung der Krankenhauspflegesätze (Bun-despflegesatzverordnung – BPflV) vom 26. September 1994,zuletzt geändert durch die Fünfte Verordnung zur Änderungder Bundespflegesatzverordnung vom 9.Dezember 1997,mitAnlage 1: Fallpauschalen-Katalog

34. Goergen H, Riedel RR, Vetter H. Sonderentgelte und Fallpaus-chalen. Köln: Deutscher Ärzte-Verlag, 1997: 29

35. Valderrábano F, Jones EHP,Mallick NP. Report of managementof renal failure in Europe, XXIV, 1993. Nephrol Dial Trans-plant 1995; 10 Suppl. 5: 1-25

36. Greiner W. Ökonomische Evaluationen von Gesundheitsleist-ungen. (Gesundheitsökonomische Beiträge, Band 31). In:Gäfgen G, Oberender P, editors. Baden-Baden: Nomos Ver-lagsgesellschaft, 1999: 240

37. Drummond MF, Stoddart GL, Torrance GW. Methods for theeconomic evaluation of healthcare programmes. Oxford: Ox-ford University Press, 1987

38. Coyle D, Tolley K. Discounting of health benefits in the phar-macoeconomic analysis of drug therapies: an issue for de-bate? Pharmacoeconomics 1992; 2: 153-62



39. Weibull W. A statistical theory of the strength of materials. IngVetenkaps Akad Handl 1939; 151: 1-45

40. Bronstein IN, Semendjajew KA. Taschenbuch der Mathematik.25, durchgesehene Auflage. Stuttgart: Teubner, 1991

41. Schädlich PK, Brecht JG. The cost effectiveness of acamprosatein the treatment of alcoholism in Germany: economic eval-uation of the Prevention of Relapse with Acamprosate in theManagement ofAlcoholism (PRAMA) study. Pharmacoecon-omics 1998; 13 (6): 719-30

42. Schädlich PK, Huppertz E, Brecht JG. Cost-effectiveness anal-ysis of ramipril in heart failure after myocardial infarction:economic evaluation of the Acute Infarction Ramipril Effi-cacy (AIRE) study for Germany from the perspective of Stat-utory Health Insurance. Pharmacoeconomics 1998; 14 (6):654-69

43. Schädlich PK, Brecht JG. Economic evaluation of specific im-munotherapy versus symptomatic treatment of allergic rhini-tis in Germany. Pharmacoeconomics 2000; 17 (1): 37-52

44. Kleijnen JPC, editor. Statistical techniques in simulation. PartI, II. New York (NY): Marcel Dekker, 1975

45. Jöckel K-H. Eigenschaften und effektive Anwendung vonMonte-Carlo-tests [dissertation]. Dortmund: University ofDortmund, 1982

46. Hartung J, Elpelt B, Klösener KH. Statistik: Lehr- und Hand-buch der angewandten Statistik. 3. Auflage. München: Olden-bourg, 1985

47. Langley PC. The future of pharmacoeconomics: a commentary.Clin Ther 1997; 19: 762-9

48. Evans C. The use of consensus methods and expert panels inpharmacoeconomic studies: practical applications and meth-odological shortcomings. Pharmacoeconomics 1997; 12 (2 Pt1): 121-9

49. Kasiske BL,Kalil RSN,Ma JZ, et al. Effects of antihypertensivetherapy on the kidney in patients with diabetes: a meta regres-sion analysis. Ann Intern Med 1993; 118: 129-38

50. Maschio G, Alberti D, Janin G, et al. Effect of the angiotensin-converting-enzyme inhibitor benazepril on the progression ofchronic renal insufficiency. The Angiotensin-Converting-En-zyme Inhibition in Progressive Renal Insufficiency (AIPRI)Study Group. N Engl J Med 1996; 334: 939-45

51. Ruggenenti P, Perna A, Gherardi G, et al. Renal function andrequirement for dialysis in chronic nephropathy patients onlong-term ramipril: REIN follow-up trial. Lancet 1998; 352:1252-6

52. Salvetti A, Mattei P, Sudano I. Renal protection and antihyper-tensive drugs: current status. Drugs 1999; 57: 665-93

Correspondence and offprints: Dr Peter K. Schädlich, InFor-Med GmbH – Outcomes Research & Health Economics,Bureau Itzehoe, Conrad-Roentgen-Strasse 58 c, D-25524Itzhoe, Germany.E-mail: [email protected]



cost effectiveness of ramipril in patients with non-diabetic nephropathy and hypertension

Documents