C o s t -of Pravustatin inSecondary Prevention ef Coronary Artery

DiseaseTalat Ashraf, MD, MS, Joel W. Hay, PhD, Bertram Pitt, MD, EllisonWittels, MD,

John Crouse, MD, Michael Davidson, MD, Curt D. Furberg, MD,and l-art-yRadican, MPH

Thisstudyanalyzedthecast-effectivenessafpravastatin infarction(Ml)far incrementalmalepatientsin thepla-in secondarypreventionof coronaryarterydisease cebogroupwhohadMl.A Markovprocesswasusedto(CAD).Thepr+cted riskmodalin 445 malepatients estimaklife-yearssavad,and decisionanalysiswaswithestablishedCAD and moderatelyelevatedserum usedtoestimatecost.Dependingonthepatient-riskpro-Iow-densitylipoproteincholesterolused resultsdata file,themidrangeestimatedcostperlife-yearsavedwithfram2 piaceba-controlledplaqueregressiontrials:Pra- pravastatininsecondarypreventionofCADvariedframvastatinLimitationofAtherosclerosisintheCoronaryAr- $7,124to$12,665,whichisfavomblecomparedwithteriesandPravastatin,Lipids,andAtherosclerosisinthe otherwidelyacceptedmedicalintwventians.Carotids.FmminghamHeartStudydatawereusedto (AmJCardiol1996;78:409-414)pro@cttheriskof mortality10 yearsaftermyocardial

ncreasing evidence suggests that secondary pre-vention of coronary artery disease (CAD) with a

3-hydroxy-3-methylglutaryl coenzyme A reductaseinhibitor reduces both morbidity and mortality. Tworecently completed 3-year placebo-controlled plaqueregression studies—Pravastatin Limitation of Ath-erosclerosis in the Coronary Arteries (PLAC 1)1andPravastatin, Lipids, and Atherosclerosis in the Ca-rotids (PLAC II)2—demonstrated a statistically sig-nificant difference in morbidity for patients with amean low-density lipoprotein cholesterol level of160 mg/dl. The present study examines the cost-ef-fectiveness of pravastatin when used in the second-ary prevention of CAD. Data from the PLAC I andPLAC II studies provided the basis for the projectedrisk model used in this economic analysis.

PIAC I AND PIAC II STUDY RESULTSIn addition to assessing the primary end point of

atherosclerotic plaque regression in the coronary andcarotid arteries, the PLAC studies also measured 4clinical outcomes: CAD death, non-CAD death, fatalmyocardial infarction (MI), and nonfatal MI. Re-sults from these trials were pooled for this analysisbecause of similar inclusion criteria, comparablebaseline demographics, and baseline risk factors. Apooled statistical analysis of clinical end points forboth PLAC trials was prespeciiied in the clinicalanalysis plan. The resulting pooled sample (n =

From Emron, Inc., Warren Township, New Jersey; the University ofSouthern California, 10S Angeles, California; the University of Michi-gan, Ann Arbor, Michigan; Ba Ior College of Medicine, Houston,

YTexas; theBowman Gray Schoo of Medicine, Winston-Salem, NorthCarolina; and Rush-Presbyterian-Saint Luke Medical Center, Chi-cago, Illinois. This study was supported by Bristol-Myers Squibb,Plainsboro, New Jersey. Manuscript received October 10, 1995;revised manuscriptreceivedand acceptedFebruary 28, 1996.

Address for reprints: Talat Ashraf, MD, MS, PPD Inc., 115 NorthThird Street, Suite 500, Wilmington, North Carolina 28401.

559) consisted of 445 men and 114 women, andthere were no statistically significant differences atbaseline between the PLAC I and PLAC II studygroups.

The sample size for women was not large enoughto permit inferences to be drawn; therefore, differ-ences in clinical outcomes of the pooled PLAC studyresults between the pravastatin and placebo groupswere tested for statistical significance only in men(mean age 60 years). Over the 3-year study period,the number of men having a nonfatal MI was lowerin the pravastatin group (9 vs 20, p <0.05). Furtheranalysis revealed no significant difference in menhaving nonfatal MIs in the pravastatin versus the pla-cebo group in year 1 (7 vs 8), but the differences inyears 2 ( 1 vs 6) and 3 ( 1 vs 6) were both significant(p <0.05). No statistically significant difference inall-cause mortality was observed at the end of the 3-year PLAC study period, possibly owing to, thestudy’s small effective sample size. The pooled 3-year frequency and incidence rates of men having aCAD death, fatal MI, non-CAD death, and nonfatalMI by study group and study year are listed in TableI. The total number of men who had events for eachclinical outcome over the 3-year study period is alsoprovided.

METHODSModeling for this economic analysis was con-

ducted only for those subgroups (men), clinical out-comes (nonfatal MIs), and PLAC study years (years2 and 3) that were statistically significant. Becausethe PLAC trials demonstrated a statistically signifi-cant difference between the placebo and treatmentgroups in morbidity (nonfatal MI) and not mortality,it was necessary to model mortality so that years oflife saved with pravastatin could be estimated. Like-wise, because the PLAC trials did not provide long-term follow-up for patients who survived an infarc-

C21996 by ExcerptaMedico,Inc.All rights reserved.

0002-9149/96/$15.00 409Pll S0002-9149(96)00328-1

TABLE1 PLAC I and II Male Patients With Clinical Events* (frequenciesand rates)

Placebo (n = 22o) Pravastatin (n = 225)

Year 1 Year 2 Year 3 Total Year 1 Year 2 Year 3 Total

CAD deathFrequency 2 0 0 2 1 2 1 4Rate 0.0091 0.0000 0.0000 0.0044 0.0089 0.0044

Fatal MlFrequency 1 1 1 3 1 0 0 1Rate 0.0045 0.0045 0.0045 0.0044 0.0000 0.0000

Non-CAD deathFrequency o 1 3 4 0 2 0 2Rate 0.0000 0.0045 0.0136 0.0000 0,0089

Nonfatal Ml0,0000

Frequency 8 6 6 20 7 1 1 9Rate 0.0364 0.0273 0.0273 0.0311 0.0044 0.0044* Some patients had >1 event; values in this chart represent the numberof patienk, not the numberof events. Specifically, in the placebogroup, there were 9 male

nonfatal Mls in year 1 and 7 in year 2, for a total of 22 male nonfatal Mk. Economicmodelingwas conductedon patients, however, and not events.CAD = coronaty artery disease; Ml = myocardialinfarction.

tion, subsequent morbidity and resulting medicalcare costs were required to be modeled as well.

Estimatingmortality:After the incremental differ-ence in nonfatal MIs between the pravastatin andplacebo groups was determined, mortality was mod-eled proportionately in a hypothetical group of 1,000patients (mean age 60 years). Because the PLACstudies followed patients for 3 years, which is aninsufficient time period in which to observe anymeaningful long-term morbidity and mortality afterMI, a Markov process was used to estimate the num-ber of patients alive or dead in each year of a 10-year post-nonfatal MI follow-up period, beginningfrom either year 2 (mean age 61 years) or year 3(mean age 62 years) of the PLAC studies.

The Framingham Study, Section 35,3 was used toestimate subsequent annual morbidity and mortalityrates for patients with nonfatal MIs. In both the treat-ment and placebo groups, it was assumed that pa-tients who experienced nonfatal MIs would have thesame yearly transition probabilities for morbidityand mortality. A 10-year follow-up period was mod-eled because patients who survive for a 10 years af-ter a nonfatal MI have a similar chance of mortalityas patients in the same age cohort who have neverhad a nonfatal MI.4

In the Markov process, 3 possibilities exist whenpatients experience a nonfatal MI during the secondor third year of the PLAC study. The patients canremain alive, die from noncardiovascular diseasecauses, or die from cardiovascular disease during thefollowing year. Patients are moved through themodel in this manner for a total of 10 years after MI,and the probability of remaining alive, dying fromnoncardiovascular disease causes, or dying from car-diovascular disease is the same every year. The tran-sition probabilities used for cardiovascular diseasedeath considered 3 scenarios based on the presenceof additional risk factors for CAD. Three indepen-dent post-MI annual cardiovascular disease mortalityrates—3.4%,3 5~0,5 and 7.5%5-were assumed be-cause all patients enrolled in the PLAC studies were

410 THE AMERICAN JOURNAL OF Cardiology@ VOL 78

at high risk with a historv of CAD. and additionalrisk Factors (e.g., diabete~ mellitus,’obesity, smok-ing, sedentary lifestyle) would effect their survivalafter MI.

Because all patients modeled were men withCAD, each of the 3 transition probabilities for deathwere assumed to be equated with differing levels of.additional risk factors; 3.4$Z0,570, and 7.5$10annualcardiovascular disease mortality rates were assumedfor male CAD patients with 1, 2, and 3 additionalrisk factors, respectively. The yearly noncardiovas-”cular disease mortality rate ( 1.290)was derived fromthe Framingham data and was constant. The yearlyprobability of a patient living was computed as: 1 –(cardiovascular disease mortality risk + noncardio-vascular disease mortality risk).

For patients who had a nonfatal MI in the secondyear of the PLAC studies (mean age 61 years), theMarkov process started at age 62 years and contin-ued until the patient reached 71 years of age. Forpatients who had a nonfatal MI during the third yearof the PLAC studies (mean age 62 years), the Mar-kov process started at age 63 years and continueduntil the patient reached 72 years of age. As patientsage.in the Markov process, they have different ratesof survival and average years of remaining life.

life-tableanalysis:Patients who experienced anonfatal MI in the PLAC studies would have a sig-nificantly higher risk of mortality than those whonever had a nonfatal MI, and fewer remaining yearsof life on average than these patients. Therefore, anactuarial or life-table analysis was performed to de-termine the survival curve and the average remainingyears of life for each age cohort for patients who hada nonfatal MI. Once the average years of remaininglife was computed by age cohort, it was used in theMarkov process to determine expected life-yearssaved with pravastatin therapy (LYS ). Owing tolack of data, these years were not adjusted by qualityof life. National life tablesGwere adjusted using the3 different CAD risk group annual cardiovasculardisease mortality rates of 3.4%, 5%, and 7.5%.

AUGUST 15, 1996

Incrementalnonfatalmyocardialin-farctions:When the results of the 3-year PLAC I and II study periodswere simulated for n = 1,000 pa-tients, the annual incremental inci-dence of nonfatal MIs in the placebogroup over the treatment group was22.83 in both years 2 and 3, Thisnumber was derived by applying theincidence rates in Table I to n =1,000 patients for both the prava-statin and placebo groups and thencomputing the difference.

TABLE II Hospito[ization and Post-EventFollow-up IncrementalCosts

Congestive HeartTime Period Ml Angina Pectoris Stroke* Failure

Initial hospitalization $18,614 $17,140 $7,366 $12,949Year 1 follow-up $2,142 $3,1A0 $5,434 $2,881Year 2 follow-up $898 $1,896 $892 $1,637Year 3 follow-up $898 $1,896 $892 $1,637Year 4 Follow-up $812 $1,724 $606 $1,465Year =5 follow-upt $812 $1,724 $806 $1,465

1

then added for uatients with complications. The

‘ FQkaw-upcost for stroke dues nut include possible nursing homeor home health core costs.T Only Ml has >5 years of followwp,Ml = myacardialinfarction.

Costofdrugtherapy:The perspec-tive of this study was societal; however, indirectcosts were not calculated because of the lack of data.Although most patients in PLAC I and PLAC 11wereprescribed pravastatin 40 mg once daily throughoutthe trial, differences in dose titration resulted in aweighted average dose per patient of 36.64 mg. TheUnited States 1995 average wholesale price of pra-vastatin 40 mg’ was used to estimate drug acquisi-tion cost for the 36.64 mg dose used in the model.

Only incremental costs of pravastatin therapy ini-tiation and follow-up were examined, because pa-tients enrolled in the PLAC studies had CAD andwere already being followed-up for their existingdisease condition. Therefore, therapy initiation andfollow-up costs were those costs over and above theresources consumed for their CAD. These costs, ad-justed to 1995 dollars using the medical care con-sumer price index, were derived from the literature8-10and included incremental physician office visits, li-poprotein profiles, creatine kinase tests, liver enzymetests, and laboratory setup fees associated with pra-vastatin therapy.

Pravastatin was prescribed for all patients whoremained both alive and enrolled in the PLAC stud-ies during each year of the 3-year study period, re-gardless of whether they experienced a coronaryevent. Because there was no significantdifference inthe development of adverse drug reactions betweenthe treatment and placebo groups of the PLAC stud-ies, incremental costs to treat pravastatin adversedrug reactions were not warranted in this analysis.Pravastatin was assumed to be dispensed 4 times peryear and the dispensing fee was the 1995 Medicaidnational average.11

Hospitalizationcostsof myocardialinfarctionandsubsequentevents:The hospitalization costs of MIand its possible subsequent events, including angina,stroke, congestive heart failure, and second MI, werecomputed using a clinical decision algorithm basedon pyramidal logic. All probabilities in the decisionalgorithm were the result of an extensive literaturereview. The bottom of the pyramid included themedical diagnosis-related group national averagecharge for uncomplicated MI 12;the probabilities andcosts of complications (angina, congestive heart fail-ure, arrhythmias, and stroke) and procedures (cath-eterization, percutaneous transluminal coronary an-gioplasty, and coronary artery bypass surgery) were

complication m-s of the algorithm ‘were also usedto estimate the hospitalization costs for the compli-cations alone. Hospital surgical and diagnostic inter-vention costs associated with MI and its complica-tions were obtained from the Agency for Health CarePolicy and Research13and the literature10and wereadjusted to 1995 dollars. Daily physician fees forhospital care of MI and its complications as well asphysician procedural fees for diagnostic and surgicalinterventions were taken from the 1995 PhysiciansFee and Coding Guide.*

Postmyocardialinfarctionand subsequenteventfol-low-upcosts:The post-MI and subsequent event fol-low-up units and frequencies of service were basedon the opinion of an expert panel composed of sev-eral clinicians. Costs for relevant medical serviceswere taken from the literature10and were adjustedto 1995 dollars. These costs included physician of-fice visits, exercise tests, laboratory tests, chest x-rays, cardiac rehabilitation, and drug therapy.7’14Pa-tient follow-up of nonfatal MI was assumed to covera period of 10 years, regardless of the time of theinfarction. In patients who experienced a subsequentevent after a nonfatal MI, a 5-year follow-up periodwas used owing to the poor survival data for patientsexperiencing >1 event. Because patient follow-upswould vary from year to year, annual follow-up costswere computed separately.

A clinical decision algorithm was created to es-timate annual subsequent events and the related hos-pitalization and follow-up costs for patients who hada nonfatal MI. For patients who had a nonfatal MI,chance nodes included no further events or eventscomposed of angina, congestive heart failure, stroke,or second MI. The probabilities of subsequent eventswere estimated from the Framingham data3; the an-nual probabilities that a patient would develop an-gina, congestive heart failure, stroke, or second MIwere estimated to be 3.7%, 1.8%, l.1%, and 2.8?70,respectively. Patients who experienced a secondevent had a chance of living or dying after that event;therefore, the total number of deaths in the algorithmwere calibrated so that they never exceeded the totalnumber of deaths estimated in the Markov process.

Totolincrementalcostestimation:The following for-mula was used to calculate the total incremental costof pravastatin therapy (TICp) including the cost of

CORONARY ARTERY DISEASE/COST-EFFECTIVENESS OF PRAVASTATIN 411

TABLE Ill Increase in life ExpectancyWith Pravastatin (discountedvalues]

Discounted Life Expectancy Discounted Life Expectancy ChangeRisk Profile Without Treatment (yr) With Treatment (yr) (yr] Average Increose

Male, CAD + 1 risk factor 11.44 11.55 0.11 1.0%Male, CAD + 2 risk factors 11.27 11 .A2 0,15 1 ,3%Mole, CAD + -3 risk Factors 10.97 11,18 0.21 1 .9%

1 risk factor = 3.4% ❑nnual post-Ml cardiovasculardisease mortality; 2 risk factors = 5% annual post-Mlcardiovasculardisease mortali~; >3 risk factors = 7.5%annual post-Ml cardiovasculardisease mortality. Noncardiovascular disease annual mortali~ = 1.2%,

Abbreviations as in Table 1.

TABLE IV Cost Per Life-Year Saved for Men Based on RiskFactor Status

Pravastatin 36.64 mg(weighted mean dose)–

Risk Profile Cost Per Life-Year Saved

Male, CAD + 1 risk factor $12,665Male, CAD + 2 risk factors $9,368Male, CAD + =3 risk foctors $7,124

See Table Ill for explanation of risk factors.Abbreviations as in Table 1.

therapy initiation and follow-up over placebo: TICP= Cfd) + C(mc) – C(scad),-where C(d) = incr~-mental cost of pravastatin therapy; C(mc) = incre-mental cost of medical care; and C(scad) = costsavings resulting from pravastatin-related reductionin CAD morbidity.

Cost-efkctivenessanalysis:The following formulawas used to calculate the cost effectiveness: Cost perlife-year saved = TICp/LYS, where LYS = yearsof life saved due to the reduction in nonfatal MIs.

Assumptionsof themodel:First, the probability ofmorbidity or mortality for patients who had a non-fatal MI was the same for both the treatment andplacebo groups. Second, patients who survived 10years after a nonfatal MI had no incremental prob-ability of morbidity or mortality relative to patientswho did not have a nonfatal MI. Third, patients whohad no subsequent events after their nonfatal MIwere followed for 10 years, and patients who expe-rienced a nonfatal MI and subsequent events werefollowed for 5 years after the event. Finally, the in-cremental cost of pravastatin therapy was adjustedeach year to account for treatinent group deaths anddropouts, but the clinical benefit of pravastatin wasbased on the intent-to-treat, as opposed to available,study population each year.

Present-valueanalysis:A present-value analysiswas performed to adjust for the relative value of dol-lars spent now, compared with dollars to be spent inthe future. According to this analysis, a life-yearsaved early in the treatment would be valued morehighly than one saved later. Thus, given a societalperspective, calculations of future costs and years oflife saved incorporated a discount rate of 5%/year.

Sensitivityanalysis:To determine whether reason-able variations in the model assumptions would sig-nificantly alter the results, several sensitivity analy-ses were performed. First, 25% variations (both

412 THE AMERICAN JOURNAL OF CARDIOLOGY@ VOL 78

increased and decreased) in the hospitalization costof MI were introduced into the calculations. Second,the benefits of pravastatin therapy for 10 years wereestimated by projecting the observed clinical benefitsof pravastatin therapy (at 3 years) for 7 more years.The mean morbidity (nonfatal MI) and mean mor-tality (CAD death, non-CAD death, and fatal MI)rates observed in years 2 and 3 of the PLAC studieswere applied through years 4 through 10. The meth-ods used for estimating the 3-year hospitalizationand follow-up costs of MI and its subsequent eventswere carried out for an additional 7 years to obtain10-year cost estimates. Pravastatin therapy was ad-ministered to patients who remained alive each year,for a period of 10 years. Third, to account for thefact that some CAD patients might have a signifi-cantly more favorable long-term prognosis than oth-ers after an MI, the effect of a 2% post-MI annualcardiovascular disease mortality rate was tested onthe model. Finally, a confidence interval-derivedsensitivity analysis was conducted using the ob-served reduction in risk of nonfatal MI for men ad-ministered pravastatin in the PLAC studies. The per-cent change from the 3-year male nonfatal MIrelative risk reduction point estimate (60Yo) to theupper and lower bounds of the 95~o confidence in-terval ( 17%, 8690) was applied to the annual inci-dence of incremental male nonfatal MIs.

RESULTSCosts:Estimated average hospitalization and an-

nual incremental follow-up costs for patients withMI, angina, congestive heart failure, and stroke arelisted in Table II. Hospitalization costs (composedof general or special care units, ancillary services,surgical and diagnostic procedures, medical thera-pies, and physician fees) were computed using theclinical decision algorithm, and the annual follow-up costs were based on units and frequencies of ser-vice and drug therapies identified by the expertpanel.

The overall annual estimated acquisition costof once-daily 36.64 mg of pravastatin therapy(weighted average dose) was $1,054.85. Incremen-tal costs associated with therapy initiation and annualfollow-up were estimated at $298.50 (year 1 only)and $163, respectively. Incremental dispensing costsamounted to $17.80 annually.

Increaseinlifeexpectancy:Analysis of the averagediscounted gain in life expectancy (Table III) re-

AUGUST 15, 1996

TABLEV Sensitivity Analysis Results: Cost Per Life-Year Saved

Pravastatin 36.64 mg (weighted mean dose)

Ml Hospital Cost Decreased Ml Hospital Cost Increased Pravastatin forRisk Prafile by 25% by 25% 10 Years

Male, CAD + 1 risk factor $14,666 $10,664 $9,632Male, CAD + 2 risk factors $10,822 $7,915 $7,156Male, CAD + >3 risk factors .$8,200 $6,050 $5,473

See previous Tables for explanation of risk factors,Abbreviations as in Table 1,

Pravastatininsecondarypreventionofcoronaryartetydisease(male,coronsityadeiydisease+ 1 additionalriskfactor)

Perouteneoustransluminalcoronaryangioplasty

Hydrachlorothiazideinthetreatmentofsystemichypertension

CoronaVatierybypassgrafling

Coronaryarterydiseasescreening(exerciseelectrocardiogram)

Hearttransplantation

, , , , ( ,$0 $10 $20 $30 S40 $50 $80 $70 $80

Cost per Life-YearSaved (1995 Dollars,000s)J

FIGURE1. Estimates(adiustedto 1995 dollars)of othermedicalinterventionsforcardiovasculardise&e.

vealed that pravastatin therapy in secondary preven-tion of CAD increased life expectancy. These gainsranged from 0.11 years/patient. in men with CADplus 1 additional risk factor to 0.21 years/patient inmen with CAD plus =3 additional risk factors.

Costperlife-yearsavedanalysis:The cost per life-year saved estimates for pravastatin in the secondaryprevention of CAD decreased as the number of riskfactors increased, from $12,665/life-year saved formale CAD patients with 1 additional risk factor to$7,124/life-year saved for male CAD patients with=3 additional risk factors (Table IV).

Modelsensitivityanalysis:The cost-effectivenessratios varied by approximately 169%when the hos-pitalization costs of MI were increased or decreasedby 25%; and when the benefits of pravastatin therapywere projected for 7 additional years, the cost perlife-year saved decreased by approximately 24%,implying additional net benefits with continued useof pravastatin over longer periods of time (Table V).The cost-effectiveness ratio increased from $12,665to $20,101/life-year saved when the 3.4% post-MIannual cardiovascular disease mortality rate used inthe main analysis was replaced by a 2% annual rate.The confidence interval derived cost per life-yearsaved estimates ranged from $77,267 for male CADpatients with 1 additional risk factor to $3,068 formale CAD patients with =3 additional risk factors.

The results of this cost-effectivenessstudy showed that depending on the pa-tient’s risk profile, the estimated cost perlife-year saved with pravastatin in sec-ondary prevention of CAD varied from$7,124 to $12,665. The model for thisanalysis used the observed, in-hospitalmortality from the PLAC I and PLAC IIstudies, but owing to the short durationof the PLAC studies, the post-MI annualmortality rates were derived from theFramingham data and estimates from theliterature. It was assumed that additionalrisk factors would likely increase the an-nual mortality rate, hence the rationalefor 3 different risk profiles.

Although the mortality data from theFrarningham Study does not reflect cur-rent advances in reperfusion therapy, itshould be noted that a recent study found

that only 34% of patients with an MI received throm-bolytic therapy in the United States.15The net effectof reperfusion therapy on overall post-MI mortalityrates are, therefore, relatively small. The rates usedin this analysis are comparable to those in recentlypublished studies.lc-zlThese studies, which followedpatients from 2 to 15 years after MI, provided crudeannual mortality rates between 3.8Y0and 6.1% inpatients <70 years of age. Nonetheless, becausepost-MI annual mortality is an extremely sensitivevariable, a sensitivity analysis was performed usinga 2Y0post-MI annual cardiovascular disease mortal-ity rate, which is compatible with rates seen in pos-treperfusion cohorts. Considering the clinical bene-fits of pravastatin reported in the PLAC trials and thefavorable estimates of this cost-effectiveness analy-sis, the secondary prevention of CAD with pravas-tatin is justifiable from both clinical and economicperspectives.

Attempts were made in this analysis to maintaina conservative perspective and include the followingassumptions: ( 1) patients who experienced a non-fatal MI while receiving pravastatin for secondaryprevention did not benefit from further drug treat-ment; (2) patients who had a subsequent event afterMI accrued follow-up costs for only 5 years after theevent; and (3) the clinical benefit of pravastatin wasbased on the intent-to-ti.eat,as opposed to available,

CORONARY ARTERY DISEASE/COST-EFFECTIVENESS OF PRAVASTATIN 413

study population each year. Also noteworthy is thatthe cost-effectiveness ratios were computed usingthe average wholesale price for pravastatin (eventhough most health care organizations pay less thanwholesale) and the costs for stroke did not includeexpenditures ti’sing from possible nursing home ad-missions.

This analysis demonstrates that pravastatin iscost-effective compared with published estimates(adjusted to 1995 dollars) of other widely acceptedmedical interventions for cardiovascular disease(Figure 1) such as percutaneous translurninal coro-nary angioplasty,22hydrochlorothiazide for the treat-

ment of systemic hypertension,23coronary artery by-pass grafting, 22 CAD screening (exerciseelectrocardiogram for a 60-year-old man),24 andheart transplantation.22

1. Pitt B, Ellis SG, Mancirri J, Rosmarr HS, McGovern ME. Design arrd re-cruitment in the United States of a mrdticenter, qurmtitative, angiographic trialof pravastatin to limit atherosclerosis in the coronary arteries (PLAC I). Arn JCardiol 1993;72:31–35.2. Crouse JR, Byington RP, Bond MG, Espelarrd MA, Sprinke JW, McGovernM, FrrrbexgCD. Pravastatin, lipids and atherosclerosis in the carotid arteries:design features of a clinical trial with carotid atherosclerosis outcome (PLACff). Control ClirrTrials 1992;13:495-506.3. f%mingham Study, Section 35. Survival Following Initial CardiovascularEvents. U.S. Department of Heafth and Humarr Services; NIH Pub. No. 88-2969. 1988:37–204.4. Sigler LH. Long-term survival following myocardial infarction: report on255 patients living 10 years or longer after the first attack. ArrrJ Cardiol1962;9:547-557.5. Goldmrm L, Sia STB, Cook EF, Rutherford JD, Weinstein MC. Cost arrdeffectiveness of routine therapy with long-term beta-adrenergic arrtagonists afteracute myocardird infarction. N ErzglJMed 1988;319:152–157.6. National Center for Health Statistics. Vital Statistics of the U.S., 1987, Mor-tality, Part A. Section 6: Life Tables. Washington, DC: Government PrintingOffice, 1990;2:1–19.7. Redbook Update. Montvale, New Jersey: Medicrd Economics Data, Inc.;December 1995; vol. 14, no. 12.

8.1995 Physicians Fee and Coding Guide. Augusta, Georgia: Health Care Con-sultants of America, Inc.; 1995,9. Goldman L, Weinstein MC, Goldmarr PA, Williams LW. Cost effectivenessof HMG-COA reductase irrhibition for primary arrd secondary prevention ofcoronary heart disease. JAMA 1991;265:1145–1151.10. Wittels EH, Hay JW, Gotto AM. Medical costs of coronary artery diseasein the Urrited States. Am J Cardiol 1990;65:432–440.11. Pharmaceutical Benefita Under State Medical Assistance Programs. Reston,Virginia: National Pharmaceutical Council, Inc.; 1995,12. The DRG Handbook: Comparative Clinical arrdFinancial Wurdards. Cleve-land, Ohio: HCIA, Inc., arrdErnst& Young LLP; 1995.13. Elixfrauser A, Andrews RM, Fox S. ClirricaJClassifications for Health Pol-icy Research: Discharge Statistics by Principal Diagnosis arrdProcedure. Rock-ville, Marykmd: Agency for Health Care Policy arrd Research Pub. No. 93-0043; 1993.14. Massie BM, Sokolow M. Cardiovascular disease, In: Schroeder SA, TiemeyLM, McPhee SJ, Papadakis MA, Kmpp MA, eds. Current Medical Diagnosisarrd Treatment. Norwalk, Connecticut: Appleton arrdLarrge; 1992:257–356.15. Rordeau JL, Moye LA, Pfeffer MA, Arnold JM, Bernstein V, Cuddy TE,Dagenais GR, Geltman EM, Goldman S, Gordon D, Hamrn P, Klein M, LamasGA, Mc@rs J, McEwarr P, Menapace FJ, Parker JO, Sestier F, Sussex B,Braunwald E, A comparison of mrmagement patterns after acute myocardialinfarction in Carrada arrd the United States, N Engl J Med 1993;328:779-784.16. de Vreede JJ, Gorgels AP, Verstraaten GM, Vermeer F, Dassen WR, Wel-lens HJ. Did progrrosis after acute myocardiaJ infarction change during Orepast30 years? A meta-analysis. JAm Coil Cardiol 1991;18:698-706.17. Goldberg RJ, Gorak EJ, Yarzebski J, Hosmer DW, Dalen P, Gore JM, AlpertJS, Dalen JE. A communitywide perspective of sex differences and temporaltrends in the incidence arrd survival rates after acute myocardial infarction andout-of-hospital deaths caused by coronary heart disease. Circukztion1993;87:1947-1953.IS. Gottlieb S, Moss AJ, McDermott M, Eherly S. Comparison of post-hospitalstrrvival after acute myocardial infarction in women and men. Am J Cardiol1994;74:727–730.19. Merrilees MA, Scott PJ, Norris RM. Prognosis after myocardial infarction:results of 15 year follow-up, Br Med J 1984;288:356–359.20. Launbjerg J, Fnrergaard P, Madsen JK, Mortensen LS, Hansen JF. Ten-year mortality in patients with suspected myocardial infarction, JJr &fed J1994;308:1196–1199.21. Herfitz J, Karlson BW, Hjalmarson A. Ten-year mortality among patientswith suspected acute myocardird infarction in relation to early diagnosis. Car-diology 1994;84:114–120.22. Tengs TO, Adams ME, Pliskirr JS, Safrarr DG, Siegel JE, Weinstein MC,Graham JD. Five-hundred life-saving interventions and their cost-effectiveness.Jfisk Arial 1995;15:369-390.23. Edelson JT, Weinstein MC, Tosteson A, Williams L, Lee TH, Goldman L.Long-term cost-effectiveness of various initial monotherapies for mild to mod-erate hypertension. JAMA 1990;263:407-413.24. Sox HC, Littenherg B, Garber AM. The role of exercise testing in screeningfor coronary artery disease. Ann Intern Med 1989;110:456–469.

414 THE AMERICAN JOURNAL OF Cardiology@ VOL. 78 AUGUST 15, 1996