AJH 1998;11:618 – 629

REVIEW

Cost-Minimization and the Number Needed

to Treat in Uncomplicated Hypertension
Kevin A. Pearce, Curt D. Furberg, Bruce M. Psaty, and Julienne Kirk

Downloaded from https://academic.oup.com/ajh/article/11/5/618/177180 by guest on 10 October 2023

The goal of this study was to compare the direct death) ranged from $4730 to $346,236 among middle-
costs associated with the prescription of thiazide aged patients, and from $1595 to $116,754 in the
diuretics, b-receptor blockers (b-blockers), elderly; generic diuretic or b-blocker therapy was
angiotensin converting enzyme inhibitors (ACEI), more economical than treatment with an ACEI, a-
a-receptor blockers (a-blockers), and calcium blocker, or CCB. The associated 5-year NNT was 86
channel blockers (CCB) for the prevention of for middle-aged patients and 29 for elderly patients.
stroke, myocardial infarction (MI) and premature Diuretic therapy remained more cost-effective even
death in uncomplicated hypertension. under the unlikely assumption that the newer drugs
We performed a cost-minimization analysis based were 50% more effective than diuretics at preventing
on numbers-needed-to-treat (NNT) derived from the these major events. The costs associated with
metaanalysis of 15 major clinical trials of potassium supplementation did not eliminate the
hypertension treatment, and the average wholesale advantage of diuretics.
prices of both the most commonly prescribed and the Treatment costs to prevent major hypertensive
least expensive drugs in each class. The inclusion complications are much lower with diuretics and
criteria for clinical trials were that they be b-blockers than with ACEI, CCB, or a-blockers,
randomized, controlled trials of drug therapy of especially in middle-aged patients. Am J
uncomplicated mild-to-moderate hypertension with Hypertens 1998;11:618 – 629 © 1998 American
stroke, MI, or death as endpoints. The wholesale Journal of Hypertension, Ltd.
drug costs and the total direct outpatient treatment
costs to prevent a stroke, MI or death among middle- KEY WORDS: Cost analysis, hypertension treatment,
aged and elderly hypertensives were our outcome drug choice, diuretic, b-blocker, angiotensin
measures. The estimated wholesale drug acquisition converting enzyme inhibitor, a-blocker, calcium
cost to prevent one major event (MI or stroke or channel blocker.

T
reatment of uncomplicated hypertension those trials, angiotensin converting enzyme inhibitors
with diuretics and b-adrenergic blockers (b- (ACEI), calcium channel blockers (CCB), and long-
blockers) has been shown to reduce morbid- acting a-receptor blockers (a-blockers) have become
ity and mortality among middle-aged and available for the treatment of hypertension. These
older patients in randomized clinical trials.1–3 Since newer drugs are useful alternatives for selected pa-

Received January 28, 1997. Accepted October 20, 1997. Address correspondence and reprint requests to Kevin A. Pearce,
From the Departments of Family and Community Medicine (KAP, MD, MPH, Department of Family and Community Medicine, Bow-
JK), and Public Health Sciences (CDF), Bowman Gray School of Med- man Gray School of Medicine, Medical Center Blvd., Winston-
icine, Winston-Salem, North Carolina; and the Cardiovascular Health Salem, NC 27012.
and Research Unit, Departments of Medicine, Epidemiology and
Health Services (BMP), University of Washington, Seattle, Washington.

© 1998 by the American Journal of Hypertension, Ltd. 0895-7061/98/$19.00

Published by Elsevier Science, Inc. PII S0895-7061(97)00488-3
AJH–MAY 1998 –VOL. 11, NO. 5 COST-MINIMIZATION IN HYPERTENSION TREATMENT 619

tients who have a medical contraindication to diuret- Event rates for each study were calculated as the
ics and b-blockers, cannot tolerate them, have failed to number of events divided by the person-years of ob-
respond adequately, or have a comorbid condition for servation, and were based on intention-to-treat analy-
which the newer medications have proven efficacy. ses. The duration of observation was based on the
The popularity and prescription rates of these newer mean follow-up interval published for each clinical
agents have grown steadily over the past decade,4,5 trial. The risk difference (RD) was used to calculate the
but their effectiveness in patients with mild-to-mod- number-needed-to-treat (NNT) to prevent a major hy-
erate hypertension for the primary prevention of myo- pertensive complication.40 Because the average dura-
cardial infarction (MI), stroke, or premature death, tion of the clinical trials was 5 years, the NNT in
person-years was then converted to the number of
compared to older agents, remains uncertain.6 – 8 Their
patients treated for 5 years to prevent one event (5-
widespread use as initial monotherapy in uncompli-
year NNT).
cated hypertension is controversial and has spawned

Downloaded from https://academic.oup.com/ajh/article/11/5/618/177180 by guest on 10 October 2023

Direct medication costs were calculated based on
considerable debate.9 –15 It has been estimated that the 1996 average wholesale prices (AWP).41 The mean
medications account for 50% to 90% of the direct costs price for all listed generic formulations (when avail-
of hypertension treatment.16 –18 The choice of drug(s) able) of each drug was used. Brand-name prices were
has a major effect on direct treatment costs because used for drugs not available in generic form in the
many of the estimated 43 million Americans who have United States. The representative drugs for each class
hypertension must be treated to prevent major com- were chosen by two criteria: 1) the most-commonly
plications in relatively few. prescribed in the United States42, and 2) the least
The purpose of this study was to compare the cost- expensive, based on the 1996 AWP. All drug costs
effectiveness of first-line antihypertensive drug classes were based on the AWP per 100 doses, then expressed
for the prevention of stroke, MI, or premature death in as the dose-specific cost per patient for 5 years of
patients with uncomplicated mild-to-moderate hyper- treatment. This model required the assumption that
tension. the 1996 AWP for each drug would be stable over a
5-year period.
METHODS
Costs were calculated for each drug based on equi-
We performed a cost minimization analysis from the potent doses in terms of BP reduction [compared with
perspective of providers and payors interested in the 25 mg/day hydrochlorothiazide (HCTZ)] as deter-
direct costs required to prevent major hypertensive mined by a survey of five leading clinical hyperten-
complications. We use the term cost-effectiveness in sion researchers. The mean dose chosen for each drug
reference to the costs of drugs and other direct medical was rounded down to the nearest available dose. The
expenses required to prevent one MI, stroke, or death. final doses were: 50 mg/day atenolol, 80 mg propran-
Our drug cost estimates are based on national survey olol twice daily, 5 mg/day terazosin, 4 mg/day dox-
data, and we calculated our estimates of effectiveness azosin, 60 mg/day nifedipine GITS, 20 mg/day nisol-
directly from major clinical trials, thus avoiding com- dipine, and 10 mg/day enalapril. At the time of our
plex assumption-laden economic models. Separate survey, benazepril was the least-expensive ACEI, but
analyses were done for middle-aged and elderly pa- later in 1996 trandolapril became the lowest-priced
tients. Sensitivity analyses were used to address un- ACEI, with the same AWP across its recommended
certainties about the comparative efficacy among an- dose range of 1 to 4 mg/day. There were no significant
tihypertensive drug classes, equipotent drug doses, discrepancies between the final dosages used in this
and direct nondrug treatment costs, including those analysis and those recommended in two standard
associated with diuretic-induced hypokalemia. drug information publications.43,44
We combined data from all randomized clinical tri- The main outcome of interest was the cost to pre-
als published in English evaluating drug treatment vent a stroke, MI, or death based on the wholesale
(versus placebo or usual care) of uncomplicated mild- drug acquisition costs. The cost-effectiveness of each
to-moderate hypertension that included MI, stroke, or drug (CEdrug) was calculated as CEDrug 5 (AWP for 5
death as a priori endpoints in two metaanalyses: one years of treatment) 3 (5-year NNT). We assumed that
for middle-aged patients, and one for elderly patients. the effectiveness of the alternative drugs in reducing
These clinical trials were identified through previous the rate of MI, stroke, and death was equal to that
overviews,1–3 the MEDLINE electronic literature data- observed for diuretics and b-blockers in the major
base, and review of references in the papers found. clinical trials. We also assumed that a morbid event or
Recommended methods for study selection and data death averted due to treatment with any given class of
synthesis for metaanalyses were followed.19 –22 Ulti- antihypertensive drug resulted in direct and indirect
mately, 15 studies fitting our criteria were identified cost savings equal to those associated with any other
and combined.23–39 class of drug. Our primary analysis thus focused on
620 PEARCE ET AL AJH–MAY 1998 –VOL. 11, NO. 5

medication costs without modeling other potential di- lemia, and the costs to correct it, were also tested in
rect or indirect costs, ie, those related to potential this manner.
differences in quality of life. Equipotent Doses The doses of alternative drugs that
However, we did additional analyses that included were considered to be equipotent with 25 mg/day
the direct costs of routine outpatient physician visits HCTZ were varied from one-half to twice the original
and laboratory tests in order to derive more complete estimate.
estimates of cost-effectiveness. These other direct costs
for hypertension management in an established pa- RESULTS
tient were estimated (based on current fees at a large Effectiveness of Treatment and the NNT In the
family practice clinic) at $1340 per patient as follows: clinical trials meeting the selection criteria, diuretics
Three routine office visits per year @ $70 5 $1050 over and b-blockers were the main therapies and mean
5 years and two serum chemistry panels per year @ duration of follow-up was approximately 5 years.

Downloaded from https://academic.oup.com/ajh/article/11/5/618/177180 by guest on 10 October 2023

$29 5 $290 over 5 years, leading to a total of $1340 There were 30,268 middle-aged patients (57% men) in
over 5 years. The overall cost-effectiveness of outpa- seven trials, and 15,990 elderly patients (42% men) in
tient treatment to prevent an MI, stroke, or death was eight trials. (Details of individual study characteristics
then calculated as: CE 5 (AWP for 5 years of treat- are on file with the author and presented in brief in the
ment 1 $1340) 3 (5-year NNT). Appendix.) Table 1 shows the combined risk ratio and
The cost of diuretic therapy was calculated with and the 5-year NNT for each outcome of interest, stratified
without potassium chloride (KCl) supplementation. by age group. The 5-year NNT to prevent one major
The cost of KCl supplementation included the AWP of nonfatal event or death among middle-aged patients
40 mEq/day of generic KCl @ $45/year,41 plus two was 86; for elderly patients the 5-year NNT was 29.
additional serum potassium levels/year @ $20 each, The validity of the combined results was supported by
for a total additional cost of $425 per KCl-treated statistical evidence of homogeneity among the indi-
patient over 5 years. It was assumed that 25% of vidual study results. The only exception was stroke in
patients treated with HCTZ at 25 mg/day would ac- middle-aged patients, for which there was statistical
tually require KCl, making the cost of KCl supplemen- evidence of heterogeneity (P 5 .042). Risk ratios were
tation averaged across all patients $106.25 per patient similar in the two age strata, but the NNT for the
over 5 years. This exercise was repeated for the alter- elderly were one-third to one-fifth those for the mid-
native of adding spironolactone at a dose of 25 mg dle-aged patients, due to the higher absolute risk with-
twice daily (instead of KCl) to HCTZ, at the AWP of out treatment in the elderly.
$57/year.41 Assuming that 25% of patients would Cost of Antihypertensive Medications Table 2
need this, the additional cost of spironolactone and shows the representative drugs, their equipotent
potassium monitoring averaged across all patients doses, and their AWP for 5 years of treatment, based
would be $121.25 per patient over 5 years. on 1996 prices. For equipotent doses of the most com-
monly prescribed drugs, the AWP varied 73-fold; for
Sensitivity Analyses Sensitivity analyses were done the least expensive drugs it varied 32-fold. The AWP
to examine the effects of our assumptions about equal for doses that were one-half and twice the estimated
drug effectiveness, equal nondrug direct outpatient equipotent doses are also shown.
costs (including correction of hypokalemia), and equi-
potent drug doses on the comparative cost-effective- Drug Costs to Prevent One MI, Stroke, or Death
ness of the drugs in question. Drug-specific cost-effectiveness ratios, expressed as
the wholesale drug cost to prevent one nonfatal MI,
Treatment Effectiveness We examined the impact on nonfatal stroke, or death are compared in Table 3.
cost-effectiveness of assuming that ACEI, CCB, and These results reflect only drug costs and the observed
a-blockers were up to twice as effective (because of effectiveness of diuretics and b-blockers, assuming
better compliance or better pharmacologic effects) as that all of the drugs shown are equally effective at
diuretics or b-blockers. This decreased the estimated reducing morbidity and mortality. The estimated ex-
NNT for each of the newer drugs to as little as one-half cess wholesale drug cost associated with using an
of the NNT for diuretics and b-blockers. ACEI, CCB, or a-blocker instead of HCTZ to prevent
one major event (MI, stroke, or death) ranged from
Nondrug Treatment Costs Our assumptions of equal- $30,160 to $115,159 in the elderly, and from $89,440 to
ity were assessed via a threshold analysis of the $341,506 in middle-aged patients. Under the assump-
amount of nondrug costs that alternative drugs would tion that 25% of patients treated with diuretics require
have to save, compared with those associated with potassium replacement, these excess costs were only
HCTZ prescription, to meet the cost-effectiveness of modestly reduced ($27,879 to $112,078 for elderly and
HCTZ. Assumptions about the prevalence of hypoka- $80,302 to $332,368 for middle-aged patients).
AJH–MAY 1998 –VOL. 11, NO. 5 COST-MINIMIZATION IN HYPERTENSION TREATMENT 621

TABLE 1. NUMBERS-NEEDED-TO-TREAT (NNT) TO PREVENT CARDIOVASCULAR EVENTS OR DEATH IN

PATIENTS WITH MILD-TO-MODERATE HYPERTENSION

P for P for
Risk Ratio (95% CI) Heterogeneity† 5-Year NNT Heterogeneity†
Event [Treated/Control] of RR (95% CI) of NNT

Middle-aged patients (based on 145,595 person-years experience in seven major clinical trials)
Fatal or nonfatal CHD 0.91 (0.79–1.04) .626 390 (158–NA)†† .597
Fatal or nonfatal stroke 0.56 (0.30–0.71) .071 135 (81–424) .042
Death, any cause 0.87 (0.77–0.99) .253 271 (97–NA)†† .100
Nonfatal event* or death 0.82 (0.75–0.90) .581 86 (47–503) .146
Elderly patients (based on 73,523 person-years experience in eight major clinical trials)
Fatal or nonfatal CHD 0.82 (0.73–0.92) .841 70 (44–167) .885
Fatal or nonfatal stroke 0.65 (0.57–0.75) .848 45 (32–75) .292

Downloaded from https://academic.oup.com/ajh/article/11/5/618/177180 by guest on 10 October 2023

Death, any cause 0.89 (0.82–0.98) .458 72 (39–462) .385
Nonfatal event* or death 0.84 (0.78–0.90) .415 29 (19–61) .233
* Nonfatal MI or nonfatal stroke.
† P . .05 indicates homogenous risk ratios across combined studies.
†† The upper 95% confidence limit cannot be estimated for NNT because the confidence limits of the risk differences include zero.
CHD, coronary heart disease; CI, confidence interval; RR, risk ratio; MI, myocardial infarction; NA, not applicable.

The overall direct outpatient treatment costs to pre- Sensitivity Analyses

vent one event were estimated by adding $1340 to Drug Efficacy Superior drug efficacy would lower the
cover 5 years of physician and laboratory fees to the NNT, thereby improving cost-effectiveness through
price. The results are illustrated in Figure 1, which reductions in both drug and nondrug costs per event
plots the cost to prevent one major event against any averted. Figure 2 illustrates the impact on direct costs
nominal 5-year outpatient treatment cost per patient. to prevent an event of increasing the assumed efficacy
The positions on the plot of the most commonly- of ACEI, CCB, and a-blockers, while holding the effi-
prescribed drugs are shown. cacy of diuretics and b-blockers constant at the level

TABLE 2. AVERAGE 1996 WHOLESALE PRICES FOR SELECTED ANTIHYPERTENSIVE DRUGS: DRUG
COSTS PER PATIENT FOR 5 YEARS OF TREATMENT

Drug Class Most Commonly Prescribed Least Expensive

Diuretic HCTZ HCTZ

25 mg daily $55 25 mg daily $55
b-Blocker Atenolol Propranolol
25 mg daily $1097 40 mg twice daily $563
50 mg daily $1222 80 mg twice daily $637
100 mg daily $1745 160 mg twice daily $1274
ACE inhibitor Enalapril Trandolapril
5 mg daily $1734 1 mg daily $1095
10 mg daily $1820 2 mg daily $1095
20 mg daily $2590 4 mg daily $1095
a-Blocker Terazosin Doxazosin
2 mg daily $2260 2 mg daily $1675
5 mg daily $2260 4 mg daily $1758
10 mg daily $2260 8 mg daily $1847
Calcium blocker Nifedipine GITS Nisoldipine
30 mg daily $2327 10 mg daily $1495
60 mg daily $4026 20 mg daily $1495
90 mg daily $4645 40 mg daily $1495
HCTZ, hydrochlorothiazide.
Doses in boldface type represent equipotent doses.
622 PEARCE ET AL AJH–MAY 1998 –VOL. 11, NO. 5

TABLE 3. WHOLESALE DRUG ACQUISITION COSTS TO PREVENT ONE MI, STROKE OR DEATH
AMONG PATIENTS WITH UNCOMPLICATED MILD-TO-MODERATE HYPERTENSION

Most Common Treatment Least Expensive Treatment

Drug Class Treatment Middle-Aged Elderly Treatment Middle-Aged Elderly

Diuretic HCTZ $4730 $1595 HCTZ $4730 $1595

b-Blocker Atenolol $105,092 $35,438 Propranolol $54,782 $18,473
ACE inhibitor Enalapril $156,520 $52,780 Trandolapril $94,170 $31,755
a-Blocker Terazosin $194,360 $65,540 Doxazosin $151,188 $50,982
Calcium blocker Nifedipine GITS $346,236 $116,754 Nisoldipine $128,570 $43,355

Downloaded from https://academic.oup.com/ajh/article/11/5/618/177180 by guest on 10 October 2023

observed in clinical trials. Assuming a 50% increase in Nondrug Direct Treatment Costs Figure 3 shows the
efficacy above that observed for HCTZ and $1340 per results of our threshold analysis of the amount of
patient over 5 years in nondrug costs, the excess direct difference in nondrug costs between HCTZ and alter-
outpatient treatment cost associated with using a CCB, native drugs that would be required to meet the cost-
ACEI, or a-blocker instead of HCTZ ranged from effectiveness of HCTZ. This figure allows the reader to
$6,622 to $63,288 per major event averted in the el- vary the assumption of comparative drug efficacy.
derly, and from $19,637 to $187,681 in middle-aged Estimates of the cost to correct and monitor diuretic-
patients. Under the extremely unlikely assumption of induced hypokalemia can be included in the nondrug
doubled efficacy, only the least expensive ACEI and costs. For example, if one assumed that enalapril had
CCB met the cost-effectiveness of HCTZ. equal efficacy with HCTZ, enalapril use would have to

FIGURE 1. Total 5-year direct outpatient treatment costs to prevent one major event (nonfatal MI or nonfatal stroke or death by any
cause). Vertical bars show per-patient costs for commonly-prescribed drugs, assuming $1340 over 5 years in nondrug costs and
assuming stable drug prices. ‘‘HCTZ 1 KCL’’ bar includes 40 mEq/day KCl supplement plus two extra serum potassium levels/year.
AJH–MAY 1998 –VOL. 11, NO. 5 COST-MINIMIZATION IN HYPERTENSION TREATMENT 623

Downloaded from https://academic.oup.com/ajh/article/11/5/618/177180 by guest on 10 October 2023

FIGURE 2. Sensitivity of comparative cost-effectiveness to assumptions about drug efficacy (middle-aged hypertensives). Total 5-yr
outpatient costs to prevent a nonfatal CV event or death, based on AWP of drug plus $1340/pt nondrug costs. The HCTZ 1 KCl bar
assumes that 25% of diuretic-treated patients require KCl supplements and K1 monitoring. The HCTZ 1 SP bar assumes that 25%
of diuretic-treated patients require spironolactone and K1 monitoring. HCTZ, hydrochlorothiazide; HCTZ 1 KCl, hydrocholorothia-
zide 1 potassium chloride; HCTZ 1 SP, hydrocholorothiazide 1 spironolactone; PROP, propranolol; ATEN, atenolol; TRAN,
trandolapril; NISOL, nisoldipine; DOXA, doxazosin; ENAL, enalapril; TERA, terazosin; NIFED, nifedipine.

achieve savings of $353/patient/year in nondrug be treated to benefit a few, especially among middle-
costs (compared with HCTZ) to meet the economy of aged patients, and choosing more expensive drugs
HCTZ in terms of preventing morbid events. If enala- results in significantly higher direct treatment costs to
pril had 1.5 times the efficacy of HCTZ, this figure prevent one major hypertensive complication. Drug
would be $142/patient/year. selection in the United States can result in a more than
Equipotent Drug Doses The AWP for all nondiuretic 70-fold variation in AWP. An increase in drug charges
drugs for one-half and twice the doses used in this of just $100/year/middle-aged patient raises the cost
analysis are shown in Table 2. The only drug for which to prevent one major event by about $43,000; a cost
lowering the estimated effective dose had any appre- that must be borne by people who pay for health care.
ciable effect on the cost to prevent an event was ni- Based on previous metaanalyses and this cost minimi-
fedipine GITS, but its cost was still very high. Raising zation analysis, we conclude that diuretics and
the dose of atenolol, propranolol, enalapril, or nifedi- b-blockers should be the mainstay of therapy for un-
pine GITS would significantly reduce the cost-effec- complicated hypertension. They are economical and
tiveness of each. have proven efficacy. At their current prices, the alter-
native drugs substantially increase the cost to prevent
DISCUSSION cardiovascular events and death without clear-cut jus-
This cost minimization analysis shows that generic tification.
diuretics and b-blockers can be expected to prevent Economic analyses of hypertension treatment have
major cardiovascular events at a much lower cost than varied widely by the models chosen and their under-
ACEI, CCB, or a-blockers among patients with un- lying assumptions.18,45–50 Most have been done by
complicated hypertension. Diuretic therapy, with or combining data from one or two clinical trials with
without potassium management, retains a significant cardiovascular risk estimates derived from observa-
advantage, even if the other drugs are assumed to tional epidemiologic studies.47,48 A few have used risk
have 50% higher efficacy than was observed in major ratios from metaanalyses of multiple clinical trials in
clinical trials of diuretics and b-blockers. Many must middle-aged patients combined with epidemiologic
624 PEARCE ET AL AJH–MAY 1998 –VOL. 11, NO. 5

Downloaded from https://academic.oup.com/ajh/article/11/5/618/177180 by guest on 10 October 2023

FIGURE 3. Threshold sensitivity analysis: minimum drug-specific cost savings required to make alternative drugs as cost-effective
as HCTZ. Assumed Efficacy Ratio for CV Event Prevention (Ratio of Alternative Drug to HCTZ). Model includes fixed cost of
$268/patient/year for medical visits and serum chemistries.

data.16,17,49,50 These more comprehensive simulations of age to cost-effectiveness through separate analyses
have been informative, but they depend on complex for middle-aged and elderly patients.
models that require many assumptions. The effect of
age on cost-effectiveness has been addressed, but with Scope and Limitations Our conclusions are limited
conflicting results.49 to the treatment of mild-to-moderate hypertension
This cost minimization analysis represents a more and are primarily intended to help guide decisions
direct, if limited, approach that distinguishes it from involving first and second attempts at therapy in rel-
previous economic analyses on the same subject. It atively well patients. These results should not be ex-
rests on the NNT derived from major controlled clin- trapolated to patients with medical contraindications
ical trials representing over 219,000 patient-years of to diuretic and b-blocker therapy or a clear history of
observation combined in intention-to-treat analyses. It intolerance to these drugs. Likewise, they may not
thereby eliminates major assumptions about treatment apply to patients with congestive heart failure (CHF),
effectiveness. The definition of effectiveness is limited renal insufficiency, unstable angina, or history of an
to the prevention of major cardiovascular events or MI or stroke within the last 6 months. The results of
death within a 5-year treatment period. Projections of randomized controlled trials support the use of ACEI
benefit or measures of effectiveness beyond this re- in the presence of CHF,51 type I diabetes mellitus,52
stricted time frame are not made because data are and in post-MI patients with asymptomatic left ven-
lacking to quantify such assumptions. These features tricular dysfunction.53 The CCB amlodipine may have
form the basis for a simple model that emphasizes favorable effects as adjuvant therapy in a subset of
what is known about the quantitative effects of anti- patients with severe CHF.54 Because most of the clin-
hypertensive therapy on cardiovascular risk, rather ical trials in this analysis drew their participants from
than relying on educated guesses derived from com- primary care practices or community-based screen-
bining observational data with more restricted clinical ings, and excluded patients with major hypertensive
trials data. We have also emphasized the importance complications, we believe that these results are appli-
AJH–MAY 1998 –VOL. 11, NO. 5 COST-MINIMIZATION IN HYPERTENSION TREATMENT 625

cable to the majority of patients with mild-to-moder- Additional costs unique to HCTZ prescription would
ate hypertension treated in the primary care setting. have to be $200 to $790/patient/year to eliminate the
Baseline risk status must be taken into account when- advantage of HCTZ over the alternative drugs. In
ever generalizing absolute risk reduction, and cost- terms of tolerability, switching, and side effects (in-
effectiveness of treatment increases with risk. cluding hypokalemia), evidence to support this
We have addressed the possibility of superior toler- amount of excess cost is lacking.
ance and efficacy for ACEI, CCB, and a-blockers com- The costs of generic diuretics and b-blockers to pre-
pared with diuretics and b-blockers. However, our vent an event are probably slightly underestimated to
sensitivity analysis showed that under reasonable as- the degree that either was augmented with a second-
sumptions of increased effectiveness for the alterna- line drug in the clinical trials. That information was
tive drugs, HCTZ remained significantly more cost- not reported for most of the trials and we are not
effective. We did not entertain the possibility that any aware of any other randomized trial data involving

Downloaded from https://academic.oup.com/ajh/article/11/5/618/177180 by guest on 10 October 2023

of the alternative drugs might be less effective than uncomplicated hypertensives that can be used to com-
diuretics. pare rates of second-drug augmentation required
The differences in NNT between middle-aged and among the five drug classes considered here. Also, the
elderly patients may be confounded by diuretic dose drug prices used in our calculations are underesti-
because middle aged patients generally received high- mated by the amount of retail mark-up and dispens-
er-dose diuretic therapy than did elderly patients. ing fees, which vary widely by locality, pharmacy, and
Concerns about toxicity without added benefit associ- health plan. However, our calculated differences in
ated with high dose diuretics have been raised.55 Di- prices among the drugs are conservative, because the
uretic dose and age group were too highly correlated absolute price increase with any percentage mark-up
in these clinical trials to calculate the effect of dose on would increase with AWP.
efficacy and, therefore, NNT. Low-dose diuretic ther- Although this study provides important informa-
apy in middle-aged patients is probably more cost- tion as to the most economically efficient monothera-
effective than the main results of this analysis suggest. pies for uncomplicated hypertension, indirect and in-
Improved quality of life over that experienced with tangible costs and benefits of various drugs are not
diuretic or b-blocker therapy might justify higher addressed. Patient preference studies of these issues,
drug costs. We did not directly address quality of life paired with the results of ongoing randomized clinical
on treatment, but available evidence argues against its trials comparing the effectiveness of various antihy-
being an issue in our comparisons. The randomized, pertensive drugs,59,60 are needed to resolve these is-
double-blinded Treatment of Mild Hypertension sues. Once available, that information can be incorpo-
Study (TOMHS) found no significant differences in rated into a comprehensive evidence-based cost-effec-
the overall quality of life indices among the five drug tiveness analysis.
classes addressed herein, with one exception: the
ACKNOWLEDGMENTS
b-blocker (acebutolol) was superior to the a-blocker
(doxazosin) in that study. Also, the b-blocker and the We wish to thank Professor Michael F. Drummond, Center
diuretic were each associated with a better overall for Health Economics, University of York, England, and
Mary Ann Sevick, ScD, Bowman Gray School of Medicine,
quality of life index than placebo, whereas the overall Winston-Salem, North Carolina, for their comments and
quality of life on the other drugs did not differ from suggestions during the preparation of this manuscript.
placebo.56 Furthermore, a recent overview found no
significant differences in the effects on overall quality REFERENCES
of life among the five antihypertensive drug classes 1. Collins R, Peto R, MacMahon S, et al: Blood pressure,
compared in this analysis.57 stroke, and coronary heart disease. Lancet 1990;335:
To keep this analysis straightforward, we have 827– 838.
avoided modeling assumptions (and related costs) 2. Pearce KA, Furberg CD, Rushing J: Does antihyperten-
concerning drug switching, drug-specific side effects sive treatment of the elderly prevent cardiovascular
events or prolong life? Arch Fam Med 1995;4:943–950.
(except hypokalemeia), and changes in AWP over
time. Rates of discontinuation and medication switch- 3. Psaty BM, Smith NL, Siscovick DS, et al: Health out-
comes associated with antihypertensive therapies used
ing have been shown to be similar among diuretics,
as first-line agents. JAMA 1997;277:739 –745.
b-blockers, ACEI, and CCB.58 Our results are re-
stricted to data based on 5 years of treatment, and 4. Manolio TA, Cutler JA, Furberg CD, et al: Trends in
pharmacologic management of hypertension in the
reflect United States drug prices in effect in 1996. Our United States. Arch Intern Med 1995;155:829 – 837.
cost comparisons will require adjustment as data on
5. Psaty BM, Koepsell TD, Yanez ND, et al: Temporal
longer periods of antihypertensive treatment accumu- patterns of antihypertensive use among older adults,
late, and as more classes of drugs go off-patent; or if 1989 through 1992: an effect of major clinical trials on
the prices of patent-restricted drugs fall precipitously. practice? JAMA 1995;273:1436 –1438.
626 PEARCE ET AL AJH–MAY 1998 –VOL. 11, NO. 5

6. Joint National Committee on Detection, Evaluation, 24. Daugherty S, Berman R, Entwisle G, Haerer A: Cere-
and Treatment of High Blood pressure: The fifth report brovascular events in the Hypertension Detection and
of the Joint National Committee on Detection, Evalua- Follow-up Program. Prog Cardiovasc Dis 1986;29:63–
tion and Treatment of High Blood Pressure (JNC V). 72.
Arch Intern Med 1993;153:154 –183. 25. Stamler J: Risk factor modification trials: implications
7. Staessen JA, Fagard R, Thijs L, et al: Randomized dou- for the elderly. Eur Heart J 1988;9:9 –53.
ble-blind comparison of placebo and active treatment 26. The Management Committee: Treatment of mild hy-
for older patients with isolated systolic hypertension. pertension in the elderly. Med J Aust 1981;2:398 – 402.
Lancet 1997;350:757–764.
27. European Working Party on High Blood Pressure in
8. Borhani NO, Mercuri M, Borhani PA, et al: Final out- the Elderly: Mortality and morbidity trial. Lancet 1985;
come results of the Multicenter Isradipine Diuretic Ath- i:1349 –1354.
erosclerosis Study (MIDAS): a randomized controlled
trial. JAMA 1996;275:785–791. 28. Coope J, Warrender T: Randomised trial of treatment of
hypertension in elderly patients in primary care. Br

Downloaded from https://academic.oup.com/ajh/article/11/5/618/177180 by guest on 10 October 2023

9. Laragh JH: Government guidelines suppress valuable Med J 1986;293:1145–1151.
treatment options. Am J Hypertens 1994;7:857– 858.
29. Dahlof B, Lindhold L, Hansson L, et al: Morbidity and
10. Tobian L, Brunner HR, Cohn JN, et al: Modern strate- mortality in the Swedish Trial in Old Patients with
gies to prevent coronary sequelae and stroke in hyper- Hypertension (STOP-Hypertension). Lancet 1991;338:
tensive patients differ from the JNC V consensus guide- 1281–1285.
lines. Am J Hypertens 1994;7:859 – 872.
30. Perry H, Smith W, McDonald R, et al: Morbidity and
11. Swales JD: Guidelines for treating hypertension: im- mortality in the Systolic Hypertension in the Elderly
proved care or retarded progress? Am J Hypertens Program (SHEP) pilot study. Stroke 1989;20:4 –13.
1994;7:873– 876.
31. SHEP Cooperative Research Group: Prevention of
12. Menard J: Oil and water? Economic advantage and stroke by antihypertensive drug treatment in older per-
biomedical progress do not mix well in a government sons with isolated systolic hypertension. JAMA 1991;
guidelines committee. Am J Hypertens 1994;7:877– 885. 265:3255–3264.
13. Furberg CD, Berglund G, Manolio TA, Psaty BM: Over- 32. MRC Working Party: Medical Research Council trial of
treatment and undertreatment of hypertension. J Intern treatment of hypertension in older adults: principal
Med 1994;235:387–397. results. Br Med J 1992;304:405– 412.
14. Furberg CD, Psaty BM: Should dihydropyridines be 33. Veterans Administration/National Heart, Lung, and
used as first-line drugs in the treatment of hyperten- Blood Institute Study Group for Cooperative Studies
sion? Arch Intern Med 1995;155:2157–2161. on Antihypertensive Therapy: Mild hypertension:
15. Epstein M: Calcium antagonists should continue to be Treatment of mild hypertension: preliminary results of
used for first-line treatment of hypertension? Arch In- a two-year feasibility trial. Circ Res 1977;40(suppl 1):
tern Med 1995;155:2150 –2156. 180 –187.
16. Fletcher A: Economics of hypertension control: a state- 34. Hypertension Detection and Follow-up Program Coop-
ment by the World Hypertension League. J Hum Hy- erative Group: Five-year findings of the Hypertension
pertens 1994;8:789 –795. Detection and Follow-up Program. I. Reduction in mor-
tality in persons with high blood pressure, including
17. Kawachi I, Malcolm LA: The cost-effectiveness of treat- mild hypertension. JAMA 1979;242:2562–2571.
ing mild-to-moderate hypertension: a reappraisal.
J Hypertens 1991;9:199 –208. 35. Helgeland A: Treatment of mild hypertension: a five-
year controlled drug trial: The Oslo Study. Am J Med
18. Hilleman DE, Mohiuddin SM, Lucas BD Jr, et al: Cost- 1989;69:725–732.
minimization analysis of initial antihypertensive ther-
apy in patients with mild-to-moderate essential dia- 36. Australian National Blood Pressure Management Com-
stolic hypertension. Clin Ther 1994;16:88 –102. mittee: The Australian therapeutic trial in mild hyper-
tension. Lancet 1980;i:1261–1267.
19. Sacks HS, Berrier J, Reitman D, et al: Meta-analysis of
randomized controlled trials. N Engl J Med 1987;316: 37. Medical Research Council Working Party: MRC trial of
450 – 455. treatment of mild hypertension: principal results. Br
Med J 1985;291:97–104.
20. Rothman KJ: Modern Epidemology. Little, Brown, and
38. Veterans Administration Cooperative Study Group on
Co, Boston, 1986.
Antihypertensive Agents: Effects of treatment on mor-
21. Laird NM, Mosteller F: Some statistical methods for bidity in hypertension. II. Results in patients with dia-
combining experimental results. Int J Technol Assess- stolic blood pressure averaging 90 through 114 mm Hg.
ment Health Care 1990;6:5–30. JAMA 1970;213:1143–1152.
22. Berlin JA, Laird NM, Sacks HS, Chalmers TC: A com- 39. US Public Health Service Hospitals Cooperative Study
parison of statistical methods for combining event rates Group: Treatment of mild hypertension: results of a
from clinical trials. Stat Med 1989;8:141–151. ten-year intervention trial. Circ Res 1977;40(suppl 1):
23. Cressman M, Gifford R: Clinicians’ interpretation of the 98 –105.
results and implications of the Hypertension Detection 40. Laupacis A, Sackett DL, Roberts RS: An assessment of
and Follow-up Program. Prog Cardiovasc Dis 1986;29: clinically useful measures of the consequences of treat-
89 –97. ment. N Engl J Med 1988;381:1728 –1733.
AJH–MAY 1998 –VOL. 11, NO. 5 COST-MINIMIZATION IN HYPERTENSION TREATMENT 627

41. 1996 Drug Topics Red Book. Medical Economics Co., 52. Lewis EJ, Hunsicker LG, Bain RP, Rohde RD: The effect
Inc., Montvale, NJ, 1996. of angiotensin-converting enzyme inhibition on dia-
42. The top 200 drugs. Am Druggist 1997:30 –37. betic nephropathy. N Engl J Med 1993;329:1456 –1462.
43. Drugs for hypertension. Med Lett 1995;37:45–50. 53. Pfeffer MA, Braunwald E, Moye LA, et al: Effect of
captopril on mortality and morbidity in patients with
44. Drug Facts and Comparisons. Wolters-Kluwer Co, St. left ventricular dysfunction after myocardial infarction.
Louis, 1996. N Engl J Med 1992;327:669 – 677.
45. Drummond M, Coyle D: Assessing the economic value 54. Packer M, O’Connor CM, Ghali JK, et al: Effect of
of antihypertensive medicines. J Hum Hypertens 1992; amlodipine on morbidity and mortality in severe
6:495–501. chronic heart failure. N Engl J Med 1996;335:1107–1114.
46. Luce BR, Simpson K: Methods of cost-effectiveness 55. Siscovick DS, Raghunathan TE, Psaty BM, et al: Di-
analysis: areas of consensus and debate. Clin Ther 1995; uretic therapy for hypertension and the risk of primary
17:109 –125. cardiac arrest. N Engl J Med 1994;330:1852–1857.

Downloaded from https://academic.oup.com/ajh/article/11/5/618/177180 by guest on 10 October 2023

47. Johannesson M, Jonsson B: A review of cost-effective- 56. Treatment of Mild Hypertension Study Research
ness analyses of hypertension treatment. PharmacoEco- Group: Treatment of Mild Hypertension Study final
nomics 1992;I:250 –264. results. JAMA 1993;270:713–724.
48. Johannesson M, Dahlof B, Lindholm LH, et al: The 57. Beto JA, Bansal VK: Quality of life in treatment of
cost-effectiveness of treating hypertension in elderly hypertension: a metaanalysis of clinical trials. Am J
people—an analysis of the Swedish Trial in Old Pa- Hypertens 1992;5:125–133.
tients with Hypertension (STOP Hypertension). J In- 58. Jones JK, Gorkin L, Lion JF, et al: Discontinuation of
tern Med 1993;234:317–323. and changes in treatment after start of new courses of
49. Johannesson M: The impact of age on the cost-effective- antihypertenstive drugs: a study of a United Kingdom
ness of hypertension treatment: an analysis of random- population. Br Med J 1995;311:293–295.
ized drug trials. Med Decis Mak 1994;14:236 –244. 59. Davis BR, Cutler JA, Gordon DJ, et al: Rationale and
50. Littenberg B, Garber AM, Sox HC Jr: Screening for design for the Antihypertensive and Lipid Lowering
hypertension. Ann Intern Med 1990;112:192–202. Treatment to Prevent Heart Attack Trial (ALLHAT).
51. Gorg R, Yusuf S: Overview of randomized trials of Am J Hypertens 1996;9:409 – 411.
angiotensin-converting enzyme inhibitors on mortality 60. Hansson L: Major antihypertensive intervention trials
and morbidity in patients with heart failure. JAMA conducted outside the United States. Am J Hypertens
1995;273:1450 –1456. 1996;9(suppl):45S– 49S.
628 PEARCE ET AL AJH–MAY 1998 –VOL. 11, NO. 5

APPENDIX: CHARACTERISTICS OF STUDIES COMBINED

TABLE. CHARACTERISTICS OF THE CLINICAL TRIALS INCLUDED IN THIS
COST-EFFECTIVENESS ANALYSIS

Mean Blood
Pressure at Mean
Number of Age Range Baseline Percent Follow-Up
Study (Ref) Participants (Years) (mm Hg) Male (Years) Principal Drug(s) Secondary Drug(s)
1
VA-NHLBI 1012 21–50 ?/93 81 1.5 Chlorthalidone Reserpine
HDFP*2 7503 30–59 S-I 152/96*† 55† 5.0† Chlorthalidone Reserpine
S-II 159/101*† Methyldopa
Oslo3 785 40–49 156/97 100 5.5 Hydrochlorothiazide Propranolol
Methyldopa

Downloaded from https://academic.oup.com/ajh/article/11/5/618/177180 by guest on 10 October 2023

ANBPS*4 2845 30–59 157/100† 63† 4.0† Chlorothiazide Methyldopa
Propranolol
Pindolol
MRC5 17,354 35–64 161/98 52 5.0 Bendrofluazide or Methyldopa
propranolol
VA6 380 Mean 5 51 163/104 100 3.3 Hydrochlorothiazide Hydralazine
1 reserpine
USPHS7 389 21–55 148/99 80 7.0 Chlorothiazide
1 rauwolfia
HDFP8 2374 60–69 171/101 53 5.0 Chlorthalidone Reserpine
ANBPS9 582 60–69 165/101 55 3.8 Chlorothiazide Methyldopa
Propranolol
Pindolol
EWPHE10 840 60–97 182/101 30 4.7 Hydrochlorothiazide Methyldopa
1 triamterene
Coope11 884 60–79 197/100 21 4.4 Atenolol Bendroflumethiazide
Methyldopa
STOP12 1627 70–84 195/102 37 2.0 b-blockers or b-blocker 1 diuretic
hydrochlorothiazide
1 amiloride
SHEP-PS13 551 60–.80 172/75 37 2.8 Chlorthalidone Metoprolol
Reserpine
SHEP14 4736 60–.80 170/77 43 4.5 Chlorthalidone Atenolol
Reserpine
MRC15 4396 65–74 185/91 42 5.8 Hydrochlorothiazide Nifedipine
1 amiloride or
atenolol
S-I, BP stratum I; S-II, BP stratum II.
* The HDFP and ANBPS study results are stratified by age. Mean baseline BP levels among middle-aged participants in HDFP are available only for
BP strata.
† These figures apply to the entire cohorts in the HDFP and ANBPS studies.
VA-NHLBI, Veterans Administration/National Heart, Lung, and Blood Institute Cooperative Study on Antihypertensive Therapy; HDFP, Hypertension
Detection and Follow-up Program; Oslo, The Oslo Study; ANBPS, Australian National Blood Pressure Study; MRC, Medical Research Council Study;
VA, Veterans Administration Cooperative Study Group on Antihypertensive Agents; USPHS, US Public Health Service Hospitals Cooperative Study
Group; EWPHE, European Working Party on High Blood Pressure in the Elderly; STOP, Swedish Trial in Old Patients with Hypertension; SHEP-PS,
Systolic Hypertension in the Elderly Program pilot study; SHEP, Systolic Hypertension in the Elderly Program.

APPENDIX REFERENCES erative Group: Five-year findings of the Hypertension

Detection and Follow-up Program. I. Reduction in mor-
1. Veterans Administration/National Heart, Lung, and tality in persons with high blood pressure, including
Blood Institute Study Group for Cooperative Studies mild hypertension. JAMA 1979;242:2562–2571.
on Antihypertensive Therapy: Mild Hypertension:
Treatment of mild hypertension: preliminary results of 3. Helgeland A: Treatment of mild hypertension: a five-
a two-year feasibility trial. Circ Res 1977;40(suppl 1): year controlled drug trial: The Oslo Study. Am J Med
180 –187. 1989;69:725–732.
2. Hypertension Detection and Follow-up Program Coop- 4. Australian National Blood Pressure Management Com-
AJH–MAY 1998 –VOL. 11, NO. 5 COST-MINIMIZATION IN HYPERTENSION TREATMENT 629

mittee: The Australian therapeutic trial in mild hyper- 10. European Working Party on High Blood Pressure in
tension. Lancet 1980;i:1261–1267. the Elderly: Mortality and morbidity trial. Lancet 1985;
5. Medical Research Council Working Party: MRC trial of i:1349 –1354.
treatment of mild hypertension: principal results. Br 11. Coope J, Warrender T: Randomised trial of treatment of
Med J 1985;291:97–104. hypertension in elderly patients in primary care. Br
6. Veterans Administration Cooperative Study Group on Med J 1986;293:1145–1151.
Antihypertensive Agents: Effects of treatment on mor- 12. Dahlof B, Lindhold L, Hansson L, et al: Morbidity and
bidity in hypertension. II. Results in patients with dia- mortality in the Swedish Trial in Old Patients with
stolic blood pressure averaging 90 through 114 mm Hg. Hypertension (STOP-Hypertension). Lancet 1991;338:
JAMA 1970;213:1143–1152. 1281–1285.
7. US Public Health Service Hospitals Cooperative Study 13. Perry H, Smith W, McDonald R, et al: Morbidity
Group: Treatment of mild hypertension: results of a and mortality in the Systolic Hypertension in the
ten-year intervention trial. Circ Res 1977;40(suppl 1): Elderly Program (SHEP) pilot study. Stroke 1989;20:
98 –105. 4 –13.

Downloaded from https://academic.oup.com/ajh/article/11/5/618/177180 by guest on 10 October 2023

8. Cressman M, Gifford R: Clinicians’ interpretation of 14. SHEP Cooperative Research Group: Prevention of
the results and implications of the Hypertension De- stroke by antihypertensive drug treatment in older per-
tection and Follow-up Program. Prog Cardiovasc Dis sons with isolated systolic hypertension. JAMA 1991;
1986;29:89 –97. 265:3255–3264.
9. The Management Committee: Treatment of mild 15. MRC Working Party: Medical Research Council trial of
hypertension in the elderly. Med J Aust 1981;2:398 – treatment of hypertension in older adults: principal
402. results. Br Med J 1992;304:405– 412.