SOUNDING BOARD

SHOULD MILD HYPERTENSION BE
        TREATED?

'I'HERE is a growing body ofopinion that all patients
with hypcrtcnsion - no mattrr how mild or uncom-
plicated -should be trcatcd. In a rcccnt report to
Congrrss, the Assistant Secretary for Health, Edward
N. Brandt, Jr., stated that the Hypcrtcnsion Dctcction
and Follow-up Program had dcmonstratcd uncquivo-
tally that cffcctivc trcatmrnt could prolong lift in both
mild and bordcrlinc hypcrtcnsion. Moser has also
claimed that sufficient data have been accumulatrd to
justify reduction of blood pressure in all patients with
diastolic prcssurrs above 89 mm Hg.'
If put into practice this recommendation will have
rather awrsomc implications. The most rrccnt csti-
maw of` thr prevalcncc of hypcrtrnsion in the United
States" is 60 million persons, at least 40 million of
whom have a diastolic blood prcssurc of 90 to 99 mm
Hg (mcasurcd on one visit). Patients with such mild
and bordcrlinc hypcrtcnsion. who ha\,r ,gcncrallv not
been trratcd in the past, constitutr apprositnatcli one
fifth of the gcnrral adult population.
Largrly hccausc of poor compliance, the long-term
cfl'cctivcncss oflow-sodium or low-caloric dirts has not
been drmonstratcd in a gcnrral population, and prob
ably the ,grrat majority of patirnts will nrcd drugs to
control their hypcrtcnsion. In considering drug trcat-
mcnt in such a large segment of the population, the
disadvantages must bc wcighrd against the possihlr
advantages. Drug trratmcnt may have toxic rB`ccts,
cspccially in paticnrs who do not bccomc normotcnsi\c
with a simple drug rcgimrn but rcquirc a combination
ofdrugs. In addition to overt toxicity. most drugs have
subjective c`ffccts that, thouah not lift%-thrcatcning. arc
disiurbing to the person's quality of lift. hlorco\,cr,
thcrc arc patients, particularly among thr rldcrly. who
do not feel normal or f'unction normally when their
blood prcssurcs arr rcduccd.

The problem would bc ,greatly magnified if 40 mil-
lion or more essentially asvmptomatic persons wcrc
cxposcd to drugs. AMany patients dislike taking pills or
forget to take them. Disturbing also is the financial
rxpcnsc that would be involved in adding 40 million
patients to lifelong drug-treatment programs: not only
dru<g costs but also fees for professional scrviccs and
laboratory tests would be included. With 40 million
patients, even a conservative estimate of $500 per pa-
tient per year would yield a total cost of $20 billion
a scar.

If it has been demonstrated, however, beyond rca-
sonablc doubt, that dru,q treatment is highly effective
in preventing cardiovascular complications in cvcn the
mildest forms ofhypcrtcnsion, these negative consider-
ations arc outwcighcd and universal treatment is indi-
catcd. If the evidcncc is not so conclusive, the possible
advantages of treatment must bc weighed against the
disadvantages, bcarimg in mind that with so many
millions of patients subjected to drugs, the possibilit)l

ofdoing harm is greatly magnified. It is crucial, thcrr-
fort, to cxaminr the cvidcncc on which the claims of'
thcrapcutic bcncfit arc based.

THE HYPERTENSION DETECTION AND FOLLOW-UP
            PROGRAM

Thr chicfpillar supporting the aggrcssivc trcatmcnt
of mild hypcrtrnsion is the Hypcrtcnsion Detection
and Follow-up Pro,y-am. ",' This large multicrntcr trial
involved approximately 1 1,000 patients. of kvhom
about 8000 had an initial diastolic blood prvssurr in
the range of90 to 104 mm Hg. `I'hc design of the trial
was unorthodox. It was not dcsignrd to test thr cni-cts
of drug trcatmrnt itself on mortality. Instead. it \vas
supposed to drtcrminc whcthcr thr availability ofcom-
plctc, intrnsivc. and fiw mc.dical (`arc in special clinics
would br associated with a lower cardiovascular mor-
tality than that associated with the health-carr scrviccs
usually provided in the community. `I'hc, so-callrd con-
trol patients (the "rcfcrrcd-cat-r" group) w-crc rcfcrrrd
to whatcvcr medical cart thry could find or affi)rd in
the community. `I'hus, gcnrral medical cart and spccifl
ic antihvpc.rt~nsivr trcatmcnt in thcsc "control" pa-
tirnts \.aricd markedly. Somr wrrr trcatcd bvith drugs
and othrrs ~crc not. Some had no mrdical supcr\ision
of any kind whrrras others did. hlan) rcfcrrcd-car-c
patients had to pay for their medical cart, Mhcrcas
nonr of thr rxpcrimcntal (step-cat-c) group did. Alf the
patients in the step-cart group wcrc lblloucd closrl~~ in
well-stafrcd hypcrtcnsion clinics hackrd up by large
teaching hospitals. The availability, quality. and cost
of medical cart, thcrcforc, wrrc \cry diKcrcnt fi)r strp-
cat-r and rcfcrrcd-care patients and could have a~-
counted for much of thr diKcrcncc in the mortalit)
rates ol-)scr\,rd.

A major finding of the study was that in bordcrlinc
and miid hypcrtcwsion (diastolic prcssurc, 90 to 104
mm H,g). mortality from cardiovascular causrs lvas
26 per cent lowrr in the step-cart patients than in
the rcfcrrcd-care group. Howcvcr, noncardio\.~~scular
mortality, includin,? cancer and accidents. bras also
rcduccd (by 14 per cent) in the step-cart patients. I`his
rcsuft again calls into question the validity of the so-
callrd control group. Mrdical prohlrms, including car-
diovascular complications, w-ould have bern more
quickly rccognizrd and mow promptly and effcctivcl)
treated under the suprrior follow-up conditions avail-
able to thr step-cart group. It is not possible. thcrcfijrc,
to dctcrminc how much of the impro\,cd cardiovascu-
lar mortality was due to more rKcctivc antihyprrtcn-
sivc-drug trratment and how much to bcttcr gcnrral
medical cart.

Onr of the most striking results of thr study was the
45 per cent reduction in fatal myocardial infarction
found in the step-cart patients with mild hyprrtcnsion.
Unfortunately. thr causes ofdrath wcrc dctcrmincd by
death crrtificatcs, which arc notoriously unrcliablc.
An additional large numbrr of cardiovascular deaths
wcrc rcportcd under the classification of "other ischr-
mic heart disease." which would still rrprcscnt deaths


thought to be associated with coronary heart disease.
In this category there were 10 per cent more deaths in
the step-care than in the referred-care group. When
the two diagnostic categories arc combined they indi-
cate a 20 per cent, rather than a 45 per cent, reduction
for all deaths rclatcd to coronary heart discasc. includ-
in<g those labeled as myocardial infarction.

A further difficulty lies in the handling of refcrred-
cart patients whose hypertension progressed to a more
severe stage. In other controlled trials approximately 2
to 3 per cent of untreated patients per year had pro-
grcssion from mild hypertension to a more severe
stage. ",' Since in these other trials the control patirnts
wcrc followed as closely as the treated patients, the
incrcascd severity of the hypertension was promptly
recognized, and the patient was removed from the trial
to be treated openly. Such a procedure tends to cause
an underrstimatc of the effectiveness of trratmcnt,
since the patients at high risk because of incrcascd
blood prcssurc arc sclcctivcly rcmovcd from the con-
trol group before a morbid cvrnt occurs. In the Hypcr-
tension Detection and Follow-up Program, however,
the patients in the referred-care group who may have
rcccivcd inadcquatc medical care or no care and who
progressed to more scvrrc hypcrtcnsion may not have
been seen for a period ofmonths to more than a year. If
so, the number of patients who died in the control
group with mild hyprrtension was inflated by patients
whose hypertension had in reality progressrd from a
mild sta,gc to a more sevcrc stag?. In contrast to man-
agcmcnt in conventionally dcsigncd trials, this typr of
managcmcnt would result in overstating the cffcctivr-
ness of trcatmcnt, since thcsc control patients would
still bc counted as mild hypertcnsivcs.

Diastolic blood pressure was rcduccd from an aver-
age of 96.4 to 87.8 mm Hg in the rcfcrrcd-care patients,
many ofwhom receivrd some trcatmcnt in the commu-
nity.' Diastolic pressure was rcduccd from an avcragc
of 96.3 to 83.4 mm Hg in the step-care patients. Br-
cause of this difference it has been suggcstcd that a
reduction in diastolic pressure to below 90 mm Hg is
not enough, and that it should bc reduced to below 85
mm Hg. It is interesting to speculate on the frcqucncy
and scvcrity of thr side effects that would rrsult if this
advice wrre implemented. The 4.4-mm Hg difrerrncc
in diastolic pressure after treatment was only one of
many din`crences between thr step-care and referred-
care groups. Hence, it is not justifiable to ascrihr the
diffcrcncc in mortality to this one factor.

If it is true that the greater the reduction of blood
pressure, the fewer the complications, there should hc
a correlation betwcrn the degree of blood-pressure
lowering and the reduction of morbid events in simi-
larly trcatcd patients. Unfortunately, a correlation be-
tween blood-prcssurc reduction and death rates was
not reported in the step-cart group. Such an analysis
would have had the advantage of minimizing extranr-
ous therapeutic influences, since all the step-care pa-
tients were treated more or less similarly. This type of
analysis was carried out in the Veterans Administra-

tion study, and it failed to show an influcncc of diff`cr-
cnt dcgrccs ofblood-pressure reduction on morbidity."
Moreover, in the Australian trial, as pointed out by
Kaplan" and others,"`,' ' trcatcd patients whose dia-
stolic prcssurc was lowered had more trial end points
than control patients at the same lcvrl of blood pros-
sure - that is. lowering thr blood pressure with drug
trratmcnt did not confer the same dcgrcc of protection
against complications that occurred in untreated pa-
tients at similar levels. Unlike the Hypcrtcnsion Dc-
tcction and Follow-up Program, the Australian study
also found that in patients with diastolic prcssurcs
averaging <95 mm Hg during the trial, thcrc was no
relation brtwecn the lcvcl of diastolic prcssurc and the
incidcncc of cardiovascular complications' ' - that is.
a reduction to 80 mm Hg was no more c.Kcctivc than a
reduction to 90 mm Hg.

THE AUSTRALIAN NATIONALBLOOD PRESSURE
            STUDY
The Australian trial is generally rcgardcd as con-

firming the Hypertension Dctcction and Follow-up
Program with rcspcct to the treatment ofpaticnts with
mild hypcrtcnsion. The design ofthc Australian trial is
acccptablc in that, except for antihypcrtcnsivc trcat-
mrnt, the conditions of follow-up care wcrc similar in
the control and treatment groups. Furthcrmorc, no
one in the control group rcccivcd antihyprrt~tlsi~,~-
drug treatmrnt. Approximatrly 3500 patients wcr(
randomized into the trial.

The Australian trial, howcvcr, provided no infi)rma-
tion on the important group ofpaticnts with bordcrlinc
hypertension - diastolic blood pressures of 90 to 94
mm Hg. Diastolic pressure on entry ranged bctwrcn
95 and 109 mm Hg only. It is more or less gcncrally
agreed that treatment is effcctivc in patients with dia-
stolic prcssurcs averaging 100 mm Hg or above. Th(
controversy is with respect to treating patients with
diastolic levels below 100 mm Hg. Thcrcforc, the most
important aspect of the Australian trial is thr sub-
group with prcssurcs of 95 to 99 mm Hg on entry.

The Australian study led to two reports - one on
the results at three years and eight months offollow-
UP, " and another on the findings at four years.' In thr
first report trratment was cffcctive only in the patients
with initial diastolic prcssurcs of 100 mm Hg or higher.
The difference in trial end points between control and
trcatcd patients in the group entering with diastolic
prcssurcs of95 to 99 mm Hg was not significant. How-
cvcr. after four more months of follow-up the diff`crcncc
bccamc significant. Would the result have rcvcrscd
again if the study had been continued for another four
months? In this regard it is noteworthy that bctwccn
3% and four years the incidence of trial end points in
the trcatcd group decreased from 16.5 to 15.6 per thou-
sand person-years of risk. It would seem important to
know why patients with trial end points were rcmovcd
from the treated group bctwccn the first and the scc-
ond publication, since this decision could bc crucial in
determining whether the effectiveness of trcatmcnt


achicvcd significance in the group with prcssurcs of'95
to 99 mm Hg. Thus, for patients with diastolic prcs-
surcs below 100 mm Hg the confirmatory evidcncc
supplied by the Australian trial is oprn to question,
hccausc the important group with bordcrlinc hypcr-
tension (diastolic prcssurcs of90 to 93 mm Hg) were
not included. and bccausc in the 95-to-99 group the
significance ofthc result does not stand up convincing-
ly vvhcn subjcctcd to close analysis.

OTHER TRIALS

`I'hr most recent trial is the Oslo study.`" `l'his invcs-
tigation included 785 men with systolic pressures bc-
tvvccn IS0 and 179 mm Hg and diastolic pressures
below 1 10 mm Hg. `I'hc design of thr trial was ortho-
dox in that patients were randomly assigned tither to
activ-e drugs or to a control group and both groups
vccrr fiillowcd for five years. In contrast to the Hy-
pcrtcnsion I)c.tcction and Follow-up Program, the
Oslo study found that treatment had no effect on car-
tlio\zascular morbidity or mortality, although blood
prcssurc was rcduccd by an avcragc of 17/10 mm
Hg in the treated, as compared with the control?
group - considerably more than the diffcrcnce
obscr1,c.d in the Hypertension Dctcction and Follow-
up Pro;ram.

`I`hc cvidcncc supporting the v~aluc of treating bor-
dcrlinc and mild hypcrtcnsion with antihypertcnsivc
dru,gs, thcrcforc, is not as clearly cstahlishcd as many
bc1icv.c. `Ihe most favorable results come from thr
study ofmost qucstionablc design, in which interprcta-
tion 1s diilicult because the trial was not planned to test
drug trratmrnt but rather global medical cat-r. `I`hc
bcttrr-controlled trials aimed specifically at drug
trcatmcttt in mild hypcrtcnsion either obtained results
that fluctuated from insignificant to significant or
li)und no indication of hcncfit at any time. In this
connrction two earlier and smaller controlled trials -
thr V~trrans Administration Study';and the U.S. Pub-
lic Health Service Hospitals trial -also found no
signilicant diffcrcncc in cardiovascular morbidity or
mortality bctwccn actively trcatcd patients and pa-
tirnts rccciving placebo for mild or bordcrlinc hyper-
tension.

EFFECT OF TREATMENT ON CORONARY HEART
    DISEASE IN MILD HYPERTENSION

`l'hcrc is a tcndoncy to assume that because the risk
of cardiovascular complications is rclatcd to the height
of thr blood pressure in untreated patients,`" the tn-
crcascd risk in mild hypcrtcnsion, which is due pri-
marily to coronary heart disease, can be reversed to
normal by lowering the blood pressure. For example,
according to this assumption the cardiovascular risk of
a patient with a diastolic pressure of99 mm Hg will be
rcduccd to that of a normal person if the diastolic
prcssurc is lowcrcd to 85 mm Hg or less. However, as
noted above, a correlation betwren blood-pressurr rc-
duction and morbid rvrnts has not been drmonstratcd

in similarly treated patients," and reduction to a given
lcvcl with treatment does not confer the same protcc-
tion observed in untreated patients with the same dia-
stolic pressures. ' '

What can the physician hope to achieve by prescrib-
ing drugs for any and all patients with diastolic blood
pressures above 89 mm Hg? Will the benefits, ifany, in
mild hypertension outweigh the disadvantages of
medical treatment? The Veterans Administration con-
trolled trial indicated that the benefits ofantihyperten-
sivc drugs are much reduced in patients with mild
hypertension, as compared with those with higher
blood pressures." Antihypertensive-drug treatment is
most effective in prevcntmg complications ofhypertcn-
sion such as hemorrhagic stroke, renal failure, conges-
tive heart failure, and aortic dissection, and it is least
cffcctivc in prcvcnting athcrosclcrotic complications,
including coronary heart disease," the major complica-
tion in mild hypertension.

The Hypertension Detection and Follow-up Pro-
gram found the opposite result - that treatment was
more effective in mild hypertension than in moderate
or severe hypertension. This finding was attributed to
the more aggressive treatmrnt by outside physicians
of rcfcrrcd-care patients with moderatc and severe
hypertension. Although this may be true, it demon-
strates again that the results of this study cannot be
acccptcd at face value but must be interpreted in the
light of the unusual design of the trial.

The most controversial question about antihyper-
tcnsive-drug treatment is whether it significantly re-
duces the incidence of coronary-artery disease. Of the
various trials, only the Hypertension Detection and
Follow-up Program, as noted above. showed a signifi-
cant reduction of fatal coronary-artery disease with
treatment. The investigators did not report thrir ex-
perience with nonfatal myocardial infarction. In the
Oslo trial more myocardial infarction occurred in the
trcatcd patients.`" Some of the other trials, such as the
Veterans Administration study" and the Australian
trial,' showed a reduction in fatal coronary-artery
events with treatment, but the number of events was
small and did not reach the level of significance. On
the other hand, nonfatal myocardial infarction oc-
curred more frequently in trcatcd than in control pa-
tients in both the Veterans Administration and the
Australian trials, so that combined morbidity plus
mortality due to coronary-artery disease was about the
same in the treated and control patients. It is possible
that the more favorable effects of treatment on fatal
myocardial infarction were due to drug-induced hemo-
dynamic changes, such as a reduction in myocardial
oxygen demand as a result of a lowered afterload, the
effects of diuretics in preventing congestive heart fail-
ure, and the influence of beta-adrenergic blocking
drugs in reducing serious cardiac arrhythmias.

RECOMMENDATIONS

Epidemiologic investigations, such as the Fra-
mingham Study,ls ' indicate that the risk of myocardial


infarction, the most frcqucnt complication in mild
hypertension, varies markedly, dcpcnding on the num-
brr of other risk factors present. For example. the risk
that a myocardial infarction will occur over a six-year
period in a il5-year-old man with a systolic prcssurc of
165 mm Hg (equivalent in risk to a diastolic prcssurr of
95 mm Hg) is 3.1 per cent if no other risk factors arc
prcscnt. With a systolic pressure of 135 mm Hg the risk
is 2.1 per cent. The most that one could rxpcct horn
antihypcrtcnsive-drug trcatmcnt in such a patient
would bc to reduce the risk by 1 per cent. On the other
hand, with multiplr risk factors prcscnt [such as ciga-
rette smoking. electrocardiographic clidcncc of Icft
ventricular hypcrtrophy, hypcrcholcstrrolcmia. and
glucose intolerance) the risk in men incrcascs to 26.6
per cent with a systolic pressure of 165 mm Hg and to
20.0 per cent with a systolic prcssurc of 135 mm Hg.
The risk is now considcrablc, and the incrcmrnt in risk
due to hypertension alone is 6.6 per cent. Thcrrforc, if
antihypcrtrnsivc trcatmcnt is bcnrficial in reducing
the risk ofmyocardial infarction, it is most bcncficial in
patients with multiple risk factors. Parcnthctically, it is
worth noting that stopping cigarcttc smoking will have
nearly as great an e8`cct on reducing thr risk as lifelong
drug treatment. The National Health Intcrvicw Sur-
try showed that only one in three hypcrtcnsi\,c smok-
crs had bcrn advisrd by a doctor to stop smokin,y. '

In view of the unccrtaintirs, WC may bc doing more
harm than #ood by giving lifelong dru%g trcatmcnt to
patients with borderline or mild hyprrtcnsion. How-
ever, because of the possibility of bcncfit, cvcn though
it is unprovrd. a compromisr position, as suggcstcd by
others. may bc most appropriate."`," Patirnts with
diastolic pressures of 90 to 99 mm Hg (avcragc of at
least three visits) arc trcatcd or not. according to the
numbrr of risk factors prcscnt. Patients with few- other
risk factors arc given rcducin%g or low-sodium diets but
not drugs. Thr Australian study found a gradual fall in
blood pressure in many of thrir placebo control pa-
ticnts. I1 By the third year of follow-up, 18 prr cent of
the patients who began with diastolic prrssurcs 295
mm Hg had pressures below this lcvcl, 12 prr cent had
progressed to a more severe stage, and only 32 per cent
remained in their initial range of 95 to 109 mm Hg.
This cxpericnce demonstrates the wisdom of waitin,g
for an extcndcd period bcforc initiating antihy-prrtcn-
sivc-drug trratmcnt in mild hypertension. Patients

with many risk Factors may havr their blood prcssurc
rrduccd with dru%gs ifnecrssary. Ifdrugs arc used the),
should br given by the strp-care method, brginnirlg
with a diuretic alone and a\Gding complicated multi-
plr-drug rcgimrns. LAll patients with rlrvatd blood
prrssurcs should hc followed periodically to dctcct an)
cvidcncrs of progrrssion to a morr scvrrc staqr of`
hypcrtcnsion. By such a discriminativcl approach.
many millions of people could br spared ncrdlrss Iif&
long cxposurr to drugs.

Vrtrrans .Admint\tration

Medical Center
Wn$hmgton. DC 20422

b\V,2Rl) 1). I'RKIS, x1.1).

I
7

3




4




5
h




7

x

Y

I 0

II

I?



I.3

`4



IS



16.

17.

REFERENCES

Idevern II Munahty hy race-\ex and ape JAMA. 1479. 242.2572-i
Veteran\ Admim\tmtlon Cwperatlvr Srudq Group on Antlhyptxuww
Agerw Effects 01 trealment on morhidlty m h~pertcnxwn. II Rewlt\ m
patent\ with d~a\tollc blood pre\\urc averag,n~ YO !hmugh I I4 I,,,,, HF.
JAMA 1970; ?`3.1IJ3-S?.
Manrlgement Comm~ttec of rhs Au~twhan Ilypertenson Tnol The Au\tr&
~iin Therapcutir Trial in Mold Hypertcns~oo. Lancct. IYXO. I 1261-Y
Taguchi J. Freih ED Par&d reductwn of blood prr~~urr and preuentmn 01
complications in hqpcrten\ion N Engl J Med lY74: ZY I .i7%il
Kaplan NM. Whom to treat' the dilemma ut mdd hypcrtenwn. Am Hurt J
IYX?. 101:X67-70
W H.O.:l S.H Mild tIypertenuon Lla~wn Comm!ttrr. Trial\ 01 the treat-
ment ot rntld hypertsn\ion: an mtrrim analyus Lancct I%?. I IJO-56.
Manqement Cornmwx of the Australian Therapeuuc Tnal I" Lllld Hypcr-
ten\,on. Untrratrd mid hyperternon. Lancet. IYX?. I 1X5-9 I.
Management Commmre of the Awral~an Hypcrtun~on Tnnl. lmtul rewlt\
ofthe Australian Therapeutic Tnal m Mdd Hypertenwn Clm SC]  lY74: 27
Suppl 5:44YS-52s.
Helgeland A. Trtwment 01 mild hypenenwn. 3 five year controlled dnl$
trial: The Oslo Study Am J Mcd IYXO: hY:7?5-32
Smith WMc. United States Public Health Service Ho\pltal\ Conperatw
Study Group. Treatment of mild hypertension. results ol'a ten-year ~ntcrwn-
tmn trial. Circ Re$. 1977. 40 Suppl ]:I-YX-105.
Kannel WB, Sorlw P Hyperten~wn m Framingham In: Paul 0. cd Epide-
miology and control 01 hypertenrion New York Stratton Intercontmental
Medral Book. `475.553-92
lnsull W Jr. Coronary risk handbook: rstmutmg rl\k 01 coronary heart
diseae in dally practtcc. New York: American Hcan Arwc~at~im. lY73.
Madhavan S. Alderman MH. The potential effect <rt hlwd pre\hure reduc-
twn on cardiovawular disease. Arch Intern Med. 14x1, 141: 15X3-6

Wopynght. I')XZ. by the Ma\\achusett\ Medtcal Societ>
          Printcd in the U.S.A