® 1987 S.

Karger AG, Basel

Comparative Studies in Hypertension. Proc. Sat. Symp. llth Sci. Meet. ISH, Titisee-Neustadt, 1986 0028-2766/87/0477-0Ö99S2.75/0
Nephron 47: suppl. I, pp.99-103 (1987)

Smoking Habits and Antihypertensive Treatment

Siegfried Heyden ', Kenneth A. Schneider', J. George Fodorb

■'Duke University Medical Center. Durham, N. C., USA; bFaculty of Medicine. Memorial University of Newfoundland,
St. John’s. Nfld., Canada

Key Words. Smoking hypertensives • Beta-blockers and smoking • Antihypertensive therapy

Abstract. Five hypertension intervention trials (HDFP, MRFIT, Australian National BP Study, IPPPSH. MRC)
were analyzed for the effect of smoking on antihypertensive therapy and final outcome in coronary and all-cause
mortality. In addition, an observational study of primary screenees for MRFIT was reviewed. Thus, the hypertensive
population evaluated in this paper amounts to 135,851 patients. HDFP revealed that smokers had about twice the
mortality rates compared to nonsmokers regardless of the treatment group to which they were randomized. The
annual incidence of events in the Australian Study among nonsmokers in the placebo group was even slightly lower
than in smokers under active therapy. The results of the MRFIT showed that smoking had a particularly deleterious
impact on those hypertensives whose cholesterol levels were elevated. In this group, the coronary death rates were 10
times higher than in nonsmokers with lower cholesterol levels. Although the treatment with beta-blockers reduced the
coronary event rates in the MRC and in IPPPSH, this beneficial effect was absent in smokers. However, in trials in
which diuretic treatment is effective in nonsmokers, it is equally effective in smokers.

Introduction
The editorial accompanying the British MRC report
[I] concluded; ‘In advising hypertensive patients, we must
continue to emphasize the great importance of stopping
smoking, for this may turn out to be a more important
therapeutic maneuver than the prescription of blood
pressure lowering drugs." The authors of the Australian
National Blood Pressure Intervention Study [5] were
equally blunt by commenting: ‘Smokers who are obese
would seem to benefit least [from antihypertensive treat­
ment] and perhaps not at all.’ Although the high risk of
cigarette smoking for triggering coronary heart disease
and the fast disappearance of that risk in exsmokers have
been documented in many clinical and epidemiological
studies, the influence of smoking on antihypertensive
therapy has not been investigated. A sufficient number of
observation and intervention studies has been completed
in the past 6 years to allow examination of the interaction
of smoking, hypertension and antihypertensive therapy.
I nterpretation of these retrospective subgroup analyses is
were only recorded at baseline and were not part of the
treatment regimen.
International Studiesof the Interaction of Smokingand
Hypertension
Hypertension Detection and Follow-Up Program
(HDFP)
The HDFP was a therapeutical trial of 5 years’ dura­
tion. The patients were randomized either to a stepped
care group, intensively treated in specialized centers or
they were referred to their usual source of care. The
largest group of patients in this study consisted of hyper­
tensives in the mild blood pressure range, DBP
90-104 mm Hg at baseline (BL). Figure 1 shows that the
all-cause mortality rates per 1,000 men and women, ad­
justed for race, age and sex, are almost doubled in smo­
kers to nonsmokers, regardless of the treatment group.
However, the all-cause mortality rates are lower in
stepped care than in referred care patients for both non-
130.237.122.245 - 1/10/2019 9:11:39 AM
Karolinska Institutet, University Library

to be done with the usual precaution since smoking habits smokers and smokers. Thus, one can expect a reduction
Downloaded by:
100 H eyden/Schneider/Fodor

100 tilia Nonsmokers, n = 4,915

td 80 □ Smokers, n = 2,910
o c
s:
60

¡C ^ 40-

20-
¿ .T 3
<3 0
Stepped care Referred care

Fig. 1. 5-year observation of 7,825 men and women 30-69 years Fig. 2. Coronary mortality among 100,032 hypertensive primary
old with DBP 90-104 mm Hg in the HDFP. screenees of MRFIT aged 35-57 years.

Table 1. Number of deaths after 6 years per 1,000 men (35-57 examined in nonsmoking participants (table I) and in
years) smokers (table II) for two groups: those hypertensives
with cholesterol levels below and above 250 mg/dl. The
Nonsmokers (SI) Cholesterol (BL), mg/dl
data document that the coronary death rates among smo­
<250 >250 kers with lower cholesterol levels are 7 times higher and,
among hypercholesterolemic smokers, 10 times higher
Men, n 620 1,188
than in nonsmokers with lower cholesterol levels
Coronary deaths, n 2(3) 23(19)
All-cause mortality, n 14(23) 42(35) (tables 1,11). No conclusion on the efficacy of antihyper­
tensive therapy among smokers can be drawn from this
Data from MRFIT [7). trial.

M RFIT - Prospective Data on the Primary Screenees

(1986)
Table II. Number of deaths after 6 years per 1.000 men (35-57 The 100,032 men in this program [8] had a DBP
years) >90 mm Hg, and 87% of these men had DBP in the range
of 90-104 mm Hg with the remaining 13% having DBP
Smokers (SI) Cholesterol (BL). mg/dl > 104 mm Hg. None of these hypertensives had evidence
<250 >250 of myocardial infarction at BL. Figure 2 depicts the
6-year CHD death rates per 1,000 hypertensive males by
Men, n 1,388 823 BL cholesterol levels and smoking history. This addi­
Coronary deaths, n 30 (22) 25 (30)
All-cause mortality, n 68(49)
tional analysis is a valuable complementary study of the
50 (61)
two previous intervention studies because of the ex­
Data from M R FIT [7], traordinary size of the cohort and because cholesterol
levels were not limited to men with moderate hypercho­
lesterolemia as defined by current American standards
(>220 mg/dl). The risk of fatal CHD increased from a
in all-cause mortality even among smokers, given the low of 3.7/1,000 for nonsmokers in the lowest cholesterol
stepped care intervention compared to the referred care quintile to 21.4/1,000 for smokers in the highest choles­
approach. terol quintile, an almost 6-fold increased risk in 6 years.
Of this cohort of hypertensive men, only 9,612 (9.6%)
Multiple Risk Factor Invervention Trial (1982) were nonsmokers and at the same time in the lowest
The patients in this study [7] were randomized to either cholesterol quintile < 182 mg/dl. ‘Thus, relatively low-
a special intervention (SI) group or referred to their risk hypertensive men were rare exceptions... Treatment
family physicians. For this review, relevant data are only must be comprehensive and not limited to measures for
130.237.122.245 - 1/10/2019 9:11:39 AM

available for the SI group. The risk of mortality was

Karolinska Institutet, University Library

controlling high blood pressure’[8].

Downloaded by:
Smoking and Hypertension
¡a Nonsmokers, n = 2,570
□ Smokers, n = 857
Fig. 3. Incidence of trial end-points in 3,247 men and women
30-69 years old with DBP95-109 mm Hg in the Australian National
BP Study.
■ Treatment
1 1 Piacebo
Fig. 4. Efficacy of antihypertensive treatment among smokers
and nonsmokers in the Australian National BP Study.
Australian National Blood Pressure Study (1984)
Hypertensive patients were treated with active ther­
apy or placebo for 31/: years [5]. Figure 3 demonstrates the
differences in the disease and death rates per 1,000 per
year in nonsmokers and smokers. The rates of nonsmo­
kers in the placebo group were slightly lower than in
smokers under active therapy. However, among smokers
in the placebo group, the rates were over 2-fold higher
than in nonsmokers under active therapy. Close inspec­
tion of figure 3 reveals evidence of efficacy of antihyper­
tensive therapy among smokers. Thus, for the smokers
randomized to active treatment, the annual incidence
rate was 23.4 and among smokers in the placebo group,
the rate was 35.7, the relative risk (23.4/35.7) is 0.66
(fig. 4). Therefore, there is a one-third reduction in annual
incidence rates among smokers if they were placed on
active therapy. This evidence of efficacy of antihyperten­
sive treatment among smokers is the same as for nonsmo­
kers: the relative risk among the nonsmokers (from 21.1 to
15.4) is 0.72. What is not evident from this general over­
view is the conclusion by the Australian investigators that
101
Table 111. Trial end-points' in the Australian Study in two
treatment groups; numbers and rates per 1,000 men according to
weight and smoking at BL
Active Rs Placebo
n rate n rate
BMI <2.6 g/cm ;
Nonsmokers 19 18.6 27 23.6
Smokers 13 34.5 18 50.0
BMI > 2.6 g/cm-
Nonsmokers 23 15.5 33 24.3
Smokers 12 24.1 13 17.0
There were only between I and 14 trial end-points in all four
categories among women. Data for women are not shown.
1 Trial end-points were unrelated to body mass index (BMI) in
nonsmokers.
the benefit of treatment ‘would occur much greater than
average, particularly in smokers with low levels of body
mass... Strikingly higher rates in untreated male and
female smokers with low BMIs and marked benefit from
treatment in these groups were apparent’ (table III).
The authors quote from the Chicago Peoples Gas
Company Study where a poorer prognosis in smokers of
low compared with those of high body weight has been
noted: ‘It seems likely that the decreasing body weight
reflects the extent of damage to health resulting from
smoking, although why this should interact so strikingly
with hypertension and be favorably affected by antihy­
pertensive treatment is not clear.’
International Prospective Printary Prevention Study in
Hypertension (IPPPSH, 1985)
This intervention study [4] was a randomized clinical
trial involving 6,351 hypertensive men and women fol­
lowed for periods between 3 and 5 years. Its purpose was
to compare merits of two antihypertensive treatment regi­
mens, one with beta-blockers and the other with diuret­
ics. The results did not show any significant difference
between the two treatment modalities. However, smo­
king was a major predictive factor for both coronary and
cerebrovascular events, i.e., an increased risk of 90 and
82%, respectively, among smokers versus nonsmokers.
Table IV indicates a strong interaction effect between
smoking and coronary events for hypertensive men but
not for hypertensive women (p = 0.316). ‘The expected
benefit of beta-blockage was seen in nonsmoking men
130.237.122.245 - 1/10/2019 9:11:39 AM
Karolinska Institutet, University Library
who showed half the cardiac event rate of their peers in
Downloaded by:
102 Heyden/Schneider/Fodor

Table IV. Critical cardiac events in IPPPSH in two treatment Table V. Influence of two drug regimens on cardiovascular
groups: numbers and rates per 1,000 patient-years according to deaths in the MRC according to smoking status at BL
smoking status at BL
Treatment Trial outcome
Beta-blockers Diuretics
stroke CHD
events rate events rate
smokers nonsmokers smokers nonsmokers
Men
Diuretics + + _ _
Nonsmokers 20 5.4 39 11.6
Smokers 34 18.1 Beta-blockers - + - +
25 14.5
Women
Nonsmokers 17 4.1 9 2.1 Symbols indicate no effect ( - ) or a positive effect ( + ) on
Smokers 8 6.6 8 8.0 reduction of cardiovascular deaths.

■ Nonsmokers n = 2,057
□ Smokers, n - 1,137
Fig. 5. Comparison of the influence of smoking and nonsmoking
on two active treatment regimens among 3,194 men 40-64 years old
with DBP 100-125 mm Hg in the IPPPSH.
the non-beta-blocker group... Any benefit [from beta-
blockage compared to diuretics in men] appears to de­
pend on smoking status’ (fig. 5). Because of the small
number of events, no conclusion should be drawn as to
the interaction of smoking and antihypertensive treat­
ment in women.
Medical Research Council (MRC) Trial o f Mild
Hypertension (1985)
The British trial [6] followed 17,354 men and women
with DBP 90-109 mm Hg over a period of 51/? years using
three treatment regimens: placebo, beta-blockers and
diuretics. The complexity of the results may be reduced to
a summary tabulation (table V), dividing patients into
smokers and nonsmokers at BL. The fact that proprano­
lol apparently reduced the coronary event rate in non-
smokers and not in smokers is confirmation of the find­
ings from the IPPPSH trial. Cigarette smoking is known
to increase the rate of metabolic degradation of propran­
olol, and hence decrease its plasma concentration, ac­
cording to Feely et al. [3] and Dawson and Vestal [2],
There is a good chance that smoking hypertensive pa­
tients ceased smoking for several hours when coming to
the clinic. Their BP levels averaged 4 mm Hg SBP and
1.5 mm Hg DBP higher than in nonsmokers. However, on
nonclinic days, these smokers had no restrictions on their
smoking habits and. therefore, it is possible that their BP
levels actually may have been higher for a longer period

Table VI. Relative efficacy of diuretic therapy in three trials among smokers and nonsmokers

Trials Nonsmokers Smokers

active stepped referred care or relative active stepped referred care or relative
care placebo efficacy care placebo efficacy

Australian Study' 15.4 21.1 0.72 23.4 35.7 0.66

HDFP2 48 58 0.83 84 103 0.82
MRC5 7.4 7.5 0.99 12.6 12.7 0.99

1 Annual incidence of morbidity and mortality.

2 5-year all-cause mortality.
130.237.122.245 - 1/10/2019 9:11:39 AM

5 5'/2-year incidence of coronary morbidity and mortality.

Karolinska Institutet, University Library
Downloaded by:
Smoking and Hypertension
of time [Dollery, pers. commun.j. ‘The difference in the
rate of strokes and coronary events between nonsmokers
and smokers was impressive for these differences were
much greater than the effects of treatment’ [1], In contrast
to IPPPSH, the differences among smokers and nonsmo­
kers were not limited to men.
Discussion
The strong and graded relationship between increas­
ing serum cholesterol concentrations and hypertension
on one hand and the risk of CHD on the other was shown
clearly in the extended MRFIT cohort [8], This relation­
ship was heavily influenced by smoking status at BL. The
six recent studies (1982-1986), reviewed in this paper,
have demonstrated unequivocally that hypertensive
smokers do experience a 2-fold higher risk of CHD than
nonsmokers. From two intervention trials, I PPPSH [4]
and MRC [6], it appears warranted to conclude that any
beneficial, i.e. efficacy, effect of beta-blocker as com­
pared to diuretic therapy was wiped out by smoking.
However, antihypertensive therapy with diuretics among
smokers is more complex. When diuretic therapy proves
efficacious (table VI), as in the Australian and HDFP
studies, the beneficial effect of treatment is the same for
smokers and nonsmokers. The MRC failed to show effi­
cacy of diuretic therapy for either smokers or nonsmok­
ers.
103
References
1 Breckenridge. A.: Treating mild hypertension (Editorial). Br.
med. J. 291:89 -90 (1985).
2 Dawson, G.W.; Vestal, R.E.: Smoking and drug metabolism.
Pharmacol. Ther. 75:207-222 (1981).
3 Feely, J.: Crooks, J.; Stevenson, I.H.: The influence of age,
smoking and hypertension on plasma propranolol steady-state
concentration. Br. J. clin. Pharmacol. 72:73-78 (1981).
4 I PPPSH Collaborative Group: Cardiovascular risk and risk
factors in a randomized trial of treatment based on the beta-
blocker oxprenolol: The International Prospective Primary Pre­
vention Study in Hypertension (IPPPSH). J. Hypertens. 3:
379-392 (1985).
5 Management Committee of the Australian National Blood Pres­
sure Study: Prognostic factors in the treatment of mild hyperten­
sion. Circulation 69:668-676 (1984).
6 Medical Research Council Working Party: MRC trial of treat­
ment of mild hypertension: principal results. Br. med. J. 297:
97-104(1985).
7 Multiple Risk Factor Intervention Trial Research Group: Multi­
ple Risk Factor Intervention Trial; risk factor changes and
mortality results. J.Am.med.Ass. 248:1465-1477 (1982).
8 Stamler, J.; Wentworth, D.; Neaton, J.D.: Prevalence and prog­
nostic significance of hypercholesterolemia in men with hyper­
tension. Am. J. Med. S0:suppl. 2A, pp. 33-36 (1986).
Siegfried Heyden, MD, PhD
Professor
Department Community and Family Medicine
Duke University Medical Center
Box 2914
Durham. NC 27710 (USA)
130.237.122.245 - 1/10/2019 9:11:39 AM
Karolinska Institutet, University Library
Downloaded by: