SlImmuy

Drugs 31 (Suppl. 1): 23-28 (1986) ()() 12-6667/86/01 00-0023/$3.()()/0 © ADIS Press Limited All rights reserved.

Further Analyses of the Hypertension Detection and Follow-up Program

H.G. Langford Endocrine and Hypertension Division, University of Mississippi, Jackson, Mississippi

The Hypertension Detection and Follow-up Program was a randomised trial to com­pare all-cause mortality of patients receiving antihypertensive therapy in special clinics with those referred to the usual sources of care. All-cause mortality was significantly re­duced overall, and in the mildest hypertensives, by stepped care. This specificity of the antihypertensive effect was shown by the proportionate lowering of stroke deaths, and the persistence of the mortality effect, when analysed by time-dependent co-variants, which took into account the amount of antihypertensive therapy the patients were receiving. Car­diovascular and coronary heart disease mortality were reduced in stepped care, as judged by death certificates. The benefits of stepped care were still present when analyses were confined to those with baseline ECG abnormalities. The 5-year incidence of angina pec­toris and myocardial infarction, as judged by the Rose Questionnaire, was decreased in stepped care. Serum alkaline phosphatase fell in thiazide treated patients, suggesting a favourable influence on calcium balance. Eight-year analyses suggest that the favourable influence on mortality persisted after the end of the program for all except the eldest participants.

Statistics from life insurance companies aug­mented by prospective studies from a number of areas, including Framingham, Massachusetts, Chi­cago, Illinois, and Los Angeles, California, in the United States demonstrated the unequivocal na­ture of hypertension as a risk factor for cardiovas­cular morbidity and mortality. The Veterans Administration study of the effect of antihyperten­sive therapy on hypertensive events demonstrated with equal certainty the value of therapy in the sev­erely hypertensive patient. The Hypertension De­tection and Follow-up Program was designed to

determine the value of therapy in the less severely ill hypertensive, as they existed in the community.

As valuable as the Veterans Administration trial was, it was restricted to individuals identified at a Veterans Administration hospital in the United States with a blood pressure which remained con­sistently elevated during hospitalisation. These specifications meant that a very severe group was enrolled in the trial. It was felt that milder indi­viduals needed to be studied, milder both in the sense of lower blood pressures and that they were not identified in the hospital setting, which tends

Hypertension Detection and Follow-up Program

to attract patients with much more end-organ damage.

1. Design

The desirability of studying almost all individ­uals with elevated blood pressures as they existed in the community led to the unique design of the Hypertension Detection and Follow-up Program. We emphasise now, and we will again, that this program did have randomised controls, but these were not blinded, placebo-receiving controls.

Fourteen cooperating clinical centres across the United States enrolled hypertensive individuals who were first identified in probability samples of the local areas (or in Chicago industries). If their fifth-phase seated diastolic blood pressure was ~ 95mm Hg, they were referred to special study clinics. If their diastolic blood pressure was ~ 90mm Hg, as determined by a bias-free blood pressure apparatus, the zero muddler, they were in­vited to participate in the program, and randomly assigned to either antihypertensive care in the spe­cial clinics of the program or referred back to the usual sources of care. This design had certain con­sequences, positive and negative:

1) Because the patients would be seen more fre­quently in the stepped care group than in the re­ferred care group, the primary end-point had to be a consequence of hypertension that could be de­termined without bias in both groups. Accordingly, death was the primary end-point.

2) If the null hypothesis was rejected, and ad­vantage was demonstrated for stepped care therapy, the strength of the positive effect would have to be inferred from the collected data, but as some con­siderable percentage of the referred care patients would be treated, it would be difficult to determine relative mortality with exactitude. In other words, the percentage benefit for randomisation to stepped care would be attenuated by the considerable num­ber of patients who were receiving antihypertens­ive therapy in the referred care group.

3) The possibility that some aspect of referral to stepped care, other than antihypertensive therapy, would be the effective aspect of the study would

24

have to be considered, and appropriate post hoc analyses done to see if this possibility were so.

These disadvantages would be balanced by a very major advantage. A placebo trial must be a trial of complete volunteers. Volunteers are health­ier than non-volunteers and, moreover, the pres­ence of some indication for antihypertensive treat­ment which would exclude entry into a standard volunteer trial would not exclude membership in the Hypertension Detection and Follow-up Pro­gram, for individuals referred to usual sources of care would not be blocked from receiving care.

An illustration of the strength of the 'volunteer' phenomenon can be derived by comparing the ra­tio between the placebo group's mortality and the population mortality in the Australian National Blood Pressure Trial. The placebo group's mortal­ity was approximately one-third that of the age-sex matched mortality for the parent Australian popu­lation. In contradistinction, the referred care par­ticipants in the Hypertension Detection and Fol­low-up Program died at approximately the same rate as the American population. The program dealt with sicker people than the Australian National Blood Pressure Trial (although not as sick as those in the Veterans Administration study), and it was able to do that because of its unique design.

2. Results and Discussion

The major results of the Hypertension Detec­tion and Follow-up Program were reported in 1979 and they are now repeated as a baseline (HDFP 1979). There was a significant reduction in all-cause mortality in the stepped care group compared to the referred care group. The reduction was also sig­nificant in the mild group with diastolic blood pressure 90-105mm Hg on entry into the program. There was a significant reduction in death from cardiovascular disease in stepped care compared with referred care patients, as determined by no­sologist classification of death certificates. The Hypertension Detection and Follow-up Program Investigative Group felt, and still feels, that they had demonstrated the efficacy of antihypertensive

Hypertension Detection and Follow-up Program

therapy in reducing one's chance of death and, in all probability, one's chance of cardiovascular dis­ease.

However, discussion since that time by some individuals has implied that the favourable effect of assignment to the stepped. care group was due to 'the general medical care' rather than to the spe­cific effect of antihypertensive therapy. As one of my colleagues points out, this demonstration would be extraordinarily interesting; no properly random­ised study has addressed this question (of the bene­ficial effect of general medical care) yet. Unfortu­nately, I do not believe that we can look to the Hypertension Detection and Follow-up Program as demonstrating the favourable effects of general medical care, apart from antihypertensive therapy, for the following reasons:

1) Hardy & Hawkins (1983) determined the in­fluence of the post-randomisation events of (a) the mean blood pressure; (b) whether the patient was at goal blood pressure; and (c) whether they were receiving antihypertensive therapy on prognosis. These data were available at 2 points after the in­itial randomisation. 70% of the mortality differ­ence between stepped and referred care could be 'explained' by the difference between the 2 groups of these 3 components. Considering the known variability of blood pressure, I am willing to accept this as evidence for all, or at the very least almost all; of the difference in mortality between stepped and referred care being due to the blood pressure control by pharmacological means.

2) Cerebrovascular accidents are the most spe­cific of the hypertensive complications. In other words, while coronary artery disease is strongly in­fluenced by blood cholesterol and smoking, in ad­dition to blood pressure, cerebrovascular accident is more closely tied to blood pressure than to either of the other 2 major risk factors. Therefore, re­duction in morbidity and mortality from cerebro­vascular accidents is an excellent indicator of the effect of antihypertensive therapy. Moreover, sev­eral studies in the United States have shown that a death certificate diagnosis of death from stroke is usually correct. Also, the patients' statements that they have had a stroke usually indicate that they

25

have had a cerebrovascular accident. The history of cerebrovascular accident seems to be both sen­sitive and specific. The 5-year incidence of stroke was significantly reduced in stepped care partici­pants, compared with referred care participants (HDFP 1982) [fig. 1]. The frequency was reduced whether or not the individuals had end-organ dam­age when they entered the program, and whether or not they were on medication. Those who were not on medication initially, and had no end-organ damage, had the largest proportional reduction in cerebrovascular accidents, with a decrease of 50%. However, absolute reduction was just as great in those who had end-organ damage, but they ended up with a much higher percentage of individuals having strokes. In other words, therapy was worth­while even if end-organ damage was present, but the chance of having a cerebrovascular accident over the next 5 years remained high if end-organ damage was present when the individual came into the program.

3) There was no difference between stepped care and referred care participants in change in smoking or change in cholesterol, factors that one would presume would be affected by a general program.

4) There was no difference in the rate of hos­pitalisation.

All of these factors suggest to us that 'general medical care' had little to do with the results of the Hypertension Detection and Follow-up Program.

2.1 Symptomatic Coronary Artery Disease

As we reported in the first of our publications giving end-point data, myocardial infarction was significantly reduced overall, as well as in Stratum I, diastolic blood pressure 90 to l04mm Hg. We have recently reported on the effect of stepped care therapy on the incidence of several indices of cor­onary artery disease, the Rose Questionnaire, ECG evidence of myocardial infarction, and history of myocardial infarction (HDFP 1984). A standar­dised questionnaire for angina pectoris and myo­cardial infarction (the Rose Questionnaire) was ad­ministered to each participant at the beginning of the study and in the second and fifth year for sur-

Hypertension Detection and Follow-up Program

6

5

4

3'

0 0

2 ~

& $ os a: 1

0

6.3

II RC

Dsc

26

No Medication medication

No end-organ damage

End-organ damage

No end-organ damage. No medication

Fig. 1. Five-year incidence of stroke by medication status and end-organ damage at entry for stepped care (SC) and referred care (RC) participants from all blood pressure strata. Data adjusted for race, sex and age.

viving participants. ECGs were performed at the same interval, and read centrally by the Minnesota code. Individuals positive for any of these abnor­malities at baseline were over twice as likely to die over the ensuing 5 years than individuals negative for these findings.

We, therefore, feel that these baseline findings, even though they were not the standard history of angina pectoris or diagnosis of myocardial infarc­tion of clinical practice, were indicative of disease with attendant increased mortality. The 5-year in­cidence of developing a positive history of angina pectoris was significantly reduced in stepped care compared with referred care, from 8.9% in referred care to 6.4% in stepped care, highly significant (p < 0.001). Moreover, the reduction was signifi­cant in each of the 3 blood pressure strata, being reduced by 54.3% in those with diastolic blood pressure higher than 115mm Hg, by 43.3% in those with initial diastolic blood pressure of 105 to

114mm Hg, and by 15.2% in those with initial dia­stolic blood pressure of 90 to 104mm Hg. Simi­larly, the 5-year incidence of non-fatal myocardial infarction was reduced from 8.3% in referred care participants to 7.0% in stepped care participants, also highly significant. These results are highly sug­gestive that the stepped care regimen has reduced the 5-year incidence of symptomatic coronary ar­tery disease.

2.2 Effect of Stepped Care on ECG Evidence of Left Ventricular Hypertrophy and Cardiomegaly by X-ray

The 5-year mortality was 3-fold greater in in­dividuals who had left ventricular hypertrophy on admission to the Hypertension Detection and Fol­low-up Program as in those with normal ECG (HDFP 1985). The percentage benefit of random­isation to stepped care was about the same in those

Hypertension Detection and Follow-up Program 27

Table I. Five-year mortality in stepped care and referred care participants according to baseline left ventricular hypertrophy and cardiomegaly· status

Baseline Sample size No. of deathsb Life-table rates per 100 (SE) p value characteristic

SC RC SC RC SC RC

Electrocardiogram

Normal 4545 4528 259 313 5.7 (0.3) 6.9 (0.4) 0.019

Tail R waves only 590 571 40 45 6.8 (1.0) 7.9 (1.1) 0.473 LVH 260 275 40 53 15.4 (2.2) 19.3 (2.4) 0.232 Unknown 90 81 11 9 12.2 (3.5) 11.1 (3.5) 0.823

Chest x-ray film

CTR < 0.50 3472 3342 185 232 5.3 (0.4) 7.0 (0.4) 0.003 CTR > 0.50 1161 1117 104 111 9.0 (0.8) 9.9 (0.9) 0.463

CTR> 0.55 285 264 29 47 10.2 (1.8) 17.8 (2.4) 0.010

Unknown 852 996 61 77 7.2 (0.9) 7.7 (0.8) 0.683

a CTR > 0.5 by chest x-ray film.

b Since the publication of the Hypertension Detection and Follow-up Program's 5-year mortality findings in 1979, 2 additional

5-year deaths have been identified from among persons lost to follow-up at that time (1 death in each care group).

Abbreviations: CTR = cardiothoracic ratio; LVH = left ventricular hypertrophy; RC = referred care; SC = stepped care.

with left ventricular hypertrophy as in those with a normal ECG, although a small sample size pre­vented a definitive test of benefit in individuals with left ventricular hypertrophy. Cardiomegaly by chest x-ray was also associated with a marked in­crease in mortality, and the benefit of stepped care was statistically significant in those with a cardi­othoracic ratio greater than 0.55 (table I). Random­isation to 'Stepped care was associated with a greater incidence of improvement in the ECG abnormality than randomisation to referred care. Of individuals with tall R waves or left ventricular hypertrophy 54.3% manifested improvement, compared to 42.9% of those in referred care (p < 0.001). Like­wise, there was regression from> 0.5 cardiothor­acic ratio to < 0.5 in 37.3% of stepped care par­ticipants and 28.5% of referred care participants (p = 0.029). Moreover, the progression of tall R waves to left ventricular hypertrophy were signifi­cantly less in stepped care than referred care par­ticipants (4.1% vs 8.6%, p = 0.001).

3. Conclusions

We feel that these findings represent a coherent and impressive demonstration of the value of anti-

hypertensive medication. The null hypothesis was conclusively rejected; randomisation to stepped care was associated with a significant decrease in all­cause mortality. That the decrease in all-cause mortality was primarily due to antihypertensive therapy is indicated by the analysis of time-de­pendent co-variates. These showed that the data related to blood pressure control at the 2 post-ran­domisation visits, explain the majority of the mor­tality differences between the 2 groups. Further indication of the benefit was provided by the sta­tistically significant reduction in both fatal and non­fatal cerebrovascular accidents. Further confirma­tion of benefit is provided by the reduction in an­gina pectoris as coded by the Rose Questionnaire, and the history of myocardial infarction. The pres­ence of indicators of myocardial disease, such as ECG changes of a relatively nonspecific type, or left ventricular hypertrophy, were associated with a 2- to 3-fold increase in mortality as was cardio­megaly by chest x-ray. However, these groups, though their mortality did not return to that of in­dividuals without such abnormality, remained less in stepped care than in referred care. Moreover, there was more regression in stepped care of ra­diological and ECG evidence of cardiac disease, as

Hypertension Detection and Follow-up Program

well as less progression of the disease. All of these findings join to indicate that antihypertensive therapy significantly lowers mortality and morbid­ity, even including quite mild hypertensive indi­viduals.

References

Hardy RJ, Hawkins CM. The impact of selected indices of anti­hypertensive treatment on all-cause mortality. American Jour­nal of Epidemiology 117: 566-574, 1983

HDFP (Hypertension Detection and Follow-up Program) Co-op­erative Group. Five-year findings of the Hypertension Detec­tion and Follow-up Program. I. Reduction in mortality in per­sons with high blood pressure, including mild hypertension. II. Mortality by race, sex and age. Journal of the American Medical Association 242: 2562-2577, 1979

HDFP (Hypertension Detection and Follow-up Program) Co-op­erative Group. Five-year findings of the Hypertension Detec-

28

tion and Follow-up Program. III. Reduction in stroke inci­dence among persons with high blood pressure. Journal of the American Medical Association 247: 633-638, 1982

HDFP (Hypertension Detection and Follow-up Program) Co-op­erative Group. Effect of stepped care treatment on the inci­dence of myocardial infarction and angina pectoris. Hyperten­sion 6 (Suppl. I): 198-206, 1984

HDFP (Hypertension Detection and Follow-up Program) Co-op­erative Group. Five-year findings of the Hypertension Detec­tion and Follow-up Program. Prevention and reversal of left ventricular hypertrophy with antihypertensive drug therapy. Hypertension 7 (Suppl. I): 105-112, 1985

Address for correspondence and reprints: Dr H.G. Langford, Endocrine and Hypertension Division, University of Mississippi, Jackson. Mississippi (USA).