Randomised, double blind, multicentre comparison of hydrochlorothiazide, atenolol, nitrendipine, and enalapril in antihypertensive treatment: results of the HANE studyBMJ 1997; 315 doi: https://doi.org/10.1136/bmj.315.7101.154 (Published 19 July 1997) Cite this as: BMJ 1997;315:154
- Thomas Philipp, directora,
- Manfred Anlauf, professorb,
- Armin Distler, directorc,
- Heinrich Holzgreve, professord,
- Joerg Michaelis, directore,
- Stefan Wellek, director the HANE Trial Research Groupf
- a Zentrum für Innere Medizin, Universitätsklinikum Essen, Hufelandstrasse 55, D-45122, Essen, Germany
- b Zentralkrankenhaus Reinkenheide, D-27574 Bremerhaven, Germany
- c Medizinische Klinik and Poliklinik, Benjamin Franklin Klinikum, Freie Universität Berlin, D-12203 Berlin, Germany
- d Medizinische Poliklinik, University of Munich, D-80336 Munich, Germany
- e Institut fur Medizinische Statistik, University of Mainz, D-55131 Mainz, Germany
- f Abteilung Biostatistik, ZI Seelische Gesundheit, D68072 Mannheim, Germany
- Correspondence to: Dr Philipp
- Accepted 15 April 1997
Objective: To compare the effectiveness and tolerability of hydrochlorothiazide, atenolol, nitrendipine, and enalapril in patients with mild to moderate hypertension.
Design: Randomised multicentre trial over 48 weeks with double blind comparison of treatments.
Setting: 48 centres in four countries.
Patients: 868 patients with essential hypertension (diastolic blood pressure 95-120 mm Hg)
Interventions: Initial treatment (step 1) consisted of 12.5 mg hydrochlorothiazide (n=215), 25 mg atenolol (n=215), 10 mg nitrendipine (n=218), or 5 mg enalapril (n=220) once daily. If diastolic blood pressure was not reduced to <90 mm Hg within four weeks, doses were increased to 25 mg, 50 mg, 20 mg, 10 mg, respectively, once daily (step 2) and after two more weeks to twice daily (step 3). The eight week titration phase was followed by an additional 40 weeks for patients who had reached the target diastolic pressure.
Main outcome measure: Blood pressure by means of an automatic device with repeated measurements.
Results: After eight weeks the response rate for atenolol (63.7%) was significantly higher than for enalapril (50.0%), hydrochlorothiazide (44.7%), or nitrendipine (44.5%). After one year atenolol was still more effective (48.0%) than hydrochlorothiazide (35.4%) and nitrendipine (32.9%), but not significantly better than enalapril (42.7%). The treatment related dropout rate was higher (P<0.001) in the nitrendipine group (n=28).
Conclusions: There is no evidence of superiority for antihypertensive effectiveness or tolerability of the “new” classes of antihypertensives (calcium channel blockers and angiotensin converting enzyme inhibitors). As these drugs are now widely used as treatment of first choice, our results further emphasise the need for studies confirming that they also reduce morbidity and mortality, as has been shown for diuretics and ß blockers.
Calcium channel blockers and angiotensin converting enzyme inhibitors as initial monotherapy in the treatment of hypertension
Reduction in mortality has been shown only with established ß blockers and diuretics
Comparison of treatment with hydrochlorothiazide, atenolol, nitrendipine, and enalapril showed no superiority of the new drug classes
Atenolol was the most effective drug, while nitrendipine showed the highest drop out rat
Elderly patients respond better to hydrochlorothiazide and nitrendipine, and women better to enalapril, although in both subgroups the highest rate of response was with atenolol
In the past three decades remarkable progress has been made in the treatment of hypertension by introduction of new well tolerated drugs. In large scale and properly controlled trials ß adrenergic blockers and thiazide diuretics have been shown to reduce clinical end points related to strokes, coronary artery disease, left ventricular failure, and mortality.1 2 3 4
The fact that other classes of drugs exert a favourable influence on a number of important but substitute end points—such as lipid concentrations, left ventricular mass, insulin resistance, microalbuminuria—can affect the choice for an individual patient, but it does not eliminate the need for unequivocal proof that the incidence of clinical end points is in fact reduced. Several national and international committees have nevertheless recommended the use of angiotensin converting enzyme inhibitors and calcium channel blockers for initial single drug therapy as an alternative to the established treatment with ß blockers and diuretics,5 6 7 8 but only a few studies have been carried out to compare these four classes of antihypertensive drugs in long term controlled trials with sufficiently large populations.9 10 11 One of the studies referred only to men10 and another referred to hypertensive patients with a diastolic pressure <100 mm Hg (mean 92 mm Hg).9
We therefore evaluated the effectiveness and the tolerability of these four important therapeutic principles in hypertensive patients in a double blind controlled study over one year. Hydrochlorothiazide, atenolol, nitrendipine, and enalapril were selected as representatives of these antihypertensive classes and are recommended for once daily application. All these drugs have been well known for many years, and all have been studied in large controlled trials.
Patients and methods
The HANE (hydrochlorothiazide, atenolol, nitrendipine, enalapril) study was designed as a double blind, randomised, multicentre trial with the primary objective of comparing the antihypertensive effectiveness and tolerability of the four drugs. Specifically, we sought to determine the ability to control blood pressure over time and the incidence of premature terminations of treatment for medical reasons. Secondary objectives were to analyse the influence of characteristics of patients that determine the effectiveness of treatment.
The study design of the HANE study was similar to that of the VERDI (verapamil diuretic) study, in which we compared verapamil and the diuretic drug hydrochlorothiazide.12 The study comprised a single blind placebo phase lasting two weeks followed by double blind active treatment. All antihypertensive medication used before enrollment in the trial was discontinued within two weeks of the initial visit. The study population comprised male and female outpatients with an age range of 21-70 years and a resting (sitting) diastolic blood pressure of 95-120 mm Hg.
Exclusion criteria were as follows. Firstly, contraindications with regard to the drug classes under evaluation: frank congestive heart failure; atrioventricular blocks; sick sinus syndrome; sinus bradycardia (<50 beats/min); pregnancy; known renovascular hypertension; and known intolerance to the drugs. Secondly, patients with unacceptable risks with regard to the wash out and placebo phase: a diastolic blood pressure <95 or >120 mm Hg at the beginning or at the end of the placebo phase; myocardial infarction within the past six months. Thirdly, possible interactions of the primary and secondary objectives of the trial: concomitant use of any other drug with antihypertensive potency (that is, for other diseases); hyperuricaemia, hyperlipidaemia or diabetes mellitus requiring treatment with drugs; the use of oral contraceptives; serum creatinine ≥150 μmol/l; hypokalaemia <3.6 mmol/l. Finally, we excluded any other patients who had a history of or evidence for malignancy or other serious diseases sufficient to interfere with long term adherence to trial protocol and those over 30% overweight.
Patients were included only after consent had been obtained after detailed written explanation of the nature and purpose of the investigation. The study protocol was approved by the ethics committee of the German Society of Hypertension.
Measurement of blood pressure
To ensure standardised blood pressure recordings at all participating centres the blood pressure was measured by means of an automatic instrument (Tonoprint electronic, Speidel und Keller, Jungingen, Germany) as described in the VERDI study.12 The blood pressure and pulse rate values are printed out with the date and time of day. The Tonoprint has been tested and evaluated as described.12 After 5 minutes of rest six recordings were taken every 2 minutes while the patient was sitting. All decisions about admission to the study and about changes in treatment were based on the average of the first five recordings. The printout dates were included in the follow up records and checked by the central office.
Between 1991 and 1993 a total of 1218 patients were screened at 48 centres (see Appendix 1) for possible inclusion; of these, 287 patients could not be considered for randomisation because the diastolic pressure had dropped below the limit of 95 mm Hg or was over 120 mm Hg under placebo at any visit. Four centres with a total of 38 patients were excluded in an early phase of the trial because of incorrect use of the automatic blood pressure device. Twenty five patients were excluded for other reasons, including over 30% overweight (n=14), concomitant use of antihypertensive drugs (n=3), and other violations of the protocol (n=8). Of the remaining 868 patients, 215 were randomised to hydrochlorothiazide, 215 to atenolol, 218 to nitrendipine, and 220 to enalapril.
Study course and drug doses
Treatment was started with a dose titration phase during which the blood pressure was taken at two week intervals. The initial low dose was maintained for four weeks and was then increased stepwise if the diastolic blood pressure had not been lowered to below 90 mm Hg. Specifically, the dose of hydrochlorothiazide was increased from 12.5 mg (step 1) to 25 mg once daily (step 2) and then to 50 mg (step 3), this last dose being given as 25 mg twice daily; atenolol was increased from 25 mg to 50 mg once daily and then to 50 mg twice daily; nitrendipine from 10 mg to 20 mg once daily and then to 20 mg twice daily; and enalapril from 5 mg to 10 mg once daily and finally to 10 mg twice daily. The eight week titration phase was followed by a long term phase of an additional 40 weeks for those patients who had reached the target of diastolic pressure below 90 mm Hg. Patients who did not reach the target blood pressure at the end of the titration period were considered to be non-responders and were withdrawn. During the long term phase, blood pressure was checked after 4, 8, 16, 28, and 40 weeks.
If at one visit during the study the target blood pressure was not achieved after at least two weeks on the highest dose (step 3) the patients were withdrawn from the main trial. Other reasons for discontinuation of treatment were specified as a diastolic pressure >120 mm Hg in the course of the study, severe adverse events, non-compliance, development of a new serious disease unrelated to the test treatment, and emergence of a new exclusion criterion. Patients withdrawn because of blood pressure, adverse events, or non-compliance were counted as non-responders. Withdrawals because of one of the remaining reasons were considered non-evaluable with respect to patients' response to the respective treatment.
The drugs were specially manufactured for study purposes and tested for bioequivalence with commercially available standard formulations. As all active treatments could not be provided in the same manner, two different tablets were made up—one kind for blinding of hydrochlorothiazide and enalapril and the other for blinding of atenolol and nitrendipine.
Randomisation was performed centrally in the biometrical trial office.
At each visit blood pressure and pulse rate were measured as close as possible to 24 hours after the last medication during the placebo period and during steps 1, 2 (once daily treatment), or as close as possible to 12 hours after the last medication during step 3 (twice daily treatment).
The confirmatory part of the analysis consisted of tests for differences in the response rates (diastolic pressure <90 mm Hg) between all pairs of treatments obtained after 8, 24, and 48 weeks of single drug treatment. To ensure a multiple significance level of 5% for all comparisons referring to the same trial period, the P values obtained from the respective set of six pairwise χ2 tests were assessed by means of HOLM's sequentially rejective multiple test procedure.13
With 200 patients per group the power of this multiple test to detect a difference of 0.15 between the responder rates when one of them equals 0.50 was computed to be about 70%. (The exact numerical value of the power depended on the magnitude of the differences between all other pairs of responder rates.)
Numerous additional analyses for subpopulation comparisons were performed only in an explorative manner with respect to treatment effects. P values obtained from these additional tests are reported for descriptive information only.
Characteristics of the patients
Table 1 gives the characteristics of patients on entry into the study. All four treatment groups were well balanced in terms of systolic and diastolic blood pressure, heart rate, sex, age, weight, and previous treatment.
Titration phase—Eighty four patients did not finish the titration phase because of adverse events and administrative causes (table 2). Of the 868 patients available for the titration period, 440 (50.7%) achieved the target diastolic pressure <90 mm Hg. With all regimens of steps 1 to 3, target blood pressure at the end of the titration period was more often achieved with atenolol (63.7%; 23.3% and 40.7% on step 1 and 2, respectively) than with enalapril (50.0%; 17.0% and 32.8%), hydrochlorothiazide (44.7%; 12.1% and 30.3%), or nitrendipine (44.5%; 14.1% and 28.5%). At the multiple 5% level the response rate to atenolol was significantly higher than that to the three other drugs. The differences between enalapril, hydrochlorothiazide, and nitrendipine were not significant (table 3).
Long term phase—Four hundred and forty patients were included in the long term phase and 324 patients finished the study after 48 weeks with blood pressure below the target: 96 (74 reached target blood pressure) treated with hydrochlorothiazide, 137 (98) treated with atenolol, 97 (69) with nitrendipine, and 110 (93) with enalapril. After 48 weeks of treatment the diastolic pressure was below the target in 48.0% of those initially treated with atenolol. The response rate to atenolol was no longer significantly different from that to enalapril (42.7%), but atenolol was still significantly better than hydrochlorothiazide (35.4%) and nitrendipine (32.9%) (P<0.001; table 3).
Sex, age, and blood pressure before treatment
Figure 1 shows the response rates for various subgroups after week 8 according age, sex, and blood pressure of the patients. Age and diastolic blood pressure were dichotomised by using the respective median values (54 years and 102 mm Hg) as the cut points.
In younger patients (≤54 years) atenolol and enalapril were more effective (P<0.001) than hydrochlorothiazide and nitrendipine after eight weeks and after 48 weeks. Within the pairs (atenolol and enalapril versus hydrochlorothiazide and nitrendipine) there were no clear differences. In older patients (>54 years) no differences could be found within the therapeutic regimens.
Both nitrendipine (descriptive P value <0.001) and hydrochlorothiazide (P=0.004) were significantly more effective in older than in younger patients (fig 1). This result could not be explained by a baseline difference in blood pressure (mean diastolic pressure 103.4 mm Hg in younger and 103.7 mm Hg in older subgroup).
In men the higher effectiveness of atenolol compared with enalapril, hydrochlorothiazide, and nitrendipine was found only at week 8 (P<0.001), whereas at week 48 no difference could be found. In women the response rate to atenolol was higher than that to nitrendipine at week 8 (P<0.001) and week 48, whereas no substantial differences between the other regimens were observed.
Women responded better in general to antihypertensive treatment than men (55.0% v 47.7%). Among these enalapril was clearly more effective in women (59.0% v 42.5%; P=0.015) (fig 1).
As expected the response rate was significantly higher for atenolol and hydrochlorothiazide (P<0.001) at lower levels of pretreatment blood pressure (≤102 mm Hg), whereas the effect of enalapril was only slightly more pronounced (P=0.042). This, however, did not apply to nitrendipine, which turned out to be equally effective in patients with higher blood pressures. In the subgroup with higher blood pressures no drug was superior.
Overall, patients treated in Israel responded less well (44.0% v 52.3% for the other patients), but atenolol was again the most effective drug (fig 1).
Withdrawals and adverse events
One hundred and thirty seven patients had to be withdrawn from treatment for various reasons (table 2), 84 of them during the dose titration period. In the nitrendipine group the incidence of withdrawals because of adverse events was significantly higher (P<0.001) than in the other groups. The main reasons for withdrawal with nitrendipine were headache, flush, or palpitations (22/28) and intolerable oedema (3). The major reasons for dropout in the hydrochlorothiazide patients were gastrointestinal complaints (4/9) and muscle cramps (2); in the atenolol group they were fatigue (6/11) and cold hands (3), and in the enalapril group they were cough (6/12) and gastrointestinal complaints (4). An age dependence of dropouts related to adverse events was observed only in patients treated with nitrendipine (23 in younger patients v 5 in older patients) and atenolol (3 v 8, respectively).
The HANE study was designed to compare the four most commonly used classes of antihypertensive drugs in the initial treatment of mild to moderate hypertension. The individual representatives of these classes were selected on the grounds of long term experience and on their recommendation for once daily application. The comparison was undertaken because up to now there have not been any long term intervention studies providing evidence for a reduction in mortality and morbidity with the use of calcium channel blockers and angiotensin converting enzyme inhibitors. In the absence of such comparisons a minimum requirement for recommending them as antihypertensive agents of first choice should be that they are as effective and as tolerable as the established classes of diuretics and ß blockers in populations of patients of sufficient size treated for a sufficiently long time.
The size of our study population, with more than 200 patients per treatment group, and the duration of the study of nearly one year ensured adequate power to detect differences between the test drugs in antihypertensive effectiveness and in tolerability.
Qualitative differences between antihypertensive drugs can be assessed principally by comparing first their antihypertensive potency (the responder rates) and then their tolerability (the incidence of adverse events and dropout rates). On this basis two compounds which do not differ in these two respects could be said to be “equivalent” antihypertensive drugs. The comparison becomes more difficult, however, if superior antihypertensive potency of one drug is associated with poorer tolerability or vice versa.
According to these definitions and to the primary objective of this study, atenolol was superior to the other three reference drugs after eight weeks of treatment because it showed the highest responder rate and a treatment related rate of dropout that was similar to the rates of withdrawals under the alternative drugs. On the other hand, our assessment of nitrendipine would be more negative as its clearly higher dropout rate related to treatment was not associated with higher antihypertensive effectiveness. After 48 weeks the responder rate of atenolol was no longer different from that of enalapril, but atenolol was still more effective than hydrochlorothiazide and nitrendipine (table 3).
This kind of global assessment has to be qualified to some extent, however, because certain characteristics of patients interacting with the relative effects of the study drugs necessitate a more detailed assessment. More precisely, atenolol is superior in patients with mild to moderate hypertension but not in those with more severe hypertension (diastolic blood pressure >102 mm Hg). In this subgroup all the treatment differences seem to vanish.
A surprising finding was the influence of sex on the responder rates of the individual agents. Enalapril in particular was obviously more effective in women than in men (responder rate 59.0% v 42.5%; P=0.015), a finding that to our knowledge has not been reported previously. A tendency towards a higher responder rate in women was similarly observed for hydrochlorothiazide (fig 1). In other comparable studies sex differences were either not evaluated9 11 or the study population comprised only male veterans.10 The reason for the superior effectiveness of enalapril in women is unclear; differences in plasma renin activity are obviously not responsible because we did not observe any differences between men and women in the baseline plasma renin activity, as will be reported in detail elsewhere.
The question of whether the patient's age could be a determinant of the response to some antihypertensive drugs has often been discussed.14 15 16 In line with the results of other studies in large patient populations9 10 we did not observe a superior antihypertensive effectiveness of the ß blocker atenolol in younger patients. Differences relevant to treatment between younger and older patients were found only with hydrochlorothiazide and nitrendipine. These two drugs seem to be more effective in elderly patients (fig 1), as has been reported by other authors.12 16 There has been much speculation about the apparently enhanced effectiveness of diuretics and calcium channel blockers with advancing age, possibly because of differences in the activity of the sympathetic nervous system,17 plasma renin activity,16 sodium turnover,17 and age related differences in pharmacokinetics. On the basis of our data (not presented here) we cannot say that differences in plasma renin activity between subgroups could be responsible for this apparent enhancement as plasma renin activities were similar in older and younger patients.
Age may also be an important determinant in the assessment of tolerability. With nitrendipine, withdrawals because of adverse events were much less common in elderly patients (5 v 23), whereas in the atenolol group 8 out of 11 withdrawals were in the older subgroup. Consequently, we conclude from our data that in elderly patients nitrendipine is as effective as the three other antihypertensives, as effective as in patients with higher blood pressure, and better tolerated than it is in younger patients.
During the first eight weeks of treatment, atenolol was superior to all other drugs used in this trial, although during the long term treatment phase of the study the difference from enalapril was no longer significant (table 3). There are only a few studies in which three or more antihypertensive drugs derived from different classes have been compared during long term treatment.9 10 11 18 In these studies the ß blocker proved to be at least as effective as the reference drugs in lowering blood pressure in white patients, the only race we investigated. In our investigation the extent of superiority of ß blocker treatment depended on the patient's age, sex, and initial blood pressure.
We are aware that our conclusions about the comparative effects of the drugs studied are influenced by the selected doses and that different results might have been obtained with different doses. On the other hand the doses chosen correspond to the recommendations of the selling companies and are in keeping with the anatomical therapeutic chemical classification index19 of the World Health Organisation, which includes “defined daily doses” for plain substances.
Moreover, with regard to the possibility that different pharmacodynamic properties could be responsible for the outcome of this study, we decided that with the second increase in dose (step 3) every drug should be given twice daily, although all of these drugs are recommended for once daily application. It should also be considered that blood pressure and pulse rate were measured as close as possible to 24 hours after the last medication during once daily treatment and as close as possible to 12 hours during twice daily treatment.
One important methodological objective of this study was to measure the blood pressure as objectively as possible and to perform those measurements in a way that would allow some verification even after they had been taken. For this purpose all participating centres used the same automatic instrument, which measured blood pressure in preset intervals and printed out the results with the date and the time of day (see methods).
With this method of objective and verifiable measurement we were unable to detect any relevant reduction in blood pressure during the placebo run-in phase, in full accord with the results of our previous VERDI study,12 in which we used the same automatic blood pressure instrument. These results suggest that with this technique there is little if any placebo effect, in keeping with the experience with 24 hour ambulatory blood pressure measurements.20
Our results do not provide any evidence for a superior antihypertensive effectiveness or superior tolerability of the new classes of antihypertensive products—namely, calcium channel blockers and angiotensin converting enzyme inhibitors. As these new drugs are now widely accepted as treatments of first choice our results further emphasise the need for research to confirm that they do reduce morbidity and mortality, as has been shown for diuretics and ß blockers.
We thank Dr U Philipp, Essen, for her invaluable help in the supervision of the study secretariat.
Funding: Bayer, Boehringer, Ingelheim. MSD was funded by Zeneca and ICI and the former Bundesgesundheitsamt.
Conflict of interest: None.
The members of the HANE Study Group were:
Advisory board—R Bestehorn, Munich; O Fischbach, R Klan, Leverkusen; J Rosenfeld, T Rosenthal, Tel Aviv; J Scholze, Berlin; R v Lucke, Ingelheim.
Investigators from Germany—P Angersbach, H P Nast, Offenbach; V Arnold, Dortmund; K J Aumann, Kamp-Lintford; K Baumann, A Matthiesen, H Richter, Hamburg; K Bischoff, W Dutz, Postdam; K Bornikoel, C Credner, Mönchengladbach; C Burck, G J Meyer, Kiel; M Claus, W Kreuzenbeck, F Pietruk, C Plogmann, R Wegehaupt, Essen; L Djurdjevicz, Eschwege; K D Dück, Erfurt; E Dundendorf, Frankfurt; K Eckard, H Seibold, Moers; D Franzen, Waldshut; R Godau, Fulda; R Hempel, Laatzen; G Hilgenstock, I Kirchhof, D Kraft, J Scholze, Berlin; B Hinkelmann, H D Klimm, Kuppenheim; J Kanziora, Troisdorf; E Keidl, Geisenheim; M Klüglich, Munich; M Kramar, Cologne; D Kroll, Hamm; M Le Claire, Solingen; G E v Mannteuffel, M Stafunsky, Marburg; A Müller, Klingenmünster; K Osiecki, Lahnsrein; M Pfich, Nuremberg; W D Pfund, Lauf; A Philippi, Bammental; H Pichotka, Ihringen; H P Rischke, Obertshausen; F Schardt, Würzburg; B Schmaltz, Bingen; G Stein, Jena; K Zenker, N Dammann, Görlitz; G Ziegler, Stuttgart.
Invesigators from other countries—R Fidel, J Rosenfeld, T Rosenthal, S Yodaft, Tel Aviv, Israel; A Graf, Zurich, Switzerland; H Hitzenberger, D Magometschnigg, Vienna, Austria; F Paran, Jerusalem, Israel.