A month ago, results of the largest hypertension trial (ALLHAT) ever conducted were published. Researchers randomized over 30,000 patients with hypertension (plus one or more other risk factors) to chlorthalidone, amlodipine or lisinopril and followed them for approximately 5 years. Overall, this trial showed, at least for the primary outcome (combined fatal CHD or non-fatal MI), that there was no difference between the three representatives of the drug classes studied. A number of secondary outcomes did, however, favour patients treated with thiazide diuretics. In particular, patients randomized to chlorthalidone had fewer strokes, less combined CVD, and less angina when compared to patients receiving lisinopril and less heart failure when compared to amlodipine.
The ALLHAT Officers and Coordinators for the ALLHAT Collaborative Research Group. Major outcomes in high-risk hypertensive patients randomized to angiotensin-converting enzyme inhibitor or calcium channel blocker vs. diuretic: the Anti-hypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT). JAMA. 2002;288:2981-97.
Overall Study Question
These investigators compared chlorthalidone, amlodipine and lisinopril for the treatment of hypertension. The primary outcome was combined fatal coronary heart disease (CHD) or non-fatal myocardial infarction. Secondary outcomes were all-cause mortality, stroke, combined CHD (primary outcome, coronary revascularization, or angina with hospitalization), and combined cardiovascular disease (combined CHD, stroke, treated angina without hospitalization, heart failure, and peripheral arterial disease).
33,357 patients with hypertension and 1 or more risk factors. Their mean age was 67. 47% were women, 36% had diabetes, 25% had a history of heart disease, 22% were smokers, and 12% had an HDL < 0.9 mmol/L.
Patients were randomized in a blinded fashion to receive either chlorthalidone, amlodipine or lisinopril. Second line therapy was atenolol, clonidine or reserpine.
The mean duration of follow up was 4.9 years.
Blood pressure differences at 5 years
Mean systolic blood pressures – amlodipine 0.8 mmHg higher than in the chlorthalidone group; lisinopril 2.0 mmHg higher than in the chlorthalidone group
Mean diastolic blood pressures– amlodipine 0.8 mmHg lower higher than in the chlorthalidone group, lisinopril no difference compared to the chlorthalidone group
* p <0.05 lisinopril vs. chlorthalidone
** p <0.05 lisinopril vs. amlodipine
*** p = 0.05 lisinopril vs. chlorthalidone, p = 0.06 amlodipine vs. chlorthalidone
# p = 0.06 amlodipine vs. chlorthalidone
Diabetic subset – Compared with Chlorthalidone – absolute numbers not provided
More stroke, more combined cardiovascular disease and more heart failure with lisinopril
More heart failure with amlodipine
How does this study contribute to the drug therapy of hypertension?
This study suggests that, in terms of the the primary outcomes evaluated, there were no differences between chlorthalidone, amlodipine and lisinopril when used for the the treatment of hypertension. However, chlorthalidone did produce a greater reduction in a number of secondary outcomes when compared to amlodipine and lisinopril. This study confirms the abundance of previous other evidence that thiazide diuretics are the preferred first choice agents for hypertension in patients with hypertension including those patients with diabetes.
Interestingly, chlorthalidone was also more effective in the diabetic patient subset, which comprised 36% of the cohort. Diabetic patients randomized to lisinopril had more strokes, more combined cardiovascular disease and more heart failure when compared to patients receiving chlorthalidone. Unfortunately, absolute numbers were not provided in the most recent report.
The strengths of this trial include its design (randomized, blinded), the diverse patient populations studied (mean age 67; 47% women; 36% diabetics; 25% with a history of heart disease; 22 % smokers), and the overall numbers of patients enrolled (N=33,357).
The results of this trial led the trial investigators and editorialists to suggest that “diuretics …should be the preferred for first-step antihypertensive therapy” and that “the most effective therapy was also the least expensive [diuretic]”. (1,2) Given that this type of trial will likely never be repeated, this is now the best evidence we have when comparing different classes of antihypertensive agents and their impact on clinically important outcomes. So what should clinicians do with these results? Well, let’s first look at what has happened over the last twenty years or so to antihypertensive prescribing patterns.
1. How have prescribing habits for hypertension changed over the last 20 years?
Over the last two decades, the use of thiazide diuretics for the treatment of hypertension has decreased. In 1982 in the US, thiazide diuretics accounted for 52% of all prescriptions, but by 1993 thiazide diuretic use had declined to 27%. The use of beta-blockers decreased from 20% to 13% over this time period while calcium channel blockers (CCBs) and ACE inhibitors increased from 0.3% to 27% and 0.8% to 24%, respectively. In another report from the US, of all antihypertensive prescriptions from 1992 to 1995, CCBs increased from 33% to 38%, ACE inhibitors increased from 25% to 33%, beta-blockers decreased from 18% to 11%, and thiazide diuretic use decreased from 16% to 8% during that period. In Canada, similar trends have been seen. In a study from Halifax involving hypertensive patients who were surveyed in 1985, and then again in 1995, an increase was seen in the use of CCBs (from 2% to 20%) and ACE inhibitors (from 5% to 25%), while the use of beta-blockers remained unchanged at 22%. The proportion of patients receiving thiazide diuretics however, decreased from 31% to 17% over that ten-year period. In 969 patients who visited an Edmonton clinic from 1993-1995, the initial therapy prescribed to patients with hypertension was ACE inhibitors (34% of total), CCBs (17%), beta-blockers (12%), and thiazides (11%). In 1999, in an outpatient clinic in India, thiazide diuretics accounted for only 13% of prescriptions compared to 47% for beta-blockers, 34% for calcium channel blockers, and 30% for ACE inhibitors. In 1996, 1,200 American physicians were asked what drug they would use as initial therapy in a white patient between the ages of 40-60. Forty-four percent chose ACE inhibitors, 19 % chose beta-blockers, 15% thiazide diuretics and 14% chose CCBs.
In summary, it appears that the use of thiazide diuretics as initial therapy for hypertension has diminished over the past several years.
2. How have guidelines for the initial therapy of uncomplicated hypertension changed over the years?
A synopsis of some of the published guidelines for the initial therapy of uncomplicated hypertension is provided in the table below. (9-17)
Despite the decrease in the use of thiazide diuretics, most guidelines still, as they have always done, suggest that thiazide diuretics can be first choice for the treatment of hypertension. I believe clinicians who develop guidelines should attempt to provide more definitive guidance. A clear statement should be made as to which drug to use as initial therapy. For instance, how can two or even four classes of drugs be considered “the initial therapy of choice”? They should review all the evidence and come out with a clear statement as to which drug should be started first. If there are no differences in efficacy or side effects, then the least expensive agent should be chosen first.
3. What have some “experts” said about the recommendation of thiazide diuretics as first-line therapy for the treatment of uncomplicated hypertension?
In regard to the recommendation of the Joint National Committee (JNC) V consensus guidelines that thiazide diuretics and beta-blocker be first line therapy, the following are some of the comments made by hypertension “experts”. As you will see, these sorts of comments and expert opinion may be one of the reasons thiazide diuretic use has fallen over the last 10-20 years.
Two “experts” have suggested that the “JNC V does not present scientifically or clinically sound approaches to the management of hypertensive patients”.
In 1994, the Executive Council of the American Society of Hypertension convened a group of seven highly skilled hypertension experts to find out their individual strategies for the most effective management of hypertension and, “not a single one of the seven used as initial treatment the drugs (diuretics and beta-blockers) preferred by the JNC”. One of these 7 “experts” suggested that adopting thiazide diuretics and beta-blockers as “preferred drugs would result in sub optimal hypertensive therapy for a strong majority of hypertensive patients”. Another one of these “experts” in their concerns with the JNC V guidelines went as far as to say that it “might have been wiser to ignore trials and concentrate on well-defined hard pathophysiological endpoints”. And finally, one “expert” provided a somewhat interesting statement when he said “it is very surprising that the current JNC V is taking a large step backward in promoting diuretics and beta-blocker as optimal first steps in therapy. The rationale for this disappointing opinion of the JNC appears to be based on their interpretation of hypertension clinical trials”.
4. What is the evidence base for the selection of agents other than thiazide diuretics as initial therapy for hypertension?
The primary reason we treat hypertension is in an attempt to decrease the likelihood of a patient developing clinical endpoints such as heart attacks, strokes etc. Therefore, it would stand to reason that drug selection should primarily and first be on the basis of evidence that looks at changes in these outcomes. Prior to 1997, anyone suggesting the use of drugs other than thiazide diuretics or beta-blockers was not making these recommendations on the basis of any published evidence from trials examining the impact of these drugs on these particular clinical endpoints. This is because the first large randomized trials looking at the cardiovascular effects of CCBs and ACE inhibitors in the treatment of hypertension were not published until 1997 and 1999, respectively. (22,23) We now also have a trial that looks at the use of an angiotensin receptor blocker for the treatment of hypertension. (24) In other words, the change in prescribing habits for hypertension over the years appears to have been based on relatively small trials that only looked at surrogate markers (e.g. blood pressure, albuminuria, and creatinine changes) – not on more relevant clinical endpoints. Given that until 1997 there was little, if any published evidence to support the use of these agents for hypertension, a lot of experts who were convinced agents like CCBs and ACE inhibitors should be used as first line resorted to justifications other than efficacy (reduction in clinical endpoints), tolerability, convenience and cost to promote CCBs and ACE inhibitor use. Here are some examples:
- “Diuretics don’t reduce left ventricular hypertrophy”Increased left ventricular hypertrophy (LVH) has been shown to be a risk factor for cardiovascular disease. In the early 1980’s it was widely reported by some hypertension experts that one of the disadvantages of thiazide diuretics was that they don’t decrease LVH.
- RealityThere have been two definitive trials conducted which examine the effect of antihypertensives on this surrogate endpoint. The first trial designed to compare a number of different classes of antihypertensive agents head-to-head (thiazide diuretics/beta-blockers/ACE inhibitors/CCBs/alpha-antagonists) was published in 1995. These investigators found that all classes produced reductions in left ventricular mass over the 4 year study period. They also found however, that mean left ventricular mass changes were greatest in the chlorthalidone treated patients and this was statistically significantly greater than that observed for those patients receiving ACE inhibitors. In another trial involving a comparison of six antihypertensive agents, thiazide diuretics were shown to be as good as any other class of antihypertensive at reducing left ventricular mass.
- “Diuretics increase cholesterol and glucose and decrease potassium and this is why they may not reduce coronary heart disease” Prior to ALLHAT, experts have suggested that because of the metabolic effects of thiazides on glucose and lipids “a larger proportion of the hypertensive population should have the advantage of being treated with CCBs, ACE inhibitors, or alpha-blockers than is the case today”. Some “experts” have suggested that these metabolic changes “pose an additional risk for some” patients and they have “concerns about their potential dangers”.
- Reality for cholesterol
When patients receive a thiazide diuretic, the average increase in cholesterol has been reported to be less than 0.3 mmol/L and these changes appear to diminish after 6 to 12 months. In the ALLHAT trial, mean cholesterol levels at 4 years were statistically higher in the thiazide diuretic group than the other treatment groups, but the mean difference was only 0.025-0.05 mmol/L (1-2 mg/dL).
- Reality for glucoseIn the ALLHAT trial, fasting glucose at 4 years was statistically different between the lisinopril and the chlorthalidone group, but again the difference was small 0.2 mmole/L (5 mg/dL). (1) Overall, 4 % more patients in the chlorthalidone group had a fasting glucose greater than 6.2 mmole/L (126 mg/dL) at 4 years. Similar changes were seen in both the diabetic and non-diabetic patients.
Reality for potassium
Thiazide diuretics do lower potassium. In the ALLHAT trial, potassium was, on average 0.3 to 0.4 mEq/L lower in the thiazide group and approximately 9% of patients in the chlorthalidone group had a potassium level less than 3.5 mEq/L at 4 years compared to 1-2% in the amlodipine and lisinopril groups.
Have these metabolic differences had an impact on clinical outcomes?
To some degree yes – but not in the direction that would be predicted by believers that these metabolic differences would be harmful. Even though thiazide diuretics produce these so-called “metabolic abnormalities”, all trials prior to ALLHAT showed that thiazide diuretics were at least as effective as other agents at reducing clinical endpoints. In ALLHAT thiazide diuretics were better than amlodipine and lisinopril for some secondary outcomes. Similar results were even found in the diabetic patient subset; a patient population that many have suggested ACE inhibitors are the initial drugs of choice. The bottom line, as shown by the ALLHAT trial and others, is that the small changes in cholesterol, blood glucose and potassium caused by thiazide diuretics have not led to a reduction, compared to other agents, in clinical benefit.
- “Thiazides cause gout”
In one of the only studies to provide an estimate of the frequency of initiation of anti-gout therapy secondary to the use of thiazides, investigators suggested that it was less was than 1 per 100 person years of exposure (when compared to patients receiving non-thiazide antihypertensives. These investigators also stated that in patients receiving doses less than 25 mg/day there was no increased risk for the initiation of anti-gout therapy. Unfortunately, the ALLHAT trialists did not provide any further information about the development and treatment of gout in this study. Hopefully, this will appear in further publications.
- “ACE inhibitors and CCBs have an important impact on other mechanistic issues of hypertension compared to older agents”
Regression of small artery remodeling and endothelial dysfunction, reduction in oxidative stress in the vascular wall, a decrease in cell migration growth, and improvement in endothelial dysfunction have all been touted as benefits of ACE inhibitors and CCBs. Yet, as one set of reviewers commented “there is a quantum leap from mechanistic studies to outcome trials”. (37) Clinicians should use trials evaluating hard clinical endpoints rather than adopting new therapies on the basis of potentially interesting mechanistic differences. The ALLHAT results suggest that thiazide diuretics, no matter what the mechanism, are at least as effective as any of the other classes of antihypertensive drugs.
5.Were diuretics compared to other drugs in large trials before the publication of ALLHAT?
The following table presents results of some of the larger trials that compared thiazide diuretics to other agents and were published prior to the release of ALLHAT. (Green favours diuretic or diuretic/beta-blocker, yellow indicates no difference, red favours the drug compared to diuretics/beta-blockers.
In addition to the results of these individual trials, prior to the publication of ALLHAT, two meta-analyses of all the trials comparing the different classes of antihypertensives were published. One of them came to the conclusion that “all antihypertensive drugs have similar long-term efficacy and safety”. Another meta-analysis came to somewhat similar conclusions, “in the trials comparing ACE-inhibitor-based regimens with diuretic-based or ß-blocker-based regimens, …there were no detectable differences between randomised groups in the risks of any of the outcomes studied” and “in the trials that compared calcium-antagonist-based regimens with diuretic-based or ß-blocker-based regimens, … patients assigned calcium-antagonist-based therapy, there was a significant 13% [0.6% absolute difference] lower risk of stroke than among those assigned diuretic-based or ß-blocker-based therapy. Additionally, there was a 12% [0.6% absolute difference] greater risk of coronary-heart-disease events of borderline significance among those assigned calcium-antagonist-based therapy. There were no significant differences between randomised groups in the relative risks of heart failure, major cardiovascular events, cardiovascular deaths, or total mortality”.
6. So what should you do differently now that the results of ALLHAT are out?
Nothing. That is, not if you were practicing evidence-based pharmacotherapy all along. That is why I entitled this editorial “ALLHAT, So What?”. Despite ALLHAT being the largest hypertension trial ever published, the results really should not change what you should have been doing. ALLHAT simply re-affirms the previous evidence. In my opinion, based on all the information I have seen, thiazide diuretics are, and have always been the first class of drugs that should be tried. If they are ineffective or not tolerated, other classes of agents would then be selected. And yes, that even applies to patients with Type 2 diabetes. The only possible exceptions where thiazide diuretics would not be first line would be for those patients with a concurrent symptomatic condition that could be managed (along with their blood pressure) with one drug (e.g. beta-blocker for essential tremor, migraines, ACE inhibitors or beta-blockers post-myocardial infarction or for congestive heart failure).
In conclusion, if you have been following the evidence and guidelines that supported the use of thiazide diuretics as first line therapy, good for you. If you weren’t, it’s time for you to change. What a great resolution for the New Year.
James McCormack, B.Sc (Pharm), Pharm.D.
Associate Editor, Journal of Informed Pharmacotherapy