ASCOT: a tale of two treatment regimens
BMJ 2005; 331 doi: https://doi.org/10.1136/bmj.331.7521.859 (Published 13 October 2005) Cite this as: BMJ 2005;331:859All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
EQM, Evidence Quality Methods in Primary Care
SIMG - Società Italiana di Medicina Generale
A CRITICAL APPRAISAL
“In God we trust, everyone else has to have data”: the necessity of an
evidence based medicine, based on strong and rigorous data, is a common
patrimony of modern western medicine.
Nonetheless this necessity seems to be a wish, a trend rather than a
reached goal.
“ASCOT” is probably a confirmation of this impression.
ASCOT is a well-conducted trial with an high methodological quality: a
superiority trial, built to test the hypothesis an antihypertensive
therapy based on amlodipine+perindopril is more effective than
atenolol+bendroflomethazide to reduce incidence of coronary death and non
fatal myocardial infarction (primary outcome).
A correct randomization gave two perfectly comparable groups at the
starting point with respectively 9639 and 9618 patients. Sample size was
correctly tailored to primary outcome assuming an HR of 0,84, a study
power of 80% (=0,20) at a two-sided significance level of 5%
(=0,05). All statistical analysis techniques were correct
(“intention to treat” analysis, log-rank procedure and Cox’s proportional
hazards model).
Even if the study was “double open”, outcomes detection was blind, saving
methodological quality.
Follow-up was well conducted, with a low drop-out rate (1,51%).
Transferibility to real practice is fair in a western population, despite
of 77% of male.
The trial was halted after a mean follow-up of 5,5 years because of
secondary outcomes’ (particularly all-cause mortality) excess in control
group (atenolol + bendroflumethiazide).
At the end of the study incidence of primary outcome (coronary mortality +
non fatal myocardial infarction) was not significantly different between
the two groups (HR=0,90 with CI 95% 0,79-1,02).
There was a significant difference between groups, against “atenolol”
group, in many secondary outcomes (all-cause mortality, cardiovascular
mortality, stroke, total cardiovascular events and procedures, non fatal
MI excluding silent IM + fatal CHD). Also the analysis of secondary
combine outcome “total cardiovascular events and procedures” in 18
subgroups shows a significant higher incidence of events in atenolol
group.
Authors conclude that a therapy based on amlodipine+perindopril is more
effective than a therapy based on atenolol+bendroflumethazide, because it
reduces the incidence of total cardiovascular events, total mortality, and
diabetes developing risk (tertiary outcome); therefore they recommend not
to prescribe as first line blocker-diuretic, if it is possible to
prescribe amlodipine+perindopril.
It is exactly here that our “EBM-wishes” crash into reality: authors built
their conclusions on data regarding secondary outcomes and emphasize
subgroups analysis of a secondary and composite outcome, even if it is (or
“should be” by this time) a common patrimony to consider methodologically
correct only conclusions based on primary outcomes analysis.
This is the reason because FDA requires randomized and well-conducted
trials in which statistical power is tailored to a primary (and not
secondary) outcome to authorize marketing of new drugs.
Methodology teaches us that sample size is tailored to the primary outcome
(1): once chosen primary outcome, the whole statistical power of the trial
is spent in the analysis concerning it. Secondary outcome analyses and
subgroups analyses should be consider “value added” to support conclusions
of primary outcome analysis or at least to produce working hypothesis.
Multiple comparisons inside the same sample presents two risks:
a) a higher probability to detect differences between groups where there
are no difference ("false positive", Type I Error).
b) a frequent undersize of the number of subjects evaluated, with an
higher probability not to detect differences where there are differences
("false negative, Type II error).
So, to analyse many secondary outcome (even statistically significant)
expose to the danger that results detected represent a pure effect of
chance: this danger increase with the number of comparisons applied to one
sample. In the original sample ASCOT-BPLA authors made 16 comparisons in
secondary outcome analysis and 18 comparisons in 9 categories of
subgroups.
Secondary outcomes (composite and not-indipendent) analyses in 62,5%
(10/16) of the cases has inadequate sample size to demonstrate the
differences they detected.
The authors’ choice to base their conclusions on secondary outcome and not
on primary outcome - that is hidden and minimized - is what Moyè calls
“'to lock the crazy aunt in the attic” (2).
Authors in 18 subgroups emphasized results of a secondary large outcome,
“total cardiovascular events and procedures”: it is a skilful trick to
shift results towards an increase of effects’ dimensions (3). Moreover
they do not report results of primary outcome analysis in the same 18
subgroups, even though they declare they made it: did it end in the attic
as the aunt?
Probably a correct reading of the trial should only answer to this
question: “Was the primary hypothesis of the trial satisfied or not
(superiority of treatment based on amlodipine+perindopril vs
atenolol+bendroflumethazide in terms of reduction of incidence of coronary
deaths and non fatal myocardial infarction)?”
On the basis of results reported the answer is “Not”, because values of
CI95% of HR for primary outcome were 0,72-1,02.
You know...Data, by means of torturing, at the end admit also they did not
commit.
REFERENCES
1)-Freemantle N. “Interpreting the results of secondary end-points and
subgroup analyses in clinical trial: should we lock the crazy aunt in the
attic?” BMJ 2001 322:989-991
2)- Moyè LA “End-point interpretation in clinical trial: the case for
discipline” Control Clin.Trials 1999 20:40-9
3)- Sackett DL “ Why randomized controlled trials fail but needn’t:
failure to employ phisiological statistics, or the only formula a
clinician-trialist is ever likely to need ( or understand )” JAMA 2001
165(9): 1226-1237
Competing interests:
None declared
Competing interests: No competing interests
McDougall et al,1 in their commentary on the recent ASCOT-BPLA study,2 unfairly conclude that thiazides should be relegated to an adjunctive rather than first-line role in the treatment of hypertension.
As they say, ASCOT compared amlodipine +/- perindopril against atenolol +/- bendroflumethiazide, finding better outcomes for the former combination. This comparison may have been reasonable at the time that the study was planned, but in recent years, a meta-analysis of studies on atenolol in hypertension has clearly shown that it performs poorly in the prevention of cardiovascular endpoints.3 Caution is therefore needed in interpreting these results.
Imagine, if you will, a comparative study of wines that compared a blend of two fine New World pinots with a blend of an excellent red Burgundy and a larger quantity of vin ordinaire. Clearly, the French wines would suffer in comparison, but to suggest that this proved the general superiority of New World wine would be to unfairly malign the Burgundy.
In the context of current hypertension literature, the study design of ASCOT is similarly unfair on the thiazides. When studied on their own, thiazides prove their worth: the large and well-designed ALLHAT trial showed that a thiazide was the equal of amlodipine or an ACE inhibitor.4 ASCOT does add to the existing evidence that atenolol is a poor first-line choice for uncomplicated hypertension, but thiazides should remain the agent of first choice for many.
References
1. McDougall C, Brady AJB, Petrie JR. ASCOT: a tale of two treatment regimens. BMJ 2005;331:859-860.
2. Dahlof B, Sever PS, Poulter NR, Wedel H, Beevers GD, Caulfield M, et al. Prevention of cardiovascular events with an antihypertensive regimen of amlodipine adding perindopril as required versus atenolol adding bendroflumethiazide as required, in the Anglo-Scandinavian cardiac outcomes trial-blood pressure lowering arm (ASCOT-BPLA): a multicentre randomised controlled trial. Lancet 2005;366:895-906.
3. Carlberg B, Samuelsson O, Lindholm LH. Atenolol in hypertension: is it a wise choice? Lancet 2004; 364:1684-9.
4. 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 antihypertensive and lipid lowering treatment to prevent heart attack trial (ALLHAT). JAMA 2002;288: 2981-97.
Competing interests:
None declared
Competing interests: No competing interests
I read with interest the editorial of ASCOT BPLA in the October 15
issue of BMJ. Unfortunately, I must disagree with the authors’
interpretations about old versus new drugs in hypertension. Before we
classify all old drugs as bad, let’s remember a recent study, ALLHAT[1],
which compared the ‘old’ drug, chlorthalidone, to two newer drugs,
lisinopril and amlodipine. After almost five years of follow up,
chlorthalidone provided blood pressure control and cardiovascular
protection that was as good as or better than newer agents. The findings
of ASCOT do not trump ALLHAT and thiazide diuretics should remain a
mainstay of therapy.
Similarly, we have seen a comparison of atenolol with a new therapy
before. In the LIFE study[2], atenolol was compared with losartan as
initial therapy in patients with hypertension and LVH. Compared to
atenolol, the absolute risk reductions in favour of the ‘new’ therapy were
very similar to those observed in ASCOT. For example, the absolute risk
reduction (ARR) in total mortality was 1.1% in LIFE versus only 0.8% in
ASCOT; for death due to a cardiovascular cause, the ARR was 0.7% versus
0.9%; for stroke, the ARR was 1.7% versus 1.0% respectively. Despite
equal or greater absolute differences in these endpoints, only the stroke
endpoint reached statistical significance in LIFE, whereas in ASCOT, all 3
endpoints reached statistical significance on their own. The statistical
differences in ASCOT were undoubtedly due to its power from a sample size
at least twice that of LIFE.
In the absense of compelling indications, guidelines from both
Canada[3] and the UK[4] discourage using beta-blockers as initial therapy
in elderly patients such as those found in LIFE, mean age 67, and ASCOT,
mean age 63. Therefore, it is not clear to me how ASCOT should change
guidelines so drastically, considering we have seen this all before.
1. 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 Antihypertensive and Lipid-Lowering Treatment to
Prevent Heart Attack Trial (ALLHAT). JAMA. December 18, 2002
2002;288(23):2981-2997.
2. Dahlof B, Devereux RB, Kjeldsen SE, et al. Cardiovascular morbidity and
mortality in the Losartan Intervention For Endpoint reduction in
hypertension study (LIFE): a randomised trial against atenolol. The
Lancet. 2002;359(9311):995-1003.
3. Evidenced Based Recommendations Task Force of the Canadian Hypertension
Education Program. 2005 Canadian Hypertension Education Program
Recommendations: The bottom line version. Available at:
http://www.hypertension.ca/recommendations_2005/execsummary2005.pdf.
Accessed March 31st, 2005.
4. Williams B, Poulter NR, Brown MJ, et al. British Hypertension Society
guidelines for hypertension management 3004 (BHS-IV): summary. BMJ.
2004;328:634-640.
Competing interests:
None declared
Competing interests: No competing interests
I see that the authors note that not all antihypertensive agents are
equal, but what about the class effect. Are all beta-blockers the same?
What about the newer beta-blockers? Should there use be discouraged as
first line?
Many prescribers are using bisoprolol as a first line agent. What
should we advise?
Thanks
Competing interests:
None declared
Competing interests: No competing interests
I agree with messers Sekkides and Spence. Where in the ASCOT trial is
the proof that angiotensin II blockers are not worse (ie better) than the
"older" agents at treating hypertension? So is it just me that feels the
conclusion to this editorial, about using angiotensin II blockers as
possible first line treatment, may smack of "big pharma" influence?
Competing interests:
None declared
Competing interests: No competing interests
I hope the BHS and NICE take good notice of the rapid responses here.
I hear many saying thiazide first and then probably an ACE, but above all
strive to find something which suits the patient and brings down the blood
pressure.
These seem practical resposes by the practical people who are looking
after normal people with blood pressure. They are mostly GPs (but also
thanks to Neal Maskrey for a useful critical summary).
Could we have a less "grasshopper" responses to a new trial? I seem to
remember Richard Lilford asking for a more Bayesian approach some years
ago.
Could we also (please Mr Editor!) ask for NNTs in all papers? Lots of us
have done it on this paper; it is the only way to make sense of the
relative risks they all quote.
Competing interests:
None declared
Competing interests: No competing interests
McDougall, Brady and Petrie's overview of the ASCOT trial was
succint, well written but clearly misinformed. The article infered GPs
were largely to blame for the excess hypertension related deaths in the
UK, a comment which only served to widen the hospital/GP gap and perhaps
raise a few eyebrows amongst community doctors.
So let's set the matter straight. I agree, management of the
hypertension burden does largely fall on the GP. In order for the blood
pressure to be controlled, the patient has to visit their doctor on a
number of occasions, change their lifestyle, take medication on a regular
basis and adhere to follow up. Personal choice not to take medication,
side effects and continuing adverse lifestyle behaviours are common
reasons why GPs can not manage to treat all patients with hypertension and
reduce cardiovascular risk.
Also in reference to the article, not all PCTs offer prescribing
incentives. Admittedly, the quality outcome framework (QOF) provides an
incentive to be agressive in hypertension management and the evidence as
to whether this will impact on cardiovascular disease is awaited.
I agree, there may be GPs who are unsure of the drug management of
hypertension but there are also many who do.
Nevertheless, patient choice and side effect profile may well influence
drug making decisions. I wonder if there is any recent evidence to support
the authors' comment as to whether GPs may be unsure of what drug to use
first?
The message here is simple, before making statements regarding the
working practises and effectiveness of our colleagues, perhaps we should
make an effort to be properly informed and consider issues on the other
side of the fence.
Dr Raj Thakkar, Principal
Competing interests:
None declared
Competing interests: No competing interests
ASCOT-BPLA is a large randomised trial in an important clinical area
and deserves both careful analysis and setting in the context of the rest
of the evidence about the treatment of hypertension. We believe the BMJ's
editorial describing the results and implications of ASCOT-BPLA paints an
extreme interpretation.
What questions was ASCOT-BPLA set up to answer?
The main objective of ASCOT ¡V BPLA was to compare the effects of two
antihypertensive regimens on the prevention of CHD in a relatively high-
risk primary care population with hypertension. The primary endpoint was a
composite of non-fatal MI + fatal CHD; secondary endpoints included all-
cause mortality, cardiovascular (CV) mortality, stroke, heart failure, and
total cardiovascular events + procedures.
Two further, post-hoc combined endpoints were added during analysis
of the results. These were CV mortality + MI + stroke and non-fatal MI +
fatal CHD + coronary revascularisation procedures. The authors give
reasons for these additions, but any post-hoc analyses need to be viewed
with caution, especially in studies such as this where differences in the
primary endpoint did not reach statistical significance.
19,257 patients aged 40-79 years with hypertension and at least three
other cardiovascular risk factors were included and followed up for a
median of 5.5 years. Risk factors included left ventricular hypertrophy,
type 2 diabetes, peripheral arterial disease, previous stroke/TIA, male
sex, >55 years, microalbuminuria or proteinuria, smoking, TC:HDL ratio
>6, family history of premature CHD. Therefore these were people with
hypertension at high risk of an event given their other multiple risk
factors. However, It is not possible to identify a baseline risk fort his
population since there was no placebo group. Those with previous
myocardial infarction (MI), angina, a cerebrovascular event within the
past three months, triglycerides >4.5mmol/l, heart failure,
uncontrolled arrhythmias were excluded.
Patients were mainly white and male, mean age was 63 years, BMI
almost 29kg/m2, TC 5.9mmol/l and BP 164/95mmHg. 80% of patients were
taking previous antihypertensive drugs, but only 11% lipid-lowering drugs
and 19% aspirin - not very high proportions considering the relatively
high risk of the patient population.
A regimen based on a beta-blocker (atenolol) + diuretic
(bendroflumethiazide plus potassium) as required was compared with one
based on a calcium channel blocker (CCB, amlodipine) + ACE inhibitor
(perindopril) as required. Patients were stepped through monotherapy of
either atenolol or amlodipine at increasing doses, then the addition of
either the thiazide diuretic or perindopril, respectively, at increasing
doses, then the addition of doxazosin in order to reach target blood
pressures.
Initial dose titration of atenolol was from 50mg to 100mg. Initial
dose titration of amlodipine was from 5mg to 10mg. With this regimen, the
relatively flatter dose response curve of atenolol (the BNF says 50mg as
the dose for hypertension with 100mg being "rarely necessary") would have
been expected to reduce BP less, and this difference would be expected to
be greater in the first part of the trial. The next titration being to
1.25mg bendroflumethiazide rather than 2.5mg may also have contributed to
the less impressive early BP lowering in the atenolol-based group.
At the end of the trial, 78% of patients were taking at least two
antihypertensive drugs. Only 9% were taking atenolol monotherapy and 15%
amlodipine monotherapy. The average number of antihypertensives used was
2.3 in the atenolol/diuretic group and 2.2 in the amlodipine/perindopril
group. Given the results from the doxazosin arm of ALLHAT , it is
interesting that how many patients were taking doxazosin is not stated in
the paper.
What happened to blood pressure in ASCOT?
BP reduced substantially in both treatment groups from a mean of
164/95mmHg to a mean of 137/78mmHg. This suggests a systematic approach to
managing hypertension with regular follow-up and a stepped approach to
drug treatment (regardless of which drugs are used) can produce results.
However, the intensity of treatment should be guided by patients,
balancing their acceptance of treatment (concordance, polypharmacy, side
effects) with the desired reduction in BP. Even in this trial environment,
only 32% of patients with diabetes and 60% of those without diabetes had
reached the BP targets set by the end of the trial.
BP was reduced more in the amlodipine/perindopril group compared with
the atenolol/thiazide group with an average difference throughout the
trial of 2.7/1.9mmHg. This difference was statistically significant
(P<_0.0001. these="these" differences="differences" were="were" largest="largest" at="at" three="three" months="months" _5.9="_5.9" _2.4mmhg.="_2.4mmhg." this="this" difference="difference" is="is" crucial="crucial" to="to" interpreting="interpreting" ascot.="ascot." it="it" could="could" be="be" that="that" there="there" some="some" factor="factor" in="in" newer="newer" drugs="drugs" means="means" they="they" have="have" advantages="advantages" compared="compared" older="older" _--="_--" as="as" the="the" authors="authors" claim="claim" an="an" accompanying="accompanying" paper="paper" based="based" on="on" complicated="complicated" modelling="modelling" techniques.="techniques." however="however" well="well" established="established" a="a" population="population" basis="basis" very="very" small="small" blood="blood" pressure="pressure" can="can" exert="exert" large="large" effects="effects" cv="cv" events.="events." likely="likely" key="key" driver="driver" of="of" seen="seen" morbidity="morbidity" and="and" mortality="mortality" endpoints="endpoints" but="but" may="may" also="also" confounded="confounded" by="by" other="other" factors="factors" differed="differed" between="between" groups="groups" such="such" patients="patients" mean="mean" bmi="bmi" hdl-="hdl-" cholesterol="cholesterol" levels.="levels." p="p"/> What about the morbidity and mortality endpoints?
There was no difference between the groups in the primary endpoint,
non-fatal MI + fatal CHD. This occurred in 4.5% of the
amlodipine/perindopril group and 4.9% of the atenolol/thiazide group (HR
0.90; 95%CI 0.79-1.02). There were statistically significant differences
between the antihypertensive groups in some secondary endpoints favouring
the amlodipine-based regimen (and the post-hoc endpoints that need to be
viewed with caution). However, all of these differences were fairly small
when viewed in absolute terms rather than in relative terms - a c1%
absolute difference in all-cause mortality, CV mortality, fatal and non-
fatal stroke, and in fatal and non-fatal heart failure.
Total CV events + procedures occurred in 14.1% of the
amlodipine/perindopril group and 16.7% of the atenolol/thiazide group (HR
0.84; 95%CI 0.78-0.90). This is an absolute difference of 2.6% or a number
needed to treat (NNT) of 38 over 5.5 years. CV death + MI + stroke
occurred in 8.3% and 9.7% (HR 0.84; 95%CI 0.76 to.0.92). This is an
absolute difference of 1.4% or a NNT of 71 over 5.5 years. There must be a
real possibility that these important differences just reflect the
differences obtained in blood pressure, which may be, at least in part, a
consequence of the regimens and especially the doses selected. When the
results of ASCOT are set in the context of the rest of the literature of
hypertension, it is grossly over-simplistic to think of drug selection in
terms of "new drugs good, old drugs bad". And in terms of absolute
benefits, the numbers in ASCOT are important but likely to be far less
impressive in a population with hypertension but no or fewer additional
risk factors.
Why was the trial stopped early and what does this mean?
ASCOT was stopped early when interim analysis showed significant
disadvantages for patients on the atenolol-based regime. These were all-
cause mortality 8.5% with atenolol/thiazide and 7.7% with
amlodipine/perindopril (P=0.0247), NNT=115 over 5.5 years and CV mortality
of 3.6% with atenolol/thiazide and 2.7% with amlodipine/perindopril
(P=0.0010), NNT=121 over 5.5 years
Stopping an RCT early creates problems. The power of the study is
calculated on the primary end point -- in this case the composite of non-
fatal MI + fatal CHD. In the case of ASCOT the above findings were
secondary end points.
As with all major trials, endpoint mortality and morbidity events
were reported to a Data Safety Monitoring Board (DSMB), who decide, before
the trial has started, statistical boundaries that, if exceeded, would
require the trial to be stopped early. In October 2004, the DSMB
recommended the trial be stopped as the atenolol/thiazide arm had
significantly higher mortality than the amlodipine arm.
When trials are stopped early, there is a possibility that follow up
will not be long enough for sufficient endpoints to be reached, leaving
the trial underpowered for these endpoints. When ASCOT was set up it was
estimated that at least 18,000 patients would need to be followed up for
an average of five years. This was based on an anticipated 1150 patients
experiencing a primary event in the study overall, a hazard ratio (HR) for
this primary event of 0.84 and a power of 80%. When the trial was stopped,
fewer primary endpoints than this had been reached, 903 in total but the
trial had continued for 5.5 years with 19,257 patients.
The authors of ASCOT argue that this has happened here for the
primary endpoint. This may be the case, but it does not mean that if the
trial had continued and was powered for the primary endpoint that this
would have reached statistical significance. The answer here is - we just
don't know - and we shouldn't be accepting that there was no difference in
the primary event rate just because the trial was stopped early. There may
still have been no difference between the two antihypertensive groups if
the study had continued for its planned duration. On the other hand, it is
unfair to call the trial a negative study. Perhaps it should be described
as a trial that did not produce a positive result for the primary
endpoint.
What about the diabetes issue?
There was a significant increase in likelihood of developing diabetes
in patients on the beta-blocker regimen compared with the amlodipine
regimen (HR 0.70, 95%CI 0.63 to 0.78). The Kaplan-Meier curve suggests an
absolute difference in the proportion of patients who develop diabetes of
about 2% at 5 years (approx. 6% vs 8%).
The clinical significance of such a finding is unclear. An analysis
of the long-term SHEP data found that patients who were taking a diuretic
-based regimen were more likely to develop diabetes (defined according to
raised blood glucose levels) but actually had a decrease in patient-
oriented outcomes (living longer or living better) than those on placebo.
In keeping with ALLHAT and several other trials, ASCOT-BPLA confirms
that a treatment strategy based on a beta-blocker and diuretic can
increase the risk of new-onset diabetes and have other unfavourable
metabolic side effects. This is recognised in the NICE guideline for
hypertension, which suggests the use of an ACE inhibitor, and not a beta-
blocker in combination with a diuretic is the preferred option in those at
high risk of developing diabetes.
What about the side effects and serious adverse events?
25% of patients stopped therapy because of an adverse event, with no
significant difference between allocated treatment groups. Patterns of
adverse events were mostly predictable - joint swelling 14% on amlodipine
vs 3% on atenolol; peripheral oedema 23% on amlodipine vs 6% on atenolol;
cough 19% on amlodipine vs 8% on atenolol; peripheral coldness 1% on
amlodipine vs 6% on atenolol. Erectile dysfunction was 6% on amlodipine vs
7% on atenolol. Safety and side effects are important determinants of
prescribing strategy in individual patients. That 25% of patients stopped
therapy because of an adverse event points us towards an individual-based
policy for drug selection in hypertension rather than a "new drugs good,
old drugs bad" philosophy.
So what does the publication of ASCOT-BPLA mean for practice?
The findings of ASCOT should be carefully considered in the context
of the rest of the evidence on hypertension treatment, which is
substantial in volume and has been accumulated over more than 40 years.
Just because ASCOT is the latest RCT does not mean therapy should now be
based on it. ALLHAT remains the largest hypertension RCT, and arguably has
a more relevant design than ASCOT. NICE recommends a thiazide as first
choice for hypertension. ASCOT provides no direct evidence that the NICE
guideline is inappropriate.
Furthermore, the use of an atenolol-based regimen using a dose not
routinely recommended for the management of uncomplicated hypertension
seriously questions the validity of any results. The alternative BHS
guidelines would advocate a calcium channel blocker or a diuretic as first
choice in people over 55 years or who are black (these two group
comprising the vast majority of people with hypertension in the UK).
Setting aside that fact that the BHS ABCD approach has not been shown to
have morbidity or mortality benefits in large RCTs, if we were to consider
the evidence for ASCOT, ALLHAT, and the Pahor and BPLT meta-analyses of
calcium channel blockers together - could we really say we should select
amlodipine over a thiazide routinely, all other things being equal?
Nevertheless, the study provides considerable new information on the
effect of drug treatment on outcomes that matter to patients with
hypertension. It strengthens the case of those who urge caution in the
widespread use of beta blockers (or at least atenolol) in hypertension.
But whether the NICE or BHS guidelines are followed most people without
compelling co-morbidities should not be receiving atenolol as their first
drug for hypertension. Hence the major thrust of ASCOT is irrelevant in
the context of both UK guidelines.
It is appropriate that the results of ASCOT are considered together
with all the other evidence by NICE in due course. Only if and when
revised guidance is issued, should a revised algorithm for use of
antihypertensive drugs to that recommended by NICE be followed.
So what does the publication of ASCOT-BPLA mean for individual
patients?
When managing patients we believe it is important to consider the
treatment paradox - for individuals, unless the blood pressure is
significantly raised, a few mmHg more or less make very little difference
to that person¡¦s overall CV risk . However, these small differences in
individual risk when multiplied by thousands produce significant
reductions in CV endpoints when these are counted at the population level.
One way of resolving this paradox is to manage patients on an individual
basis, using a choice of regimen which the patient can tolerate, remember
to take regularly and which produces a satisfactory set of blood pressure
readings. It remains important to take the blood pressure precisely using
immaculate technique and a recently calibrated and maintained
sphygmomanometer.
And most importantly - it has very recently been shown than on a
population basis it is possible to reduce CV events by drug therapy, but
the same paper shows conclusively that greater benefits in a population
come from smoking cessation and dietary changes.
Yours,
Neal Maskrey
Medical Director
National Prescribing Centre.
Jonathan Underhill
Assistant Director, Education and Development
National Prescribing Centre.
1. ALLHAT Officers and coordinators for the ALLHAT Collaborative
Research Group. Diuretic versus alpha-blocker as first-step
antihypertensive therapy. Final results from the antihypertensive and
lipid-lowering treatment to prevent heart attack trial (ALLHAT).
Hypertension 2003; 42: 239-246.
2. Staessen JA, Birkenhager WH. Evidence that new antihypertensives are
superior to older drugs. Lancet 2005;366:869-870.
3. Kostis JB et al. Long-term effect of diuretic-based therapy on fatal
outcomes in subjects with isolated systolic hypertension with and without
diabetes. Am J Cardiol 2005; 95: 29-35
4. Pahor M, et al. Health outcomes associated with calcium antagonists
compared with other first-line antihypertensive therapies: a meta-analysis
of randomised controlled trials. Lancet 2000; 356: 1949-1954.
5. Blood Pressure Lowering Treatment (BPLT) Trialists' Collaboration.
Effects of ACE inhibitors, calcium antagonists, and other blood-pressure-
lowering drugs: results of prospectively designed overviews of randomised
trials. Lancet 2000; 356: 1955-1964.
6. Hansson L, et al. Effects of intensive blood-pressure lowering and low-
dose aspirin in patients with hypertension: principal results of the
Hypertension Optimal Treatment (HOT) randomised trial. Lancet 1998; 351:
1755-1762.
7. Unal B, Critchley JA, Capewell S. Modelling the decline in coronary
heart disease deaths in England and Wales, 1981 ¡V 2000: comparing
contributions from primary prevention and secondary prevention. BMJ 2005;
331: 614 - 7.
Competing interests:
None declared
Competing interests: No competing interests
Thiazides have proven themselves time and again to be one of the
safest of anti-hypertensives, and one of the best tolerated; certainly
better than drugs like amlodipine and perindopril(1).
Should we really be thinking of relegating thiazides from first line
treatment to adjunctive therapy on the basis of the ASCOT study?
One glaring omission in the authors of this editorial was of the only
other study to claim a drug regime better than with "older drugs"; the
LIFE study (2). Both studies where remarkably similar in that they took a
particularly high risk group of patients, and they were essentially both
studies of "two regimes"; the LIFE study was even more so than ASCOT.
Both also used a high dose atenolol with add in thiazide as the comparator
regime. The combination was used on 54.9% of days in ASCOT but on 72.4% of
days in the LIFE study.
The study regime in ASCOT was, as we now well know, was amlodipine
with perindopril added in, with the combination used 49.5% of the time. In
LIFE it was losartan with hydrochlorothiazide added in, used 73.9% of the
time.
Both studies showed statistically better results, of a similar
magnitude, with the "newer regime", than against the atenolol based
regime.
The LIFE study gives considerable support to the view that thiazides
still have a central role in the management of hypertension, even in
combination regimes with newer drugs; so much so that following the trial
a combination therapy of losartan and hydrochlorthiazide was brought out
with FDA approval.
The obvious explaination should be that atenolol based treatment is
less effective than treatments where atenolol is not used; but then
Carlberg et al (3) have already told us that.
1. The ALLHAT Officers and Coordinators for the ALLHAT Collaborative
Research Group. Major outcomes in high-risk hypertensive patients
randomised to ACE inhibitor or calcium channel blocker vs diuretic – The
ALLHAT Trial. JAMA 2002; 288: 2981-2997.
2.Dahlof B, Devereux RB, Kjeldsen SE, Julius S, Beevers G, Faire U,
Fyhrquist F, Ibsen H, Kristiansson K, Lederballe-Pedersen O, Lindholm LH,
Nieminen MS, Omvik P, Oparil S, Wedel H. Cardiovascular morbidity and
mortality in the Losartan Intervention For Endpoint reduction in
hypertension study (LIFE): a randomised trial against atenolol. Lancet
2002;359(9311):995-1003.
3.Carlberg B, Samuelsson O,Lindholm LH, Atenolol in hypertension: is
it a wise choice? Lancet 2004;364:1684-89
Competing interests:
None declared
Competing interests: No competing interests
The ASCOT-study – a more balanced view
EDITOR - On the basis of the ASCOT-study findings, an editorial
comment in BMJ recommends “more frequent prescriptions of ACE inhibitors
or angiotensin receptor blockers plus calcium channel blockers as first
line treatments and in combinations”.1 We do not agree that one should
jump to this conclusion.
Was atenolol a wise choice as a β-blocker reference drug?
Certainly not, as some clinical trials have found that atenolol compared
to placebo did not lower the stroke frequency, and was without effect on
cardiovascular mortality.2 Moreover, in spite of similar blood-pressure
reductions, cardiovascular mortality tended to be higher in atenolol-based
treatments than in treatments with other antihypertensive drugs, and with
a raised risk of stroke. 2,3 Other β-blockers have fared better than
atenolol in clinical trials.3
As atenolol seems to have low efficacy with regard to prevention of
cardiovascular events, it should not be used as a reference drug in future
hypertension studies. In addition, when two antihypertensive drug regimens
are compared, it is mandatory that they are equipotent with respect to the
blood pressure lowering effect. That was not the case in the ASCOT-study.
In conclusion, the results of the ASCOT-study do not lend support to
an uncritical endorsement of the newer antihypertensive drugs. Rather,
given some apparent limitations of the comparator drug regimen, the
amlodipine-based treatment did not stand out in a particularly remarkable
way.
Knud Landmark
professor emeritus
Ivar Aursnes
professor
Åsmund Reikvam
professor
Department of Pharmacotherapeutics, University of Oslo, PB 1065
Blindern, 0316 Oslo, Norway
k.h.landmark@labmed.uio.no
Competing interests: KL has received a fee for speeking from LEO
Pharma A/S.
1 McDougall C, Brady AJB, Petrie JR. ASCOT: a tale of two treatments
regimens. Better blood pressure, fewer deaths, and less diabetes with
newer antihypertensive agents. BMJ 2005; 331: 859-60.
2 Carlberg B, Samuelsson O, Lindholm LH. Atenolol in hypertension: is it a
wise choice? Lancet 2004; 364: 1684-9.
3 Lindholm LH, Carlberg B, Samuelsson O. Should β blockers remain
first choice in the treatment of primary hypertension? A meta-analysis.
Lancet 2005; 366: 1545-53.
Competing interests:
KL has received a fee for speeking from LEO Pharma A/S
Competing interests: No competing interests