The ASCOT BPLA study is certainly stirring up the exciting (yawn) world of hypertension management.
At first glance it would appear to be a resounding success for "newer" treatments, which romped home ahead of the "older" ones.
But wait...........................I detect the odour of spin! Could it be that there is another possible explanation for the results? Well, lets wait and see. The Great and the Good of UK hypertensionology are due to gather soon to decide on a way forward. That's assuming they can find a room big enough!
Find out more about ASCOT BPLA here:
http://www.eshonline.org/journal_highlights/2005_set_art1.asp
ps Check out the "Editorial Comment". I hope Robyn did as asked.
And here is another view from No Free Lunch UK:
http://www.nofreelunch-uk.org/ASCOT.doc
1 comment:
The ASCOT trial had two parts: a blood pressure study and a lipid
lowering study. The LL side was reported ages ago and essentially told
us what we know: giving statins to people at high risk of CVD is a good
idea. Unfortunately the BBC item has mixed this up in their reporting
and gives the wrong impression of today's study!
On to the BP study. Briefly about 20,000 adults aged 40-79 who all had
hypertension plus at least 3 of the following risk factors (LVH,
abnormal ECG, type 2 DM, peripheral arterial disease, previous stroke
or TIA, male, age > 55, microalbuminuria or proteinuria, smoker, TC:HDL
ratio of > 6, premature family history of CHD) were randomised to one
of two BP lowering regimes. Those who had had an MI, a history of
angina or a recent stroke were excluded.
This was predominantly a high risk, primary prevention population, but
perhaps about 15% or so would be classed secondary prevention. As there
was no placebo arm we can't estimate the average risk level these
people were at, but the CHD event rate was around 4-5% over 5 years.
Strangely despite these risks only 20% were on aspirin and 10% on lipid
lowering at baseline.
The two regimes were:
Amlodipine based: Start on 5mg then increase through 10mg, 10mg + 4mg
perindopril, 10mg + 8mg perindopril, add doxazosin etc... as needed to
reach the target BP
Atenolol based: start on 50mg the increase through 100mg, 100mg +
1.25mg bendroflumethiazide, 100mg + 2.5mg bendroflumethiazide, add
doxazosin etc as above
78% of patients ended up on both main treatments, 15% stayed on
amlodipine alone and 9% on atenolol alone.
The primary endpoint they set was a combined measure of non-fatal MI
and cardiac death. All the various elements of this were analysed
separately and then after they'd done the trial they decided to add a
new outcome of non-fatal MI, non-fatal stoke and fatal CVD.
Although the populations were balanced at baseline there were a number
of important differences at the end of the study. The amlodipine-based
group had:
Lower average BP (about 2.6mmHg systolic)
Higher pulse rate (11bpm)
Higher HDL (0.1mmol/l)
Lower BMI (0.3kg/m2)
Lower triglycerides (0.3mmol/l)
Lower glucose (0.2mmol/l)
No difference in total or LDL cholesterol
25% in both groups stopped due to adverse effects - no significant
difference overall, but some differences in the actual adverse effect
rates.
The study was stopped early after 5.5 years of follow up. So what did
it find?
Well, in contrast to the media hype, there was NO difference in the
rate of the primary outcome, ie the question the study was set up to
answer. In other words, both regimes were equally effective at
preventing this outcome. There are some differences in the rates of the
individual elements and the post-hoc outcome they defined.
In the discussion the authors argue that the lower than expected event
rate meant that the study was underpowered and that's why it didn't
come out well for the primary outcome. One does now have to ask why it
was stopped early if this was the case. The cynic would also suggest
that the addition of a post-hoc analysis which turned out to be
significant is just a reflection of the fact that the primary outcome
wasn't achieved and they had to go data-dredging for something.
Perhaps the most important outcome of those remaining is the post-hoc
defined one: fatal or non-fatal stroke or MI. You would need to treat
68 people for 5.5 years to avoid one of these with the amlodipine-based
regime rather than the atenolol-based one. A huge 1.5% absolute
difference in event rate, which the authors themselves admit are
"fairly small absolute benefits". Yes this is a 14% reduction and yes
there are 23% reductions in stroke and 15% reductions in MI, but
although these sound big, the absolute rate is low so a big % of a
small amount is still a small amount. Typical media presentation of the
data that leads to misinterpretation.
There was also a lower rate of new onset diabetes in the amlodipine arm
(as we already knew there would be). Whether this is a problem is for
debate elsewhere - the most recent data from long-term follow up of
patients from the SHEP (I think) trial would suggest it isn't.
So what does it all mean? Well, the authors say that the BP differences
(2.6mmHg remember) would account for part of this effect (they estimate
4-8% reductions in MI and 11% in stroke), so about half what was seen.
But the accompanying (independent) editorial from some guys who know
what they're talking about with BP, states that their analysis would
support the fact that the BP difference would explain all of the event
differences seen.
We also have to contend with the other beneficial differences seen in
the amlodipine group: better lipid profiles and lower weight, as well
as a potentially less beneficial / harmful effect of greater doxazosin
use in the atenolol arm (I am assuming this - can't see it stated, but
more patients in the atenolol arm needed add on therapy).
We could also debate the suitability of using 100mg atenolol to control
BP (more likely to cause side effects and not likely to help BP
compared to 50mg). The ALLHAT trial demonstrated that a thiazide was
unsurpassed as a first-line drug for hypertension. ASCOT does nothing
to undermine that as it didn't compare to a thiazide first line.
Interestingly in ALLHAT, a beta-blocker was the first add on treatment
in all groups and didn't worsen the thiazide's effect to show
superiority with an ACEI or CCB there.
So where does that leave us? Basically halving the BP difference views
from the paper and the editorial means that the NNT would be in the
region of 130 even if the outcome was still statistically different -
ie a tiny gain. This itself could still be explained by the lipid
differences seen. So for me, there isn't anything in it at population
level.
So, the bottom line for me is to continue what I'm doing: its all about
lowering the BP to an appropriate goal. We should normally do this by
using a thiazide first and then adding drugs quickly according to BP.
The add-ons should be selected depending on the individual's profile
and risks for things like DM.
I'll be reassuring my patients that their current regimes are well
selected and doing the best they can (unless of course they aren't, but
I'd be changing them anyway if that were the case!).
Let the debate continue, but watch out for LOTS of aggressive pharma
advertising and stuff in the media! I'll finish with a quote from the
editorial which I think could be particularly pertinent...
"Unfortunately, the companion article in today's Lancet, by Neil
Poulter and colleagues, weakens the key message that in hypertensive
patients it is the lowering of blood pressure that produces most of the
benefit, and thereby opens the door for possible misinterpretation, or
even misuse of post-hoc results by drug marketers."
Magnus
--
Magnus Hird
Pharmacist Practitioner, Bloomfield Medical Centre
NPC Trainer, National Prescribing Centre
Post a Comment