My ABCD alternative
GP Dr Gulshan Kaul gives a personal view on why NICE has interpreted the evidence wrongly on hypertension drug choice and suggests an alternative ABCD model
D-Day has finally dawned. Six months after the findings of the ASCOT study were published in The Lancet, NICE has finally issued its draft consultation on hypertension. The reaction has been mixed some GPs are glad to simply get on and implement the new oversimplified ACD algorithm, while others like me are yet to be convinced we should change our practice when the guidance is based primarily on the ASCOT study. This is a trial that would possibly fare quite poorly if subjected to a critical appraisal exercise.
In short, the NICE recommendations mean ABCD becomes AbCd with calcium channel blockers and ACE inhibitors becoming the drugs of choice, while ?-blockers are scrapped as mainstream therapy and the use of thiazide diuretics is sidelined (see diagram on page 48.)
?-blockers unfairly targeted
I am bitterly disappointed that ?-blockers have been relegated to fourth-choice drugs (more so, to be found at the bottom of the pile) in patients <55, given="" the="" fact="" that="" the="" committee="" admits="">55,>
·There is a paucity of trial evidence in younger individuals and those with 'low risk'. Most of the trials in the analysis
including ASCOT looked at results in
patients deemed high risk, therefore most of the evidence derived from studies in
older people, making the results less reliable in the younger age group.
·Assumptions were made that may
influence/bias the results, for example: 'treatment effects are attributable to first-line drugs (why have a ?-blocker as fall guy? How many patients were on doxazosin?)
·The committee admits another
assumption is trials may have used 'more or less' effective treatment protocols (ASCOT-BPLA with 25mg atenolol + bendroflumethiazide 1.25mg which would naturally lead to a flatter dose response curve).
NICE looked at all evidence during the period July 2004-December 2005. But the only significant trial which has clearly had a major impact is ASCOT. Some of the others included PHYLLIS (Plaque Hypertension and Lipid Lowering Italian Study) which had a small power of around 250 and looked primarily at carotid atherosclerosis progression rather than BP lowering, and the side study JMIC-B (Japan Multicentre Investigation for Cardiovascular Disease-B), once again with a small power of around 825, patients unrepresentative of the UK population and a different primary end point.
Furthermore, some of the comparisons in the analysis appear to portray ?-blockers in a negative light. For example, although they comment on the increased protection against stroke with calcium antagonists (which may be due to a simple biochemical fact that ?-blockers like atenolol do not penetrate the blood brain barrier), they fail to mention the meta-analyses by BPLTTC2 which pointed out that CCBs were associated with a 12 per cent increase in coronary heart disease (INSIGHT, NICS-EH, STOP2, NORDIL, VHAS) with similar conclusions drawn by Pahor et al in 20007.
We are therefore left with ASCOT as the dominating influence and I have several
issues with this:
·How can an investigator-initiated and
investigator-led trial that was not double-blinded be completely reliable?
·If the investigators were already aware of the increased risk of new-onset diabetes in the atenolol and bendroflumethiazide group (as alluded to in The Lancet), was it
ethically appropriate to trial and hence
further disadvantage this group? Was
informed consent obtained?
·We are all aware of the shortcomings of a stepped-care approach and interpretation of results in trials. Certainly, with ASCOT, the 1.25mg dose of bendroflumethiazide and 100mg of atenolol would never compare equitably with a doubling of amlodipine to 10mg and of perindopril to 8mg. Furthermore, it has come to light that some patients in the atenolol arm may well have been initiated on a 25mg dose, which would weaken/
invalidate the results of the study.
·Cardiovascular risk estimation itself is far from an exact science (cf: Framingham data) and so categorisation of patients based on risks and subsequent recommendations may not always be congruous.
·81 per cent of patients in the atenolol arm were already on medication with 36 per cent of these on two drugs! We do not know if these included atenolol and/or a thiazide which would inevitably disadvantage this arm. What is interesting is that the NICE committee actually excluded three trials from its analysis due to confounded use of either ?-blocker or thiazide as first-line
therapy in the same treatment arm but let ASCOT through.
·ASCOT emphasises better 'relative' risk
reduction in the amlodipine arm but admits to no significant 'absolute' risk reduction.
The NICE search strategy attempts to answer the question: what is the efficacy of anti-hypertensive drugs in 'preventing' diabetes? This gives the impression that calcium channel blockers are not causally associated with diabetes which as ASCOT itself shows is far from the truth! Could we not simply explain to our patients that if they were on a calcium channel blocker there was a 7.5 per cent chance they would develop diabetes as opposed to a ?-blocker where there might be a 10 per cent chance and leave the informed decision to them?
Might it also not be true that the reason there was a higher incidence of new-onset diabetes in the ASCOT study was the biased selection of patients? Impaired glucose tolerance, abnormal lipid profile, mean BMI of 28.7 and evidence of metabolic syndrome right from the outset!
An ABCD alternative
I feel GPs should come together as an authoritative body of expert opinion. In the NICE context this would be grade IV evidence and grade C/D recommendation. We should propose an algorithm that is evidence based and can be universally applied and accepted by all stakeholders.
One would have hoped for the new NICE guidance to adopt a risk-stratified approach (cf European Society of Hypertension guidelines) with the possible introduction of the concept of associated clinical conditions and target organ damage. I have made an attempt to get the ball rolling by suggesting a template (see page 47) which would offer
real choice for patients with evidence to back it (including 1,3,4,5 and ASCOT).
My main inspiration stems from the fact that having spoken to several colleagues in primary care, it would seem that most GPs already use a scheme similar to this. Further, there are hardly any GPs ( new and old ) who will argue against the fact that ?-blockers are effective in lowering BP, more so on an individual patient basis. Just because there isn't evidence doesn't mean it doesn't work.
Hypertension experts will point out that results from the ASCOT study can not be extrapolated to the younger patient with fewer or no risk factors, hence my emphasis on the role of ?-blockers in this group of patients, especially those with increased adrenal drives. ASCOT investigators and NICE have already admitted that the (seemingly) negative findings in the atenolol arm can not be attributed to a class effect and further
studies are needed to evaluate other
I am naturally open to constructive criticism and ask you to bear in mind that this is a 'prototype' and I am relying on grass-roots GPs to suggest how to improve it and close any loopholes. I appreciate that policing is not something we do as GPs but I feel economic consideration is required in this case (note that the Markov model used by NICE, although popular, is based on several assumptions) especially if there isn't any statistically significant evidence that a modern and more expensive drug is better than a well-established and cheaper one.
I draw the readers' attention to the results of analyses of more than 30 hypertension
trials (including ASCOT and most of those considered by NICE) in the Blood Pressure Lowering Treatment Trialists' Collaboration, which suggest the size of absolute blood pressure reduction is a more important determinant of the relative effects on total cardiovascular events than is antihypertensive drug choice. More recently (June 20056) it stated that total major cardiovascular events were reduced to a comparable extent in individuals with or without diabetes by regimens based on ACE inhibitors, calcium antagonists, angiotensin receptor blockers or ?-blocker/ diuretics.
Gulshan Kaul is a GP and practice cardiovascular lead in Lichfield, Staffordshire
The author is a member of the Primary Care Cardiovascular Society but the views expressed here are entirely his own