Letters

Preventing stroke with ramipril

Results should have been presented in ways that help practising clinicians

Benefits were considerably overstated

Superiority of particular class of antihypertensive agent remains to be shown

Presentation of data is misleading

Summary of responses

Results should have been presented in ways that help practising clinicians

EDITORI was surprised to see that the BMJ published a trial that presented the results in a way that exaggerates the findings.¹ Stroke prevention, the topic under discussion in the paper by Bosch et al, is important for patients, doctors, and funders of care. Hence the results should have been presented in a way that would help practising cliniciansby giving numbers needed to treat (NNT) along with relative risk reductions (RRR). The authors report a relative risk reduction of 32% in all strokes and of 61% in fatal strokes. For all strokes this translates into a number needed to treat of 67 for four and a half years' treatment.

(Credit: DR E WALKER/SPL)

Evidence shows that the way results of clinical trials are presented influences both physicians and funders of health care. ^{2 3} In the randomised controlled trial by Bucher et al, doctors gave higher ratings for the effectiveness of the drug and were more inclined to prescribe lipid lowering drugs when the results were presented as relative risks.² A study from a health authority in the United Kingdom reported that health authority members' willingness to purchase services was influenced by the methods used to present results.³ Interestingly, both these papers were published in the BMJ.

The problem of biased reporting of clinical trials is not a new phenomenon. Pocock et al in their survey of three medical journals in 1987 found that, overall, the reporting of clinical trials seems to be biased towards an exaggeration of treatment differences.⁴ What do the CONSORT guidelines say?⁵ The following quote may be relevant here: "For both binary and survival time data, expressing the results also as the number needed to treat for benefit (NNTB) or harm (NNTH) can be helpful." The two citations supporting this statement in the CONSORT guidelines are from the BMJ.

What can be done to improve the quality of reporting of results of randomised controlled trials? Both reduction in relative risk and reduction in absolute risk should be reported in medical papers because exclusive emphasis on the reduction in relative risk may overstate the effectiveness of a treatment.² If general agreement is reached then the next CONSORT guidelines should include a statement that wherever applicable the results of clinical trials should include the numbers needed to treat.

Although the BMJ has published many studies on the appropriate way to present results and their impact, in future if it emphasises to the authors of clinical trials the importance of presenting the numbers needed to treat, this will help its readers and avoid criticisms of the authors, reviewers, and editors.

P Badrinath, specialist registrar in public health medicine. 
Suffolk Public Health Network, St Clements Hospital, Ipswich IP3 8LS badrishanthi{at}hotmail.com

I thank Kev Hopayian, Leiston, Suffolk, United Kingdom, for bringing this issue to my attention through a posting in the evidence based health discussion list.

Benefits were considerably overstated

EDITORIn their editorial commenting on the HOPE study, Schrader and Lüders say that ramipril substantially decreased the risk of stroke and that fatal stroke was reduced by 61% and non-fatal stroke by 24%.¹ The former statement overstates the effect of the drug, and the latter statement is quite simply incorrect.

Although the relative risk reduction in the trial was 61% for fatal stroke and 24% for non-fatal stroke, the absolute risk reduction (ARR), which is the clinically relevant outcome measure, was reduced by only 1.5% and 0.9%, respectively. Since the follow up period of the trial was an average of 4.5 years, these "substantial" results are equivalent to an overall reduction of only 0.33% and 0.2% per year, respectively, in the occurrence of fatal and non-fatal stroke. This shows the pitfalls that can arise when the results of intervention trials are presented only in terms of relative risk reduction but recommendations are made in terms of alleged clinical benefits.²

Although patients in the trial were labelled "high risk," participants showed an absolute risk of only 4.9% for any stroke over the follow up period. This confirms the importance of hypertension control as one of the main public health interventions in preventing stroke. In addition, it is already proved that antiplatelet agents are an effective secondary prevention strategy in high risk patients and that anticoagulants effectively prevent stroke in patients with atrial fibrillation. ^{3 4}

Stroke is an important cause of mortality and disability in the United Kingdom and the search for new cost effective solutions to reducing death and disability must continue. We doubt, however, that the results of the HOPE trial warrant Bosch et al's recommendation that patients who are at high risk of stroke should be treated with ramipril irrespective of their blood pressure.⁵ Perhaps the authors intuitively accept this when they choose to present the trial outcomes only in terms of relative risk reduction.

Andrew P Wakeman, director of public health. 
Burntwood, Lichfield and Tamworth Primary Care Trust, Lichfield WS13 6JB andrew.wakeman{at}talk21.com

Jacqueline G Wakeman, general practitioner. 
Langton Medical Group, St Chad's Health Centre, Lichfield WS13 7HT

1.	Schrader J, Lüders S. Preventing stroke. BMJ 2002; 324: 687-688[Free Full Text]. (23 March.)
2.	McQuay H, Moore A. Using numerical results from systematic reviews in clinical practice. Ann Intern Med 1997; 126: 712-720[Abstract/Free Full Text].
3.	Antiplatelet Trialists Collaboration. Collaborative overview of randomised trials of antiplatelet therapy1: Prevention of death, myocardial infarction and stroke by prolonged antiplatelet therapy in various categories of patients. BMJ 1994; 308: 81-106[Abstract/Free Full Text].
4.	Atrial Fibrillation Investigators. Risk factors for stroke and efficacy of antithrombotic therapy in atrial fibrillation. Arch Intern Med 1994; 154: 1449-1457[Abstract].
5.	Bosch J, Yusuf S, Pogue J, Sleight P, Lonn E, Rangoonwala B, et al. Use of ramipril in preventing stroke: double-blind randomised trial. BMJ 2002; 324: 699-702[Abstract/Free Full Text]. (23 March.)

Superiority of particular class of antihypertensive agent remains to be shown

EDITORBosch et al claim that the benefits of ramipril in subjects at high risk is unrelated to its antihypertensive effects.¹ The HOPE study showed that the risk of stroke can be reduced by 32% (95% confidence interval 16% to 44%) and of myocardial infarction by 20% (10% to 30%) with ramipril treatment. ^{1 2} Recent evidence, from the HOPE study itself, makes it likely that this is simply a benefit of further blood pressure reduction.

The claims of protective benefit from angiotensin converting enzyme inhibitors are based on estimates of probable benefit calculated from observed differences in blood pressure between the HOPE study groups. In the study, the mean reduction in blood pressure (taken at the clinic) with treatment was 3/2 mm Hg, which would predict around 13% reduction in risk of stroke and around 6% in that of myocardial infarction, substantially less than that observed. ^{2 3}

A HOPE substudy was recently published that investigated ambulatory blood pressure in 38 subjects who were treated with ramipril or placebo.⁴ In these subjects, no significant differences were found between groups in clinic blood pressure. But 24 h blood pressure was significantly lower in the subjects treated with ramipril (10/4 mm Hg), mainly as a consequence of the substantially lower night time blood pressure (17/8 mm Hg). The authors note that more of the benefits of ramipril in HOPE may be related to reduction of blood pressure (especially during night time) than was explained by the effects on office blood pressure. They also noted that the HOPE protocol is the only large trial in which an antihypertensive agent was given at bedtime, thereby making it unique in regard to variation of blood pressure between day and night.

If similar reductions in blood pressure occurred in all HOPE participants, the calculations of likely benefit would be around 35% for stroke and 15% for myocardial infarction, matching reasonably the actual benefits observed. These estimates might help resolve the HOPE "paradox," whereby the findings of an apparent protective effect of the drug class are not supported by head to head studies.⁵ The HOPE study simply shows that in high risk patients, for blood pressure as for cholesterol, the lower the better.

A demonstration of specific organoprotection from a drug or a drug class will require head to head comparisons, with both clinic, and, ideally, ambulatory monitoring, at least in a subset. It still remains to be shown whether any one class of antihypertensive agent provides superior benefit.

John S Yudkin, professor of medicine. 
Department of Medicine, Diabetes and Cardiovascular Disease Academic Unit, Royal Free and University College London Medical School, London N19 5LW j.yudkin{at}ucl.ac.uk

Presentation of data is misleading

EDITORBosch et al show a relative risk reduction of 32% in any stroke and of 61% in fatal strokes, in patients treated with ramipril compared with placebo over 4.5 years.¹ This is a marvellous study result, and the conclusion is appropriate, that patients at high risk of stroke should be treated with ramipril, irrespective of their blood pressure. But analysing the data further and calculating the reduction in absolute risk, the numbers to treat (NNT) and cost, the same data do not seem as impressive. The table shows that the cost of preventing a single stroke with ramipril over 4.5 years is in the range of C$100 000-250 000 (US$64 000-160 000; £42 000-105 000; 65 000-162 500).

The table also shows further analysis of subgroup data given in figure 4 of the long version of the paper (bmj.com/cgi/content/full/324/7339/699). The risk of stroke in ramipril subgroups is not given so my calculations are based on the relative risk reductions shown in the figure.

On the basis of this analysis I believe that ramipril is useful and possibly a cost effective alternative in preventing stroke only in three select groups of patients at high risk.

Patients who are not having aspirin or antiplatelet treatmentIdeally, these patients should receive these agents but if a patient cannot tolerate them or is allergic to these agents then ramipril is a viable alternative.

Patients who are not taking a calcium channel blocker for some other indicationIf a patient is already taking a calcium channel blocker (for any indication) then addition of ramipril has not shown any significant effect in prevention of stroke.

Patients who are taking lipid lowering agentsAlthough the risk of stroke was decreased in patients who were treated for hyperlipidaemia with lipid lowering agents, the effect of ramipril was insignificant in both groups. There was, however, a suggestion of a synergistic effect of ramipril in patients who were treated with lipid lowering agents.

Only 28% of these so called high risk patients received lipid lowering agents in the late 1990s. On the basis of the evidence given, it is currently not reasonable to recommend widespread use of an angiotensin enzyme inhibitor such as ramipril in patients at high risk of stroke, but it is useful and may be cost effective in a select group of high risk patients as mentioned above.

Malvinder S Parmar, consultant physician. 
Timmins and District Hospital, Timmins, Ontario, Canada P4N 8R1 atbeat{at}ntl.sympatico.ca

Summary of responses

The study reported by Bosch et al with the accompanying editorial by Schrader and Lüders prompted 21 responses on bmj.com and one letter to the editor. ^{1 2} Twelve responses were from Britain, the rest from Australia, Austria, Canada, France, the Netherlands, the United Arab Emirates, and the United States.

Most of the correspondents criticised the way in which the results had been presented in the study and the editorial. In the words of Brian Mansfield, a general practitioner in Beckington, Somerset: "Am I alone in finding the confusion between lack of evidence for one course of action and positive evidence for another being misinterpreted as the superiority of course B over course A disturbing?"

It was noted that only the relative risk reduction was given in the study. This should have been accompanied by data on absolute risk reduction and number needed to treat and even number needed to harm (as per CONSORT guidelines). Several of the correspondents had done these calculations and concluded that the number needed to treat over 4.5 years and the resulting costs were too high to make ramipril a viable treatment for patients with stroke.

W Hoefnagels, a neurologist from the Netherlands, questions whether the outcomes with ramipril were actually any better than outcomes achieved with aspirin. Trevor Thompson, a clinical lecturer in primary care at Bristol University, further notes that no quality of life data in the two groups were given and the incidence and nature of adverse drug reactions were not mentioned; this is also a criticism made by Yoon Loke, a clinical lecturer in pharmacology in Oxford. Two correspondents note that no measures of all cause and overall mortality were given. Loke draws our attention to the fact that the reported study is a substudy of HOPE and that the adverse data missing from this study are available from the original study.

Several correspondents draw attention to the figures. Figure 3 shows a non-significant effect of ramipril on people with previous stroke; no flow diagrams accompany the article; and the conclusions are not supported by figure 2. Furthermore, table 1 does not add up.

Peter David Burrill, a specialist in pharmaceutical public health, comments that the two different dosages of ramipril should have been examined with respect to their efficacy. Several authors find that the fact that the blood pressure lowering effect of the drug might have been responsible for the benefits had not been taken into consideration. Claudia Stöllberger and colleagues from Austria believe that the prevalence of atrial fibrillation and antithrombotic treatment should have been made known as atherothrombosis is not the only cause of stroke. John Attia and his team from Australia question whether the conflicting results of the HOPE and PROGRESS trials may be a function of the ethnic group of participants.

Three correspondents mention the importance of authors declaring their competing interests, and pessimism about modern medicine being dominated by pharmaceutical companies, as evidenced by the prodrug bias of the study.

Birte Twisselmann, technical editor. 
BMJ