Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Don't Forget Observational Studies
- Godfrey Oakley, Jr. (15 October 2004)
-
What about the study 'subjects'?
- Chris MH Hiley (15 October 2004)
-
Independent trials of new interventions
- Lawrence A Emtage (15 October 2004)
-
role of NICE guidelines?
- Chris Ducker (16 October 2004)
-
Out patient review of patients on rofexcoxib
- badal pal, hanu reddy,sujit shetty,andreas alexander (18 October 2004)
-
BMJ COX'S UP
- Faiyaz Mohammed (20 October 2004)
-
No surprises over Vioxx
- Paul Cracknell (22 October 2004)
-
Be vigilant
- Christopher Anton (22 October 2004)
-
The new French paradigm : the Public Heath Impact Assessment of Drugs
- Lucien Abenhaim (24 October 2004)
-
Withdrawal of a drug have more clinical and ethical perspectives
- Juan Carlos Maldonado, Edmundo Estevez. Biomedical Center. Central University of Ecuador (29 October 2004)
-
Vioxx controversy -- Lancet publishes unscientific meta-analysis of rofecoxib studies
- Jeffrey Mann (9 November 2004)
-
COXIBs' cardiovascular effects
- Mario Guslandi (19 November 2004)
-
Risk reduction in patients taking NSAIDS and COX-2 inhibitors
- James A McGuigan (29 November 2004)
|
|
|||
|
Godfrey Oakley, Jr., Research Professor of Epidemiology, Rollins School of Public Health of Emory University Rollins School of Public Health, 1518 Clifton Rd, NE, Atlanta, GA 30322
Send response to journal:
|
Randomized controlled trials are not the usual way serious, uncommon adverse effects of a new drug are discovered. They are usually discovered in observational studies, often case-control studies. I think there is danger in using RCTs to find adverse effects. It would feed on the current vogue from trialists that the only data that count are those from RCTs. By this standard, we would not have concluded that cigaret smoking causes lung cancer. We should not exclude observational studies, but rather continue to do them well in a timely fashion. As I understand the Vioxx story, the cardiovascular effect is strong enough to see in an RCT in 2000, although there was some creative interpretation of the data suggesting that a protective effect of control drug could have explained the result. The current unpublished RCT confirms the adverse effect some 4 years later. What if, rather than conducting a RCT on a few thousand more patients, a large case-control study had been conducted to test the hypothesis of whether or not the increase was from Vioxx rather than waiting for results of a RCT not designed to find the adverse effect. My hunch is that the case control study would have nailed the drug sooner than waiting for the RCT. At less cost than an RCT, the case-control study would likely have been powered to determine increased risk that would be missed in an RCT simply because one could not afford a large enough sample in the RCT. There is value in observational studies. They have played and they likely will continue to play an important role in the protection of the public health. Godfrey Oakley, Jr. Competing interests: Co-inventor on patent that covers putting folic acid in oral contraceptives. A paid consultant to Johnson and Johnson on this issue. |
|||
|
|
|||
|
Chris MH Hiley, Head of Policy and Research The Prostate Cancer Charity
Send response to journal:
|
I have been following the Vioxx story with interest as it was the study drug in a prostate cancer prevention trial. I think the proposals in this paper go some way to stopping future problems, but whether the pharmaceutical companies will embrace them, or our policy makers get round to forcing them to, remains to be seen. I'd like to put in a plea for the people who take part in trials. They should surely be informed of results, indeed, be the FIRST members of the public to have access to the data they helped create? Both negative and positive. Are patients not entitled to wonder just what is the point of joining trials? Do they not deserve it to be properly executed, efficiently and maybe independently evaluated, and the drug followed up over time after it has gone to market? Vioxx victims are not only the people who experienced adverse events - they include everyone who joined the original trials in good faith that their contribution would have meaning and depth. 'Subjects' in my dictionary also means people 'living under the rule of a monarch, government, or dare I suggest it ...research culture? The time has long passed when research 'subjects' should change to a concept of research 'citizen'. A citizen is a member of a state, nation or other community. The community in this case is the team looking for answers to questions of common interest - and that's both the medics and the patients with the conditon or symptom being treated. Competing interests: None declared |
|||
|
|
|||
|
Lawrence A Emtage, Consultant Urologist Russells Hall Hospital, Dudley, DY1 2HQ, UK
Send response to journal:
|
To be truly independent means that the funding of the trial must be from an unimpeachable source, which usually means NHS or Government or Charity funds in the UK. Most clinicians have no access to such funds which would have to be massive to realise a large randomised controlled trial with sufficient power statistically to answer the usual type of question; is one treatment better than the other? In the last ten years no trials have got under way in the West Midlands in Urology unless funded via the pharmaceutical industry. My second point is related to the way the pharmaceutical trial is linked to the marketing of the new treatment. Critical scrutiny at this time is usually weak and this was the case with rofecoxib, as the editorial shows that the evidence for serious adverse events was in the public domain in 2000. Perhaps we should always have our critical hats on. Competing interests: None declared |
|||
|
|
|||
|
Chris Ducker, GP Whithorn, Scotland. DG8 9QZ
Send response to journal:
|
I do wonder about the role of NICE in the spread of rofecoxib use in the UK. As a GP who could remember Opren I did not use rofecoxib at all until their guidelines were issued. Competing interests: None declared |
|||
|
|
|||
|
badal pal, consultant thrumatologist withington hospital,manchester m20 2lr, hanu reddy,sujit shetty,andreas alexander
Send response to journal:
|
Dear Sir, Rofecoxib (Vioxx) which was first marketed in the UK in 1999 was recalled worldwide on 30th of September, 2004.. In the UK some 400,000 patients have been taking this for various conditions , mainly osteoarthritis, rheumatoid arthritis and other rheumatic as well as non rheumatic conditions. This withdrawal, which was entirely voluntary by the pharmaceutical company (Merck, Sharp & Dohme), was a consequence of long term trial results which showed higher cardiovascular risks compared with placebo after continued use for 18 months, and concerns expressed recently (1,2). An advisory letter had been sent by the Medicines and Healthcare Products Regulatory Agency with advice to discontinue this drug for patients who had been prescribed this and to arrange replacement/alternative medication. It was made clear by the agency that this specific advice was relating to Rofecoxib alone and not for other Cox II inhibitors. It is understood that switching to an alternative should happen within a 28 day period. As most hospital rheumatology departments are not computerised, as is the case in general practice, it was felt that hospital clinicians such as rheumatologists will not be in a position to advice patients immediately on this issue. We reviewed our patients attending the rheumatology clinics with regard to use of Rofecoxib and whether they were still on this medication over the 2 week period following its voluntary withdrawal. Of consecutive 75 patients seen with the age range from 24-80 (predominantly female), 12 patients were prescribed Rofecoxib and all had discontinued this medication within the first few days on advice either from their general practitioner (3), from pharmacists (3), on hearing about this via the media (3), advice from the clinic (2) or advice from a friend (1). Of these 12 patients, only one actually rang our department to get further advice on hearing about this on the media. Of the 12 who were on Rofecoxib, 6 were advised to replace with another coxib ( Celecoxib 1,Valdecoxib 2, and Etoricoxib 3). In the remaining 6 patients, no further coxib or non steroidals were prescribed and they were advised to rely on simple analgesics only . Therefore, continuation of a non steroidal was not deemed essential in this half of the total 12 patients who were on Rofecoxib. It is to be noted that the cardiovascular safety concerns are related only to Rofecoxib and not generalised to other cyclo-oxygenase selective inhibitors, although in time other drugs of the same class may show problems in due course, and practitioners should be vigilant about this. There are lessons to be learned from this even (3). It had also highlighted issues concerning computerised data record keeping, or the lack of it in hospital specialist departments and at least in this regard general practice is better equipped to deal with emergency situations of this nature. Our local Primary Care Trust had advised all general practices in the area to contact patients on this drug to discontinue it immediately. This action would not be possible in most rheumatology departments in the country for lack of computerised or electronic records. Communications to patients from general practices in our experience appears to be very effective in that we received one recorded call only in our area from a concerned patient and enquiry with our local chemists revealed hardly any enquiries from concerned patients. Yours sincerely B Pal
H Reddy
S Shetty
A Alexander
Reference: 1.Mukherjee D, Nissen S E, Topol E J. Risk of cardiovascular events associated with selective COX 2 inhibitors. JAMA 2001;286:954-9. 2.Juni P,Dieppe P.Older people should NOT be prescribed ‘coxibs’ in place of conventional NSAIDs.Age Ageing 2004;33:100-4. 3.Dieppe P, Ebrahim S, Martin R M, Juni P. Lessons from withdrawal of rofecoxib.BMJ,2004;329:867-8. Competing interests: None declared |
|||
|
|
|||
|
Faiyaz Mohammed, Specialist registrar - Gastroenterology Wythenshawe Hospital, Manchester, M23 9LT
Send response to journal:
|
The BMJ has published an editorial on lessons from the withdrawal of rofecoxib.(1) So where does the BMJ place the editorial? On the page opposite to a full-page advertisement for celecoxib! Whether accidental or deliberate, the advertising executives at Pfizer (the manufacturers of celecoxib) are probably extremely pleased with this situation. Is it not time for the BMJ to consider adopting the policy of the New England Journal of Medicine – “main content uninterrupted by commercial advertising"?(2) Faiyaz Mohammed Specialist Registrar – Gastroenterology Wythenshawe Hospital, Manchester M23 9LT 1 Dieppe P, Ebrahim S, Martin R, Juni P. Lessons from the withdrawal of rofecoxib. BMJ 2004; 329:867-8 2 Drazen J.M, Anderson K.R, Curfman G.D. A New Look – Form Follows Function. N Engl J Med 2003; 348:66, Jan2, 2003 Competing interests: None declared |
|||
|
|
|||
|
Paul Cracknell, GP Killamarsh Medical Practice, S21 1DX
Send response to journal:
|
The warning signals were there from the beginning over cardiovascular adverse effects from Vioxx, for those willing to see them. Why then were GPs and consultants keen to use Vioxx and other COX-2s with such enthusiasm? I believe there were several reasons from very effective marketing by the drug companies, uncritical appraisal of the evidence by clinicians, pressure to prescribe from arthritis charities, and the effect of NICE guidelines. As a GP I certainly felt pressure from secondary care to prescribe COX-2s in preference to NSAIDs especially in the elderly (on the presumption this was safer). This episode should act as an important reminder to treat all new drugs with great caution, and drug company claims with great scepticism, until their risks and benefits over existing treatments are fully known. The black triangle is there for a reason. Competing interests: None declared |
|||
|
|
|||
|
Christopher Anton, Administrative Co-ordinator City Hospital, Birmingham B18 7QH
Send response to journal:
|
Dr Cracknell is right to draw attention to the use of the black triangle symbol to indicate that a drug is newly marketed, or being used with a novel method of delivery, or in a new patient population with a new indication, or as a combination product. However, Vioxx was not a black triangle drug when it was withdrawn (it having been first marketed in August 1999) and reporters should remain vigilant in reporting their SUSPISCIONS of serious adverse reactions to any drug irrespective of how long it has been marketed. An electronic Yellow Card is available at www.yellowcard.gov.uk Competing interests: West Midlands Centre for Adverse Drug Reaction Reporting |
|||
|
|
|||
|
Lucien Abenhaim, Prof. of Epidemiology, McGill U. (PT) and U. of Paris V, Former General Director of Health of France Faculty of Medicine, 24 rue du Faubourg Saint Jacques, Paris 75014 France
Send response to journal:
|
Sir, In their editorial following the withdrawal of rofecoxib, Dieppe and collaborators are advocating for a series of measures before definite licensing of a drug. We have been confronted with several experiences at the General Direction of Health of France regarding this issue, and would like to add to the debate by bringing some more perspectives. Dieppe and collaborators are suggesting that new drugs to be used on a large scale should undergo a 'phase introduction' before definite licensing is granted. This objective is, we believe, one of the main measures to be taken by regulators in the near future because it is by definition impossible to obtain before marketing a sufficient database to be able to assess with accuracy all drug-related risks and benefits. However, we do not believe that the "large scale randomized trials" asked for by Dieppe and collaborators would be the only solution. From the information available to date, we do not believe either, as suggested by Dieppe and colleagues, that more transparency on data was the problem at hand in this instance. In effect, to take the example of the rofecoxib, cardiovascular risks, were suggested very soon from the Vigor study, as Dieppe recalls and this did not resulted in, nor justify, stopping the marketing of the drug. And a large scale randomized trial was indeed conducted, which led to the withdrawal of the drug, but it was conducted on patients who were not fully representatives of those usually treated with the drug. The main problem with drugs is that randomized trials, and many classical epidemiology studies, conducted before or after marketing mainly for etiologic purposes, are unlikely to represent fairly the broad experience of drug utilisation in the real life. The issue is not only whether a drug could be associated with an unknown adverse event but also whether its overall impact on the public health is beneficial or not in the real life. This depends largely on the way it is used in practice, what patients are switched to it, what are the other therapeutic options and their drawbacks, what are the common co-prescriptions and of what co- morbidities do the patients prescribed the drug suffer. We have requested in France that whenever a drug is likely to be used on a large scale, pharmaceutical companies should organize a post- marketing study of the "public health impact" (PHI) of the drug. This PHI assessment goes far beyond single endpoints as is the case in trials and classical epidemiologic (etiologic) research. This is now requested, before reimbursement is granted, within the framework of a formal agreement signed in May 2003 between our Health Product Economic Committee (CEPS) and the Association of Drug Enterprises (LEEM),. What is required is that large scale epidemiologic evaluations are conducted that measure the potential shifts in the disease-related morbidity and mortality at the population level, as well as risk assessment evaluation and the utilization of other concerned drugs. Our first very example of this new policy was actually to ask for an independent, large scale cohort study of 35000 patients treated with rofecoxib, cerecoxib or classical NSAIDs (Moore and collaborators). Concerned companies participated willingly in this study. Unfortunately, due to the delay in having this policy approved by all parties and implemented, full results are not yet available in this case (preliminary results will be soon published however), but for future drugs, they should hopefully be obtained in a more timely fashioned. More than 50 such studies are in effect now signed between pharmaceutical firms and the French authorities. Some of these agreements have included the limitation of the size of the population for which the drug should be reimbursed, as a mean for a stepwise introduction of the drug pursuing similar goals as those advocated above. However, the policy is only one year on its wheels and will have to be evaluated for its efficiency in the future. The present drug-decision system is in effect presenting with a serious gap. It is schematically divided in two steps: evaluation of efficacy and tolerance mainly through randomized trials on the one hand and pharmacovigilance reporting of rare adverse events on the other hand. Yet, the true impact of drugs on the public health is not fully evaluated. Billions of dollars are spent and hundred millions of patients are exposed to many classes of drugs worldwide without a full measure, if any in some cases, of their impact of the incidence and prevalence of diseases afterwards. In order to estimate how many myocardial infarctions have been prevented by statins or GI bleeding by coxibs, one is left with mainly modelling from the results of randomized trials, which barely represent the real life experience. In this situation, one single drug may, such as in the case of rofecoxib, easily appear as the "usual suspect" or "scapegoat" following the result of this or that study, but this leaves unanswered the question of what is really happening with the other ones, and how is the public health "better" protected now? We suggest that an international task force is organized to define the objectives, methodologies and guidelines for the assessment of the Public Health Impact of Drugs and to promote it consistently towards regulatory bodies and health insurances, be they public or private. Elements of the evaluation required are listed in table 1 Table 1 Elements of the Public Health Impact Assessment of Drugs required by the General Health Directorate of France (2003) • Description of the treated population and comparison with the target population for the drug. Identification, within the population of treated patients, of the fraction who was actually poorly controlled with previous therapies • Measure of the impact of the introduction of the new drug on the incidence and prevalence of complications of the treated disease at the population level, including associated morbidity and mortality • Measure of the impact of comparative new other drugs in the same population • Evaluation of the impact of the introduction of the new drug on the health system (medical and hospital services, use of diagnostic procedures and other treatments, including other drugs. Lucien Abenhaim, MD, MSc, PhD Former General Director of Health of France (1999-2003) Professor of Public Health, University of Paris V France Full Professor of Epidemiology and Biostatistics, McGill University (PT) Competing interests: Participation or consulting for studies funded by Roche, United Therapeutics, Negma and Astra-Zeneca. None on NSAIDs or coxibs |
|||
|
|
|||
|
Juan Carlos Maldonado, Pharmacology Unit Biomedical Center. Central University of Ecuador. POBox: 17-11-6120., Edmundo Estevez. Biomedical Center. Central University of Ecuador
Send response to journal:
|
Dieppe et al, comments some lessons from the withdrawal of rofecoxib [1]. We are agree about the recommendations proposed, specially with the need to determine the cardiovascular risk for other COX-2 selective inhibitors, because this adverse drug reaction must be a class effect but with a different risk for each drug. For example, more than twenty years ago the risk of upper gastrointestinal bleeding was associated with aspirin and other non- steroidal anti-inflammatory drugs, but a study published in 1991 was the first to demonstrate a different risk for individual drugs [2]. Something similar could be happening now. Certainly in the VIGOR Study [3] the incidence of myocardial infarctions were less common in the naproxen group (0.1%) than in the rofecoxib group (0.4%), meaning a relative risk of 0.2 (CI95%= 0.1 to 0.7). Although it is true that in the CLASS study (celecoxib vs diclofenac/ibuprofen) the rates of myocardial infarction were similar between groups (0.25% vs 0.27%) in this trial patients taking aspirin were included (20.9% vs 20.4%) [4]. Then, the pharmacodynamic weakness of celecoxib (COX-2 do not inhibit platelet aggregation) was avoided. The promotion of rofecoxib was also questioned before [5], nevertheless in the last years this drug was worldwide used by a lot of people. Now, the voluntary withdrawal from the market [6] seems ethical, but it is? How many deaths could be attributed to the cardiovascular adverse drug reaction? Now, maybe a domino effect of withdrawals will never be, but it is clear that COX-2 inhibitors do not have to be used as first line drugs, because other non-steroidal anti-inflammatory drugs with the same efficacy, cheaper, and probably with a better safety profile are available in the market. Moreover, we do not know about another serious adverse drug reactions of low frequency related to Cox-2 inhibitors. For example, recently an acute temporary visual impairment was probably associated with the use of celecoxib and rofecoxib, because the inhibition of synthesis of prostacyclin and other compounds could affect the retinal blood flow [7]. In the ethics perspective, the clinical research aims to benefit individual participants and patient groups through the identification and testing of improved therapeutic treatments and to benefit society by making these treatments available. But the potential risk of harm to participants has led to widespread agreement that sound ethical standards must be observed in clinical research, irrespective of the geographic and economic setting in which it is undertaken. The rofecoxib incident lead us to think that sometimes nobody remember the lessons form other drugs such as mibefradil [8], troglitazone [9], phenylpropanolamine [10] or the stilbestrol [11] and thalidomide [12] tragedies. Clinical research sponsored or undertaken by developed countries in developing countries also rises fundamental questions about distributive justice [13]. Moreover, in these countries some unethical trials could be performed [14] because many professionals do not have the experience or the knowledge, the regulatory authorities do not have a legislative background similar to FDA or EMEA, the health systems have limited resources, and the pharmaceutical prices could affect the access to drugs. Anyway, the rofecoxib withdrawal remember us some rational drug use statements: “a new drug do not have to be always the best”, “an expensive drug is not always more efficacious”, and “the absence of evidence about insecurity does not means evidence of safety”. Again, the patient´s health was affected. The pharmaceutical company only loss money. Juan Carlos Maldonado, MD. Clinical research, Pharmacology Unit. Biomedical Center, Central University of Ecuador. Edmundo Estévez, MD, MSc. Director, Biomedical Center and Secretary, Bioethical Committee. Central University of Ecuador. REFERENCES: 1- Dieppe PA, Ebrahim S, Martin RM, Jüni P. Lessons from the withdrawal of rofecoxib. BMJ 2004; 329: 867-8. 2- Laporte JR, Carné X, Vidal X, Moreno V, Juan J. Upper gastrointestinal bleeding in relation to previous use of analgesics and non-steroidal anti-inflammatory drugs. Lancet 1991; 337: 85-9. 3- Bombardier C, Laine L, Reicin A, et al., for the VIGOR Study Group. N Engl J Med 2000; 343: 1520-8. 4- Silverstein FE, Faich G, Goldstein JL, et al. Gastrointestinal toxicity with celecoxib vs nonsteroidal anti-inflammatory drugs for osteoarthritis and reumatoid arthritis. The CLASS Study: a randomized controlled trial. JAMA 2000; 284: 1247-55. 5- Josefson D. FDA warns Merck over its promotion of rofecoxib. BMJ 2001; 323: 767. 6- Singh D. Merck withdraws arthritis drug worldwide. BMJ 2004; 329: 816. 7- Coulter DM, Clark DWJ, Savage RL. Celecoxib, rofecoxib, and acute temporary visual impairment. BMJ 2003; 327: 1214-5. 8- Po ALW, Zhang WY. What lessons can be learn from withdrawal of mibefradil from the market? Lancet 1998; 351: 1829-30. 9- Wise J. Diabetes drug withdrawn after reports of hepatic events. BMJ 1997; 315: 1559. 10- Kernan WN, Viscoli CM, Brass LM, et al. Phenylpropanolamine and the risk of hemorrhagic stroke. N Engl J Med 2000; 343: 1826-32. 11- Herbst AL, Ulfeder H, Poskanzer DC. Adenocarcinoma of the vagina: association of maternal stilbestrol therapy with tumor appearance in young women. N Engl J Med 1971; 284: 878-81. 12- Lenz W. Thalidomide and congenital abnormalities. Lancet 1962; 1: 45. 13- Lansang MA, Crawley FP. The ethics of international biomedical research. BMJ 2000; 321: 777-8. 14- Charatan F. Surfactant trial in Latin American infants criticized. BMJ 2001; 322: 575. Competing interests: None declared |
|||
|
|
|||
|
Jeffrey Mann, Retired physician Salt Lake City, UT 84103
Send response to journal:
|
The Lancet published the results of an unscientific meta-analysis of rofecoxib studies online on November 5th 2004 [1]. I think that this cumulative meta-analysis is highly flawed and I think that the authors misrepresent a number of facts. The meta-analysis is essentially a review of all randomised controlled clinical trials in adult patients with chronic musculoskeletal disorders that compared rofecoxib with placebo or other NSIADs. The authors reviewed 63 reports, and 18 clinical trials (involving 25 273 patients). They specifically studied the incidence of MI, and they concluded that the overall rate of MI was increased in rofecoxib patients (combined relative risk of 2.24, based on 52 events in the rofecoxib group compared to 12 in the control group). In particular, the authors claimed that this increased RR was already apparent in 2000, when 14 347 patients had been randomised and 44 MI events had occurred in rofecoxib treated patients (see figure 3 in the paper). However, the increased RR occurred suddenly in the year 2000 and it was due to the fact that the cumulative data was then being driven by a comparison with naproxen as the control agent (large amount of data derived from the VIGOR trial), rather than other NSIADS or a placebo (which mainly drove the cumulative data until that point in time). The choice of a comparator is critically important if the comparator agent is associated with a significantly lower, or significantly higher, risk of MI than that found in rofecoxib treated patients. The authors concluded that by the end of the year 2000, the cumulative evidence demonstrated that rofecoxib more than doubled the risk of MI. Compared to what? It may appear that the risk of MI was more than doubled compared to naproxen, but it was not doubled relative to placebo and other NSIADs. In fact, the authors published the comparative results in table 2 in their paper, and their published results demonstrate that the RR of rofecoxib compared to placebo was 1.04, and 1.55 when compared to non-naproxen NSIADS. However, the lead author Jüni apparently still made this statement as reported by Zosia Chustecka in a heartwire report [2]-: "Jüni tells heartwire that the meta-analysis also shows that this increased risk of MI with rofecoxib did not vary with controls (similar for placebo, naproxen, or a nonnaproxen NSAID), did not vary with the daily dose of rofecoxib used (12.5 mg, 25 mg, and 50 mg), and did not depend on length of treatment (it was similar for trials of duration less than 6 months and those of more than 6 months)". Why did Jüni claim that the increased risk of MI with rofecoxib did not vary with controls when their own meta-analysis specifically demonstrated that this statement was not true? In fact, Jüni et al. comprehensively dealt with the issue of naproxen as a comparator agent in their meta-analysis paper, and they specifically state-: "The difference in coronary risk in the VIGOR trial has been widely interpreted as being due to a cardioprotective effect of naproxen, rather than an adverse effect of rofecoxib. We examined this hypothesis by stratifying results from randomised trials according to the control intervention and found that the increase in risk was indeed greater in trials comparing rofecoxib and naproxen, but that this finding was probably attritutable to chance." Don't you think that this conclusion is bizarre? Chance! I can understand a meta-analyst concluding that a specific finding could be due to chance when it occurs once, or very infrequently. However, Jüni's own meta-analysis paper reviewed a number of naproxen studies (see figure 4) and they demonstrate a near-uniform result among the 11 studies -- that naproxen is associated with a reduced risk of MI compared to other controls. Therefore, using naproxen as a comparator agent presumably inflates the estimated RR of iatrogenic MI due to rofecoxib, and makes the "apparent" harmful effect greater than the likely "true" harmful effect. I think that Jüni's cumulative meta-analysis is highly flawed because it uses a heterogeneous group of control population studies as a comparator. I think that a meta-analysis of multiple clinical studies is only scientifically valid if the different control population studies have relatively homogeneous results (similar baseline risk of MI in the different control groups). I cannot understand why the Lancet decided to rush such a flawed meta-analysis into publication when this particular issue has already been discussed years ago. The issue of whether the increased risk of MI associated with rofecoxib is "real" or "apparent" was already discussed in the mainstream medical literature a few years ago. In particular, the authors of a 2001 paper [3] expressed serious concerns regarding the issue of whether the increased risk of MI in rofecoxib treated patients is due to a direct harmful thrombotic effect of rofecoxib, or whether the "apparent" harmful effect is due to a comparison with naproxen, which may have a cardioprotective antithrombotic effect. The authors concluded that they didn't have enough evidence to come to a definitive conclusion, and they therefore used another ancillary comparative technique to shed more light on this issue. They compared the risk of MI in rofecoxib treated patients (who were at low risk of MI) to the risk of MI in placebo patients from four aspirin primary prevention trials (who were presumed to have a similar low risk of MI). They noted that rofecoxib treated patients had a higher risk of MI (0.74%) compared to those placebo patients (0.52%). This fact suggested that the increased risk of MI in rofecoxib treated patients could be "real" rather than "apparent". However, the authors noted that there is no guarantee that the rofexcoxib and placebo patients had the same baseline risk of MI (because they were from totally different studies). They therefore suggested that a large randomised trial should be performed in "higher-than-normal risk" MI patients (eg. elderly OA patients who have a higher baseline risk of MI) in order to to generate a stronger signal, and hopefully a definitive answer to this dilemma. I think that this suggestion was very rational and an appropriate solution to this dilemma. I think that Merck was highly delinquent because it did not perform that much-needed RCT during the past few years. I think that the FDA was also highly delinquent because it did not insist that it be performed. And, finally, I think that the Lancet editor (and other mainstream medical journal editors) were also highly delinquent because they did not repeatedly insist, during the past four years, that this RCT be performed. I think that the available evidence strongly suggests that the increased risk of MI in rofecoxib patients may be "real", but the medical research community obviously missed a golden opportunity to prove that point by performing a well-designed large randomised trial using an appropriate homogeneous control group as the comparator group. What particularly bothers me about this Lancet published meta-analysis is that it obfuscates the issue as to whether rofecoxib significantly increases the risk of an MI, and it does not shed any useful light on the issue. Sackett [4] stated that confidence in a RCT's conclusion is directly proportional to the trial's signal/noise ratio. An increase in a trial's signal, and/or a decrease in a trial's noise, increases the trial's signal/noise ratio, thereby allowing one to become increasingly confident in the scientific validity of the trial's results. To maximise a RCT's signal, Sackett suggests that trial designers should ensure that there is a combination of a high control event rate and a large effect size. All the rofecoxib trials had a low baseline MI rate (partly because high risk coronary patients were excluded from trial participation + partly because most trials were of short duration) and this fact significantly decreased each trial's potential signal, and therefore the trials' signal/noise ratio. Theoretically, a cumulative meta-analysis could allow one to increase the "overall" signal and thereby improve the "overall" signal/noise ratio (according to Sackett, confidence in a RCT's results is inversely proportional to the square root of the sample size, which means that the any gains in confidence obtained by adding the results of multiple studies is not simply proportional to the final sample size number). However, that basic assumption is only valid if the different trials use the same control agent (or a similar control agent that results in a similar control event rate). That basic assumption certainly doesn't apply to the rofecoxib studies included in Jüni's meta-analysis, which used different control agents (placebo, naproxen and non-naproxen NSIADS) with varying control event rates. Therefore, a cumulative meta-analysis of heterogenous rofecoxib studies doesn't increase the signal/noise ratio. In fact, it decreases the signal/noise ratio by significantly increasing the noise level. Publication of this cumulative meta-analysis is therefore antithetical to the basic principle of EBM science -- because it is likely to obscure, rather than illuminate, the truth. Jeff Mann. References: 1. Peter Jüni, Linda Nartey, Stephan Reichenbach, Rebekka Sterchi, Paul A Dieppe, Matthias Egger. Risk of cardiovascular events and rofexcoxib: cumulative meta-analysis. Lancet. November 5th 2004. Published online at http://www.lancet.com 2. Heartwire report published November 4th. Available in the heartwire section of theheart.org. 3. Mukherjee D, Nissen SE, Topol EJ. Risk of cardiovascular events associated with selective COX-2 inhibitors. JAMA 2001; 286:954–59. 4. Sackett, David L. Why randomized controlled trials fail but needn't: 2. Failure to employ physiological statistics, or the only formula a clinician-trialist is ever likely to need (or understand!) CMAJ. 165(9):1226-1237, October 30, 2001. Available online at http://www.cmaj.ca/cgi/content/full/165/9/1226 Competing interests: None declared |
|||
|
|
|||
|
Mario Guslandi, Head of Clinical HepatoGastroenterology Unit Gastroenterology Dept.. S. Raffaele University Hospital. Milan, Italy
Send response to journal:
|
The article seems to hint that the cardiovascular side-effects reported with rofecoxib can be a class effect. Only time will tell if this assumption is right or wrong . I’d like to point out,however,that we already had examples in the past of drugs withdrawn from the market due to adverse reactions not shared by other compounds belonging to the same pharmacological class. Years ago two H2-receptor antagonists, oxmetidine and niperotidine have been found to induce acute liver injury and their use has been banned. Nevertheless ranitidine and various other H2-blockers have continued to be used by millions of individuals all over the world – and still are- with no harm for the patients. In August 2001 cerivastatin was removed from European and USA markets because of a high risk of rhabdomyolysis , but other statins keep being employed everywhere as a cholesterol-lowering agents , although, obviously, under careful medical surveillamce. It’s true that “absence of evidence is not evidence of absence”. On the other hand everybody is presumed innocent until found guilty. Competing interests: None declared |
|||
|
|
|||
|
James A McGuigan, Consultant Surgeon Royal Victoria Hospital Belfast
Send response to journal:
|
As a surgeon with a large oesophageal practice I am deeply concerned that the recent scare and confusion around COX-2 inhibitors, as referred to in ‘Lessons from the withdrawal of rofecoxib’ (BMJ 2004; 329:867-868), will increase the already alarming proportion of patients taking anti- inflammatories who suffer oesophageal, gastric and duodenal ulceration – with potentially fatal consequences. Prescriptions of traditional non-steroidal anti-inflammatory drugs (NSAIDs) will undoubtedly continue to rise and my worry is that this will be done without enough consideration given to their side effect profile - it is already well documented that an estimated 30% of regular users develop peptic ulcers1. Currently the debate as to whether it is appropriate to switch to another COX-2 inhibitor or switch to the older NSAIDs remains unresolved. Regardless of which side of the fence anyone sits, as a surgeon, I would strongly urge clinicians to protect their patients against the gastro- intestinal side effects associated with NSAIDs and also COX-2 inhibitors. This can be achieved effectively and simply by considering concomitant proton pump inhibitor (PPI) prescribing - particularly for those at high risk (those with a history of peptic ulcer and / or older than 60 years). Most PPIs are licensed for the prevention of gastro-intestinal ulcers and healing in those taking NSAID’s but only esomeprazole has data for concomitant use with COX-2 inhibitors2, on the basis of which it has recently been approved for use with this class of drugs. Reference: 1. Laine L. The gastrointestinal effects of nonselective NSAIDs and COX-2- selective inhibitors. Semin Arthritis Rheum 2002; 32 Suppl 1; 25-32. 2. Schieman JMet al. Esomeprazole prevents gastric and duodenal ulcers in at-risk patients on continuous non-selective or COX-2-selective NSAID therapy. Gastroenterology 2004;126(4 Suppl 2):A-82, Abs 638. Competing interests: I have been involved in Multi Centre studies sponsored by AstraZeneca and have received honoraria for lecturing at sponsored medical meetings. |
|||