> Message from: > > Wim J. van der Steen and Vincent K.Y. Ho > Faculty of Biology > De Boelelaan 1087 > 1081 HV Amsterdam > the Netherlands > E-mail: <[log in to unmask]> and <[log in to unmask]> > > Dear colleagues, > > We are sending this request to students and professionals in areas, pure > and applied, of the life sciences (biology, medicine, psychiatry). Also > included are areas that may help to uncover strengths or flaws in the life > sciences, for example philosophy and sociology of science, psychology, > technology assessment, ethics. > > The request is to submit to us an essay of at most 1000 words in response > to the text below (written by WJS). Your essay may amount to a commentary, > but you may also choose to express your own view about any particular theme > in the text, while disregarding the text itself. For example, you may > choose to focus on benefits or dangers of medication. > The text is technical since it comprises many themes in a succint > way. In contrast, we would like you to use a more popular style. > > Some essays will be included in a primer we are writing this year, which > aims to offer guidelines beyond fashionable ones, both for writing and for > interpreting texts in the life sciences. In the primer, we will argue that > knowledge accumulation currently produces chaos best revealed by showing > that disciplines covering similar themes are out of touch with eachother, > that philosophy made practical should have remedial functions (see W.J. van > der Steen, "A Practical Philosophy for the Life Sciences", SUNY Press > 1993), and that ethical concerns should intrinsically contribute to science > in progress (see W.J. van der Steen, "Facts, Values and Methodology; A New > Approach to Ethics", Rodopi 1995). > Essay not included in the primer will become part of course > materials for students. > > The text will also develop into a longer, more popular version to be > published as a seperate article. In case your essay helps shape this > article, we will mention this in the acknowledgements. > > If you wish, you may remain anonymous in the primer and the course > materials. But we should be able to mention your position (e.g., graduate > student, professor, science reporter), the discipline(s) you are working in > and/or are acquainted with, and your country. Please let us know by email > to Vincent K.Y. Ho <[log in to unmask]> if you are willing to cooperate. > Our schedule calls for receiving essays before the end of March. > > Please forward our message to persons known to you who might be interested, > or to members of internet-lists. > > Format for response: > > I will contribute an essay. > I wish to remain anonymous/I do not wish to remain anonymous > > Name: > Position: > Discipline (expertise): > Country: > > BIASED PROMOTION OF DRUGS: THE CASE OF NSAIDS > > WIM J. VAN DER STEEN > > Cultural factors affect medical science. Few researchers would quarrel with > this. The point has been made in a general way by sociologists of science. > But few sociologists would take their cues from primary sources of medical > research. Conversely, medical researchers in their writings seldom consider > how culture affects their work. I aim to close this gap by a case study > revealing profound negative effects of cultural factors in medical research > and practice in the wake of it. My subject is a class of drugs, non-steroid > anti-inflammatory drugs, NSAIDs, which are used on a massive scale although > they have serious side-effects. Overuse of these drugs is caused by several > forms of bias that are seldom investigated in medicine. Regulation has been > biased in favour of the pharmaceutical industry to the detriment of > patients. Further, researchers funded by a pharmaceutical industry > manufacturing a drug, tend to produce biased reports in favour of the drug. > Medical research is also biased by its focus on drug therapies. Diet > therapies may be equally effective and without serious side-effects. > Further, inappropriate prescriptions are common. Some forms of bias are > unavoidable. Thus, the methodology chosen for controlled studies depends on > ethical decisions. All the forms of bias deserve more scrutiny. It is > anyhow desirable that we reduce the use of NSAIDs. > > EFFECTS OF NSAIDS > Many studies indicate that NSAIDs may palliate symptoms in inflammatory > diseases such as rheumatoid arthritis. But the drugs have dangerous > side-effects also. A recent study notes that 10-40% of elderly people in > the populations studied use NSAIDs, while the drugs frequently have > gastrointestinal side-effects. NSAIDs increase the risk of peptic ulcer > complications by 3-5-fold, and in several different populations it has been > estimated that 15-35% of all peptic ulcer complications are due to NSAIDs. > In the United States alone, there are an estimated 41,000 hospitalisations > and 3,300 deaths each year among the elderly that are associated with > NSAIDs (Griffin, 1998). In another study the drugs are even estimated to > cause 20,000 deaths in the US each year (Wolfe, 1996). Any cursory > inspection of the literature will uncover many articles with a similar > message. > The alleged benefits of NSAIDs are not unchallenged. One source > describes recent data regarding the prognosis of rheumatoid arthritis with > the use of the current drug treatments as "gloomy" (Odeh, 1997), and > according to another source, most standard medical approaches to treatment > have had little or no impact on the course of rheumatoid disease (Schiff, > 1997). Further, we should not simply take positive data in the literature > for granted. Controlled trials concerning drugs in rheumatoid arthritis are > often problematic due to pitfalls to do with outcome measures, patient > populations, characteristics of the study design, and economic evaluation > (Ortiz, Tugwell and Yocum, 1997). Impressions concerning the long-term > natural history and results of therapy are strongly influenced by the types > of measures and study designs used to assess patient status and outcomes > (Pincus, 1995). Randomised controlled trials are often a poor measure of > long-term outcomes (Pincus and Stein, 1997; Simon, 1995). > > A BROAD PERSPECTIVE > How should we explain the massive use of NSAIDs? The most obvious answer > would be that controlled trials indicate that NSAIDs alleviate symptoms of > inflammation, at least in the short term, that they admittedly have > dangerous side-effects, but that the overall balance of evidence warrants > continued usage. This answer is unsatisfactory. It presupposes that the > research is unbiased and that it affects practice in an unbiased way. I > would assume that the wide-spread use of NSAIDs does not sit well with the > observed side-effects, and that it points to undesirable forms of bias. Let > me review some potential sources of bias. > Abraham offers a detailed survey spanning decades of NSAID > regulation in the US and the UK (Abraham, 1995). The survey demonstrates > that research by the pharmaceutical industry on benefits and side-effects > of NSAIDs has been biased in favour of its own products. That should not > surprise us. Surprising is incontrovertible evidence that authorities > responsible for regulation have on a grand scale acted in violation of > their own guidelines, so as to favour the industry, to the detriment of > patients. Many dangerous NSAIDs now on the market should never have been > approved for sale. > Bias in favour of the industry may also exist in medical research > itself. The chances that researchers observe positive effects of NSAIDS are > way above average if they are paid by the industry manufacturing the drugs > (Goetsche 1989). > Medical practice also appears to be biased in favour of NSAIDs, > since unnecessary prescriptions are very common (Coste, Hanotin and > Leutenegger, 1995; Tamblyn, Berkson, Dauphinee et al., 1997). Further, in > spite of the side-effects noted, the use of NSAIDs is fostered by their > being available as over-the-counter drugs. > These kinds of bias could be avoided in principle, but other forms > of bias are unavoidable. Consider the problem of false positives > (statistical type I errors) and false negatives (statistical type II > errors). Of course we should aim at tests ensuring that the probabilities > of errors of either kind are low. The trouble is that a decrease in one > type of error logically entails an increase in the other type of error > (with a given sample size and effect size). Thus, the methodological set-up > of our research depends on ethical considerations concerning the > acceptability of particular errors which are part of our culture (Cranor, > 1992). This can be generalised. Ideal experiments cannot satisfy all > methodological requirements at the same time. We have to prioritise > requirements, and that cannot be done without ethical considerations (Van > der Steen, 1995). > As I noted, some medical researchers have argued that the > methodological set-up of experiments affects conclusions about the value of > NSAIDs. We should add to this that the choice of a methodology is > determined in part by ethical issues which are seldom made explicit in > medical research. They should be brought out into the open. We are not > dealing here with bias in a pejorative sense, because we have to opt for > some methodology, but we should be aware of normative presuppositions. > Considering tests of drug effects, patients and the industry have > different interests with respect to different types of error. In the > interest of patients, we would like to minimize type I errors of falsely > attributing positive effects to drugs. But the industry would rather > benefit from low levels of type II errors of falsely attributing no effects > to drugs. In most research, significance testing is done without power > analysis. This allows for type I errors at the standard 5% level, while > type II errors remain unkown. If tests of a drug are repeated over and over > again, a common practice nowadays, we will sooner or later get a > significant result due to a type I error. To the extent that drugs are > marketed on the basis of this, the type I error level is inflated while the > type II error level is decreased, and we get a bias in favour of the > industry, to the detriment of patients. Patients and the industry also have > different interests in the study of adverse side-effects. The trouble is > here that methodologically proper tests are hardly feasible since the set > of potential side-effects is open-ended, and guidelines for the time-scale > of experiments are arbitrary. In the interest of patients, we should aim at > comprehensive short-term and long-term studies, but this would not be > palatable for the industry. In practice, comprehensive long-term studies > are rare, which again favours the industry. Next, controlled experiments > are problematic if drugs are compared with a placebo without side-effects. > The chances are that this will generate false positives when patients > conclude from experienced side-effects that they are receiving the drug > rather than the placebo. This kind of phenomenon has on a grand scale > resulted in inappropriate drug treatments in psychiatry (Jacobs, 1995). > Medical research may also be biased in that the choice of subjects > to be investigated is one-sided. Concerning therapies for rheumatoid > arthritis, research reveals a pronounced bias in favour of drugs. As I am > writing this (August 1999), Medline lists 11,017 articles on drug therapy, > 126 articles on diet therapy. This fits in with a pervasive > negligence of ecology in medical research and medical practice (Chivian, > McCally, Hu and Haines, 1993; Garrett, 1994; Van der Steen, 1998a). Diet > therapy, for that matter, does seem to have positive effects. > Particular diets, or components of diet, most notably particular > unsaturated fatty acids, reduce disease activity in rheumatoid arthritis > (Ariza-Ariza, Mestanza-Peralta and Cardiel, 1998; De Luca, Rothman and > Zurier, 1995; James and Cleland, 1997; Rothman, De Luca and Zurier, 1995). > It is indeed conceivable that poor diet is among the factors which > predispose toward the disease (Kowsari, Finnie, Carter et al., 1983; > Morgan, Hine, Vaughn and Brown, 1993). We should add to this that modern > techniques of food processing cause the destruction of natural components > of diets, not least unsaturated fatty acids including essential ones. This > fosters degenerative and systemic diseases, and it may well contribute to > aetiology in rheumatoid arthritis (Erasmus, 1995). > The emphasis in medical research on pathology is now overwhelmingly > on genetics and molecular biology. In part, this may be explained by the > history of medicine, which developed biological approaches prior to the > emergence of ecology. Further, this emphasis fits in with current > reductionistic paradigms. These paradigms foster a self-perpetuating bias. > Considering genetics, I have elsewhere explained how this works, by an > analysis of research on schizophrenia (Van der Steen, 1998b). Twin studies > have shown that concordance for schizophrenia is much higher in identical > than in non-identical twins. Researchers have concluded from this that > genes are implicated in etiology. The following counterexample shows how > problematic this is. Suppose, a number of twin pairs are on a trip in a > coastal plain. A flood wrecks the trip. Tall individuals survive, and short > individuals drown since they cannot keep their heads above the water. > Identical twins will obviously show a higher concordance than non-identical > twins. If we follow the logic of schizophrenia research, we should conclude > that genetic factors rather than the flood were causally responsible for > the drowning. Something is obviously wrong with the logic of common > research. > Researchers do admit that the environment modifies the effects of > genetic factors. But they commonly conceptualise the environment as the > psychosocial environment. Ecological factors are simply disregarded. This > happens in psychiatry and in many areas of medicine. > > CONCLUSIONS > NSAIDs are reported to alleviate symptoms in inflammatory diseases. This > may be true, but we should beware of exaggerating positive effects, and we > should realise that short-term benefits may not be extrapolatable. The > pervasive use of these drugs is indeed problematic in view of serious > side-effects. If we wish to assess NSAIDs, we should not be content with an > appraisal of existing controlled trials. Such an appraisal would not > uncover all sources of bias. > Research mostly conducted outside medicine itself has revealed many > sources of cultural bias. Drug regulation is biased in favour of the > pharmaceutical industry. Likewise for medical research funded by the > industry. Further, the emphasis in research is overmuch on drug therapy, > while less dangerous treatments such as diet therapy may be equally > effective. > It is desirable that medical journals pay more attention to these > forms of bias. More research is needed on alternative treatments, > especially diet therapy. The highest priority should go to trials that aim > to compare drug therapies with diet therapies. We might save thousands of > lives if we replace drug therapies by diet therapies. > > REFERENCES > Abraham, J. (1995) Science, Politics and the Pharmaceutical Industry: > Controversy and Bias in Drug Regulation. London: UCL Press. > Ariza-Ariza, R., Mestanza-Peralta, M., Cardiel, M.H. (1998) 'Omega-3 fatty > acids in rheumatoid arthritis: an overview', Seminars in Arthritis and > Rheumatism 27: 366-370. > Chivian, E., McCally, M., Hu, H., Haines, A. (1993) Critical Condition: > Human Health and the Environment. Cambridge, Mass.: MIT Press. > Coste, J., Hanotin, C. and Leutenegger, E. (1995) '[Prescription of > non-steroidal anti-inflammatory agents and risk of iatrogenic adverse > effects: a survey of 1072 French general practitioners] [French]', > Therapie, 50: 265-270. > Cranor, C.F. (1992) Regulating Toxic Substances: A Philosophy of Science > and the Law. New York: Oxford. > DeLuca, P., Rothman, D., Zurier, R.B. (1995) 'Marine and botanical lipids > as immunomodulatory and therapeutic agents in the treatment of rheumatoid > arthritis', Rheumatic Diseases in the Clinics of North America, 21: 759-777. > Erasmus, U. (1995) Fats That Heal, Fats That Kill. Burnaby: Alive Books > (third ed.). > Garrett, L. (1994) The Coming Plague: Newly Emerging Diseases in a World > Out of Balance. New York: Farrar, Straus, and Giroux. > Goetsche, P.C. (1989) 'Methodology and overt and hidden bias in reports in > 196 double-blind trials of nosteroid anti-inflammatory drugs in rheumatoid > arthritis', Controlled Clinical Trials, 10: 31-56. > Griffin, M.R. (1998) 'Epidemiology of nonsteroidal anti-inflammatory > drug-associated gastrointestinal injury', American Journal of Medicine, > 104: 23S-29S, 41S-42S. > Jacobs, D.H. (1995) 'Psychiatric drugging: Forty years of pseudo-science, > self-interest, and indifference to harm', Journal of Mind and Behavior, 16: > 421-470. > James, M.J. and Cleland, L.G. (1997) 'Dietary n-3 fatty acids and therapy for > rheumatoid arthritis', Seminars in Arthritis and Rheumatism, 27:85-97. > Kowsari, B., Finnie, S.K., Carter, R.L. et al. 'Assessment of the diet of > patients with rheumatoid arthritis and osteoarthritis', Journal of the > American Dietetic Association, 82: 657-659. > Morgan, S.L., Hine, R.J., Vaughn, W.H., Brown, A. (1993) 'Dietary intake > and circulating vitamin levels of rheumatoid arthritis patients treated > with methotrexate', Arthritis Care Research, 6: 4-10. > Odeh, M. (1997) 'New insights into the pathogenesis and treatment of > rheumatoid arthritis', Immunopathology, 83: 103-116. > Ortiz, Z., Tugwell, P. and Yocum, D.(1997) 'The design of clinical trials > aimed at assessing the DC-ART properties of new molecules in rheumatoid > arthritis', Clinical and Experimental Rheumatology, 15: S39-S44. > Pincus T. (1995) 'Assessment of long-term outcomes of rheumatoid arthritis. > How choices of measures and study designs may lead to apparently different > conclusions', Rheumat. Diseases in the Clinics of North America, 21: > 619-654. > Pincus, T., and Stein, C.M. (1997) 'Why randomized controlled clinical > trials do not depict accurately long-term outcomes in rheumatoid arthritis: > some explanations and suggestions for future studies', Clinical and > Experimental Rheumatology, 15: S27-38. > Rothman, D., DeLuca, P. and Zurier, R.B. (1995) 'Botanical lipids: effects > on inflammation, immune responses, and rheumatoid arthritis', Seminars in > Arthritis and Rheumatism, 25: 87-96. > Schiff, M. (1997) 'Emerging treatments for rheumatoid arthritis', American > Journal of Medecine, 102: 11S-15S. > Simon, L.S. (1995) 'Actions and toxicity of nonsteroidal anti-inflammatory > drugs', Current Opinion in Rheumatology, 7:159-166. > Tamblyn, R., Berkson L., Dauphinee, W.D. et al. (1997) 'Unnecessary > prescribing of NSAIDs and the management of NSAID-related gastropathy in > medical practice', Annals of Internal Medecine, 127: 429-438. > Van der Steen, W.J. (1995) Facts, Values, and Methodology: A New Approach > to Ethics. Amsterdam and Atlanta: Rodopi. > Van der Steen, W.J. (1998a) 'Forging links between philosophy, ethics, and > the life sciences: A tale of disciplines and trenches', History and > Philosophy of the Life Sciences 20: 233-248. > W.J. van der Steen (1998b) 'Bias in behavior genetics: An ecological > perspective', Acta Biotheoretica 46: 369-377. > Wolfe, M.M. (1996) 'NSAIDs and the gastrointestinal mucosa', Hospital > Practice, 31: 37-44, 47-48.