- Email: [email protected]
Shall we nim a horse?
THE LANCET 8965
Shall
we
nim
a
horse?
We know very little about the
science;
our
prevalence of fraud in mainly from Nevertheless, a compilation of 46
information
comes
isolated cases. documented instances of scientific misconduct in biomedical research since 1974 suggests that it is uncomfortably common.’ The estimate, published in Science in 1987, that only 1 in 1000 000 scientific publications was anything other than accurate and truthful begins to look fatuously optimistic.2 Scientific fraud has lately attracted considerable media attention. The public are understandably upset when they learn that what was trumpeted as a breakthrough is in fact a fabrication. Such a case hit the headlines in the UK earlier this month when Mr Malcolm Pearce, formerly a consultant gynaecologist at St George’s Hospital in London, was found by the General Medical Council’s Professional Conduct Committee to have deceived the medical profession and the public for his own professional gain. Mr Pearce had falsified medical records to claim a "first" with respect to relocation of an ectopic embryo within the uterus, and had invented a 3-year trial of treatment for recurrent miscarriage. His name was struck off the Medical
Register. However, focusing
on
deceive distracts from what
systematic attempts to might be a much bigger is a clear dichotomy
there between the actions of a few fraudsters and those of an overwhelming majority of honest researchersrotten apples in a barrel of otherwise untainted fruit-but it seems more likely that scientists vary in their possession of the qualities of honesty and integrity and that the examples of fraud unearthed so far represent the extreme of the distribution. The corollary is that scientific dishonesty of a less serious type might be found in much greater quantities. John Byrom, a minor eighteenth century poet, wrote some verse about a couple of likeable scoundrels discussing what to do about their lack of transport.’ Stealing a horse would clearly be wrong; filching one would be little better. Could they pilfer one? No, that would be less than honest. Yet, by the end of the poem, they decide that they need not think too badly of themselves if they nim a horse.
problem. Perhaps
June 24, 1995
Scientists practise the same sort of semantic sleightof-hand. Inventing data would clearly be wrong; suppression of inconvenient results would be less than honest. Yet they need not think too badly of themselves if they gloss over the study’s methodological shortcomings, optimise the statistical analysis, cite published work selectively, or perhaps make someone a gift of authorship. Peer review (at least as the process is operated at the moment) is unable to detect even major fraud reliably. Detection of fraud is not its main purpose. The response from funding agencies, journals, and governments has been to recommend increased vigilance and stricter audit-the recent statement from the European Medical Research Councils is an example. A few countries have introduced a sort of scientific police force. The Danish Medical Research Council has appointed standing committees to investigate scientific dishonesty, and the USA has an Office of Research Integrity. In Britain, the Royal College of Physicians announced on June 16 that it is calling together interested parties to establish a central body to evaluate allegations of fraud and misconduct in medical research. Such systems, even if they are not yet very effective, have the potential to do something. The cost of the actions may nevertheless be heavy. Allegations of data fabrication and delay in publication of results showing an adverse outcome of treatment led to the forced resignation last year of the director of the National Surgical Adjuvant Breast and Bowel Project in the USA and suspension of new patient recruitment in several studies. An approach that damages an imaginative programme of research cannot be the best way of dealing with allegations of scientific misbehaviour.’ Rather than concentrate on establishing guilt and punishing the transgressors, we should consider for a moment how the moral economy of the scientific community tolerates or even rewards fraud. We need to be aware that the everyday behaviour of researchers and what they are prepared to let pass in the behaviour of colleagues and collaborators sets a standard that determines the quantity of more extreme forms of dishonesty. If the average scientist
1585
talk himself or herself into nimming a horse, others will try to steal the whole stable. Shapin has commented on the central role of trust in science.6 Most of what we know about the natural world depends upon it. Readers who believe that the earth goes around the sun, that DNA is the genetic substance, that there are such things as electrons, and that light travels at 186 000 miles per second demonstrate the extent of their trust. They have not established these things for themselves. Scientists are of course sceptical on occasions, and sometimes try to replicate the claims of others. However, trust is apparent even in acts of scepticism. Imagine that a molecular biologist doubted reports that HIV contained RNA. He secures a supply of the virus and analyses its nucleic acid content. He justifiably feels that he has rejected trust and sought personal verification. But his act of doubt is possible only because he takes almost everything else relevant to this act on trust: the identity of the virus with which he has been supplied, the identity and claimed purity of the reagents used in the assay, the labelled speed of the centrifuge and the proper working of other instruments, and the honesty and competence of technicians and of the authors of the papers and manuals containing the assay methods. Distrust can exist only at the margins of trusting systems. Attempts to subject the biomedical research community to vigilant policing are more likely to kill the patient than eradicate the disease. Vigilance amounts to enforcement of scepticism and distrust; it damages science because trusting is essential to the making of knowledge. The Greek notion of a philosopher was someone who loved truth and would not lie. To the founder members of the Royal Society-Robert Boyle, Thomas Willis, and Christopher Wren-a gentleman was someone who valued the integrity of his word as he valued his sense of honour. Both the Greeks and the founder members of the Royal Society understood that virtue had to be practised and believed that it could be taught. They would not have found it embarrassing or old-fashioned to talk about the inspirational effect of seeing respected individuals doing science out of curiosity, for the pleasure of solving puzzles and finding out how the universe worked. These individuals were understood not to lie because nothing they wanted could have been gained from a lie. If we are serious about addressing problems of intellectual dishonesty we could do worse than try to revive such a culture of virtue. Is this sentiment unrealisably Utopian? Many will doubtless think so, but it is the only response to scientific dishonesty that will benefit rather than harm science. It can also be translated into action. can
1586
first step, institutions should recognise explicitly that junior researchers absorb the ethical standards of those who supervise their work. Heads of departments who, for example, claim authorship without contributing more than the infrastructure for research (the results of a recent survey reveal that there are many of them’) are part of an ecology in which fraud will thrive. Universities can surely find a way to clean up this environment. Another simple step would be for medical research councils and other funding bodies to insist, as a condition of the award of a research grant, that data collected during a study are archived. Within limitations imposed by considerations of confidentiality and the need to give the investigators time to publish, the data would be made available to others. This policy would signal that data are not owned in perpetuity by the investigators who first collected them but constitute a resource for the whole community, and would likewise show that
As
a
bodies who fund biomedical science expect openness and honesty from those to whom they award grants. Above all, it would encourage an approach to fraud in which scientists policed themselves. The fact that the Economic and Social Research Council in Britain already operates such a policy demonstrates its practicability. Journals could also lend
weight to this proposal, by making the archiving of data a condition of acceptance for publication.
their
The Lancet 1
2
3 4
Lock S. Research misconduct: a resumé of recent events. In: Lock S, Wells F, eds. Fraud and misconduct in medical research. London: BMJ Publishing Group, 1993: 5-24. Koshland DE. Fraud in science. Science 1987; 235: 141. Byrom J. The nimmers. In: Johnson S, ed. The works of the English poets, from Chaucer to Cowper. London: J Johnson, 1810: 198-99. Evered D, Lazar P. Misconduct in medical research. Lancet 1995; 345: 1161-62.
5 6
7
Editorial. Breast cancer: clearing trails in the forest without losing our way. Lancet 1994; 343: 1049-50. Shapin S. Trust, honesty and the authority of science. In: Bulger RE, Bobby EM, Fineberg HV, eds. Society’s choices: social and ethical decision making in biomedicine. Washington: National Academy Press, 1995; 388-408. Goodman NW. Survey of fulfilment of criteria for authorship for published medical research. BMJ 1994; 309: 1482.
Beyond all
reasonable DNA
"Once you have eliminated the
impossible, then whatever
remains, however improbable, must be the truth." So said Sherlock Holmes, and so do prosecutors argue in that hang upon the evidence of DNA The difficulty, which the great fictional detective swept under the carpet, is in deciding what remains. What are the alternative explanations of the evidence? In law, juries must be convinced "beyond all reasonable doubt"-but what doubts are reasonable? No scientific theory is ever beyond reasonable doubt, because
criminal
cases
fingerprints.
Shall
we
nim
a
horse?
We know very little about the
science;
our
prevalence of fraud in mainly from Nevertheless, a compilation of 46
information
comes
isolated cases. documented instances of scientific misconduct in biomedical research since 1974 suggests that it is uncomfortably common.’ The estimate, published in Science in 1987, that only 1 in 1000 000 scientific publications was anything other than accurate and truthful begins to look fatuously optimistic.2 Scientific fraud has lately attracted considerable media attention. The public are understandably upset when they learn that what was trumpeted as a breakthrough is in fact a fabrication. Such a case hit the headlines in the UK earlier this month when Mr Malcolm Pearce, formerly a consultant gynaecologist at St George’s Hospital in London, was found by the General Medical Council’s Professional Conduct Committee to have deceived the medical profession and the public for his own professional gain. Mr Pearce had falsified medical records to claim a "first" with respect to relocation of an ectopic embryo within the uterus, and had invented a 3-year trial of treatment for recurrent miscarriage. His name was struck off the Medical
Register. However, focusing
on
deceive distracts from what
systematic attempts to might be a much bigger is a clear dichotomy
there between the actions of a few fraudsters and those of an overwhelming majority of honest researchersrotten apples in a barrel of otherwise untainted fruit-but it seems more likely that scientists vary in their possession of the qualities of honesty and integrity and that the examples of fraud unearthed so far represent the extreme of the distribution. The corollary is that scientific dishonesty of a less serious type might be found in much greater quantities. John Byrom, a minor eighteenth century poet, wrote some verse about a couple of likeable scoundrels discussing what to do about their lack of transport.’ Stealing a horse would clearly be wrong; filching one would be little better. Could they pilfer one? No, that would be less than honest. Yet, by the end of the poem, they decide that they need not think too badly of themselves if they nim a horse.
problem. Perhaps
June 24, 1995
Scientists practise the same sort of semantic sleightof-hand. Inventing data would clearly be wrong; suppression of inconvenient results would be less than honest. Yet they need not think too badly of themselves if they gloss over the study’s methodological shortcomings, optimise the statistical analysis, cite published work selectively, or perhaps make someone a gift of authorship. Peer review (at least as the process is operated at the moment) is unable to detect even major fraud reliably. Detection of fraud is not its main purpose. The response from funding agencies, journals, and governments has been to recommend increased vigilance and stricter audit-the recent statement from the European Medical Research Councils is an example. A few countries have introduced a sort of scientific police force. The Danish Medical Research Council has appointed standing committees to investigate scientific dishonesty, and the USA has an Office of Research Integrity. In Britain, the Royal College of Physicians announced on June 16 that it is calling together interested parties to establish a central body to evaluate allegations of fraud and misconduct in medical research. Such systems, even if they are not yet very effective, have the potential to do something. The cost of the actions may nevertheless be heavy. Allegations of data fabrication and delay in publication of results showing an adverse outcome of treatment led to the forced resignation last year of the director of the National Surgical Adjuvant Breast and Bowel Project in the USA and suspension of new patient recruitment in several studies. An approach that damages an imaginative programme of research cannot be the best way of dealing with allegations of scientific misbehaviour.’ Rather than concentrate on establishing guilt and punishing the transgressors, we should consider for a moment how the moral economy of the scientific community tolerates or even rewards fraud. We need to be aware that the everyday behaviour of researchers and what they are prepared to let pass in the behaviour of colleagues and collaborators sets a standard that determines the quantity of more extreme forms of dishonesty. If the average scientist
1585
talk himself or herself into nimming a horse, others will try to steal the whole stable. Shapin has commented on the central role of trust in science.6 Most of what we know about the natural world depends upon it. Readers who believe that the earth goes around the sun, that DNA is the genetic substance, that there are such things as electrons, and that light travels at 186 000 miles per second demonstrate the extent of their trust. They have not established these things for themselves. Scientists are of course sceptical on occasions, and sometimes try to replicate the claims of others. However, trust is apparent even in acts of scepticism. Imagine that a molecular biologist doubted reports that HIV contained RNA. He secures a supply of the virus and analyses its nucleic acid content. He justifiably feels that he has rejected trust and sought personal verification. But his act of doubt is possible only because he takes almost everything else relevant to this act on trust: the identity of the virus with which he has been supplied, the identity and claimed purity of the reagents used in the assay, the labelled speed of the centrifuge and the proper working of other instruments, and the honesty and competence of technicians and of the authors of the papers and manuals containing the assay methods. Distrust can exist only at the margins of trusting systems. Attempts to subject the biomedical research community to vigilant policing are more likely to kill the patient than eradicate the disease. Vigilance amounts to enforcement of scepticism and distrust; it damages science because trusting is essential to the making of knowledge. The Greek notion of a philosopher was someone who loved truth and would not lie. To the founder members of the Royal Society-Robert Boyle, Thomas Willis, and Christopher Wren-a gentleman was someone who valued the integrity of his word as he valued his sense of honour. Both the Greeks and the founder members of the Royal Society understood that virtue had to be practised and believed that it could be taught. They would not have found it embarrassing or old-fashioned to talk about the inspirational effect of seeing respected individuals doing science out of curiosity, for the pleasure of solving puzzles and finding out how the universe worked. These individuals were understood not to lie because nothing they wanted could have been gained from a lie. If we are serious about addressing problems of intellectual dishonesty we could do worse than try to revive such a culture of virtue. Is this sentiment unrealisably Utopian? Many will doubtless think so, but it is the only response to scientific dishonesty that will benefit rather than harm science. It can also be translated into action. can
1586
first step, institutions should recognise explicitly that junior researchers absorb the ethical standards of those who supervise their work. Heads of departments who, for example, claim authorship without contributing more than the infrastructure for research (the results of a recent survey reveal that there are many of them’) are part of an ecology in which fraud will thrive. Universities can surely find a way to clean up this environment. Another simple step would be for medical research councils and other funding bodies to insist, as a condition of the award of a research grant, that data collected during a study are archived. Within limitations imposed by considerations of confidentiality and the need to give the investigators time to publish, the data would be made available to others. This policy would signal that data are not owned in perpetuity by the investigators who first collected them but constitute a resource for the whole community, and would likewise show that
As
a
bodies who fund biomedical science expect openness and honesty from those to whom they award grants. Above all, it would encourage an approach to fraud in which scientists policed themselves. The fact that the Economic and Social Research Council in Britain already operates such a policy demonstrates its practicability. Journals could also lend
weight to this proposal, by making the archiving of data a condition of acceptance for publication.
their
The Lancet 1
2
3 4
Lock S. Research misconduct: a resumé of recent events. In: Lock S, Wells F, eds. Fraud and misconduct in medical research. London: BMJ Publishing Group, 1993: 5-24. Koshland DE. Fraud in science. Science 1987; 235: 141. Byrom J. The nimmers. In: Johnson S, ed. The works of the English poets, from Chaucer to Cowper. London: J Johnson, 1810: 198-99. Evered D, Lazar P. Misconduct in medical research. Lancet 1995; 345: 1161-62.
5 6
7
Editorial. Breast cancer: clearing trails in the forest without losing our way. Lancet 1994; 343: 1049-50. Shapin S. Trust, honesty and the authority of science. In: Bulger RE, Bobby EM, Fineberg HV, eds. Society’s choices: social and ethical decision making in biomedicine. Washington: National Academy Press, 1995; 388-408. Goodman NW. Survey of fulfilment of criteria for authorship for published medical research. BMJ 1994; 309: 1482.
Beyond all
reasonable DNA
"Once you have eliminated the
impossible, then whatever
remains, however improbable, must be the truth." So said Sherlock Holmes, and so do prosecutors argue in that hang upon the evidence of DNA The difficulty, which the great fictional detective swept under the carpet, is in deciding what remains. What are the alternative explanations of the evidence? In law, juries must be convinced "beyond all reasonable doubt"-but what doubts are reasonable? No scientific theory is ever beyond reasonable doubt, because
criminal
cases
fingerprints.