Counterpoint: Full Disclosure—Where Is theEvidence for Nefarious Conflicts of Interest?Robert M. Sade, MD

Department of Surgery and Institute of Human Values in Health Care, Medical University of South Carolina, Charleston,South Carolina

Recent years have seen increasing focus on relationsbetween physicians and industry, particularly phar-

maceutical and device companies, and the conflicts ofinterest (COIs) that might arise from them [1]. Specificabuses of such relationships have often garnered na-tional attention through scholarly journals and publiccommunications media [2, 3].

For related article, see page 413

In May 2009, the American Association for ThoracicSurgery (AATS) and The Society of Thoracic Surgeons(STS) jointly adopted a report establishing standardsfor proper relationships between individual cardiotho-racic surgeons and industry [4, 5]. One of the standardsaddresses the question of disclosure: “Membersshould disclose their own or their institutions’ financialrelationship with the manufacturer of a drug or devicewhenever clinical research or experience with a partic-ular procedure or device is presented at a meeting or ispublished.”

This standard is stated in broad terms, and the precisemeaning of “disclose” is not spelled out. The AATS andthe STS require that all authors must disclose relation-ships with companies that could constitute a COI beforetheir papers can be presented at annual meetings [6, 7].Both organizations report only the name of the possiblyconflicted author, the company with which he has arelationship, and the nature of the relationship. No otherdetails are published in the annual meeting program.

The AATS and STS official journals, The Annals ofThoracic Surgery (ATS) and Journal of Thoracic and Cardio-vascular Surgery (JTCVS), respectively, have specified thedetails of industry relationships that must be disclosedbefore a paper will be published. If a conflict is disclosed,ATS places a text box on the title page of the publishedarticle: “Dr X discloses that he/she has a financial rela-tionship with company Y.” No additional details areprovided [8]. The JTCVS disclosure statement requiresreporting not only the name of the company with whicha conflict exists and the nature of the relationship, butalso the dollar amount received by the author—other

Address correspondence to Dr Sade, Department of Surgery and Instituteof Human Values in Health Care, Medical University of South Carolina,

25 Courtenay Dr, Ste 7018, MSC 295, Charleston, SC 29425-2950; e-mailsader@musc.edu; http://www.values.musc.edu.

© 2011 by The Society of Thoracic SurgeonsPublished by Elsevier Inc

details, such as the duration of the relationship, areomitted [9].

The published disclosures of COIs for these annualmeetings and journal publications include neither dollaramounts nor duration of the reported relationships.

Should Full Disclosure Include Dollar Amountsand Relationship Duration?

Cardiothoracic surgeons are familiar with the require-ments of disclosure of COIs in presentations and inpublications, as they are ubiquitous in presentations atmeetings and papers in journals. In answer to the ques-tion of whether increased levels of disclosure are needed,J. Peter Murphy responds in the affirmative in the accom-panying article, and argues that we should add dollaramounts and duration of relationships to current disclo-sure requirements [10]. He cites the work of JeromeKassirer approvingly in support of his position. This isquite appropriate, as Kassirer has been a vocal leader ofthe critics of relations between physicians and industry.Yet, he is not an admirer of disclosure in general: “Wemust be highly skeptical about the most popular ap-proach to managing conflicts of interest, namely, disclos-ing the relationships [11].” There is good reason forKassirer’s skepticism. Disclosures of conflicts are rarelyverified, making their accuracy unreliable; listeners andreaders are often unable to identify biased informationbecause of their lack of expertise; and attention may bedistracted from bias by implying that the disclosure itselfdiminishes the speaker’s or author’s bias [2]. Moreover,the plethora of disclosures at conferences and in publi-cations may dull sensitivity and interest in them by virtueof frequent repetition, inuring receivers of the informa-tion to the relevance of disclosures [12].

Much anecdotal information has been published aboutbiased presentations and publications [13]. Yet, the grav-ity of problems related to inaccurate, unbalanced, orprejudicial presentations is poorly understood owing tothe paucity of systematic investigations of the frequencyand impact of such presentations.

Arguments Against Disclosure of RelationshipDuration and Payments

Dr Murphy makes several observations with which weagree: COIs are not related only to money, but also tononfinancial considerations such as career advancement

and academic achievement; the collaboration between

Ann Thorac Surg 2011;92:417–20 • 0003-4975/$36.00doi:10.1016/j.athoracsur.2011.05.078

418 ETHICS IN CARDIOTHORACIC SURGERY SADE Ann Thorac SurgNEFARIOUS COI 2011;92:417–20

industry and physicians has been highly productive andbeneficial for patients in the past; some, but not all,financial interactions between industry and physiciansare unseemly or worse; hard evidence for the existence orseverity of the problem is in short supply; and, finally, thestatus quo is unsatisfactory.

In making the case for increasing disclosure require-ments to include dollar amounts and duration of rela-tionships, however, Murphy cites what he understandsto be the two “legitimate” arguments against such arequirement: it is an intrusive, gratuitous invasion ofprivacy, and collecting the data and reporting it aredifficult and costly. These are legitimate arguments,but they seem of minor importance compared with twomuch more important counterpoints to the case forrequiring more detail in disclosures: the real danger ofslowing if not interrupting the highly productive re-search and innovation that has come from physician-industry collaboration over the last half-century, andthe nearly complete absence of reliable evidence thatthis collaboration or any other interactions betweenphysicians and industry poses a serious threat topatient safety.

Productive Research and Innovation

To appreciate the progress that has been made throughphysician-industry collaboration, one need only men-tion the pump oxygenator developed by Dr John Gib-bon collaborating with engineer Thomas Watson ofIBM, the cardiac valve prosthesis developed by AlbertStarr working with engineer Lowell Edwards (resultingin the creation of the highly successful Edwards Lab-oratories), and external, then implantable, pacemakersdeveloped by Dr C. Walton Lillehei jointly with engi-neer Earl Bakken, creator of Medtronic. These pairsrepresent collaboration between heart surgeons andcommercial engineers, each of which led to a majoradvance in heart surgery. Many other highly produc-tive alliances could be cited in other surgical special-ties, such as orthopedics and neurosurgery.

The contributions to human welfare by cooperativework between physicians and industry—both pharma-ceutical and device companies—have been enormous. Inthe last few decades, fully half of the reduction in deathsdue to coronary artery disease can be attributed tocollaborations that have led to improvements in evi-dence-based therapies [14]. Indeed, such advances wouldnot have been possible without merging the knowledgeand skills of businessmen, engineers, and technical ex-perts in industry with the clinical expertise of physicians,whose knowledge of clinical needs and the ways in whichdrugs and devices interact with the human body isindispensible to biomedical innovation. The higher thebarriers that are erected between the creative minds ofindustry and of medicine, the less innovation and im-provement of human health are likely to occur. Admit-tedly, there is little clear and convincing evidence that

physician-industry collaboration has been harmed by

current disclosure requirements, but given the amplyproved benefits of such alliances and the absence ofscientific evidence that they cause harm, the burden ofproof rests on those who wish to impose additionalregulation.

Absence of Evidence

The facts underlying the national debate on relationsbetween industry and physicians are seen to be meagerwhen stripped of assumptions grounded in intuition andunsupported assertions. Murphy clearly documentsmuch negative publicity in the media, some emanatingfrom Congress, about disturbingly high levels of pay-ment from industry to some physicians. Beyond that,very little hard evidence indicates that physician COIsvis-à-vis industry pose a serious threat to health care.Nearly all of the experimental research underlying thisconcern comes from the psychology and sociology liter-ature, from which physician behavior is projected frominvestigations carried out in nonmedical contexts. Mostof the remaining evidence comes from surveys of theviews and experiences of physicians, residents, and pa-tients [10]. Both of these data sources have low probativevalue.

The ranking and relative value of evidence types hasbeen investigated and described by several groups, suchas the US Preventive Services Task Force (USPSTF) [15]and the Grading of Recommendations Assessment, De-velopment and Evaluation (GRADE) Working Group[16]. Rankings of the relative value of different kinds ofevidence may be helpful in thinking about the questionat hand (Table 1). All of the evidence supporting the needfor a high level of scrutiny and regulation of physician-industry relations falls into code D, quality of evidence �very low, suggesting that existing data are far fromdispositive of this issue. Moreover, the lowest of fiveranks on the USPSTF effectiveness of evidence scale isauthoritative statement: “Opinions of respected authori-ties, based on clinical experience, descriptive studies, orreports of expert committees [17].” We ought to exerciseextreme caution in using currently available information,all of which is of low quality, to make important policydecisions.

Dr Murphy agrees that the data supporting his positionis weak and points to the difficulty of carrying outrandomized controlled trials in this area, implying thatthe only alternative to anecdotes, surveys, expert opin-ions, and extrapolations from experiments in unrelatedfields is a randomized controlled trial. There are manyother kinds of evidence, however, such as nonrandom-ized prospective and retrospective studies of varioustypes. In fact, one particular study that is both feasibleand relevant to the debate has not been done. It is relatedto the critical central issue, the reason why anyone caresabout physicians’ COI: the possibility that COIs causeharm to patients. Harm to patients should be the primaryoutcome measure in studies of the concerns about COI,disclosures, and conflict management. Yet, no study,

retrospective or prospective, has been carried out to

ent an

419Ann Thorac Surg ETHICS IN CARDIOTHORACIC SURGERY SADE2011;92:417–20 NEFARIOUS COI

examine the effects of physicians’ COIs on the outcomesof patient care. In the absence of this critical information,no rational policy to regulate physician-industry rela-tions is possible.

One of the most widely cited and exhaustively re-searched papers on investigations of the effects ofrelations between physicians and the pharmaceuticalindustry was published in 2000 and states this aboutthe outcomes of such relations: “No study used patientoutcome measures [18].” To fill in that information forsubsequent years, this writer undertook an extensivesearch of the biomedical literature and was able to findonly a single paper that addressed, albeit indirectly,COI and clinical outcomes. It is a meta-analysis ofpatient outcomes after autologous chondrocyte im-plantation, comparing commercially funded with inde-pendently funded studies [19]. The authors found nodifference in patient outcomes between the twogroups. Although this evidence is from meta-analysis,and therefore of only intermediate quality, it does notsupport a need for intensified regulation of physician-industry relations.

Until we have well-designed and executed studies ofphysician-industry COI that use the outcomes of patientcare as the primary outcome measure, we will not be ableto weigh the benefits to patients (known and documentedmedical innovation and improved health outcomes)against the harms (as yet undocumented compromise ofpatient care outcomes caused by physicians’ biases inprescribing and using drugs and devices due to COI).Without such studies, we cannot create rational policyand law on physician-industry relations—that is, wecannot rationally decide whether optimal health care canbe achieved most reliably by strengthening disclosurerequirements, as Murphy suggests, or by encouragingcollaboration between physicians and industry by break-ing down barriers rather than building them. To thiswriter, the latter, although politically unpopular, seems

Table 1. Levels of Evidence and Need for Further Research

Code Quality of Evidence

A High Further research very unlikelySeveral high-quality studies wIn special cases: one large, hig

B Moderate Further research likely to havchange estimate

One high-quality studySeveral studies with some lim

C Low Further research very likely tolikely to change estimate

One or more studies with sevD Very low Any estimate of effect is very

Expert opinionNo direct research evidenceOne or more studies with ver

Source: GRADE (Grading of Recommendations Assessment, DevelopmEditorial Team) [20].

the more reasonable course.

Dr Sade’s work was supported in part by the South CarolinaClinical and Translational Research Institute, Medical Univer-sity of South Carolina’s Clinical and Translational ScienceAward Number UL1RR029882. The contents are solely theresponsibility of the author and do not necessarily represent theofficial views of the National Center for Research Resources orthe National Institutes of Health.

References

1. Sade RM. Dangerous liaisons? Industry relations with healthprofessionals. J Law Med Ethics 2009;37:398–400.

2. Brennan TA, Rothman DJ, Blank L, et al. Health industrypractices that create conflicts of interest: a policy proposalfor academic medical centers. JAMA 2006;295:429 –33.

3. Armstrong D. Cleveland Clinic had ties to maker of faulteddevice. Wall Street Journal, Dec 16, 2005:A3.

4. Mack MJ, Sade RM, for the American Association for Tho-racic Surgery Ethics Committee and The Society of ThoracicSurgeons Standards and Ethics Committee.Relations be-tween cardiothoracic surgeons and industry. Ann ThoracSurg 2009;87:1334–6.

5. Mack MJ, Sade RM, for the American Association for Tho-racic Surgery Ethics Committee and The Society of ThoracicSurgeons Standards and Ethics Committee.Relations be-tween cardiothoracic surgeons and industry. J Thorac Car-diovasc Surg 2009;137:1047–9.

6. American Association for Thoracic Surgery. Disclosure pol-icy. Abstract book, 89th annual meeting. Beverly, MA: AATS,2009:5.

7. The Society of Thoracic Surgeons. Education disclosurepolicy. Program book, 45th annual meeting. Chicago, IL:STS, 2009:6–7.

8. Annals of Thoracic Surgery, conditions for publication form.Available at: http://ats.ctsnetjournals.org/misc/pubconditions.shtml. Accessed December 30, 2010.

9. Journal of Thoracic and Cardiovascular Surgery, disclosurestatement. Available at: http://www.editorialmanager.com/jtcvs/. Accessed December 30, 2010.

10. Murphy JP. Point: full disclosure—presentations and publi-cations should include dollar amounts and duration ofsurgeon-industry relationships. Ann Thorac Surg 2011;92:413–6.

11. Kassirer JP. Commercialism and medicine: an overview.

Definition

hange our confidence in estimate of effectonsistent resultsality multicenter trialortant impact on our confidence in estimate of effect and may

nsimportant impact on our confidence in estimate of effect and is

mitationsrtain

ere limitations

d Evaluation) Working Group 2007 (modified by the EBM Guidelines

to cith ch-qu

e imp

itatiohave

ere liunce

y sev

Camb Q Healthc Ethics 2007;16:377–86.

420 ETHICS IN CARDIOTHORACIC SURGERY SADE Ann Thorac SurgNEFARIOUS COI 2011;92:417–20

12. Cain DM, Loewenstein G, Moore DA. The dirt on comingclean: perverse effects of disclosing conflicts of interest. JLegal Studies 2005;34:1–25.

13. Kassirer J. On the take: how medicine’s complicity with bigbusiness can endanger your health. New York: OxfordUniversity Press, 2004.

14. Ford ES, Ajani UA, Croft JB, et al. Explaining the decrease inU.S. deaths from coronary disease, 1980-2000. N Engl J Med2007;356:2388–98.

15. US Preventive Services Task Force. Available at: http://www.uspreventiveservicestaskforce.org/. Accessed Decem-ber 30, 2010.

16. Grading of Recommendations Assessment, Development andEvaluation (GRADE) Working Group. Available at: http://

www.gradeworkinggroup.org/about_us.htm. Accessed Decem-ber 30, 2010.

17. West S, King V, Carey TS, et al. Appendix C. Systems to ratethe strength of scientific evidence. Evidence Reports/Technology Assessments, No. 47. Rockville, MD: Agency forHealthcare Research and Quality (US), 2002. Available at:http://www.ncbi.nlm.nih.gov/books/NBK33882/. AccessedDecember 30, 2010.

18. Wazana A. Gifts to physicians from the pharmaceuticalindustry. JAMA 2000;283:2655–8.

19. Lubowitz JH, Appleby D, Centeno JM, Woolf SK, Reid JB.The relationship between the outcome of studies of autolo-gous chondrocyte implantation and the presence of com-mercial funding. Am J Sports Med 2007;35:1809–16.

20. Essential evidence plus. Levels of evidence. Available at:

http://www.essentialevidenceplus.com/product/ebm_loe.cfm?show�grade#. Accessed December 29, 2010.