See Article, page 1052

From 2017 to 2018, medical and health research and development (R&D) spending in the United States grew by 6.4%, reaching $194 billion. Of that, industry invested $130 billion in medical and health R&D (67%), and federal agencies invested a total of $43 billion (22%).a Simply said, industry invests 3-fold of research funding as compared to federal funding. This gap is steadily increasing and while industry increased their R&D spending by 39% from 2013 to 2018, over the same period, federal R&D investment grew by 30%. While there is no argument on the crucial role of industry in bringing new products and other advancements to patients care, there is discussion on possible biases in research that is sponsored by industry and the ongoing legal litigation between Pacira Pharmaceuticals and the American Society of Anesthesiologists is an example of this debate.

Table. - Industry-Sponsored Research Pro Nearly all new devices and drugs come from industry and are based on industry-funded research. Eliminating industry funding would eliminate more than two-thirds of all clinical research, which could not possibly be replaced by other funding sources. Advances in clinical trial methodology marked reduce the potential for bias; furthermore, investigators, journals, and trial registries all provide substantial protection against bias in industry-funded research. Con Conflict of interest is inherent to industry-sponsored research leading to high risk of bias in this work. Emphasis on success and aggressive pursuit of positive results influence the type of research funded, the choice of study design, data analysis, results interpretation, and language of the report. Bias can be introduced by omission and lack of publication of negative results.

In this Pro-Con commentary article, we discuss the role of industry in sponsoring research and advancing knowledge and innovation (Table). By design, the authors of this commentary represent 2 specialties, anesthesiology and orthopedics, and this article and the accompanying editorial will be published both in Anesthesia & Analgesia and the Journal of Arthroplasty. The 4 distinguished authors of this Pro-Con article have been extensively funded by both federal agencies and industry and combined have published more than 2500 peer-reviewed full papers. The article includes a Pro section followed by a Con section. Each section is written by 2 authors and includes a rebuttal by the authors of the previous position.

PRO: INDUSTRY-SPONSORED RESEARCH (D.I.S. AND M.A.M.)

Everyone celebrates innovation, and appropriately so because novel devices and drugs allow us to provide the best care for our patients. It is, therefore, worth considering how we get new devices and drugs. They do not arrive by magic.

The ideas that spark clinical developments come from various sources, and clinical and academic institutions have no monopoly. Many outstanding ideas come directly from industry. But, quite independently of where an idea germinates, there is a long and expensive road between original conceptualization, development of a viable product, and regulatory clearance. Even for a simple device, millions of dollars and hundreds of person-years are usually required. For drugs, it might be hundreds-of-millions of dollars and many thousands of person-years. Part of the product-development process—an important and expensive part—is the clinical studies necessary to guide development of the product, characterize potential benefits, refine populations of interest, and eventually provide formal validation to support regulatory clearance.

Whom do you think pays for the process of moving a rough idea into an approved product? With rare exceptions, it is neither academic institutions nor the government. Instead, research is largely supported by industry. Consider, for example, the $43 billion dollars the private sector invested in Alzheimer’s research between 1998 and 2017. And that is appropriate because there is no compelling reason for the National Institutes of Health to support for-profit companies—and politicians and the public might reasonably object if that were the institutes’ primary function. Furthermore, the amount required would exceed the institutes’ entire budget, even if they supported no other research. Of course, the National Institutes of Health, other peer-reviewed funding organizations, and academic institutions occasionally support product-related research. But, it is also mostly done when products are well past their initial development and validation stage.

Pure academic research is an important adjuvant to industry-supported development work, but it does not substitute for industry funding either in focus or magnitude. In practice, industry relies on academic investigators for most product development and validation research since they rarely have direct patient access and, therefore, cannot do the necessary studies on their own. They also require content expertise that clinicians provide. As might be expected, about a quarter of academic investigators in biomedical research have received research funding from industry, and about a third have industry financial ties. The fraction for clinical investigators is presumably much greater.

There are various clinical research funding models, and it is worth considering the major ones. Most often, ideas are generated by independent investigators who are usually in academic institutions. Of course, ideas do not develop in a vacuum. Investigators are naturally interested in novel developments and recognize that much related expertise is in industry, often leading to productive collaborations. All US academic institutions enforce contracts that provide considerable protection for investigators, including data ownership and publication rights.

Clinical research can be funded in 3 ways: (1) the institution, usually the investigator’s department, can support the work; (2) the investigator can collaborate with a relevant company that has an interest in the product or concept being tested; or (3) the work can be supported by an independent peer-review organization such as the National Institutes of Health, the Patient-Centered Outcomes Research Institute, or a foundation. All 3 are designated “investigator-initiated research”; note that the designation is independent of the funding source and refers to the origin on the research question and that academic investigators control the study. The other model is for a company to initiate, fund, and control studies (usually pivotal trials). This sort of research is referred to as “sponsored trials” and differs distinctly from investigator-initiated research in that the sponsoring company develops the protocol and controls the conduct of the trial, analysis, and manuscript preparation. Such trials are usually designed and conducted in collaboration with the US Food and Drug Administration (FDA) and are always strictly monitored.

The question, then, is whether industry-supported research is biased or otherwise compromised? The term bias, like many epidemiologic terms, has both lay and technical meanings which differ. Bias in ordinary use simply refers to opinions. But opinions are normal; anyone who thinks has opinions, and investigators are not exempt. No one does trials to test arbitrary interventions to see if there “might be a difference.” Trials are done to test specific hypotheses that specify exposure and an expected outcome. For example, “epidural analgesia reduces cancer recurrence in adults having potentially curative lung tumor resections” (it does not1). Having a hypothesis that epidural analgesia reduces cancer recurrence does not indicate bias, even though the trial was done because the investigators believed that fewer recurrences was plausible and perhaps even likely.

The same applies to industry-supported trials. People in industry, many of whom have spent years developing a product, naturally believe that their product will improve health in certain populations, and consequently promote trials that test an analogous hypothesis. That is normal and is not qualitatively or quantitatively different from the sort of opinion academic investigators might hold about a question in which they have invested decades of their lives. Neither constitutes bias in the epidemiologic sense.

Excluding flat-out fraud or fabrication, meaningful bias must substantively degrade trial inclusion (selection bias), quantification of outcomes (measurement bias), or interpretation of the results. Both industry-sponsored and investigator-initiated trials usually specify inclusion of patients likely to benefit from the exposure being tested and unlikely to suffer complications, an approach that is ethical and reduces sample size. It is common to point to tight enrollment criteria as evidence of bias. It is not bias, and cannot be, because subject selection occurs before randomization. In fact, proper randomization provides nearly complete protection against selection bias.

Measurement bias is the next major concern. Completely blinded trials, say with pharmacy randomization of identical-appearing drug and placebo, are largely immune to measurement bias. (An exception is attrition bias, but attrition originates with patients, not investigators.) The risk of bias is greater when trials are only partially blinded. But good design mandates protections against bias such as having assessors blinded even when it is impossible or impractical to blind everyone. Journal editors and regulatory agencies well understand the risk of measurement bias and take a dim view of any unnecessary design compromises in this regard. It is naïve to believe that any company would intentionally design a trial with inadequate blinding because the results would be difficult or impossible to publish in a quality journal; nor would they support regulatory approval goals.

There was a time when investigators would evaluate many outcomes without defining primary and secondary ones, and without protection against multiple comparisons. They would then select statistically significant outcomes and “promote” them to become the focus of the research report. (Industry-supported research hardly had a monopoly on that approach.) Selective outcome reporting is no longer possible for trials because by federal law, all trials must be registered at ClinicalTrials.gov and the registration specifies primary and secondary outcomes, along with other important details. (Even outside the United States, registration is effectively required because reputable journals will not publish unregistered trials.)

Furthermore, many journals now require that the entire protocol and an a priori statistical analysis plan be published as a supplement to trial reports. Even systematic reviews and database studies usually follow Preferred Reporting Items for Systematic Reviews and Metanalyses (PRISMA) guidelines and need to be registered at Prospective Register of Systematic Reviews (PROSPERO). Editors of competent journals insist that the methods actually used and the results presented are consistent with the registration and are presented with appropriate emphasis in the proper order. These protections apply equally to whatever the funding source.

The discussion section is usually the most controversial part of research reports because given results can reasonably be interpreted in various ways. A corollary is that authors can easily slant the discussion to their preferred interpretation. We remain amazed at the number of submitted manuscripts in which the discussion is based on what the investigators wanted to find, rather than what was actually observed. But this sort of wishful thinking is hardly restricted to corporate-funded research, and may even be more common in investigator-generated reports.

It remains unclear how much fraud there is in clinical research, but there is probably more than generally appreciated. That said, poor study design and execution surely contributes way more error to medical literature than deliberate fraud.2 Fraud is overwhelmingly the providence of academic investigators, and almost all major perioperative fraudsters (Rubin, Boldt, Fuji, Schietroma, etc),3–6 have been academic investigators who mostly conducted investigator-initiated studies. Fortunately, fraud by organizers of sponsored trials is almost nonexistent because these trials are inevitably monitored by parties independent of both the investigators and sponsors—and by the FDA.

With only occasional exceptions, properly done clinical research is too complicated and expensive for clinical departments to fund. Real research requires real funding, and lots of it. (If it were ever true, it certainly is no longer the case that useful research can be conducted by untrained investigators without adequate funding and little or no nonclinical time.) Research that is unfunded or funded from peer-reviewed sources does not suffer from putative commercial bias, although plenty of investigators have their own serious biases. But research is of little value if not properly done. Inadequate funding often results in compromised research, especially inadequate sample sizes based on implausible treatment-effect estimates, but also insufficient statistical sophistication and lack of trial monitoring.

In summary, industry-funded research is critical to new product development as well as evaluation of drugs and techniques. Realistically, without corporate-funded studies, perioperative research would stagnate with little innovation and few new products. Opinions are ubiquitous and normal, but do not constitute epidemiologic bias. Competent clinical research includes many protections against selection and measurement bias, and the publication process provides at least moderate protection against misinterpretation of results. Trial registries largely prevent selective data presentation. Sponsored trials are particularly protected against inappropriate corporate influence because they are usually codesigned with the FDA, analysis is based on a formal predefined statistical plan, and they are conducted with rigorous external monitoring. Novel products, which are essential for advances in clinical care, largely come from industry, and industry appropriately funds much of the required research. We should celebrate industry’s contribution to improvements in clinical care.

REBUTTAL ON PRO: INDUSTRY-SPONSORED RESEARCH (B.A. AND M.M.T.)

There are important differences between industry settings and academic environments. While academic curiosity and interest in advancing the field typically brings significance and innovation that can drive future industry discoveries, industry R&D is driven more toward bringing new products to market. Some argue that innovation would not happen without industry funding. A National Science Foundation survey examined this issue.7 It found that industry funding has substantially increased in the past few decades, where federal sources used to fund 70% of research, at the start of the last decade it funded less than half of research in the United States. Philanthropy has increased in importance and now funds over 25% of research. Industry makes up the rest. In this analysis, it was found that industry funding primarily focuses on applied research that brings already developed concepts to a product. While applied research is critical to bring discoveries to care, these data also suggest that the governmental and philanthropic funding is the prime driver of developing new advances.

Whether intentionally or unconsciously, bias is present in both academic- and industry-funded scientific work. Although intentional misconduct can happen, we argue that the main driver of differences between academically initiated research and industry-initiated research is the motivation driving the discoveries. The major issue in industry-sponsored research is conflict of interest. Harmful intervention can be minimized, side effects not studied at all, and omissions rather than reporting could be a concerning source of inadequate knowledge. A desire to produce positive results is universally present. However, the psychological influence of the driving goal can be subtle and may not even be recognized by the investigators, and there could be substantial consequences from the commitment to obtain positive results.8 The story of the development and approval of cyclooxygenase-2 inhibitors illustrates this problem. After initial approval, the cyclooxygenase-2 pain reliever rofecoxib (Vioxx) was voluntarily withdrawn from the market in 2004 after a study testing whether the pain reliever could prevent colon polyps showed that it doubled heart attack and stroke risk. The pivotal study leading to the initial approval had failed to accurately capture these serious long-term adverse events. Other examples please replace with “of publication bias include”, the observation that negative studies of lamotrigine were unpublished leading to the false impression that most randomized trials had positive results. Concerns exist for postmarketing bias. Advertisements and inducements, brought by the motivation of profit to subsidize further R&D, can bring about serious adverse results, such as in the promotion and marketing of OxyContin. It has been described as both a commercial triumph and a public health tragedy.9 Finally, there can be a bias in the type of research that industry will fund, and this can result in biases in the data available upon which clinicians make patient care decisions.

We acknowledge that academic discoveries may not have been brought to patient care in the absence of parentships external to the academic setting. However, we argue that bias can easily and often occur even in well-monitored randomized trials. Therefore, while neither industry or academia are immune to bias, the likelihood and sources bias in industry can be more nuanced and more prone to influence consequential decisions. We strongly emphasize the need for full transparency and the enforcement of a priori safeguards to protect against bias. It is critical that we recognize the sources of bias and understand their influence in the interpretation of the available data in the field. It is strongly recommended that study protocols be provided with manuscript submissions that report study results, that exploratory and hypothesis-generating analysis be prespecified and rigorously conducted, and that all key objectives be transparently reported in clinical trials. It is imperative that such safeguards be in place to protect the academic community, industry sponsors, and the public for the production and exposure to incorrect science.

CON: INDUSTRY-SPONSORED RESEARCH (B.A. AND M.M.T.)

Advancements in clinical care are built on knowledge from research and practice. Rigorous research can be costly, and sources of research funding are limited in the setting of resource-constrained environments. Industry funding is an attractive source of research dollars. Generating and reporting on high-quality unbiased data is critical to advancing clinical care in the right direction. For research to ultimately improve care, it needs to address the most clinically relevant question using the most rigorous designs. Unfortunately, industry-funded research can introduce bias into this process. Bias can arise at several levels ranging from (1) the choice of the study questions to be addressed; (2) the planning and design of the study; (3) the approach to study implementation; and (4) the dissemination of data and publication processes. These sources of bias can result in tackling less relevant questions, using less rigorous approaches, not reporting results in a timely manner, or not interpreting results impartially. As such, industry-funded research can produce outcomes that may hinder advances unless appropriate safeguards are implemented.

One potential area of bias is the actual research study that is undertaken. Industry-funded research predictably focuses on investigations establishing the effectiveness of drugs or implants that a company produces. This curtails other studies not focused on new products coming to market, or potential increased product utilization not being undertaken. Furthermore, industry-sponsored research may not choose to study the most relevant clinical comparisons. An ideal comparison might be with an alternate product produced by a competitor. Industry would also be less likely to fund research, whose results might show their products less effective or more costly.

Because of the focus on success for an industry partner, repeated publication of positive results accomplishes the role of a self-fulfilling prophecy where concordance reported in the literature can be inflated by consistency in publication of positive results whereby certain interventions are a priori expected to only have positive outcomes. In this context, publication bias can also influence the results of systematic review and meta-analysis. The evaluation of unpublished work identified from the “gray” literature resulting from clinical trials in cancer and other areas has shown that several industry-sponsored trials presented at conferences often reporting null results do not lead to publication if the results were negative, leaving a large void in knowledge regarding interventions that will not be beneficial.

Many industry-funded studies require that the company gives permission to publish and also reserve the right to edit and approve the language used by the investigators. In the event that a study that might prove a certain product is not effective, the company might not allow publication. Thus, clinically important data may never be made available to practitioners or other researchers. The fact that pharmaceutical industry tends not to disseminate data from negative studies through published scientific literature leads to publication bias, as negative treatment studies remain selectively unpublished.10 Barriers to publication of studies with negative outcomes (eg, null findings) are that the results, instead of being seen as important in their own right, are treated by the pharmaceutical industry and academic investigators as noncontributory and more challenging to publish without giving a positive spin.11 In this way, industry-funded studies can result in select information being withheld.12–14 While listing studies on clincialtrials.gov will show what kinds of studies are being undertaken, it is not a safeguard for requiring that the data be made available to others in the medical or scientific field.15

“An imbalance between commercial and public interests is another potential area that can introduce bias”16. Information made available via registration in prospective databases, such as clinicaltrials.gov, is not necessarily reviewed by the medical community as data that are published in mainstream journals. On the other end, accrual as well as the time to publications for clinical trials with positive findings in high-impact peer-reviewed journals is highly accelerated compared to the time to publication of investigator-initiated studies or research funded via other sources.17,18 The general efficiency of industry with the emphasis on productivity could explain the higher success of industry compared to academic settings in accomplishing rapid publication of positive results in high-impact journals.

Preclinical research can also suffer from undue influence exerted by industry. The choice of animal species, experimental model, and the suitability of animal models are often critical factors in improving the success of drug development to move from a promising compound to an approved, marketable drug. Shortcomings in the experimental design have led limited translation of preclinical research results into effective treatments. The critical review of animal research underlying the development of treatments for stroke in humans is a well-studied example of the failure of translation.

The bias in the types of research industry might fund can limit investigators to studies that address issues of interest to a company.2,19,20 Unlike public granting agencies, industry does not always provide funding based on unbiased peer review following an open call for proposals. Often intuitions or investigators who use a particular company’s product or have consulted with a company are able to garner research funds. In addition to biasing the kinds of studies performed, this also biases the patient populations that might be studied. Rather than selecting a random sampling of the population with a condition, only patients who see certain physicians or are seen at certain intuitions are enrolled. This can result in data not necessarily applicable to the population that most clinicians treat.21 As an example, many clinical trials for arthritis study patients who are younger than the average patient treated for this condition, raising the possibility that this published body of work does not apply to most patients with this condition.22

It is common for industry sponsors to want to spend the least amount of money to undertake a research study. Industry, with more limited budgets for research than most federal funding agencies, may be more prone to pressures to fund at lower budgets. Furthermore, in an attempt to lower costs, the study design may rely on less rigorous outcome measures, or not use the most rigorous statistical or other analyses. This can result in studies using less rigorous approaches, for instance being underpowered, or lacking ideal controls.23 While an underpowered study might find by chance a statistically significant result in many instances, adding the additional patients as predicted to be required in the prestudy power analysis will change the outcome of the result.24,25 This pattern has been observed in several phase II trials of promising drug therapies, such as tirilazad mesylate, which was extensively tested in preclinical studies and was found to improve surrogate end points in phase II trials on the basis of subgroup analyses but resulted in no effects in subsequent phase III trials.

Investigators or other participants in industry-funded studies might have financial ties with the company that funded the study. There might be shared intellectual property or consulting arrangements. Hidden conflicts can also exist in the form of industry payments to FDA advisers after drug approvals with potential ethical concerns due to potential influence on voting panel members. While it can be argued that patients are not always concerned about such relationships,26 they still can make a difference in outcome. Even if these are not directly related to the research project in question, there could be unconscious bias that can cause data to be interpreted in a different way than would have been otherwise.12,13 While unconscious biases always exist in research, such biases related to a specific product might cause conclusions of studies to look more favorably on certain interventions.

Industry plays a critical role bringing new products and other advancements to patient care. Indeed, in many cases without industry support, providing a path to commercialization therapies that can improve patient outcome can never be brought to fruition. Internal industry research also generates new therapies that advance medical care. However, biases that can arise from industry funding of research have a potential to generate evidence and result in publications that may not always result in the best outcome for patients. Although progress has been made toward improving transparency, studies show that a substantial portion of pivotal oncology trials that support the FDA registration of modern anticancer medicines are unavailable to the scientific community.27

Appropriate safeguards are needed to ensure that research addresses the most important and relevant questions, that results are available even when they do not support the use of a product produced by the funding company, that populations studied reflect those of the patients we treat, that the most rigorous approaches are applied, that studies have the appropriate power to address the question posed, and that conclusions are presented in an unbiased manner.28,29 To address these concerns, the FDA has been providing increasingly detailed guidance for industry to conduct clinical trials. Furthermore, journal editors have elevated the reporting standard to increase the rigor and transparency of the research being published. It is increasingly required that study protocols be published to better understand the context of how the data were collected, what end points were primary and secondary, what exploratory analyses were undertaken. Without these safeguards and others, industry-funded research has a potential to distort the practitioner’s knowledge of patient treatments, ultimately resulting in worse patient or population outcomes.8

REBUTTAL ON CON: INDUSTRY-SPONSORED RESEARCH (D.I.S. AND M.A.M.)

False dichotomies lead to poor decisions. If industry funding is eliminated, peer-reviewed clinical research will not increase. Instead, there will simply be less research—a lot less. The question then is whether industry-funded research, by far the majority, is on balance beneficial? Clearly it is.

B. Alman and M. M. Treggiari specify 4 potential sources of bias in industry-funded research which we will address in turn. First, they note that funded research predictably focuses on establishing effectiveness of drugs and devices. Of course! But if we are to have new drugs and devices, that research needs to be done. It does not follow that industry-funded research “curtails other studies not focused on new products.” What limits clinical research is largely funding, not a lack of patients or investigators. It simply is untrue that industry-funded research prevents publicly funded research. We need more of both!

The second issue B. Alman and M. M. Treggiari raise is of research design and implementation. All investigators aim to test hypotheses. It is usual to select an enriched population that is likely to both benefit from the treatment in question and experience the outcomes of interest. Furthermore, it is proper and ethical to exclude patients who are unlikely to benefit or suffer harm from the experimental treatment. But, this approach is universal, because it reduces the necessary sample size, and is hardly restricted to industry-funded work. Readers should understand that all research results apply most directly to actual participants, and can reasonably be extrapolated to similar patients.

An appropriate comparator group is an important part of research design. The current standard is to compare novel treatments to the best alternative treatment. But interestingly, the FDA often requires companies to compare treatments to placebo—even when an alternative reference treatment would be preferable.

It is undoubtedly true that basic science results rarely translate to humans.30 Consider nearly all nutrition research, for example, where virtually no basic science results have translated to humans. And similarly, mechanistic studies (phase 2) often fail to predict the hard outcomes that most interest clinicians. But this is a universal problem, and consequent to poor understanding of physiological mechanisms. It is not industry-specific, and there is no evidence that industry-funded basic science is less likely to translate to humans than research-funded other ways. If anything, industry is especially likely to be skeptical about basic science results because human trials are so expensive.

The final point B. Alman and M. M. Treggiari make is that repeated publication of positive outcomes or failure to publish negative outcomes influences what information is available to clinicians. There are compelling reasons to publish various aspects of trials separately, and nearly all major trials generate at least several papers exploring different outcomes and perspectives. But secondary publications always reference the original to avoid confusion about how many patients were involved and provide the correct denominator for meta-analyses. Furthermore, decisions to publish are ultimately made by editors and reviewers who determine whether presented information is sufficiently novel. No US university will agree to contracts that prohibits publication of their own data.

Publication bias is a huge problem and largely results from “negative” results not being published.31 But failure to publish equivocal results is hardly restricted to industry-funded work. Plenty of investigators do not bother to submit equivocal results for publication. And even when they do, journals are often less interested than in positive results. A subtle aspect of publication bias is that “negative” studies are often underpowered and thus should not be published, but then are not available for inclusion in meta-analyses. Industry-funded research is usually adequately powered, especially for major trials, which are designed with FDA input. Insufficient power and false negative results are especially common in unfunded research, done on the cheap, with limited departmental resources. Inadequate power is also common in student research where the sample-size estimate is often based on unrealistic assumptions and “adjusted” to available time and patients, rather than what is biologically necessary.

We agree that “the general efficiency of industry … could explain the higher success of industry compared to academic setting in accomplishing rapid publication of positive results in high-impact journals.” But this is good! Delayed publication harms everyone, and it is to industry’s credit that they get their studies published quickly. The publication venue is largely determined by the quality of the research (investigators can submit to any journal, but manuscripts will only be accepted if they meet the journal’s quality criteria). That industry-funded studies are so often published in high-profile journals that reflect the quality of the work, which includes the importance of the question, its novelty, and the strength of the design including adequate sample size.

B. Alman and M. M. Treggiari assert that “safeguards are needed to ensure that research addresses the most important and relevant questions, that results are available …, that the populations studied reflect those of the patients we treat, that the most rigorous approaches are applied, that studies have appropriate power …, and that conclusions are presented in an unbiased manner.” We agree and believe that such safeguards are largely in place. In fact, B. Alman and M. M. concede that the FDA provides detailed guidance to industry and that journal editors have greatly elevated the standards for reporting. Trial registries also provide an unmodifiable record of critical design features including anticipated sample size and defined primary and secondary outcomes. Federal law requires that results be posted to registries within a reasonable time.

In summary, we fully agree with B. Alman and M. M. Treggiari that “industry plays a critical role bringing new products and other advancements to patients’care. Indeed, in many cases without industry support, providing a path to commercialization therapies that can improve patient outcome can never be brought to fruition. Internal industry research also generates new therapies that advance medical care.” We could not put it better!

DISCLOSURES

Name: Daniel I. Sessler, MD.

Contribution: This author helped write this manuscript.

Conflicts of Interest: D. I. Sessler’s department conducts studies funding by the National Institutes of Health, the Canadian Institutes of Health Research, various foundations, and many industrial sources. He consults for various companies, none directly related to this article.

Name: Benjamin Alman, MD.

Contribution: This author helped write this manuscript.

Conflicts of Interest: B. Alman is supported by the National Institutes of Health.

Name: Miriam M. Treggiari, MD, PhD, MPH.

Contribution: This author helped write this manuscript.

Conflicts of Interest: M. M. Treggiari is supported by the National Institutes of Health.

Name: Michael A. Mont, MD.

Contribution: This author helped write this manuscript.

Conflicts of Interest: M. A. Mont is a consultant for Stryker, and various other companies not related to this research article.

This manuscript was handled by: Zeev N. Kain, MD, MBA.

REFERENCES 1. Xu ZZ, Li HJ, Li MH, et al. Epidural anesthesia-analgesia and recurrence-free survival after lung cancer surgery: a randomized trial. Anesthesiology. 2021;135:419–432. 2. Ioannidis JP. Why most published research findings are false. PLoS Med. 2005;2:e124. 3. Carlisle JB. A meta-analysis of prevention of postoperative nausea and vomiting: randomised controlled trials by Fujii et al compared with other authors. Anaesthesia. 2012;67:1076–1090. 4. Kranke P, Apfel CC, Roewer N, Fujii Y. Reported data on granisetron and postoperative nausea and vomiting by Fujii et al. are incredibly nice! (letter). Anesth Analg. 2000;90:1004–1007. 5. McHugh UM, Yentis SM. An analysis of retractions of papers authored by Scott Reuben, Joachim Boldt and Yoshitaka Fujii. Anaesthesia. 2019;74:17–21. 6. Myles PS, Carlisle JB, Scarr B. Evidence for compromised data integrity in studies of liberal peri-operative inspired oxygen. Anaesthesia. 2019;74:573–584. 7. Mervis J. Data check: federal share of basic research hits new low. Science. 2017;355:1005. 8. Fleming TR. Clinical trials: discerning hype from substance. Ann Intern Med. 2010;153:400–406. 9. Van Zee A. The promotion and marketing of oxycontin: commercial triumph, public health tragedy. Am J Public Health. 2009;99:221–227. 10. Nassir Ghaemi S, Shirzadi AA, Filkowski M. Publication bias and the pharmaceutical industry: the case of lamotrigine in bipolar disorder. Medscape J Med. 2008;10:211. 11. Pater C. Current trends in the cardiovascular clinical trial arena (I). Curr Control Trials Cardiovasc Med. 2004;5:4. 12. Anglemyer AT, Krauth D, Bero L. Industry sponsorship and publication bias among animal studies evaluating the effects of statins on atherosclerosis and bone outcomes: a meta-analysis. BMC Med Res Methodol. 2015;15:12. 13. Siegel M, Eder JSN, Wicherts JM, Pietschnig J. Times are changing, bias isn’t: a meta-meta-analysis on publication bias detection practices, prevalence rates, and predictors in industrial/organizational psychology. J Appl Psychol. 2021;107:2013–2039. 14. Twombly R. Researchers left to guess at outcomes of most cancer clinical trials. J Natl Cancer Inst. 2009;101:72–74. 15. Lenzer J, Hoffman JR, Furberg CD, Ioannidis JP; Guideline Panel Review Working Group. Ensuring the integrity of clinical practice guidelines: a tool for protecting patients. BMJ. 2013;347:f5535. 16. Furberg CD, Hall MA, Sevick MA. Balancing commercial and public interests. Curr Control Trials Cardiovasc Med. 2004;5:6. 17. Pasalic D, Tang C, Jagsi R, Fuller CD, Koong AC, Ludmir EB. Association of industry sponsorship with cancer clinical trial accrual. JAMA Oncol. 2020;6:1625–1627. 18. Ross JS, Mocanu M, Lampropulos JF, Tse T, Krumholz HM. Time to publication among completed clinical trials. JAMA Intern Med. 2013;173:825–828. 19. Ioannidis JP. How to make more published research true. PLoS Med. 2014;11:e1001747. 20. Lundh A, Lexchin J, Mintzes B, Schroll JB, Bero L. Industry sponsorship and research outcome. Cochrane Database Syst Rev. 2017;2:MR000033. 21. Nunan D, Aronson J, Bankhead C. Catalogue of bias: attrition bias. BMJ Evid Based Med. 2018;23:21–22. 22. Liberopoulos G, Trikalinos NA, Ioannidis JP. The elderly were under-represented in osteoarthritis clinical trials. J Clin Epidemiol. 2009;62:1218–1223. 23. Yuen SY, Pope JE. Learning from past mistakes: assessing trial quality, power and eligibility in non-renal systemic lupus erythematosus randomized controlled trials. Rheumatology (Oxford). 2008;47:1367–1372. 24. Bhandari M, Jin L, See K, et al. Does teriparatide improve femoral neck fracture healing: results from a randomized placebo-controlled trial. Clin Orthop Relat Res. 2016;474:1234–1244. 25. Lochner HV, Bhandari M, Tornetta P III. Type-II error rates (beta errors) of randomized trials in orthopaedic trauma. J Bone Joint Surg Am. 2001;83:1650–1655. 26. Camp MW, Mattingly DA, Gross AE, Nousiainen MT, Alman BA, McKneally MF. Patients’ views on surgeons’ financial conflicts of interest. J Bone Joint Surg Am. 2013;95:e91–e98. 27. Modi ND, Abuhelwa AY, McKinnon RA, et al. Audit of data sharing by pharmaceutical companies for anticancer medicines approved by the US Food and Drug Administration. JAMA Oncol. 2022;8:1310–1316. 28. Breault JL, Knafl E. Pitfalls and safeguards in industry-funded research. Ochsner J. 2020;20:104–110. 29. American Society of Clinical Oncology. American Society of Clinical Oncology policy statement: oversight of clinical research. J Clin Oncol. 2003;21:2377–2386. 30. Sessler DI. Lost in translation: the 2016 John W.Severinghaus lecture on translational research. Anesthesiology. 2017;126:995–1004. 31. Sessler DI. Negative trials, and what to do with them? (editorial). Anesthesiology. 2020;132:221–224.