Endometriosis: the elusive gray area between evidence-based and evidence-biased medicine

Primary (randomized controlled trials) and secondary (systematic reviews and meta-analyses) research guides clinical practice and patients' decisions. However, general and gynecologic medical journals warn that not all that glitters is gold, bringing into the spotlight the perils that may be hidden in the folds of apparently elegant methodology (1Hurd W.W. Conflicts of interest and medical publishing.Obstet Gynecol. 2013; 122: 511-512Crossref PubMed Scopus (2) Google Scholar). Now Guo addresses the characteristics and potential drawbacks of interventional trials in endometriosis (2Guo S.W. An overview of the current status of clinical trials on endometriosis: issues and concerns.Fertil Steril. 2014; 101: 183-190Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar). In general, qualitative and quantitative issues should be considered here, particularly when the effect of new medications is investigated. A specific trial outcome may be selected among the possible ones, with the objective of favoring the drug under study; the trial design may be “tailored” to facilitate demonstration of the ineluctable significant difference; the study population may be inappropriately restricted, excluding those subjects in whom the new compound might reveal less effectiveness; a placebo instead of an active comparator may be chosen if the main trial objective is pain relief; a comparator with known major subjective and metabolic untoward effects may be preferred to one with a good safety and tolerability profile; clinical equivalence may be used as an excuse not to search for embarrassing differences; several data sets regarding different outcomes may be analyzed, publishing only favorable results and concealing unfavorable ones; multiple trials may be conducted, publishing only the results of those who demonstrated the superiority of the experimental drug. If everything fails, study oblivion may constitute the best plan B. Since 2005, the International Committee of Medical Journal Editors requires researchers to deposit information about their trials into an official registry before participant enrollment as a prerequisite for publishing the trial's results in member journals. Scientists, physicians, patients, and reviewers can consult these trial registries. Registration provides information about ongoing trials and allows comparisons between original protocols and final reports regarding methodology, primary and secondary outcomes, selection criteria, preplanned power calculation, statistical analyses, and funding. Guo and Evers have recently conducted a survey of trials on endometriosis registered at ClinicalTrials.gov and found than more than two-thirds of the trials identified as completed were actually unpublished (3Guo S.W. Evers J.L.H. Lack of transparency of clinical trials on endometriosis.Obstet Gynecol. 2013; 121: 1281-1290Crossref PubMed Scopus (29) Google Scholar). In addition, trials sponsored by the pharmaceutical industry were about four times less likely to reach publication than those supported by independent institutions. This was likely due to problems with efficacy, safety, or both (3Guo S.W. Evers J.L.H. Lack of transparency of clinical trials on endometriosis.Obstet Gynecol. 2013; 121: 1281-1290Crossref PubMed Scopus (29) Google Scholar). Not surprisingly, no negative or unfavorable trials sponsored by industry have been published so far. Yet omission of publication seems to be barely perceived as a dramatic problem. To complete the picture, Guo conducted a qualitative analysis comparing 45 industry- and 35 non–industry-sponsored interventional trials on drugs, biologics, and medicated intrauterine devices for endometriosis registered at ClinicalTrials.gov (2Guo S.W. An overview of the current status of clinical trials on endometriosis: issues and concerns.Fertil Steril. 2014; 101: 183-190Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar). Industry-sponsored trials are predominantly phase II/III and have a larger size and shorter duration than non–industry-sponsored trials. Moreover, the former are more likely to target endometriosis-associated pain; to assess the effect of nonclassic hormonal compounds (GnRH antagonists, aromatase inhibitors, selective estrogen or progesterone receptor modulators) or of new nonhormonal drugs (e.g., anti-NGF antibody and CCR1 inhibitor); to be placebo controlled; and to be unpublished even though they have been completed. Several placebo-controlled trials have consistently demonstrated that any drug is better than no drug when addressing endometriosis-associated pain. Allocating suffering women to a placebo arm is thus ethically questionable, but “it would be suicidal for a for-profit enterprise to show that its drug is merely similarly efficacious as an existing, but likely cheaper, drug, or worse” (2Guo S.W. An overview of the current status of clinical trials on endometriosis: issues and concerns.Fertil Steril. 2014; 101: 183-190Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar). Because pharmacotherapy can not cure endometriosis, it is surprising that industry-sponsored trials are so short (months) when compounds to be used for disease control over prolonged periods of time (years) are needed. Planning short treatment periods may limit withdrawal of participants in case the experimental drug is associated with insufficient safety or tolerability. These doubts will probably never be clarified, because the results of unfavorable industry-sponsored trials will not see the light. In published industry-funded trials, results systematically favor the sponsor's novel compound. This is particularly concerning because health care providers can not have adequate clinical experience with new drugs and therefore take decisions based on information published in peer-reviewed journals. In addition, novel molecules are usually much more expensive than existing ones. Therefore, overestimating efficacy or overusing a specific new drug also has economic implications. Guo confirms the existence of major methodologic drawbacks and overall limited transparency in the area of interventional trials for endometriosis (2Guo S.W. An overview of the current status of clinical trials on endometriosis: issues and concerns.Fertil Steril. 2014; 101: 183-190Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar). Although the author did not investigate the existence of selective reporting (when specific outcomes are omitted) and misleading reporting (e.g., switching between primary and secondary outcomes), he could confirm omission of reporting, particularly of industry-sponsored trials, although this problem also afflicts trials with no commercial input. Many researchers and editors are convinced that “negative” studies are of no interest for readers. Investigators may be less willing to submit their reports, reviewers less objective in judging nonsignificant findings, and editors less prone to publish them. However, selective nonpublication favors statistically significant findings, thus introducing a major bias in the literature. An additional problem arises when data on the effect of nonhormonal drugs for endometriosis are not reported. In these cases, compounds are used that should interfere with the very pathogenic mechanisms. Therefore, even negative findings would be helpful for better understanding the alternative causes of development and progression of lesions. A slowdown of translational research is likely if investigators have access to only a biased sample of existing evidence. The detrimental consequences of selected or omitted publication include an effect on secondary research assessing the overall value of health care interventions for endometriosis. Systematic reviews and meta-analyses are based on data available in the public domain, and negative or unfavorable studies are essential for their validity. If only a partial picture is reviewed, estimates of safety and efficacy will be biased and less precise, carrying the risk of amplifying the irrelevant or of missing the identification of what works best for patients. As Guo (2Guo S.W. An overview of the current status of clinical trials on endometriosis: issues and concerns.Fertil Steril. 2014; 101: 183-190Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar) and Guo and Evers (3Guo S.W. Evers J.L.H. Lack of transparency of clinical trials on endometriosis.Obstet Gynecol. 2013; 121: 1281-1290Crossref PubMed Scopus (29) Google Scholar) demonstrated, trial registration alone is not the solution to all problems. If trials are registered but results are selectively or never published, the issue of reporting bias is not overcome. However, the benefit of registration is that the nonpublication becomes detectable, allowing the possibility of quantifying this bias, as Guo (2Guo S.W. An overview of the current status of clinical trials on endometriosis: issues and concerns.Fertil Steril. 2014; 101: 183-190Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar) and Guo and Evers (3Guo S.W. Evers J.L.H. Lack of transparency of clinical trials on endometriosis.Obstet Gynecol. 2013; 121: 1281-1290Crossref PubMed Scopus (29) Google Scholar) have done. Moreover, pharmacologic industries or investigators can be held responsible or contacted to clarify the reasons for not publishing the results (4Wager E. Williams P. “Hardly worth the effort”? Medical journals' policies and their editors' and publishers' view on trial registration and publication bias: quantitative and qualitative study.BMJ. 2013; 347: f5248Crossref PubMed Scopus (107) Google Scholar). Given the high estimated disease prevalence and usually chronic course, the endometriosis market is attractive for pharmaceutical companies that “are pursuing low-hanging fruits” with the objective of achieving “the brightest business prospect while minimizing the risk of failure” (2Guo S.W. An overview of the current status of clinical trials on endometriosis: issues and concerns.Fertil Steril. 2014; 101: 183-190Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar). The concerns raised by Guo (2Guo S.W. An overview of the current status of clinical trials on endometriosis: issues and concerns.Fertil Steril. 2014; 101: 183-190Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar) are not exclusive to industry-sponsored trials; non–industry-sponsored research also showed limitations. Indeed, pharmaceutical companies provided useful new therapeutics that changed patients' quality of life, and a hypothetical withdrawal from developing novel medications for women with endometriosis would be disastrous. Most of the interventional research in endometriosis is currently sponsored by pharmaceutical industry. However, data are generously contributed by patients, and even in cases of industry funding, trials are normally conducted with the use of the academic infrastructure. Therefore, the scientific community should rethink the current interpretation of the public-private partnership as “public duties and obligations” and “private privileges and advantages” (5Garattini S. Bertelè V. Bertolini G. A failed attempt at collaboration.BMJ. 2013; 347: f5354Crossref PubMed Scopus (11) Google Scholar). Now that major problems with trials on endometriosis have been unveiled (2Guo S.W. An overview of the current status of clinical trials on endometriosis: issues and concerns.Fertil Steril. 2014; 101: 183-190Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar, 3Guo S.W. Evers J.L.H. Lack of transparency of clinical trials on endometriosis.Obstet Gynecol. 2013; 121: 1281-1290Crossref PubMed Scopus (29) Google Scholar), a proactive agenda should consider several purposes for both medical journals and clinical investigators:1.Gynecologic journals should play a greater role in increasing the awareness of potential drawbacks of interventional trials on endometriosis (4Wager E. Williams P. “Hardly worth the effort”? Medical journals' policies and their editors' and publishers' view on trial registration and publication bias: quantitative and qualitative study.BMJ. 2013; 347: f5248Crossref PubMed Scopus (107) Google Scholar).2.Medical journals should require that authors of randomized controlled trials of new compounds for the treatment of endometriosis commit to making their anonymized data available on request.3.Journal policy regarding negative trials should change. Currently, only a small minority of medical journals encourage authors to submit reports of scientifically sound research regardless of the direction or strength of the results (4Wager E. Williams P. “Hardly worth the effort”? Medical journals' policies and their editors' and publishers' view on trial registration and publication bias: quantitative and qualitative study.BMJ. 2013; 347: f5248Crossref PubMed Scopus (107) Google Scholar).4.Sponsorship of trials on endometriosis by government institutions and independent foundations should increase, with the objective of achieving a more balanced scenario;5.Contracts with pharmaceutical companies can be refused when investigators have limited or no role in designing the study, selecting the study population, choosing the comparator, or determining the duration of the trial or when access to overall patient outcome data from multicenter studies would not be permitted (5Garattini S. Bertelè V. Bertolini G. A failed attempt at collaboration.BMJ. 2013; 347: f5354Crossref PubMed Scopus (11) Google Scholar). An adequate dose of free thinking and independent scientific culture is badly needed in the conception, conduct, and report of clinical trials on endometriosis, and the survey by Guo (2Guo S.W. An overview of the current status of clinical trials on endometriosis: issues and concerns.Fertil Steril. 2014; 101: 183-190Abstract Full Text Full Text PDF PubMed Scopus (47) Google Scholar) greatly helps in dissipating the fog over this scarcely scrutinized research environment. An overview of the current status of clinical trials on endometriosis: issues and concernsFertility and SterilityVol. 101Issue 1PreviewTo examine and compare differences, if any, between industry- and nonindustry-sponsored clinical trials on endometriosis and to evaluate the effect of prior published positive preclinical results, or lack thereof, on trial status. Full-Text PDF