Publication Bias in Retina

HOWARD F. FINE, MD, MHSC • JONATHAN L. PRENNER, MD • DANIEL B. ROTH, MD • SARAH J. DRISCOLL, BA • DAVID A. SALZ, BS

Selective publication is a problem whereby only a portion of initiated clinical trials are published. This issue has been well documented in other fields, such as oncology and cardiology, but has not yet been well studied in the ophthalmology field. Data from clinical trials, whether positive or negative, directly impact clinical practice and also shape future re search efforts. While the majority of clinical trials are now registered before initiation, requirements to provide summary results following completion of those trials are lacking.

HUMAN SUBJECTS: LUCKY DUCKS OR GUINEA PIGS?

Human subjects who participate in clinical trials provide an incredibly important service to society. Experimental trials pose inherent risks, and only a fraction of investigational treatments are ultimately proven safe and effective. Patients may feel in some ways that they play the role of a “guinea pig.” Conversely, though, enrolling in clinical trials can afford patients access to the latest treatments for their disease — if effective, they are truly “lucky ducks.” Most pa tients are motivated, at least in part, by an altruistic goal not just to treat their own condition, but also to advance scientific knowledge and thereby help others with the same af fliction. Many ethicists argue that there exists an inherent social contract in clinical trials: patients agree to be subjects with the caveat that the results will be used to help science and humanity. Clinical investigators and sponsors of trials therefore bear a great responsibility to disseminate the knowledge obtained from the sacrifices made by their human subjects, whether those results are positive or negative.

THE JOURNAL OF NEGATIVE RESULTS

For obvious reasons, there is no “Journal of Negative Results.” Investigators may not wish to invest the effort to publish studies that did not reach the desired conclusion. Industry sponsors may be wary of publishing results for fear of revealing their progress (or lack of progress) to competitors. Ye t failure to publish trials that do not reach their clinical endpoint or exhibit adverse outcomes can harm patients and adversely affect the standard of care. First, clinical care decisions are based on the body of published evidence. Evidence-based medicine cannot proceed if physicians only have access to a portion of the evidence. Second, if harmful effects of treatment are not publicized, there can be a deleterious effect on patient care. Horrific examples in recent history abound.

For example, the news media widely reported a lawsuit by the New York State attorney general against Glaxo-SmithKline for allegedly concealing the results of four studies that cast doubt on the effectiveness of the antidepressant paroxetine for treating adolescents and children, and that the drug might increase suicidal behavior. After legal wrangling to divulge the results of the allegedly suppressed studies, the drug now carries a “black box warning,” the FDA's most grave notification, regarding suicidal ideation and behavior.¹

Delay in publication can also have negative consequences.² In another example, an anti-arrhythmic drug trial that showed an unexpectedly high mortality rate was completed in 1980 but not published until 1993.³ Estimates suggest that up to 75,000 people died each intervening year in the US from the inappropriate use of this anti-arrhythmic for secondary prevention of heart attack.⁴ Investigators and sponsors have an obligation not only to release trial results to the public, but also to do so in a prompt fashion.

TRIAL REGISTRATION

Registration of clinical trials has been a major step forward in minimizing the effects of selective publication.⁵ The NIH and FDA established ClinicalTrials.gov, an online database, in 2000 following the FDA Modernization Act of 1997. Registration of a trial with ClinicalTrials.gov is required prior to initiation by many major medical journals. While ClinicalTrials.gov does provide researchers with the ability to link to trial results, such information is not mandated, nor is there any enforcement provision within the FDA Modernization Act.¹

In 2004, the American Medical Association petitioned the Department of Health and Human Services to “establish a comprehensive registry for all clinical trials conducted in the United States; every clinical trial should have a unique identifier; and all results from registered clinical trials should be made publicly available through either publication or an electronic data-repository.”¹ Unfortunately, the AMA's exhortation has not yet been fully realized.

Section 801 of the FDA Amendments Act was passed by Congress in 2007 and requires certain outcome measures to be reported on ClinicalTrials.gov. These requirements are limited rather than comprehensive, exempting Phase 1 device and surgical trials as well as medications that have not yet received FDA approval,¹ but provides another step forward toward transparency in clinical research.

HOW DOES RETINA FARE?

A recent presentation at the American Society of Retina specialists by Prenner and colleagues investigated the issue of selective publication for the field of retina. A comprehensive search of all trials on ClinicalTrials.gov concerning age-related macular degeneration was performed. Interventional trials concerning AMD completed by January 1, 2007 (to allow time for publication) were included; of 386 candidate trials, 64 met inclusion/exclusion criteria. The overall publication rate was 54%. Industry sponsored trials were published at a slightly lower rate (52%) than were non-industry sponsored trials (58%). Early phase 1 or 2 trials were published at a rate of 42% (n=36), slightly lower than the 71% rate (n=28) of Phase 3 and 4 trials, p=0.02. Whether trials were foreign or domestic did not statistically significantly influence publication rate.

CONCLUSION

As in other fields of medicine, selective publication is an important but often overlooked issue in retina, with roughly only half of registered trials reaching publication according to the recent study discussed. With the explosion of therapeutic trials for a number of retinal conditions, publication bias will only increase in importance in the future. Trial registration on ClinicalTrials.gov has been an important step forward. Yet registration is only part of the solution; more transparency is needed in the conduct and reporting of clinical trials for science and for our patients. RP

REFERENCES

1. Steinbrook R. Public registration of clinical trials. N Engl J Med. 2004:22;351(4):315-7.
2. Stern JM, Simes RJ. Publication bias: evidence of delayed publication in a cohort study of clinical research projects. BMJ. 1997:13;315(7109):640-5.
3. Cowley AJ, Skene A, Stainer K, Hampton JR. The effect of lorcainide on arrhythmias and survival in patients with acute myocardial infarction: an example of publication bias. Int J Cardiol. 1993 Jul 1;40(2):161-6.
4. Dickersin K, Rennie D. Registering clinical trials. JAMA. 2003 Jul 23;290(4):516-23.
5. De Angelis C, Drazen JM, Frizelle FA, Haug C, Hoey J, Horton R, Kotzin S, Laine C, Marusic A, Overbeke AJ, Schroeder TV, Sox HC, Van Der Weyden MB; International Committee of Medical Journal Editors. Clinical trial registration: a statement from the International Committee of Medical Journal Editors. N Engl J Med. 2004:16;351(12):1250-1.