The Problem of Faked Medical Research Data

Around the world, billions of people rely on the information discovered by doctors and scientists for life-changing medical research and data, but things may not be as it seems. With the competitive field of academia, the issue of faking or altering medical data is becoming increasingly apparent, as 90 out of around 5,000 clinical trials analyzed by Dr. John Carlisle, anaesthetist for England’s National Health Service, contained suspicious data, and 80% of studies involving medical devices had evidence of reporting bias. Yet, many of these researchers risk their careers and funding for future projects when results don't go according to plan, which encourages poor research practices and possibly faked research data. However, false research data has the possibility of impacting millions of lives that rely on scientists to build a brighter and healthier future for all.

According to Beate Wieseler, the deputy head of IQWiG's Drug Assessment Department, a major issue that leads to misleading medical data is due to financial reasons, as their research showed that studies with positive results correlated with more funding. One of the main ways a lab receives funding is to look into a researcher's previous publications for consistency in publishing, which leads to researchers publishing as many papers as possible, with more concern for the quantity of the papers over the quality and accuracy of the papers. A paper published in PLOS Biology states, "Employers and funders often count papers and weigh them by the journal’s impact factor to assess a researcher’s performance," which is significant as many prestigious journals prefer publishing studies with statistically significant results, as groundbreaking headlines seem much more significant than studies that show negative results. Ben Mudrak, PhD in Molecular Genetics and Microbiology, illustrates this as he states, "Imagine that the same experiment is repeated by 20 labs, but only one lab finds a significant result (say, p < 0.05). Then, that positive result is published... We might expect one significant result out of twenty by random chance alone, and the published result may in fact be spurious." With negative clinical trials being statistically less likely to be published, patients and doctors may not know the true effect of the drug. In his famous 2005 paper, Dr. John Ioannidis found that only 44% of the 45 most highly cited clinical studies could be replicated, which is important as studies that are done using the correct methods should be replicable. Incorrect studies can create a huge impact in the scientific community and in the public, as faulty study from the 1990s caused a stir when it found a link between vaccines and autism. This study has now been discredited, but it made its mark on the public as currently, around 13% of parents are still skeptical of vaccines.

This study shows how to identify flawed studies! Image Source

Some of the biases that occur include selection bias, which selects a sample group that is not random, detection bias, which pursues the results of the treatment more than the control, observer bias, where the observer makes subjective decisions about the outcome, publication bias, where the results are skewed in favor of having a satisfactory results, and more. Yet, how exactly does this happen? Often times, studies will only report the positive results, leaving out the negative or no-finding results. This is called p-hacking, or inflation bias, where researchers knowingly selects certain results until the desirable result becomes significant. By analyzing the data midway through the experiment, selecting which variables to report, and stopping the experiments when the desired results are achieved, scientists can skew their p-value, which is a measure to see if the data is real or occurred due to chance.

GlaxoSmithKline was sued after a child was born with birth defects due to his mother's use of the drug Paxil during pregnancy. Image Source

However, this can have a huge impact on other researchers, patients, and doctors. For example, in October 2009, the pharmaceutical company, GlaxoSmithKline, went to trial after a child was born with a heart defect due to his mother's use of the anti-anxiety medication Paxil. It was found that the company suppressed four clinical trials that showed no effectiveness for treatment and an increased risk for suicidal tendencies for off-label use in children and teens. In addition, doctors often refer to clinical trial data and studies to prescribe medications, but when the negative or no-finding data isn't shown, doctors and patients do not know the real risks associated with the drug. This proved to be an issue as the drug designed to prevent irregular heartbeat, Tambocor, caused thousands of people to die because the studies that showed possible dangerous side effects were not published. In fact, around 60% of studies on the effects of drugs currently on the market are not published, which shows that the effects of many drugs are not fully understood by the doctor or the patient. Yet, in addition to the issues caused for the consumer, unpublished negative results also impact other research. Not publishing failed experiments may cause other researchers to unknowingly conduct the same experiments, which wastes time and resources. If they knew about the experiments that were done previously, scientists could help each other contribute to new discoveries by publishing failed study designs, which allows other researchers to change their study design rather than conducting the same study. As the inventor Thomas A. Edison said, "I have not failed. I've just found 10,000 ways that won't work." Similarly, negative results are just as important, as it shows the study designs that didn't achieve a particular result.

More people have been demanding clinical trials to be publicly registered and published, no matter the result of the study. In June 2009, the FDA launched the Transparency Task Force, which would make information about drugs and other medical devices publicly available. This would indicate when and why a drug is being studied, when the application is submitted, any major safety concerns that caused an application to be withdrawn, and the reasons why the FDA rejected the application. With increasing demand for access to clinical trial data from the public, the FDA and other agencies have begun to look into transparency in medical research data. However, many companies are opposed to the idea of making clinical trial data publicly available, as these companies want to keep industry secrets and other information private. Even though some negative clinical trials are being published, this usually happens around a year after the positive study was published. In response to the lack of publishing for negative studies, some journals, including F1000Research, BMC Psychology, PLOS ONE, the Journal of Negative Results in Biomedicine, All Results, and more, welcome negative, inconclusive, or null results. However, while there are tens and thousands of scientific journals, there are significantly less journals that accept negative results, which creates a problem as these journals are not as popular or commonly cited as other journals, so many scientists do not submit their work for publication.

Millions of people rely on medical discoveries and devices founded by researchers everyday, but many of these people don't realize the prevalence of bias of these studies. Negative clinical trials are much less likely to be published, and this may cause a certain medication to appear to be harmless. Often times, dangerous side effects of the medication are not published, and this puts lives on the line as the risk of the medication is not fully known. Yet, within the competitive field of scientific research, the incentive for scientists to report positive results is evident, and as positive results have a higher chance of being published, this may mean better job opportunities or more funding. This systemic issue of promoting groundbreaking headlines, positive results, and frequent publications over good science and careful studies has already costed thousands of lives, so it's clear that we need to change the our current system for medical research into a better system dedicated to it's intended purpose; to help humankind.

Bibliography

Hsu, Jeremy. “Dark Side of Medical Research: Widespread Bias and Omissions.” LiveScience, Purch, 24 June 2010, www.livescience.com/8365-dark-side-medical-research-widespread-bias-omissions.html.

Odierna, Donna H et al. “The cycle of bias in health research: a framework and toolbox for critical appraisal training.” Accountability in research vol. 20,2 (2013): 127-41. doi:10.1080/08989621.2013.768931

George, Stephen L, and Marc Buyse. “Data fraud in clinical trials.” Clinical investigation vol. 5,2 (2015): 161-173. doi:10.4155/cli.14.116

ESHRE Capri Workshop Group, Protect us from poor-quality medical research, Human Reproduction, Volume 33, Issue 5, May 2018, Pages 770–776, https://doi.org/10.1093/humrep/dey056

Adam, David. How a Data Detective Exposed Suspicious Medical Trials. 23 July 2019, www.nature.com/articles/d41586-019-02241-z.

Yartsev, Alex. “Types of Bias in Medical Research: Deranged Physiology.” Deranged Physiology, 28 June 2015, derangedphysiology.com/main/required-reading/statistics-and-interpretation-evidence/Chapter 2.1.5/types-bias-medical-research.

Mudrak, Ben. “Negative Results: The Dark Matter of Research.” AJE Scholar, Research Square, www.aje.com/arc/negative-results-dark-matter-research/.

Blake, Aaron. “Here's How Many Americans Are Actually Anti-Vaxxers.” The Washington Post, WP Company, 26 Apr. 2019, www.washingtonpost.com/news/the-fix/wp/2015/02/09/heres-how-many-americans-are-actually-anti-vaxxers/.

“No MMR Vaccine-Autism Link in Large Study.” Autism Speaks, 21 Apr. 2015, www.autismspeaks.org/science-news/no-mmr-autism-link-large-study-vaccinated-vs-unvaccinated-kids.

Amici, David. “Are You Guilty of P-Hacking?” Bitesize Bio, 14 Jan. 2020, bitesizebio.com/31497/guilty-p-hacking/.

Pannucci, Christopher J, and Edwin G Wilkins. “Identifying and avoiding bias in research.” Plastic and reconstructive surgery vol. 126,2 (2010): 619-25. doi:10.1097/PRS.0b013e3181de24bc

Head ML, Holman L, Lanfear R, Kahn AT, Jennions MD (2015) The Extent and Consequences of P-Hacking in Science. PLoS Biol 13(3): e1002106. https://doi.org/10.1371/journal.pbio.1002106