The following article is part one of a two-part essay examining the current challenges facing science and medicine in particular arising from fraud, conflicts of interest, political interference, and misaligned priorities, with direct harm to patients and erosion of public trust.
“The most solid piece of scientific truth I know is that we are profoundly ignorant of nature.” Lewis Thomas
Science and medicine are in crisis. Fraud, conflicts of interest, political interference, and misaligned priorities are eroding the foundations of medical research and, with them, public trust. The crisis is not theoretical; it has direct consequences for patients, for the practice of medicine, and for the future of biomedical innovation. As clinicians, we cannot afford to look away. The purpose of this two-part essay is to examine the nature of the crisis, name its causes, and propose a path forward.
The success of modern science and medicine has become its curse. Life expectancy in the late 1700s in America was 35 to 40 years, a value extremely skewed by high infant mortality rates, with many children dying before the age of five. The development of public health initiatives providing clean water and sanitized waste, childhood vaccines, antibiotics, and modern medical practice dramatically shifted that paradigm. Current life expectancy in the U.S. is 81.4 years for women and 76.6 years for men, with infant mortality close to 0.55% of live births. The advances that have saved lives are making us complacent. Only in a world where serious childhood disease is rare and resources are readily available would an educated society entertain withholding proven vaccines and routinely practice pseudoscience as lauded by “influencers.” Historical perspective matters because it reminds us of what science and medicine have accomplished and are protecting us from. When that memory fades, the institutions that produced those gains become easy targets for skepticism, ideology, and disinformation.
Although significant criticism can be made of the disproved 1998 and 2002 publications of Andrew Wakefield trying to link autism with the administration of the measles, mumps, and rubella vaccine, issues of misinformation and fraud in the medical literature are pervasive and ongoing. They are not the product of one rogue investigator. They are a structural feature of how modern science is funded, published, and rewarded. Consider Alzheimer’s disease. Starting in 2021, accusations of scientific misconduct were leveled against the drug developer Cassava Sciences and its CUNY investigator Hoau-Yan Wang, alleging that their experimental drug Simufilam, then in clinical trials, was based on potentially fraudulent research. Despite the FDA being warned by an independent researcher that Cassava’s published claims could not be replicated and that altered or duplicated images had been identified in multiple journal articles, the clinical trials were allowed to proceed. The Phase 3 clinical trial failed; however, the questionable claims put almost 2,000 patients at risk. Wang was cited by CUNY for “egregious misconduct,” although the finding was later withdrawn by the university. He was indicted by the U.S. Department of Justice for defrauding the NIH, although charges have been subsequently dropped.
That same year, the same physician-scientist who identified discrepancies in Cassava’s data identified similar evidence of data manipulation in one of the most influential papers in neuroscience and Alzheimer’s research. The 2006 Nature paper, with first author Sylvain Lesné and corresponding author Karen Ashe, purported the identification of a previously unknown form of beta amyloid, Aβ56. The authors reported that injecting the peptide into young rats inhibited the recall of previously learned information. This was touted as the first direct link between a beta amyloid peptide and cognitive decline in an animal model and further entrenched the beta amyloid hypothesis in Alzheimer’s research and pharmaceutical development. There are currently three FDA-approved agents based on the amyloid beta hypothesis: donanemab-azbt, lecanemab-irmb, and aducanumab. A recently published Cochrane review concluded that these drugs provide no meaningful clinical benefit for patients with mild cognitive impairment or mild dementia. The amyloid beta legacy is well documented in Charles Piller’s book “Doctored: Fraud, Arrogance, and Tragedy in the Quest to Cure Alzheimer’s,” which expands on his investigative reporting and exposes what he calls the “amyloid mafia,” a network that has prioritized novelty over replication, marginalized dissenters, and benefited from the FDA’s repeated approval of drugs of questionable benefit.
The example of Alzheimer’s is dramatic but not unique. It has been estimated that as much as half of the published scientific literature may be false. There are multiple reasons why this problem exists. The primary marker of success and job placement in both academia and industry is an individual’s academic record and history of publication in peer-reviewed, high-impact-factor journals. The incentive structure produces a “natural selection of bad science.” When publication is the principal metric for career advancement, methods that increase the likelihood of publication, regardless of whether they increase the likelihood of being correct, will spread through the research community. The methods that get people hired and promoted are the ones that survive. In a 2012 report, scientists at Amgen attempted to confirm the findings of 53 landmark preclinical oncology papers, reproducing positive results in only six. The same year, Bayer Healthcare reported that, across 67 internal target validation projects, the published findings were only reproducible in 20% to 25% of cases. Reliable estimates indicate that the prevalence of nonreproducible results in preclinical research exceeds 50%. This accounts for approximately 28 billion dollars per year of research spending in the U.S. The Reproducibility Project: Cancer Biology, led by Errington and the Center for Open Science, attempted to replicate experiments from high-profile cancer biology papers and was unable to complete many of them. An analysis of retracted biomedical papers indexed by PubMed found that only 21% of retractions were attributable to honest error; 67% were attributable to misconduct, with fraud or suspected fraud accounting for 43%. The continued growth in retractions partly reflects improved oversight, but the absolute volume of bad data entering and remaining in the literature is significant.
John Ioannidis, a prominent physician-scientist, demonstrated two decades ago that most published research findings are false. His later work has shown that an estimated 85% of research resources are wasted, and that the use of p-values continues to expand even as their misuse becomes better understood. Selective reporting is the rule rather than the exception. Other authors have shown that reported outcomes in randomized trials routinely differ from those specified in the original protocols. Additionally, trials with positive results are four times more likely to be published than those with negative results. A prime example of this is that of 74 FDA-registered antidepressant trials, 31% were never published, and that selective publication and reporting together inflated the apparent effect size by 32% across the class. These are not isolated failures. They are the predictable output of a system that rewards positive results, rewards publication, and rewards speed.
In addition to these structural problems are the financial relationships that increasingly shape the questions science asks and the answers it accepts. Approximately one-quarter of biomedical investigators have industry affiliations and two-thirds of academic institutions hold equity in start-ups that sponsor research performed at the same institutions. Aggregating across more than a thousand original studies, industry sponsorship was associated with pro-industry conclusions at a pooled odds ratio of 3.6, and that industry sponsorship was also associated with restrictions on publication and data sharing. Updating this question in a Cochrane review reached the same conclusion: Industry-sponsored studies are systematically more likely to favor the sponsor’s product than independently funded studies of the same intervention.
The reach of industry into the practice of medicine is not limited to research. A review of the Massachusetts disclosure database documented the routine flow of payments from drug and device manufacturers to physicians across every specialty, from food and continuing medical education to consulting and speakers’ bureau fees. These payments are not, in themselves, evidence of misconduct, but they create relationships that bias prescribing and that, over time, blur the line between the practice of medicine and the marketing of products. Other authors have argued that the “evidence-based” brand has been progressively misappropriated by industry, which now sets much of the research agenda, defines new diseases and pre-disease “risk states” amenable to its products, and shapes which trials are run, and which are reported.
The COVID-19 pandemic illustrated how quickly these structural weaknesses can produce direct patient harm. In an environment of extreme uncertainty, low-quality observational studies were amplified by news media and political leaders, sometimes promoted by individuals with undisclosed conflicts of interest, leading to the widespread use of therapies that lacked robust evidence. The American College of Physicians, in its 2024 position paper on scientific integrity and disinformation, observed that this confusion contributed measurably to a decline in public trust in medical information. The absence of high-quality randomized data forced clinicians to rely on studies that were neither methodologically nor conceptually sound; the resulting recommendations spread rapidly through outpatient settings before they could be evaluated. None of this happens in a vacuum. It happens within a literature already inflated by selective reporting, already biased by industry funding, and already shaped by the incentive structures of academic record and history of publication in peer-reviewed, high-impact-factor journals. When a true emergency arrives, the cracks in the foundation become fault lines.
How can physicians help protect scientific integrity and rebuild trust in the evidence that guides patient care? Share in the comments.
