I entered medicine through science.
Before medical school, I spent a significant amount of time as a research fellow at the NIH, working alongside world leaders in virology during the COVID-19 pandemic. In that environment, I was taught that the scientific literature is not simply a collection of publications or a means to further my career, but a shared public good. Its integrity matters. Its accuracy matters. And when errors persist uncorrected, the consequences extend far beyond academic embarrassment: They shape policy, influence clinical decision-making, and ultimately affect patients.
Equally important, I was trained to understand how science goes wrong. I learned to interrogate methodology, to identify bias and confounding, and to recognize when conclusions outpace data. The lesson was clear: Reverence for the literature does not mean blind acceptance. It means active stewardship.
I carried that ideology with me into medical school. As my interests expanded from bench science to clinical and health services research, I continued to engage deeply with the literature. Some of my earliest independent publications were letters to the editor: brief, focused critiques of published studies that failed to account for obvious confounders or methodological limitations. These letters were not written to be provocative. They were written because something was wrong, and silence felt like complicity.
Over time, I published original research across several domains, but I retained a particular respect for the letter to the editor. When done well, it represents post-publication peer review at its best: precise, evidence-based, and directed at the work rather than the author. It is one of the few mechanisms by which the scientific record can self-correct after formal peer review has failed.
That belief was reinforced later in medical school when, working closely with surgical mentors, I encountered a meta-analysis published in a prominent specialty journal. On closer inspection, it became clear that the study was a meta-analysis of other meta-analyses, repeatedly counting the same patient populations, some up to seven times. This kind of duplication bias is not a minor technical issue. In meta-analyses, where pooled estimates are often treated as the highest level of evidence, such errors fundamentally invalidate conclusions.
Addressing this was uncomfortable. No one wants to see another researcher’s work retracted. But with careful mentorship, we drafted a narrowly focused, methodologically grounded letter outlining the problem. The journal ultimately retracted the paper in its entirety.
I felt no triumph in that outcome. But I did feel a sense of responsibility fulfilled. Correcting the record mattered, especially for a study positioned to influence clinical recommendations and patient counseling.
What troubled me more was what came afterward. As I continued to read the literature, I began to notice how frequently similar flaws, particularly patient duplication in meta-analyses, were identified and acknowledged through published correspondence, yet left formally uncorrected. Some even reported complication rates for outcomes that were not reported in the primary studies at all, effectively imputing data that did not exist. Letters would appear. Concerns would be noted. But reanalyses were rarely required, and original conclusions often remained untouched.
The message, implicit but unmistakable, was that acknowledging error was sometimes sufficient even if the data themselves remained flawed.
That realization crystallized during residency interviews, when I posed a hypothetical question to leadership at one of the most prestigious academic surgical programs in the world. I asked how faculty would respond if a trainee identified a clearly flawed, high-impact study authored by a professional colleague and requested support to submit a letter to the journal correcting the scientific record while balancing professional collegiality.
The response surprised me. I was told, unequivocally, that such public correction should be avoided in all circumstances, regardless of the severity of the flaw, because the surgical field is small and professional relationships matter. Even when I offered an example of the meta-analysis that had ultimately been retracted due to methodological failure, the answer did not change.
I do not share this to single out any institution. But the exchange was momentarily stunning for its unapologetic candor, revealing how professional considerations are often placed ahead of a responsibility to correct the scientific record. It revealed how deeply ingrained this value system has become, even at the highest levels of academic medicine. More importantly, it underscored how these norms are transmitted to trainees, shaping how the next generation of physician-scholars learns to think, write, and act.
Professional relationships do matter. The risks of reputational harm, mentor vulnerability, and political fallout are real. Human beings are imperfect, and academic medicine is not immune to pettiness or retaliation. Acknowledging those realities is necessary.
But silence is not neutral. When those of us best equipped to identify flawed research choose not to act because correction may be uncomfortable, we do more than protect relationships. We allow erroneous data to persist in the very literature that informs patient care. We conflate critique of work with critique of people. And we teach trainees, explicitly or implicitly, that preserving professional harmony outweighs scientific accountability.
Leadership in medicine has never been about avoiding discomfort. It is about making principled decisions despite it. Critiquing flawed research is not an act of hostility; it is an act of stewardship. Iron sharpens iron, but only if we are willing to engage honestly with one another’s work.
Academic medicine holds disproportionate influence over guidelines, policy, and public trust. With that influence comes responsibility. We owe it to our patients and to the credibility of our profession to correct the record when it is wrong, even when doing so carries personal cost.
Rather than discouraging critique, we should be teaching trainees how to do it well: mentor-supported, methodologically precise, and focused on the data rather than the individual. That is how I was trained as a scientist, and it is how we should be training physicians.
