Menu
Close
.svg)
The pandemic may be behind us, but the weight borne by physicians has hardly disappeared. As AI promises to reshape clinical practice, an enduring burnout crisis raises important questions about how doctors make decisions, absorb medical evidence and embrace therapeutic innovation.
At ThinkGen, the topic of physician burnout has occupied our thinking for many years. One of our colleagues, Dr. Richard Vanderveer, was particularly prescient in identifying it as a defining challenge that pharmaceutical companies could ill afford to ignore. His concern extended beyond the welfare of physicians themselves, important though that unquestionably was. For several years, he almost seemed the only one who recognized that professional exhaustion could fundamentally alter how physicians make treatment decisions, assimilate clinical evidence, engage with pharmaceutical education and ultimately care for their patients. In an industry accustomed to scrutinizing prescribing behavior, he argued stridently more a decade ago that we needed to pay considerably more attention to the context and circumstances in which that behavior occurs.
Several years after a pandemic that brought healthcare systems to the brink, and amid an artificial intelligence revolution promising to transform virtually every aspect of medical practice, his observations seem remarkably apposite. Which bring us to the main question of this essay: where does physician burnout stand in 2026? Has the profession recovered from the extraordinary challenges and privations of the COVID-19 pandemic? And might the technologies now permeating clinical workflows finally alleviate the problems that have proved so intractable?
For those of us working to support the efforts of pharmaceutical, biotechnology and medical technology companies, these questions are neither peripheral nor academic. They go to the heart of how medical innovation reaches patients.
The statistics offer grounds for cautious optimism, albeit hardly celebration. A longitudinal study published in Mayo Clinic Proceedings in 2025, involving researchers from Stanford Medicine, the American Medical Association and other institutions, found that 45.2% of American physicians reported at least one symptom of burnout in 2023, down substantially from 62.8% at the pandemic's height in 2021. Nevertheless, after adjusting for demographic and occupational differences, physicians remained approximately 82% more likely to experience burnout than workers in other professions.
The pandemic, however, was more accelerant than original cause. Physician burnout was already widespread long before COVID-19 patients overwhelmed hospitals, disrupted established routines and subjected health care providers (HCPs) to extraordinary emotional duress. Electronic health records (EHRs), proliferating administrative obligations, reimbursement complexities and diminished professional autonomy had been eroding satisfaction for decades. Indeed, the same longitudinal research found that 45.5% of physicians experienced burnout symptoms as far back as 2011. The pandemic intensified an existing malaise rather than conjuring one from nothing.
The 2026 Medscape Physician Burnout & Depression Report reinforces this uncomfortable reality. Nearly half of full-time physicians surveyed reported burnout, while almost one-quarter experienced depression. Bureaucratic responsibilities and excessive working hours remained prominent culprits. More troublingly, 14% of burned-out physicians described their condition as sufficiently severe that they were contemplating abandoning medicine altogether. These findings suggest that, despite considerable institutional attention, the underlying affliction remains stubbornly entrenched.
Of note, there is an important discrepancy between burnout and depression, just as there is between professional exhaustion and moral injury. Burnout typically encompasses emotional depletion, detachment and diminished professional accomplishment. Moral injury, by contrast, can emerge when physicians find themselves unable to deliver the care they believe patients deserve because institutional or financial constraints obstruct their judgement. Such distinctions matter because they suggest that the remedies must extend beyond counseling, resilience training or the occasional exhortation to practice mindfulness.
Few developments better illustrate the paradoxes of modern medicine than the EHR. Initially conceived to improve documentation, information accessibility and continuity of care, these systems have frequently transformed physicians into reluctant clerks. Research published before the pandemic found that, for every hour spent directly with patients, physicians in ambulatory practice devoted nearly two additional hours to electronic records and administrative work. The phenomenon acquired the rueful appellation of "pajama time", reflecting the hours spent finishing up documentation after returning home.
A 2024 study involving more than 10,000 family physicians, published in JAMA Network Open, identified two particularly consequential factors associated with lower burnout: manageable EHR responsibilities outside working hours and effective clinical teamwork. Interestingly, proficiency with EHRs was not significantly associated with burnout. The problem, in other words, was not that physicians were insufficiently adept at using technology. Rather, the technology and the organizational structures surrounding it were imposing demands that even proficient users struggled to accommodate.
That distinction has profound implications. Physicians are not necessarily resistant to innovation. They may simply be resistant to additional encumbrances masquerading as innovation. An ostensibly helpful tool that introduces another password, dashboard, verification requirement, prompt, or administrative obligation can become yet another source of exasperation. The cumulative burden matters considerably more than the merits of any individual intervention.
Enter artificial intelligence, whose advocates promise to liberate physicians from precisely the administrative drudgery that earlier generations of technology helped create. Ambient AI scribes, which "listen" to clinical consultations and generate draft documentation, have emerged as particularly promising applications. Their appeal is readily apparent: fewer hours typing notes, more attention devoted to patients and perhaps the restoration of something increasingly scarce in medicine: undivided attention during a clinical encounter.
Encouraging evidence is beginning to accumulate. A multicenter quality improvement study published in JAMA Network Open in October 2025 found that burnout among participating ambulatory clinicians declined from 51.9% to 38.8% after 30 days of using ambient AI scribes. Researchers also documented improvements in cognitive workload, after-hours documentation and attention to patients. Although the observational design and relatively short follow-up preclude definitive conclusions about causation or durability, the findings suggest that this application of AI may offer more than superficial convenience.
Yet the picture is considerably more nuanced. A larger study reported by STAT in April 2026 examined approximately 1,800 clinicians across five academic medical centers. AI scribe users saved an average of 16 minutes of documentation time and 13 minutes in EHRs per eight hours of patient care. There was no statistically significant reduction in after-hours electronic record work. Furthermore, only about one-third of users employed the technology for at least half their clinical notes, suggesting that adoption remains inconsistent even among those with access.
These apparently discordant findings invite a more sophisticated interpretation. Perhaps the value of AI cannot be measured exclusively in minutes reclaimed. Relieving physicians of an especially irksome responsibility may produce psychological benefits disproportionate to the time involved. Fifteen minutes spared from tedious documentation may be worth considerably more than fifteen minutes saved from an activity a physician finds intrinsically rewarding. Equally, the ability to maintain eye contact, listen attentively and engage meaningfully with patients may restore elements of professional satisfaction that conventional productivity metrics scarcely capture.
There is, nevertheless, a potentially disquieting corollary. If AI enables physicians to work more efficiently, will healthcare institutions allow them to enjoy the resulting respite, or simply expect them to see additional patients? The answer may determine whether AI truly alleviates burnout or merely facilitates a more technologically sophisticated version of the same exhausting system.
For those working in pharma, biotech and medtech, physician burnout deserves attention for reasons extending well beyond professional sympathy. Consider for example the extraordinary demands imposed upon a contemporary oncologist. Novel mechanisms of action, biomarker requirements, increasingly intricate treatment sequences, evolving clinical guidelines and an expanding repertoire of combination therapies compete for attention alongside a full clinical schedule. Meanwhile, pharma companies continually introduce new evidence, educational initiatives and promotional communications, each predicated on the assumption that physicians possess sufficient intellectual bandwidth to engage.
That assumption warrants closer examination. Burnout may influence not only how physicians feel about their profession but also how they process information, evaluate alternatives and navigate therapeutic decisions. Cognitive exhaustion can encourage reliance on familiar habits and routines, while time constraints may discourage the exploration of unfamiliar treatments requiring additional investigation or coordination. These are plausible behavioral mechanisms rather than universally established consequences of burnout, but they deserve serious empirical scrutiny.
Indeed, the implications for pharmaceutical commercial strategy are potentially substantial. Traditional marketing research frequently concentrates on awareness, perceptions, clinical preferences, prescribing intent and drivers and barriers to adoption. Yet comparatively little attention is devoted to understanding how physicians' workloads, administrative responsibilities and emotional circumstances influence those measures. A physician may recognize the merits of a novel therapy while lacking the practical capacity to incorporate its associated monitoring, reimbursement or patient education requirements into an already congested practice.
Moreover, the commercial response should not simply involve shortening promotional messages or producing more digestible educational materials. The more impactful and consequential opportunity may lie in reducing the practical frictions surrounding treatment adoption. Simplified administration, accessible evidence, clearer patient identification pathways, coordinated support services and fewer bureaucratic impediments can make innovation easier to implement. Medical affairs, commercial strategy, market access and patient services all have roles to play in addressing these challenges.
Artificial intelligence could amplify these efforts, provided it is deployed judiciously. Rather than inundating physicians with additional information, AI-enabled educational resources might help HCPs locate relevant evidence, clarify treatment considerations and navigate increasingly complex clinical decisions. Such applications will require rigorous scientific accuracy, appropriate safeguards and a clear separation between balanced medical information and promotional persuasion. The objective clearly should be to support professional judgement, not circumvent it.
In partnership with firms like ours, pharma, biotech and medtech spends vast resources scrutinizing prescribing behavior, striving to understand what drives (and inhibits) physicians' choices. Less thoroughly examined are the mounting pressures under which those choices must be made.
This means recognizing that physicians' attention is a finite resource, that professional autonomy matters profoundly, and that the practical burden associated with adopting a treatment can be as consequential as the information available about it. It also means appreciating that technological sophistication does not automatically translate into clinical usefulness. An innovation that demands more effort than it relieves may struggle to find favor, however impressive its fundamental attributes and capabilities.
The encouraging news is that physician burnout has receded from its pandemic peak. The less reassuring reality is that medicine remains an unusually demanding profession whose practitioners continue to contend with systemic pressures that technology alone cannot resolve. AI may prove an important part of the remedy, but its success will ultimately depend upon whether it restores physicians' capacity to practice medicine rather than merely increasing their capacity to process work.
Dr Vanderveer's early admonition remains pertinent. If the life sciences industry truly wishes to improve treatment decisions, accelerate appropriate therapeutic adoption and support better patient outcomes, it must understand not only the science confronting physicians but also the circumstances confronting the physicians themselves. Medical innovation has rarely been more abundant. The human capacity to absorb and implement it, however, remains stubbornly finite.
