Studies have quantified various specific benefits related to the use of medical scribes, finding physician workflow and productivity improvements, with some demonstrating marginal value or detrimental impact. However, this evidence base misses a critical underlying issue with the expanding number of physicians using medical scribes routinely. There are an estimated 28,000-33,000 peer reviewed biomedical journals worldwide, currently publishing an estimated 1.8-2 million scientific articles every year. Over a typical physician’s career from the 11-13 years of undergraduate through medical school and specialty/residency training as well as 34-36 practice/care delivery years beyond (to age 65), this yields 84-94+ million peer reviewed journal articles that are published in the global medical literature and to be potentially consumed/ considered over a roughly 47-year career. Clinical trial results in various stages of peer review, with 409,000 clinical trials registered in 2022, augment this massive volume of new clinical and bioscience information that clinicians might utilize to advance their care delivery by over 19 million bioscientific reports over a lifetime of training and care delivery.
Inclusive of clinical trial reports and peer reviewed journal articles, a physician might derive clinical care value from an expanding career-long evidence base of 103-113+ million scientific communications. Even if only 0.1 percent of the global output of biomedical science has clinical relevance to a highly specialized physician, the narrowed career-long total remains a staggering 103,000 journal publications and clinical trial reports. For physicians with a more general and diverse clinical focus such as family medicine, emergency medicine physicians, and hospitalists, if 1 percent of newly published evidence-based literature is pertinent, the total career-long estimate is over 1 million journal articles and clinical trials to be reviewed and clinically integrated.
As a result, a challenging issue created by the increasing role of medical scribes is not just evaluating their value (or lack thereof) for practicing physicians in their workflows and productivity. Rather it concerns the impact that medical scribes may be having by decoupling physicians from the iterative technological and cognitive progression of the electronic health record (EHR) and its evolving artificial intelligence (AI), which can facilitate the integration of the year-over-year proliferation of clinically pertinent new scientific evidence into a physician’s practice of medicine. This commentary addresses the challenge to the evolution of the AI of the EHR posed by physicians’ increasing use of and reliance upon medical scribes, and highlights how medical scribes may also, inadvertently, isolate and insulate physicians from their essential role in continuous refinement and advancement of EHR AI. Consideration is given to the broader challenge of inadequate focus and resources needed across sectors to drive the evolution of AI in the EHR, and associated health informatics research, as a US national priority.
Keywords: medical scribes; electronic health record; artificial intelligence; medical/health informatics; EHR evolution; health informatics research; health IT research
Introduction: Looking Beyond Clinician Value Derived from Medical Scribes to a Systemic Impact of Medical Scribe Use on the Advancement of EHR AI
Over the last decade in the United States, as the use of the electronic health record (EHR) has become ubiquitous, the medical scribe industry has grown dramatically. The medical scribe industry seeks to fill the understandable need and desire of some physicians to liberate and unburden themselves from extremely time-consuming EHRs with poor usability, insofar as safely possible, by having a clerical scribe complete a substantial part of the documentation component of physicians’ EHR workflow. One desired major benefit of medical scribe use is to reduce EHR-related as well as general professional burnout of physicians, nurses, and other clinicians. Current estimates suggest that there could be as many as 100,000 medical scribes employed within the US, serving the nation’s roughly one million professionally active physicians.1 However, medical scribe training remains defined primarily by the industry itself—scribes are frequently medical or nursing students, but no minimum educational background or training requirement have been defined nationally. Further, medical scribe certification, as well as scribe clinical and operational performance, are, just like the industry itself, unexamined and largely unregulated.
Published studies have quantified specific outcomes produced by the use of medical scribes, some finding broad improvements, and others detrimental impact.2-7 However, these studies mostly omit consideration of a critical underlying issue associated with the expanding and increasingly pervasive use of medical scribes by physicians. An estimated 28,000-33,000 peer reviewed biomedical science journals worldwide are currently publishing a collective annual output of an estimated 1.8-2 million peer reviewed journal articles.8,9 In a single decade of a physician’s career, therefore, the scientific literature and evidence base intended to progress the clinical effectiveness and safety of their care of patients may expand by 18-20 million reported studies.
Over the course of a typical physician’s training and delivery of patient care, often about 47 years, if the growth rate of journal articles published annually in the scientific literature remains at current levels—unlikely given recent trends—the evidence base underlying the contemporary practice of medicine may expand by 84-94 million new journal articles. In addition, as of 2022, there were 409,000 clinical trials registered globally, a prolific rate of growth from the 2,119 trials that were registered annually just 22 years ago in 2000.10 If this level of clinical trial growth sustains, this adds over 19 million scientific studies to the literature over a typical physician’s 47-year career length. For physicians seeking to follow the emerging and dynamically changing evidence base to inform and evolve their delivery of patient care, this remains an impressive volume of new journal literature and clinical research to assimilate, even excluding the substantial majority of which may have little or no direct bearing on any given physician’s clinical care and specialty focus.
Thus, inclusive of clinical trial reports and peer reviewed journal articles, a physician confronts an expanding evidence base in excess of 103-113 million scientific communications over the course of their career. Aside from primary care, family medicine, and emergency medicine physicians and hospitalists whose clinical scope is very wide, most specialty and subspecialty physicians are potentially impacted by a far narrower evidence base, given their focus on clinically managing an often much delimited, finite range of pathologies within a clinical scope where they are “learning more and more about less and less” in terms of consumption of new medical science reporting. Nonetheless, when one adds in the imperative for all physicians to keep apprised of certain epidemiological reports issued regularly by municipal, county, state, and federal health agencies/departments about the local incidence of prevalent or highly transmissible communicable diseases, and other advisories of clinical or public health importance, these numbers remain impressive and daunting. Few career endeavors require such a level of continuous integration of newly discovered specialized knowledge and practices over the course of a career.
If only 0.1 percent of articles within the global output of biomedical science has clinical relevance to a highly specialized physician, their career total information integration burden remains a staggering 103,000 reports and journal publications; or 2,191 articles or clinical trial reports per year, every year over 47 years; or six scientific articles/trial reports per day, every day of the year. If only one in 10,000 reports are pertinent to a particular narrowly focused specialist, that drops the annual consumption—including cognitive integration and potential clinical practice application—of their emerging evidence base to 219 articles or reports per year, or 4.2 per week, every week, year-round. For physicians with a more general, diverse, and broader clinical care focus such as family and emergency medicine physicians and hospitalists, if one percent of newly published evidence-based literature is pertinent, potential total career consumption is over 1.03 million journal articles and clinical trials, almost 22,000 articles and reports per year, or 423 articles and trials per week, every week, year-round.
The above estimates assume that global journal article and clinical trial report quantitative output or generation remains at current levels, which, based on trends observed during the last decades, seems counterintuitive and highly unlikely. These metrics convey only crude volumes of information to be integrated and do not consider how effectively individual clinicians can differentiate a journal article or clinical trial report with a strong methodological design and adequate statistical power from one that is weaker. For physician consumers of scientific evidence, even with the support of systematic evidence reviews, meta-analyses, and specific evidence-based clinical guidelines, integrating into practice only the most pertinent clinical implications of global evidence growth is arguably already (or will soon be) beyond human capability and capacity. In effect, the successful progression of medical science and knowledge has outpaced our individual and collective ability to systematically and comprehensively evaluate, integrate, and exploit the massive daily and annual production of biomedical science. Only an artificial intelligence (AI) can coherently and comprehensively keep up with the expansion of medical knowledge and drive its integration into the EHR in an expedited, timely manner so that it can inform every physician’s care.
Given these challenges, an issue to consider about the expanding integration of medical scribes into physician care delivery is not solely what value or negative effects scribes do or do not convey in the context of physician practice and clinical workflow per se. An unexamined question concerns the impact of medical scribes in directly undermining the influence of critical physician EHR end users, stakeholders, and resultant physician/hospital market pressure on a heavily market saturated, highly (financially) successful EHR industry. An imperative exists for the EHR industry to invest in the improvement of not just the usability of EHRs, but in efforts to drive their technological, cognitive, and scientific evolution and progress them to a level where embedded artificial intelligence can real-time surveil for and integrate the enormous year-over-year production of new scientific and clinical evidence, applying it in a clinically meaningful, physician-usable and impactful way in patient care.
AI can potentially integrate and apply this expanding evidence and knowledge base in near real time at the patient level to inform specific episodic patient care delivery, while also informed by what will soon be decades of individual patient (and populational) past medical history EHR data. This is central to the future of EHR AI. How can this AI be developed in the absence of physicians using the full capabilities of the EHR? This commentary endeavors to explore this query and address the challenge to EHR and EHR AI evolution posed by physicians’ increasing reliance on medical scribes, which effectively isolates and insulates them from both the problems—and the opportunities—implied by routine physician EHR use and engagement in continuous EHR refinement. As will become clear, while medical scribes may reduce physician EHR engagement, the rise of the scribe industry is primarily a symptom of a far greater problem in the lack of US national private and public sector investment in advancing the AI of the EHR, and its consequent stagnation.
The EHR as a Delivery Vehicle for the Evidence-Based Transformation of Medicine
The adoption of EHRs, while eliminating paper from clinical workflows and making electronic one of the last major global industries to resist digitization, was only partly about these objectives. The primary and essential value of EHR adoption has been its acceleration of the global practice of evidence-based medicine through science-driven standardization of clinical order set content and order issuance, clinical workflows, and clinical decision support, along with electronic documentation, organization, and leveraging of patient health information. Evidence-based medicine, after decades of systematic meta-analyses of the peer review medical literature led by global collaborative initiatives like the Cochrane Collaboration, revealed that many contemporary medical practices were not based on robust evidence. Substantial clinical care was supported by methodologically flawed and weak studies, many inadequately powered statistically. Yet the adoption and impact of evidence-based medicine was slow, languishing in impact on and use by physicians. Other than issuance of evidence-based guidelines, there was no vehicle to drive and ensure ubiquitous adoption of care exclusively defined by the evidence base into the practice of every physician.
That is, until the near ubiquitous adoption of the EHR in nations with moderately mature health system information technology infrastructure, initially by the most digitally advanced nations but later across a growing spectrum of nations. The EHR, through its computer-based standardization of clinical order issuance, clinical decision support, and electronic documentation functionality, serves as a highly effective mass distribution vehicle for the practice of evidence-based medicine. The order sets within computer patient order entry (CPOE) and management are driven exclusively by peer-reviewed evidence, as is the integrated clinical decision support (CDS). As millions of physicians around the world adopt the EHR, they will practice continually evolving and refreshed evidence-based medicine, and today at least a half a billion patients are realizing clinical effectiveness and safety benefits as a result. However, all the patient data captured in the EHR promises a future impact as well, where data-driven analytics and the integration of new science and evidence merge seamlessly with CPOE, e-documentation, and CDS to yield an evolution in early detection, sensitivity and specificity of disease diagnosis, and improved clinical effectiveness and safety achieved in part by the clinical integration of peer reviewed global science to benefit every patient.
The objective of advancing clinical AI in the EHR, which is to imitate, emulate, and ultimately exceed the abilities of human intelligence, including inference, decision-making, and prediction, is more complex than in many other fields.11,12 Already, AI has been utilized effectively in medicine, delivering value and advancements in speech recognition, image recognition, expert systems, intelligent tutoring, predictive clinical guidance and decision-making, adaptive neural networks, deep and symbolic machine learning, natural language processing, and complex statistical analyses for varied healthcare uses.13-21 AI has demonstrated value in disease assessment, diagnosis, clinical problem resolution, and prognostics. Illustrations include AI prescriptive and predictive analytics that improve inpatient care and reduce clinician workload,22 virtual counseling AI for training nurses in enhanced communication skills,23 AI standardized electronic care handover that improves patient safety/quality and efficiency,24 and AI medical information processing in emergency care.25 These examples only hint at the great promise of AI in the EHR, if it can systematically evolve and be harnessed.
Growth in Biomedical Science has Exceeded Human Cognitive Capacity: EHR Adoption as Just the Beginning of a Journey that Progresses Through Artificial Intelligence
No human mind or multidisciplinary, multispecialty team of minds can complete the critical integration and clinically meaningful application of the tsunami of continuously expanding medical science evidence base and soon-to-be decades of individual patient EHR/medical history data. But AI can help accomplish and accelerate this imperative to deliver effective patient care through and based on the integration of the rapidly expanding, evolving evidence base. EHR AI can only continue to grow and thrive if physicians continue to use EHRs, personally and directly, so their expertise, user experiences, and learning, including dissatisfactions as well as inspired care improvements and creative refinements, drive EHR evolution. Medical scribes, by reducing or eliminating physician interface and use of the full EHR, interrupt and eliminate physician-driven input and advancements in EHR and EHR AI functionality. This viewpoint was first articulated eight years ago,26 and in the interim, medical scribe use—and the scribe industry—have continued to grow significantly, disconnecting an increasing number of physicians from the EHR as its most critical end users and as essential drivers of EHR improvement and evolution of its AI.
The Meaningful Use of EHRs era, initiated and funded by the Obama administration, achieved its objectives of driving EHR adoption in the United States, resulting in 92 percent of American hospitals and 75 percent of office-based physicians currently utilizing EHRs. EHRs are being adopted internationally as well across Europe, the Middle East, and in parts of Asia and Latin America. However, the last decade of adoption was only the beginning, not the end, of physicians’ journey with the EHR. Now that EHRs are more widely adopted, this and subsequent generations of physicians need—or one could argue have a responsibility—to drive improvements and refinements to the EHR, and its emerging AI, with knowledge and insight that can only be derived from their clinical training and experience. This evolution of the EHR must involve physicians’ personal/individual and collective use of the EHR as clinicians. Medical scribes fundamentally disrupt the physician-EHR-AI ecosystem and the technology advancement lifecycle that are essential to driving advancement of EHR AI. The medical scribe industry is effectively relegating the clinician’s training and experience to the background through reduced EHR use in favor of use by individuals who lack the training, depth of understanding and experience needed to identify and distill the gaps, weaknesses, imperatives, and opportunities that can drive EHR advancement.
Why Physicians Are Essential to the Evolution of EHR AI
The evolutionary development of AI within EHRs cannot occur with physicians disconnected from the EHR and instead deploying college students and other clinically untrained individuals working as medical scribes. By effectively isolating and insulating the EHR’s most critical users—physicians—from its technological progression, medical scribes are contributing to the stagnation of EHR innovation. When physicians can relegate EHR interface and use to others, and when they are completely uncoupled from the current crude state of the EHR, their unique ability to drive EHR vendors unwilling to invest in improving their product vanishes. The fundamental question about medical scribes is not how well they capture information or ease and expedite physician workflows; rather, it is how any complex, highly advanced interactive technology can evolve without its primary intended end users engaged and using it.
Medical scribes have neither the training nor the experience to drive the evolution of EHR AI; only physicians, nurses, and other clinicians can do so. Continuous improvement and refinement of the contemporary EHR can only be meaningfully driven by the most critical generators and users of patient clinical information as it evolves during care delivery; by those who originate and most frequently implement clinical care orders: physicians and nurses, respectively. Thus, physician, nurse, and other clinician insights into how current EHRs are not optimized for their workflows can only be rendered by those clinicians, not by individuals such as scribes who have little or no clinical training (and no role in clinical care delivery to patients). Furthermore, as employees of medical scribe vendors, individual scribes who might perceive problematic EHR workflows or other issues face an inherent conflict of interest, because in identifying poor EHR performance, navigation, usability, or disruption in clinical workflows, they risk effectively biting the hand that feeds/pays them: medical scribe vendors. As regards evolving EHR AI, medical scribes are not only untrained and under-skilled, but financially conflicted.
Thus, while studies of medical scribes deliver insights into their impact and utilities in clinical settings and workflows,27-32 they do not address the central objective and imperative of rapidly advancing AI within EHRs so that machine capacity and intelligence can help integrate the expanding evidence base and enable it to be increasingly actionable and valuable to physicians in their care of patients. As the populations of many nations age, as a greater percentage of patients will present complex chronic comorbidities to manage clinically, and as medicine innovates and the evidence base continues to expand voluminously, physicians will need their individual—and inherently limited—clinical intelligence augmented with the EHR’s artificial intelligence more than ever. Thus, every physician who chooses, for understandable personal professional reasons, to opt out of using the EHR by employing a medical scribe is, effectively, choosing to disregard the immense need—one could argue every physician’s professional obligation—to advance the EHR and its evolving AI to provide more clinically effective and safe care to their patients.
Recognizing and Reducing the Contribution of Current EHR Technology to Clinician Burnout
As a chief medical information officer working to support and improve 15,000 physicians’ work lives with EHRs, our team captured every concern, issue, and unmet need physicians articulated about the EHR.33 Some we could resolve, many we could not, as only the EHR vendor could do so, but often not until a next version or release of its platform, if then. Current and future generations of physicians and nurses have been forcibly placed on a kind of heroic journey not only to work with the rudimentary EHR of this early period, but to advance it. Massive improvement in the technology is coming, and we asked physicians to recall that only 15 years separated the first generation of single (and poorly) functioning mobile telephones in the 1990s and the beginning of the smartphone era that connected us to the internet, placing the digital world in our pockets. The technological evolution of the smartphone has conveyed almost limitless applications and value, transforming how we live, work, and play through an elegant, intuitively navigable, mobile device at lesser cost.
Of course, 15 years or longer is a substantial part of any individual physician’s career life. Nonetheless, it is only through capturing clinicians’ individual and collective frustration with EHRs, where their time is lost and where their diagnostic/treatment and cognitive workflow needs are unmet, that the opportunities for point-of-care AI-driven clinical enhancement of the EHR’s potential to empower physicians can be realized. It is through physicians sharing their actionable insights and recommendations that today’s EHR AI will evolve. Physician EHR users have a critical role to play in working collaboratively with medical informaticists, data and AI scientists, and the EHR industry to drive the evolution of the EHR and its emerging AI. This future AI will not only make medicine more clinically effective, safer, and cost-effective; it will evaluate, distill, integrate, synthesize, and inform physicians of the clinical care application derived from the massive continuing expansion of medical science evidence and best practices. EHR AI will empower physicians in a way that they are unable to accomplish as individuals, and that the existing healthcare system and medical science infrastructure is unable to convey.
Evolving AI within the EHR will also help make the practice of medicine more satisfying and can potentially eliminate the EHR as a contributor to clinician professional and EHR burnout. Advancing EHR AI will make use of the EHR more intuitive, less clerical, faster, and more efficient. It will make future physicians more clinically powerful and enable precisely what physicians using medical scribes are seeking: more time to deliberate about patient diagnosis and care, and to engage with patients and their families. But evolution of EHR AI is threatened by the increasing decoupling of physicians from the EHR as its primary user due to reliance on medical scribes. By deploying a new professional role to shield physicians from the EHR, we are throwing out the baby with the bath water, giving up at the precise moment when EHRs have achieved ubiquity in the US, and handing over a major part of EHR functionality to a clerical scribe who is inherently unable to improve the technology.
Much as the test pilots of supersonic jets in the post-WWII era facilitated the evolution of subsequent engine design technology that was used in the exploration of space, generations of physicians will need to be in the pilot’s seat of the EHR to truly evolve AI that clinicians increasingly need and that their patients deserve. Is this analogy overly drawn? Perhaps, but less so if one considers not physician-EHR pilot lives lost, but those of their patients in an era when an estimated 110,000 to 400,000 patients in the US are killed annually due to medical errors,34 arguably the third leading cause of death in the nation prior to the COVID-19 pandemic.35 One of the greatest obstacles to the evolution of AI within the EHR is the increasing abandonment of the EHR by the expanding number of physicians who have rationalized giving up on it, dismissed their duty to use it and through that use, drive the continuous refinement and evolution of what should and must become one of the most powerful innovations in medical history.
Recommendations: Converging Critical Imperatives and Opportunities to Move Forward
Given the above factors and realities, what is the best way forward? There is little doubt that a substantial component of EHR utilization by physicians and other clinicians is work well below the scientific and clinical “top of license” functioning of these care providers. Given the volume of data captured by EHRs—some of which mediates the financial reimbursement of care delivery and revenue cycle management not directly pertinent to patient clinical care—a role for “new data occupations” in healthcare around the EHR such as medical scribes may have emerged.36 While information capture and clerical coding of care delivery processes are central to the transactional financial components of healthcare, and are driven by decisions and orders issued by physicians, this must be disaggregated from efforts to advance the artificial intelligence embedded within EHRs that can improve clinical care effectiveness, cost-effectiveness, and patient safety.
The better ultimate answer to the problems of poor EHR navigability, excessive physician time consumption/inefficiency, physician EHR burnout, and others that clinicians rightly have with the contemporary EHR is to improve the technology so these problems are mitigated, not to divorce physicians from the EHR. If physicians disengage from the EHR, their use-based dissatisfactions, critical perceptions, and insights toward advancing AI in the EHR will be lost and thus unable to drive progress of the EHR and its AI toward its potential to enable better, more scientific and evidence-based, clinically effective, and safer care. Medical scribes partly close the door of this opportunity for advancement of EHR AI to converge and distill the expanding medical science evidence base to empower physicians in improving patient outcomes and safety.
Almost a decade into its existence, the medical scribe industry remains a rapidly growing frontier business that is unregulated, where training and performance standards are defined and monitored not by independent and objective third parties but by the industry itself and specific medical scribe vendors. This conflict of interest is not tolerated in any of the healthcare professions and must be eliminated. Medical scribes remain an ill- and undefined, industry-trained healthcare system role held to no objective performance standards beyond satisfying the clinicians and care delivery organizations who “rent” them, and meeting what medical scribe companies deem the minimum necessary skill level and training to sell their services to physicians and care delivery organizations. Furthermore, while the impact of medical scribes has been and continues to be studied with respect to patient throughput, cost-effectiveness, easing physician work burden and burnout, no system of monitoring for EHR inaccuracies and clinical errors caused by medical scribes exists, and our understanding of who medical scribes are professionally and with respect to competencies remains very limited.
In order to differentiate the clerical roles of medical scribes from the critical role physicians must play in experiencing/using the EHR with its clinical decision support and AI and advancing it, the following is recommended:
1. Differentiating What Medical Scribes Can Contribute Without Undermining Advancement of EHR AI
A national multicenter and evidence-based process should systematically examine the role and function of medical scribes in the continuous development of AI within the EHR, involving the multidisciplinary expertise and insights of physicians, health informaticists, data and AI scientists. The objective and focus of this process should include defining critical areas where direct EHR utilization by clinicians is imperative to advancing EHR AI, and where medical scribe use should be prohibited and physician engagement required, versus those areas that are comprised purely of non-clinical clerical functionality and impact that can be relegated to a new role such as medical scribes (with standardized appropriate training, certification, performance evaluation, and continuing education requirements).
2. Funding Needed for Research Driving Accelerated Development and Deployment of EHR AI
Minimizing the negative impact of medical scribes on the development of EHR AI by ensuring essential engagement by clinicians can mitigate detrimental effects introduced by medical scribes, but does not by itself ensure rapid evolution of EHR AI. The value proposition of the EHR at the outset of the Meaningful Use era was at least threefold. The EHR was to take the last major American industry electronic, seizing the varied efficiencies and utilities gained through digitization, including advancing health information management and access, as well as facilitating healthcare financial transactions. Second, the EHR was to enable—and has enabled—the mass distribution and effective enforcement of and compliance with the science and practice of evidence-based medicine as it continuously evolves through the EHR, which, prior to EHR ubiquity, no such pervasive or systematic vehicle existed. And third, the immense volume of individual patient and population data captured by the EHR was to drive a transformative human/patient outcomes, safety and financial return on investment by enabling valuable analytics and precision in healthcare delivery in general, and at the point of care hitherto not possible.
The first two of these objectives have been achieved. However, the third area of value delivery, which focuses on enabling advanced data aggregation, synthesis, analysis, and evidence-based clinical care guidance, is linked inextricably to the development of the AI within the EHR, whose fruition has been unacceptably slow. Despite a plethora of health IT industry actors, including and beyond the EHR industry, trying to achieve and/or claiming achievement of this critical objective, the transformative advancement of healthcare precision, patient outcomes, and safety envisioned to emerge from EHR ubiquity have not substantially materialized. How can this be in an era when AI is trusted to fly aircraft, drive road vehicles, perform robotic surgery, and land exploratory devices on other planets?
After investing massively to achieve EHR adoption in more than 90 percent of US hospitals, increasing EHR vendor growth to remarkable levels, investment in the next critical phase of AI advancement—of driving/evolving EHRs to deliver their greatest potential value—has dwindled. We continue to be slow in recognizing that because health information technology is so central to everything that occurs in a modern healthcare system at all levels—national and institutional—engagement and investment in health IT research on the digital transformation of healthcare should be equal to that which supports laboratory basic biomedical research and clinical trials. The ongoing evolution and expansion of telemedicine, virtual health, wearables and remote monitoring, and how these will be integrated with EHR AI, have ensured the centrality of IT in the US healthcare system—in its performance and outcomes at all levels. Moreover, health IT and informatics research constitute a discrete and critical area of population health and public health research. Just as the conduct of clinical trials research benefits hospitals, health IT and informatics research should as well.
What is required to enable this potential transformation is a systematic, multidisciplinary, multicenter, and well-funded national research investment to advance EHR AI to achieve improved clinical impact on diagnostic accuracy, diagnostic and care timeliness, and improved therapeutic and patient outcomes, as well as operational and financial efficiencies. The longstanding year-over-year impact of healthcare-related errors as a leading cause of death in the US,37 as well as unacceptably high levels of avoidable morbidity, care utilization and human suffering across the nation—despite the cliché—warrants a Manhattan Project or Apollo Moonshot level of US national commitment, focus, intensity, and resourcing.
This research effort should be managed and implemented by a consortium of academic medical centers, non-academic public and private community hospitals, and health IT, informatics, data and AI scientists and thought leaders, with the integral participation of EHR vendors. It is quite clear, however, that the critical imperative of advancing AI within the EHR is not actively being driven solely or even primarily by the EHR industry/marketplace and market forces. Despite years of high EHR industry revenues and growth, with hospitals and health systems often spending tens or hundreds of millions of dollars on the purchase, implementation, and maintenance of EHRs, the industry has not demonstrated it can drive meaningful advancement of EHR AI. Indeed, there appears to be little or no financial incentive and/or sense of urgency within the EHR industry to advance the AI within its technology product. Perhaps it is unreasonable to expect that industry alone could accomplish such a massive evolution of technology, as the task likely demands cross-sector and public-private partnership and collaboration.
The rise of the medical scribe industry has also insulated or shielded the EHR industry from the kind of consumer dissatisfaction and resulting market pressures that normally drives innovation in industry by systematically eliminating the most important, demanding and disenchanted customers-end users of their product—physicians—from the impact of consumerism and usual market forces. Given the ongoing national crisis in healthcare related errors and patient safety, and the modest impact thus far of the EHR broadly on patient care outcomes, as well as ongoing physician dissatisfaction with EHR demands and performance, a multi-billion dollar, independently operating fund should be developed to invest in advancing the AI of the EHR.
This fund should be financed in part by the EHR industry, along with US federal government investment of its institutional medical science and population/public health research assets, capabilities and resources at the National Institutes of Health (NIH) and the Centers for Disease Control and Prevention (CDC). Its goals should be to drive multicenter research and government-industry collaboration to evolve EHR AI and advance the technology in a non-proprietary manner that will benefit all EHR vendors, patients, and physicians across the US and the globe. American industry, medical and public health science, and government are well positioned to collaboratively foster a technological transformation that will eventually touch the lives of people everywhere with impact and value equal to that of the personal computer, the internet and smartphones, its forbearers and component vehicles. Developed in tandem with the ongoing disruption brought by telemedicine, remote monitoring and wearables, and embraced creatively by multilateral organizations such as the World Health Organization (WHO) and major donors, a truly intelligent and personalized/personally connected EHR can potentially be deployed encompassing the Internet of Medical Things (IOMT) to also help overcome the entrenched and resistant health inequities and outcomes disparities existing in most nations.
3. Preparing and Educating Physicians and Population Health Scientists to Understand and Engage Their Critical Role in Advancing the AI of the EHR
The central and pivotal role of physicians in advancing the AI in the EHR should be a focus in the undergraduate and professional training of physicians in the medical school curriculum, complemented with efforts dedicated to preparing them for their roles in EHR use and advancement so that they can contribute to their resolution. The appropriate and inappropriate uses of medical scribes, as determined by the above discussed research, should be an element considered within such curricular content.
For graduate students in public health and related population health sciences, a training focus on the EHR as both a vehicle for capturing population data, and for potentially for program development aiming to achieve and document improved population health outcomes, is similarly warranted. Along with physicians, these domain experts and stakeholders will help the nation mine value from the EHR as a strategic public health improvement capability/vehicle, and as a population outcomes data source and asset. There is a population/public health AI to be developed in the EHR in addition to a clinical AI, including in areas such as expanded, heightened disease/risk surveillance, earlier detection of population health/disease trends in both infectious and chronic non-communicable diseases, systematic population health engagement and behavioral change/risk reduction, and population impact outcomes evaluation.
4. Evaluating Patient Safety When Medical Scribes are Used
At present, we have no systematic data collection to evaluate the extent to which problems in communications or clinical workflows, or other factors, can be attributed to the use of medical scribes. Because medical errors occurring where malpractice has been claimed often end in negotiated settlements which are sealed from scrutiny, we have no objective sense of the extent to which medical scribe use contributes to errors or negative patient outcomes. Thus, we have inserted a new element in care delivery across the nation interposed between physicians and their EHR documentation—and between physicians and patients some might argue—without monitoring for and assessing its downside risks, in addition to its upside benefits. Systematic research is needed to compare the patient safety performance of care delivery organizations using medical scribes versus those that do not.
5. Medical Scribe Industry and Practice Oversight
Healthcare regulatory organizations and authorities such as The Joint Commission and local, state and federal healthcare agencies focused on the healthcare professions should ensure that their evaluation and oversight of healthcare delivery performance and standards includes review/assessment of the appropriateness of medical scribe use and performance, once the parameters of appropriate scribe use are identified and standardized.
6. Standardizing Medical Scribe Training Curriculum and Minimum Competencies
Statewide and nationally consistent or standardized training, and a curriculum with defined minimum competencies, needs to be defined for the role of medical scribes in assuming the purely clerical functions of EHR documentation, and clearly differentiating the latter from any scribe EHR engagement/use that contributes to critical clinical care workflows and executive clinical decisioning of physicians in the diagnosis and treatment of patients. While the input, knowledge, and experience of the medical scribe industry can be utilized in defining this training curriculum, its definition, implementation and monitoring should be completed by a third party, independent agency, and organizations with no vested financial interests in the scribe industry.
7. Standardizing Medical Scribe Certification, Continuing Education, and Performance Monitoring
A certification examination and process for individuals who have completed a nationally defined and standardized curriculum of training for medical scribes should be established in order to ensure that all medical scribes in the US have achieved and maintain minimum required competencies. The training, evaluation, professional certification, and periodic re-certification of medical scribes should be completed by an organization that is not funded or sponsored exclusively by the medical scribe industry, and rather by a fee or other financial mechanism, much as the examination and certification processes of physicians, nurses, and other professions in healthcare are funded. As patient facing and impacting roles within healthcare delivery, medical scribes should be required to engage in continuing education and recurrent examination and certification, much as other clinical care roles in healthcare delivery are required to (such as physicians, nurses and ancillary care providers).
In sum, with respect to the evolution of EHR AI, it is clear that the rise of the medical scribe industry is both a cause of—but also a symptom of—the current slow progress toward this goal. While medical scribes may be of value to some physicians by narrowing or delimiting their EHR use, it is in the ultimate interest of patients and physicians to evolve the AI within EHRs such that medical scribe use is not regarded as, and does not become, a long-term solution filling existing critical needs. During this transition period, and until EHR AI and general EHR usability/efficiency achieves its required performance level and potential, the use of medical scribes should be highly circumscribed, and must abide by evaluation, monitoring and regulatory standards similar to other professional healthcare roles that impact patient outcomes and safety. It is imperative that medical scribe use must not decouple physicians’ critical engagement in evolving the AI of the future EHR, and thereby impede the realization of its potential and implicit promise to improve healthcare delivery and clinical as well as population/public health outcomes.
Conflict of Interest Statement: The author has no conflict of interests to report.
Funding: No funding supported this work.
Ethics Statement: No patients or patient data were used in this work.
1. Advisory Board, With scribes, doctors think medically instead of clerically. The Daily Briefing. January 14, 2014. https://www.advisory.com/daily-briefing/2014/01/14/with-scribes-doctors-think-medically-instead-of-clerically. Accessed September 2, 2022.
2. Bossen C, Chen Y and Pine KH, The emergence of new data work occupations in healthcare: The case of medical scribes. International Journal of Medical Informatics 2019;123:76-83. doi:10.1016/j.ijmedinf.2019.01.001
3. Mishra P, Kiang JC and Grant RW, Association of medical scribes in primary care with physician workflow and patient experience. JAMA Internal Medicine 2018;178(11):1467-1472. doi:10.1001/jamainternmed.2018.3956
4. Gottlieb M, Palter J, Westrick J and Peksa GD, Effect of medical scribes on throughput, revenue, and patient and provider satisfaction: A systematic review and meta-analysis. Annals of Emergency Medicine 2021;77(2):180-189. doi:10.1016/j.annemergmed.2020.07.031
5. Gidwani R, Nguyen C, Kofoed A, et al., Impact of scribes on physician satisfaction, patient satisfaction, and charting efficiency: A randomized controlled trial. Annals of Family Medicine 2017;15(5):427-433. doi:10.1370/afm.2122
6. Ullman KM, L, Bart B, Park GM, R, Linskens E and Wilt T, The effect of medical scribes in cardiology, orthopedic, and emergency departments: A systematic review. Evidence Report, Health Services Research and Development Service OoRaD, Department of Veterans Affairs; 2020:98. VA ESP Project #09-009. September 2020. Accessed September 2, 2022. https://www.hsrd.research.va.gov/publications/esp/reports.cfm 85
7. Florig ST, Corby S, Rosson NT, et al., Chart completion time of attending physicians while using medical scribes. AMIA Annual Symposium Proceedings 2021:457-465.
8. Gower T, What are medical journals? WebMD May 12, 2021. Medical Journals: What You Should Know (webmd.com). https://www.webmd.com/a-to-z-guides/medical-journals#:~:text=There%20are%20about%2030%2C000%20medical,and%20JAMA%2C%20are%20household%20names.
9. Eveleth R, Academics write papers arguing over how many people read (and cite) their papers.
10. Smithsonian Magazine, March 25, 2014. Academics Write Papers Arguing Over How Many People Read (And Cite) Their Papers | Smart News| Smithsonian Magazine. https://www.smithsonianmag.com/smart-news/half-academic-studies-are-never-read-more-three-people-180950222/
11. Stella NN and Madhu, BK. Impact of expert system as tools for efficient teaching and learning process in educational system in Nigeria. International Journal on Recent and Innovation Trends in Computing and Communication 2017: 5(11), 129-133.
12. Tang KY, Chang CY and Hwang GJ, Trends in artificial intelligence supported e-learning: A systematic review and co-citation network analysis (1998– 2019). Interactive Learning Environments 2021: 1-19.
13. Stella NN, 2017.
14. Tang KY, Chang CY and Hwang GJ, Trends in artificial intelligence supported e-learning: A systematic review and co-citation network analysis (1998– 2019). Interactive Learning Environments 2021: 1-19.
15. Lu H, Li Y, Chen M, Kim H and Serikawa S, Brain intelligence: Go beyond artificial intelligence. Mobile Networks and Applications 2018: 23(2), 368-375.
16. Azzi S, Gagnon S, Ramirez A and Richards G, Healthcare applications of artificial intelligence and analytics: A review and proposed framework. Applied Sciences 2020: 10(18), 6553.
17. Seibert K, Domhoff D, Bruch D, Schulte-Althoff M, Fürstenau D, Biessmann F and Wolf-Ostermann K, Application scenarios for artificial intelligence in nursing care: Rapid review. Journal of Medical Internet Research 2021: 23(11), E26522.
18. Shorey S, Ang E, Yap J, Ng ED, Lau ST and Chui CK, A virtual counseling application using artificial intelligence for communication skills training in nursing education: Development study. Journal of Medical Internet Research 2019: 21(10), e14658.
19. Moreno‐Fergusson ME, Guerrero Rueda WJ, Ortiz Basto GA, Arevalo Sandoval IAL and Sanchez–Herrera B, (2021). Analytics and lean health care to address nurse care management challenges for inpatients in emerging economies. Journal of Nursing Scholarship 2021:53(6), 803-814. https://doi.org/10.1111/jnu.1271. Accessed September 2, 2021.
20. Zhou J, Zhang, F, Wang H, Yin Y, Wang Q, Yang L et al., Quality and efficiency of a standardized e‐handover system for pediatric nursing: A prospective interventional study. Journal of Nursing Management Journal of Nursing Management 2022. https://doi.org/10.1111/jonm.13549. Accessed September 2, 2022.
21. Liu Q, Yang L, and Peng Q, (2022). Artificial intelligence technology-based medical information processing and emergency first aid nursing management. Computational and Mathematical Methods in Medicine 2022. https://doi.org/10.1155/2022/8677118. Accessed September 2, 2022.
22. Azzi S, 2020.
23. Shorey S, 2019.
24. Zhou J, 2022.
25. Liu Q, 2022.
26. Gellert GA, Ramirez, R, and Webster SL, The rise of the medical scribes industry: Implications for the advancement of electronic health records. Journal of the American Medical Association 2015;313:1315-16.
27. Bossen C, 2019.
28. Mishra P, 2018.
29. Gottlieb M, 2021.
30. Gidwani R, 2017.
31. Ullman KM, 2020.
32. Florig ST, 2021.
33. Gellert GA, Hill V, Maciaz G, Bruner K, Saucedo L, Catzoela L, Ramirez R, Jacobs WJ Nguyen P, Patel L and Webster SL, Successful implementation of clinical information technology: Seven key lessons from CPOE. Applied Clinical Informatics 2015; 6: 698–715.
34. James JT, A new, evidence-based estimate of patient harms associated with hospital care. Journal of Patient Safety 2013;9:122-128.
35. Makary MA and Daniel M, Medical error—the third leading cause of death in the US. British Medical Journal 2016;353:2139 doi: 10.1136/bmj.i2139.
36. Bossen C, 2019.
37. Makary MA, 2016.