Abstract
This study explores clinicians’ evolving information needs and evaluates the potential of Generative Artificial Intelligence (Gen AI) to address these gaps by reassessing and extending the Currie et al. (2003) taxonomy. Despite advancements in electronic health records (EHRs), unresolved information needs persist, impacting clinical efficiency and patient care. A cross-sectional survey conducted at Columbia University Irving Medical Center (CUIMC) analyzed clinician-generated Gen AI prompts, comparing them against the 2003 taxonomy. Findings reveal that while 80% of prompts align with existing categories, 20% represent emerging needs, including AI-driven workflow optimization and fairness-related inquiries. These findings highlight the necessity of adapting clinical decision support frameworks to integrate AI-driven solutions, ensuring that modern tools meet evolving clinician needs. By formally extending the Currie et al. taxonomy, this study provides a foundational framework for leveraging Gen AI to bridge long-standing information gaps and enhance patient outcomes in an increasingly complex healthcare environment.
Introduction
Timely and accurate information retrieval in clinical settings is critical for informed decision-making, directly influencing patient outcomes, diagnostic accuracy, and treatment efficacy. Clinicians rely on various sources, including electronic health records (EHRs), institutional protocols, peer consultations, and medical literature. However, as medical knowledge continues to expand, doubling approximately every 20 years, unresolved information gaps remain a persistent challenge1. Studies indicate that less than 50% of clinicians’ questions are adequately addressed during patient care, limiting the ability to apply evidence-based knowledge in real time2,3. This issue, initially quantified by Covell et al. (1985) and later categorized by Currie et al. (2003), persists despite significant advancements in EHR adoption and digital health innovations4-7.
While digital health tools have enhanced data accessibility, they have also introduced unintended cognitive burdens. Clinicians spend a significant portion of their workday interacting with EHRs, with family physicians averaging 5.9 hours per day, including 1.4 hours outside of work for documentation, order entry, and data retrieval tasks8. The complexity and volume of digital documentation contribute to decision fatigue and workflow inefficiencies, further complicating clinical decision-making. Existing clinical decision support tools, such as UpToDate, PubMed, and EHR integrations, offer valuable insights, but information overload is associated with increased risk of medical errors and negative patient safety outcomes9.
As healthcare has evolved, so have the tools available to address these challenges. The Currie et al. (2003) taxonomy was developed to categorize clinician information needs systematically, highlighting areas where timely access to relevant knowledge is essential. However, since its creation, clinical workflows have undergone significant changes. The rising documentation burdens and advancements in information retrieval tools point to the need for a reassessment of the taxonomy to better align it with the modern clinical environment9. One of the most notable innovations is the emergence of Generative Artificial Intelligence (Gen AI), which holds the potential to revolutionize clinical decision support by synthesizing vast medical knowledge, analyzing multimodal clinical data, and generating tailored responses.
Recent studies have demonstrated the proficiency of large language models (LLMs) in medical question answering, suggesting potential applications in evidence synthesis, patient case analysis, and real-time decision support10,11. However, despite these advancements, the integration of AI into clinical practice remains contentious. Most existing research prioritizes technical accuracy over clinician workflow considerations, often overlooking clinician perspectives on how AI should be integrated12. Additionally, concerns regarding algorithmic bias, misinformation risks, lack of transparency, and ethical implications further underscore the need for a clinician-centered evaluation of AI’s role in healthcare13,14.
To address these gaps, this study reassesses clinician information needs in the modern era, incorporating a sub-analysis of survey data from Ruan et al. (2024) collected at Columbia University Irving Medical Center (CUIMC), a diverse clinical environment with increasing AI adoption15. It examines the evolution of clinician information needs since Currie et al. (2003) and evaluates Gen AI’s potential to address gaps by analyzing clinician-proposed Gen AI prompts, bridging AI capabilities with real-world needs to enhance workflows, efficiency, decision-making, and patient outcomes. We specifically focus on Gen AI to conceptually understand how emerging tools might address unresolved and novel gaps in decision-making today. To our knowledge, this is the first study to update the 2003 taxonomy of clinician information needs or propose a similar framework that leverages Gen AI to enhance clinical practice.
Methods
A cross-sectional survey was conducted among clinical staff at CUIMC to assess their perspectives on the integration of Gen AI into clinical workflows (Appendix A). Participants were recruited between December 15, 2023, and February 15, 2024, via email invitations and word-of-mouth referrals. The survey was self-administered through Qualtrics, with an introductory information sheet ensuring voluntary participation and anonymity. The study protocol and survey instrument were approved by the Columbia University Institutional Review Board.
The survey comprised 32 questions addressing four primary domains:
Familiarity with Generative AI: Assessing clinicians’ baseline knowledge and exposure to Gen AI
Perceived Usefulness of Generative AI in Clinical Workflows: Exploring clinician perspectives regarding AI-augmented decision support in various types of clinical tasks
Characteristics and Motivations for Adoption: Identifying factors influencing Gen AI uptake in clinical practice
Sentiments Regarding the Impact of Generative AI on Clinical Practice: Examining clinician perspectives on potential impact of various modalities of Gen AI in healthcare.
In addition to structured responses, participants were invited to propose prompts they believed would be valuable for Gen AI to answer in their clinical practice.
To systematically assess clinician information needs, submitted Gen AI prompts were categorized based on the Currie et al. (2003) taxonomy, which organizes clinical information needs into the following domains:
Subject (Patient): Information directly related to individual patient care
Institution: Healthcare system or institutional protocols
Domain: Specialty-specific medical knowledge
Subject-Institution: Intersection of patient care and institutional factors
Subject-Domain: Patient-specific questions within medical specialties
Domain-Institution: Relationship between medical specialties and institutional practices
Subject-Domain-Institution: Overlapping aspects of patient care, medical specialty, and institutional protocols
Other: Gen AI prompts that did not align with the predefined categories
Four additional categories from Currie et al. (2003) (Foreground, Background, Explicit, Implicit) were initially considered but ultimately excluded due to differences in study design, particularly the focus on AI-driven clinical decision support rather than traditional reference material searches.
The remaining survey responses were analyzed to identify new categories of information needs that were not identified in the 2003 work by Currie et al. These unmapped information needs were initially examined to see if they could be aligned with or extend Currie et al.’s existing categories. When prompts did not cleanly fit existing domains, SKT and SCR discussed and either expanded the taxonomy by identifying hybrid or overlapping areas, or created entirely new classifications through thematic analysis, ensuring that novel information needs were adequately represented and distinguished from previous work. Initial mapping to Currie et al. (2003)’s information need categories and identification of new categories was done by SKT and confirmed by SCR.
Survey responses were analyzed using R, employing the dplyr, tidyr, readr, and kableExtra packages. Descriptive statistics were computed to summarize key survey trends. Additionally, a comparative analysis was conducted to evaluate shifts in clinical information needs from the Currie et al. 2003 taxonomy to the CUIMC 2024 survey results.
Results
A total of 185 survey responses were received, with 83 respondents (44.9%) providing at least one Gen AI prompt suggestion. On average, each respondent submitted 1.56 prompts, yielding a total of 130 collected prompts. These Gen AI prompts were manually reviewed and classified based on the Currie et al. (2003) taxonomy to assess how clinician information needs have evolved over time.
Of the 130 submitted Gen AI prompts, 104 (80%) aligned with one of the predefined categories in the Currie et al. (2003) framework (Table 1). However, 6 prompts (4.6%) addressed emerging knowledge gaps that, while still fitting within the broader taxonomy, reflected new contextual demands. The remaining 20 prompts (15.4%) were identified as novel, dynamic needs that could not be categorized within the existing framework, suggesting the emergence of new domains of clinician information needs (Table 1).
Of the 104 (80%) Gen AI prompts that mapped to the Currie et al. (2003) framework, there were discrepancies found in the prevalence of each category. The Subject-Domain category, which accounted for only 4/154 (2.6%) of classified needs in 2003, represented 39/130 (30%) of Gen AI prompts in the 2024 CUIMC survey (Table 2). In contrast, Institution-based information needs declined significantly, from 35/154 (22.8%) in 2003 to 6/130 (4.6%) in 2024.
Among the 26 Gen AI prompts (20%) that did not align with the original 2003 taxonomy, 6 prompts (4.6%) were categorized into three newly emerging categories, while 20 prompts (15.4%) represented dynamic, AI-driven needs that required entirely new classifications (Table 1). The three extended taxonomy categories include Subject-Third Party Institution, Subject-Fairness, and Institution-Subject-Education.
Queries classified under Subject-Third Party Institution included clinician access to financial and insurance-related information, with examples such as “Will medication B be covered by Y insurance, and if so, how much is the co-pay?”. These knowledge gaps related to patient interactions with third parties, such as insurers, represented 3/6 (50%) of the extended taxonomy prompts (Table 3).
Similarly, Subject-Fairness Gen AI prompts (2/6, 33.3%), such as “Where can I find bowel prep instructions in Spanish?”, pertained to language accessibility and social determinants of health in clinical decision-making (Table 3). The Institution-Subject Education category (1/6, 16.7%) emerged as clinicians sought professional development resources, with questions such as “How can I improve my communication skills?”. Primary care clinicians identified these extended taxonomy knowledge gaps at the highest rate (3/6, 50%), followed by surgical specialists (1/6, 16.7%).
The remaining 20 prompts (15.4%), such as “Complete daily assessment tools” and “Clean up the EMR”, were classified as dynamic needs that largely revolved around AI-driven clinical workflow optimization (Table 3). These queries focused on reducing administrative burdens and improving operational efficiency, with frequent mentions of automating daily assessment tools, optimizing EHR documentation, streamlining billing processes, and improving patient scheduling systems. Physicians were the most likely to submit AI-driven workflow augmentation requests, whereas nurses were evenly split between seeking AI education and optimizing job functions (Table 3).
Discussion
The findings of this study suggest that clinician information needs have evolved since 2003, yet the majority (80%) of Gen AI prompts still align with the original Currie et al. (2003) taxonomy, indicating that the fundamental types of clinical queries remain largely unchanged. Gen AI, however, may be better able to answer some of the previously identified questions. Further, 20% of submitted prompts fell outside the original taxonomy, reflecting emerging information needs that go beyond solutions offered by traditional decision-support tools. Six (4.6%) of these new domains largely relate to financial considerations and health fairness, highlighting a need to expand existing information needs classification frameworks to better address modern clinical practice.
The original Currie et al. (2003) taxonomy categorized information needs into discrete domains, such as understanding a patient’s latest laboratory results (Subject), accessing updated institutional protocols (Institution), or identifying potential medications for treatment (Domain). These needs have traditionally been met through clinician consultations, medical textbooks, and institutional databases.
However, more complex knowledge gaps have emerged at the intersections of these categories, potentially driven by the emergence of dynamic technologies such as Gen AI. For example, Gen AI prompts, such as ‘Please identify the most likely sources of my patient’s fever or leukocytosis’ (Subject-Domain), generating patient-specific discharge instructions (Subject-Institution), or ‘Please generate orders for central line infection prevention for patient Doe, Jane’ (Subject-Domain-Institution) require a more dynamic and context-specific approach. This is further underscored with the significant increase in Subject-Domain queries, which accounted for only 4/154 (2.6%) of categorized needs in 2003, but now represent 39/130 (30%) of clinician-submitted Gen AI prompts. The increasing potential of Gen AI to address context-aware, patient-specific decision support is likely a significant contributing factor in this finding. Conversely, Institution-based queries accounted for 35/154 (22.8%) information needs in 2003 and only 6/130 (4.6%) prompts in our data set, suggesting that existing institutional guidelines are sufficient for addressing institution-related questions, without the need for Gen AI.
Additionally, new domains of information needs have emerged, particularly in areas that extend beyond direct patient care. The introduction of Subject-Third Party Institution (e.g. insurance-related inquiries), Subject-Fairness (e.g. language accessibility and social determinants of health), and Institution-Subject Education (e.g. clinician communication resources) suggests that clinicians are now explicitly engaging with – and/or struggling with - a broader spectrum of knowledge that encompasses financial, administrative, and social factors affecting healthcare delivery.
The rise of health insurance complexities, particularly following the implementation of the Affordable Care Act in 2010, likely contributed to the growing need for information about patient financial responsibilities and insurance interactions16. The ability to ask these types of questions may reflect shifted expectations beyond traditional clinical care decisions, as digital tools and resources have made it possible to address complex concerns that were previously difficult to access or manage.
Another notable trend is the high volume of clinician queries related to workflow efficiency, particularly AI-driven administrative support. Among the 20% of Gen AI prompts that did not align with the original taxonomy, the majority (20/26) reflected a desire to reduce documentation burden, automate routine tasks, and improve EHR usability. Requests for automated daily assessment tools, optimized billing processes, and more efficient patient scheduling suggest that clinicians are actively seeking tools that enhance efficiency and reduce cognitive overload and recognizing that Gen AI tools may offer this ability in a way that was previously unavailable.
This shift may underscore the significant documentation burden placed on clinicians. Over the past two decades— except for a brief reduction during the COVID-19 pandemic in 2020—documentation requirements have steadily increased17. Clinicians must now record not only clinical details but also extensive administrative information, which does not necessarily enhance patient care9. Our findings indicate that clinicians are now seeking real-time, workflow-conscious solutions in addition to what are seen now as traditional, and when originally introduced were transformative, decision-support tools like EHR-integrated search functions, EHR info buttons, UpToDate, and PubMed.
These findings suggest that existing taxonomies of clinician information needs should be periodically reassessed to remain aligned with the realities of modern healthcare. While the Currie et al. (2003) framework remains relevant, new and evolving clinical queries highlight gaps in current information retrieval systems that are ripe for Gen AI to fill. Our identification of extended domains to the Currie et al. (2003) framework suggests that opportunity for future efforts to focus on designing knowledge systems that integrate multiple domains of information, allowing clinicians to seamlessly retrieve clinical, administrative, financial, and social determinants of health data in a single, intuitive interface.
The incorporation of Gen AI into clinical workflows offers this more adaptive and context-sensitive approach, capable of synthesizing multimodal data from patient history, medical literature, and institutional guidelines. Gen AI, when integrated into the EHR, has the potential to address these complex, multidimensional knowledge gaps dynamically by adapting to the specific patient, institution, and specialty in question. This represents a paradigm shift to comprehensively address clinicians’ information needs in a more sophisticated and holistic manner to improve efficiency and enhance patient-centered care. As we embark on implementation science research for Gen AI, we recommend the continued formal classification of evolving clinician information needs as one method, among many, to guide best practices for design, use, and evaluation of Gen AI clinical use cases. In particular, our findings highlight two promising directions: (1) user-centered strategies to ensure that Gen AI tools target high-impact workflow and documentation tasks; and (2) iterative, real-world usability testing within EHR environments to optimize fit, performance, and clinician satisfaction.
Despite AI’s potential, challenges related to bias, explainability, and clinician trust remain key barriers to adoption11,14. Our findings highlight emerging needs in domains like Subject-Fairness and Institution-Subject-Education, where clinicians seek more inclusive solutions. AI-generated recommendations can reflect biases in training data, raising concerns about fairness in patient care11,14. To address these challenges, future research should focus on developing transparent AI models with clinician oversight to ensure they align with best practices and support equitable, patient-centered care.
Limitations
Currie et al. (2003) framework classified clinician information needs based on observations in the clinical setting. Our data set used in this study was from a survey focused on clinician’s perspectives on the integration of Gen AI into clinical workflows. Therefore, the counts and distribution per each category are not expected to be similar across the two data sources and as such our primary goal was to identify extensions to the framework. However, we do report those counts and distribution as context given the majority of the information needs identified by clinicians continued to be consistent with the Currie et al. (2003) framework, even in a survey that was distributed 10 years later (2023-2024) and was focused on Gen AI.
We also acknowledge that, as reported in Ruan et al. (2024), survey participants had varying levels of familiarity with Generative AI which may have driven variability in the submitted prompts, particularly the appropriateness of the prompt for a Gen AI tool versus a traditional information retrieval tool.
Finally, while this study draws valuable insights from survey-based data, future observational studies embedded within clinical workflows could provide richer context on how information needs arise in real-time. Such studies would allow for the identification of latent or unarticulated needs, better capture interactions with existing decision-support tools, and validate whether Gen AI prompts align with actual clinician behavior at the point of care.
Conclusion
This study highlights the evolving information needs of clinicians and the potential of Generative AI (Gen AI) to address these gaps. While most Gen AI prompts align with the original Currie et al. (2003) taxonomy, 20% reflect emerging needs such as AI-driven workflow optimization, financial concerns, and fairness-related issues. These findings suggest that the taxonomy must be updated to capture the growing complexity of clinical practice. Gen AI offers an opportunity to provide context-specific, patient-centered decision support, bridging clinical, administrative, and social factors. By integrating these advanced tools, healthcare can evolve to meet the dynamic needs of clinicians, ultimately transforming the way care is delivered and enhancing both efficiency and patient outcomes.
Appendix A. Relevant Prompt(s) from Survey Instrument
Please suggest at least one prompt that you would ask a generative AI tool in order to assist you with a specific clinical task. Prompts could be in the form of a request or a question. Please enter each prompt on a new line.
Figures & Tables
Table 1.
Mapping of CUIMC Clinician Prompt Suggestions to Predefined and Manually Defined Taxonomies.
| Taxonomy | Count |
|---|---|
| Currie et al. 2003 | 104 (80%) |
| Extension of Existing Domains | 6 (4.6%) |
| Novel AI-Driven Domains | 20 (15.4%) |
| Total | 130 (100%) |
Table 2.
Comparison of Information Need Events Within Currie et al. (2003) Taxonomy.
| Currie et al. 2003 Observational Study | CUIMC 2024 Survey | |
|---|---|---|
| Currie et al. 2003 | ||
| Subject | 62 (40.3%) | 14 (10.8%) |
| Domain | 37 (24%) | 25 (19.2%) |
| Institution | 35 (22.8%) | 6 (4.6%) |
| Subject-Domain | 4 (2.6%) | 39 (30%) |
| Subject-Institution | 11 (7.1%) | 8 (6.2%) |
| Domain-Institution | 4 (2.6%) | 4 (3.1%) |
| Subject-Domain-Institution | 1 (0.6%) | 8 (6.2%) |
| Extension of Existing Domains | ||
| Subject-Third Party Institution | 0 (0%) | 3 (2.3%) |
| Subject-Fairness | 0 (0%) | 2 (1.5%) |
| Institution-Subject Education | 0 (0%) | 1 (0.7%) |
| Novel AI-Driven Domains | ||
| AI Education | 0 (0%) | 6 (4.6%) |
| Augmenting Clinical Workflows | 0 (0%) | 14 (10.8%) |
| Total | 154 (100%) | 130 (100%) |
Table 3.
Newly Identified Information Need Categories for Extension to Currie et al. (2003) Taxonomy.
| Role | Specialty | ||||||
|---|---|---|---|---|---|---|---|
| RN/APRN | Physician | Other | Primary | Surgery | Critical | Other | |
| Extension of Existing Domains | |||||||
| Subject-Third Party Institution (N = 3) | 1 (10%) | 2 (15.4%) | 0 (0%) | 1 (16.7%) | 0 (0%) | 0 (0%) | 2 (33.3%) |
| Subject-Fairness (N = 2) | 1 (10%) | 1 (7.7%) | 0 (0%) | 1 (16.7%) | 1 (11.1%) | 0 (0%) | 0 (0%) |
| Institution-Subject Education (N = 1) | 0 (0%) | 1 (7.7%) | 0 (0%) | 1 (16.7%) | 0 (0%) | 0 (0%) | 0 (0%) |
| Novel AI-Driven Domains | |||||||
| AI Education (N = 6) | 4 (40%) | 0 (0%) | 2 (66.7%) | 0 (0%) | 3 (33.3%) | 2 (40%) | 1 (16.7%) |
| Augmenting Clinical Workflows (N = 14) | 4 (40%) | 9 (69.2%) | 1 (33.3%) | 3 (50%) | 5 (55.5%) | 3 (60%) | 3 (50%) |
| Total (N = 26) | 10 (38.5%) | 13 (50%) | 3 (11.5%) | 6 (23.1%) | 9 (34.6%) | 5 (19.2%) | 6 (23.1%) |
References
- 1.Wyatt JC, Sullivan F. What is health information? Bmj. 2005 Sep 8;331(7516):566–8. doi: 10.1136/bmj.331.7516.566. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Ely JW, Osheroff JA, Chambliss ML, Ebell MH, Rosenbaum ME. Answering physicians’ clinical questions: obstacles and potential solutions. Journal of the American Medical Informatics Association. 2005 Mar 1;12(2):217–24. doi: 10.1197/jamia.M1608. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Del Fiol G, Workman TE, Gorman PN. Clinical questions raised by clinicians at the point of care: a systematic review. JAMA internal medicine. 2014 May 1;174(5):710–8. doi: 10.1001/jamainternmed.2014.368. [DOI] [PubMed] [Google Scholar]
- 4.Covell DG, Uman GC, Manning PR. Information needs in office practice: are they being met? Annals of internal medicine. 1985 Oct 1;103(4):596–9. doi: 10.7326/0003-4819-103-4-596. [DOI] [PubMed] [Google Scholar]
- 5.Currie LM, Graham M, Allen M, Bakken S, Patel V, Cimino JJ. Clinical information needs in context: an observational study of clinicians while using a clinical information system. In AMIA Annual symposium proceedings. 2003;2003:p. 190. [Google Scholar]
- 6.Timmins L, Kern LM, O’Malley AS, Urato C, Ghosh A, Rich E. Communication gaps persist between primary care and specialist physicians. The Annals of Family Medicine. 2022 Jul 1;20(4):343–7. doi: 10.1370/afm.2781. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Cohen DJ, Wyte-Lake T, Dorr DA, Gold R, Holden RJ, Koopman RJ, Colasurdo J, Warren N. Unmet information needs of clinical teams delivering care to complex patients and design strategies to address those needs. Journal of the American Medical Informatics Association. 2020 May;27(5):690–9. doi: 10.1093/jamia/ocaa010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Arndt BG, Beasley JW, Watkinson MD, Temte JL, Tuan WJ, Sinsky CA, Gilchrist VJ. Tethered to the EHR: primary care physician workload assessment using EHR event log data and time-motion observations. The Annals of Family Medicine. 2017 Sep 1;15(5):419–26. doi: 10.1370/afm.2121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Levy DR, Withall JB, Mishuris RG, Tiase V, Diamond C, Douthit B, Grabowska M, Lee RY, Moy AJ, Sengstack P, Adler-Milstein J. Defining Documentation Burden (DocBurden) and Excessive DocBurden for All Health Professionals: A Scoping Review. Applied Clinical Informatics. 2024 Oct;15(05):898–913. doi: 10.1055/a-2385-1654. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Singhal K, Azizi S, Tu T, Mahdavi SS, Wei J, Chung HW, Scales N, Tanwani A, Cole-Lewis H, Pfohl S, Payne P. Large language models encode clinical knowledge. arXiv preprint arXiv:2212.13138. 2022 Dec 26.
- 11.Wang D, Zhang S. Large language models in medical and healthcare fields: applications, advances, and challenges. Artificial Intelligence Review. 2024 Sep 20;57(11):299. [Google Scholar]
- 12.Reddy S, Shaikh S. The long road ahead: navigating obstacles and building bridges for clinical integration of artificial intelligence technologies. Journal of Medical Artificial Intelligence. 2024 Sep 10.
- 13.Karimian G, Petelos E, Evers SM. The ethical issues of the application of artificial intelligence in healthcare: a systematic scoping review. AI and Ethics. 2022 Nov;2(4):539–51. [Google Scholar]
- 14.Van Dis EA, Bollen J, Zuidema W, Van Rooij R, Bockting CL. ChatGPT: five priorities for research. Nature. 2023 Feb 9;614(7947):224–6. doi: 10.1038/d41586-023-00288-7. [DOI] [PubMed] [Google Scholar]
- 15.Ruan EL, Alkattan A, Elhadad N, Rossetti SC. Clinician Perceptions of Generative Artificial Intelligence Tools and Clinical Workflows: Potential Uses, Motivations for Adoption, and Sentiments on Impact. medRxiv. 2024 Jul 31:2024–07. [Google Scholar]
- 16.American Medical Informatics Association. 25x5 Recommendations to Reform Prior Authorization [Internet] American Medical Informatics Association [cited 2025 Mar 21] Available from: https://brand.amia.org/m/333808f24a85b327/original/2023-25x5-Prior-Authorization.pdf.
- 17.Moy AJ, Schwartz JM, Chen R, Sadri S, Lucas E, Cato KD, Rossetti SC. Measurement of clinical documentation burden among physicians and nurses using electronic health records: a scoping review. Journal of the American Medical Informatics Association. 2021 May 1;28(5):998–1008. doi: 10.1093/jamia/ocaa325. [DOI] [PMC free article] [PubMed] [Google Scholar]