Building the evidence-base to reduce electronic health record–related clinician burden

Abstract

Clinicians face competing pressures of being clinically productive while using imperfect electronic health record (EHR) systems and maximizing face-to-face time with patients. EHR use is increasingly associated with clinician burnout and underscores the need for interventions to improve clinicians’ experiences. With an aim of addressing this need, we share evidence-based informatics approaches, pragmatic next steps, and future research directions to improve 3 of the highest contributors to EHR burden: (1) documentation, (2) chart review, and (3) inbox tasks. These approaches leverage speech recognition technologies, natural language processing, artificial intelligence, and redesign of EHR workflow and user interfaces. We also offer a perspective on how EHR vendors, healthcare system leaders, and policymakers all play an integral role while sharing responsibility in helping make evidence-based sociotechnical solutions available and easy to use.

INTRODUCTION

More than half of clinicians and healthcare workers in the United States currently experience burnout symptoms.¹ A wide range of factors contribute to clinician burnout including excessive workload, inefficiencies in the work environment, increased regulatory burden, and reduced clinician autonomy.² A significant factor is the increased amount of time and effort clinicians spend using electronic health records (EHRs).³ A study in 4 different specialty practices found that 27% of a clinician’s time was spent in clinical face-to-face time while 49% was spent using the EHR and addressing other deskwork.¹

In order to meaningfully address EHR-related burden, we need to understand the areas of greatest burden. Arndt et al⁴ developed a taxonomy to help specify the percentages of time that clinicians spent on key EHR tasks. Table 1 recaps findings from this publication that identified the top 6 time-consuming EHR tasks.⁴ Based on this taxonomy, 3 main areas needing attention from a time perspective are documentation, chart review, and inbox tasks.

Table 1.

Where does the time go?

Daily total (%)	EHR task category	Task classification
23.7	Documentation	Clerical
16.9	Chart review—notes	Medical care
15.5	Refills and results management	Inbox
12.1	Order entry	Clerical
7.3	Chart review—medications	Medical care
5.6	My chart portal	Inbox

Source: Arndt et al.⁴

We need to better understand how to focus innovation to address EHR-related physician burden, building on what is already known.⁵ Within each of the 3 burden areas mentioned previously, we (1) describe the nature of the burden, (2) share promising evidence-based approaches and pragmatic strategies that are ready for implementation, and (3) suggest further research directions. Our perspective provides guidance on how EHR vendors, healthcare system leaders, and policymakers can contribute to these solutions and make evidence-based solutions widely available and easy to use.

DOCUMENTATION

Documentation is one of the most time-consuming EHR tasks for clinicians⁶ and contributes to clinician frustration and burnout.⁷ Many physicians are not satisfied with the time required to create electronic notes. Perception of time spent on documentation has risen over time: in 1988, 10% to 21% of physician time was spent on documentation; in 1997, case review and documentation was reported to consume 38% of time on call.⁸ In 2010 and 2012, studies showed internal medicine physicians spend between 40% and 49% of their time using a computer, and 70% of this time is spent in documentation and order entry.^9,10 In a survey of internal medicine physicians, more than two-thirds (67.9%) reported spending in excess of 4 hours per day on documentation.¹⁰

Many clinicians are unable to complete daily documentation requirements during clinic hours and use their personal time to finish.¹¹ Nurses and physicians spend large parts of each day entering notes, with less time spent in other activities.¹² Documentation in the United States is almost 4 times longer than in other countries using similar EHR software.¹³ Many existing documentation requirements were developed before the EHR era and have not been updated for many current health information technology systems.¹¹ Some key burden areas identified include excessive documentation (or “note bloat”), overstandardization of “one-size-fits-all” documentation tools found in EHRs, and poor usability and functionality documentation features that cause clinicians to spend more time on the computer.¹⁴ Efforts are underway to reduce unnecessary regulatory requirements, but informatics solutions can address remaining documentation burdens.^11,15

A promising approach

Speech recognition (SR) software that helps create text from voice commands can enable clinicians to spend less time on documentation and more time interacting with patients.¹⁶ Use of voice for EHR documentation is growing,¹⁷ and its acceptance depends greatly on how well it is integrated with the EHR and fits into workflow.¹⁸ One system called the Voice-Generated Enhanced Electronic Notes (VGEENS) used SR integrated with natural language processing (NLP) to generate inpatient progress notes, and to extract concepts within the text, in a way that fits rounding workflow.¹⁹ VGEENS has been integrated with commercial EHRs and with hospital physician workflow and provides an alternative for clinicians who prefer voice to typing to create notes.

Ambient virtual scribes that use devices such as Amazon Alexa or Google Assistant and combine artificial intelligence to parse dictated information and detect structured data are also gaining interest.²⁰ Some EHR vendors are either partnering with others or developing their own ambient virtual assistants to provide this option to clinicians.²⁰ Although SR usage in EHRs is becoming more common, research studies in this area remain heterogeneous and are limited to specific care settings. Most physicians who currently use voice to enter notes use commercial products, which do not always match their workflow.

EHR vendors could encourage development of documentation methods tailored to clinician workflow by supporting application programming interfaces similar to those used by transcription vendors for innovators to include notes into the record. Clinicians can invest time to explore alternative methods that are best aligned with their aptitudes. Policymakers should continue to provide incentives for physicians to document visits in a manner that benefits their patients without requiring non–value-added time.

Future research directions

Research on how SR technologies affect documentation quality, efficiency, cost, and user satisfaction over time across different clinical settings is needed to fully understand SR impact on reducing EHR burden. For example, in a randomized controlled trial comparing the use of VGEENS to typing notes, VGEENS was associated with longer length of time for progress notes to become available in the EHR compared with the usual method.¹⁹ The study team noted that this result was unexpected and hypothesized that clinicians did not dictate their notes at the bedside or immediately after leaving the bedside (as had been anticipated prior to study), contributing to the time increase. Additionally, some clinicians in the study were averse to creating notes using VGEENS, citing inexperience with dictation and reluctance to learn a new skill during their busy clinical training. Findings from this study show the potential of using a SR system to create notes that capture clinician thinking as close to rounds as possible. However, more research is needed to improve the usability, speed, and accuracy of the notes.

Also of interest is recent work on patient participation in note creation.^21,22 While some evidence indicates that patient participation may increase patient engagement and improved accuracy of notes,^23,24 more research is needed to determine how to incorporate patient-generated documentation efficiently into clinical workflow and whether doing so lessens clinician documentation burden.

CHART REVIEW

Chart review of an individual patient to understand their underlying history and diagnostic test results has always been a time-consuming task, even for proficient EHR users.²⁵ Clinicians are constantly “drowning” in data and have limited time to interact with patients during clinical visits while reviewing salient and relevant information in their charts. This is particularly true for clinical notes, in which a common behavior in documentation, copying and pasting from previous notes and test results, contributes to the significant amount of redundant information in medical charts. One study found 76% of redundancy in both inpatient and outpatient clinical notes.²⁶ Reviewing redundant information increases cognitive burden and causes increased risk of errors in diagnosis and treatment.^27,28

A promising approach

The use of NLP and machine learning (ML) provides a newer approach to reduce chart review burden by quickly identifying new or relevant information in the chart.²⁹ For example, Goss³⁰ developed a novel way to automatically compile clinically and contextually relevant information using NLP and ML. This tool helps automatically process, filter, and rank free-text information that clinicians need to know for a patient’s complaint (eg, pain) and presents this alongside with structured or coded information so that providers have a “snapshot” of relevant information. This study demonstrated that NLP and unsupervised ML can provide a reasonably accurate, low-effort, and scalable method for clinical relevancy ranking, thereby increasing the value of information available for providers’ medical decision making and improving the quality and safety of patient care.

The use of note visualization tools and graphical representations that better synthesize patient information and improve navigation to new, clinically important details can accompany the NLP and ML approach mentioned previously.²⁶ Some evidence indicates that charts may not be organized optimally to facilitate efficient clinician review of information. Typical patient chart progress notes are documented in SOAP (subjective-objective-assessment-plan) order. However, one study identified variations in chart reading and retrieval styles based on clinicians’ preferences and workflow demands.³¹ It appears that the flexibility in note organization creates additional burden for clinicians reviewing notes.^32,33 Most clinicians actually preferred to review the assessment and plan sections first.

Ultimately, our ability to improve the process of chart review particularly for clinical documentation requires a multipronged approach. From a policy perspective, tackling the volume and streamlining the requirements of clinical documentation will improve the process of chart review. In addition, adding EHR mechanisms for validating the review of a patient’s history should be established—as opposed to copy and paste of this information into notes—to encourage shorter, easier-to-read documentation. Similarly, EHR vendors should develop and implement functionality to standardize note presentation—particularly with the ordering of note sections, which appears to have value and will help decrease the cognitive burden of chart review.

Future research directions

Additional research should explore (1) how automated methods (eg, ML and NLP) can improve the information retrieval and note presentation process in charts, particularly clinical notes; and (2) how accompanying graphical representations can best facilitate rapid interpretation of patient data. Also, comparing different chart review styles with the traditional SOAP order warrants more investigation. Ideal EHR clinical note interfaces should align with clinicians’ mental models, task needs, and workflow demands.³¹ For example, an EHR interface designed with a holistic view of the patient and flexible navigation across different sections will improve the chart review process.¹⁴

INBOX TASKS

Handling inbox tasks is an emerging burden for clinicians.^34–36 A study that quantified clinician inbox notifications estimated that primary care clinicians spent nearly 67 minutes per day processing notifications, adding substantial burden to their workday.³⁷ Poor EHR inbox design, content, and workflows can lead to inefficiencies and safety issues (eg, clinicians missing important information due to notification fatigue or information overload).^38,39 A clinician can receive several types of notifications, including test results, messages from other clinicians and patients, and refill requests. It is often not easy to reduce the types and quantities of information, because clinicians find several of these messages important for clinical care. Clinicians may not also agree on the types of messages they should not receive and which types of messages can be routed to other team members. These issues make inbox burden a challenging problem to solve. Studies suggest that all improvement approaches should be sociotechnical.⁴⁰ Also, because most EHRs are not optimally designed to support team-based care, we need a greater understanding of the factors that will best enhance the usability of EHRs by teams—particularly as applied to handling inbox tasks.

Promising approaches

Team-based staffing models and better EHR inbox designs might assist with inbox burden.^41,42 One team-based care model reduced burnout rates from 53% to 13% within 6 months after launch.⁴³ There is also evidence to support other types of care redesign interventions as potential solutions to reduce inbox burden, including expanding roles and responsibilities for nurses and medical assistants under physician-written standing orders.⁴⁴

Approaches garnering preliminary success for better inbox design include involving practicing clinicians in inbox redesign and utilizing quality improvement methodologies. In one mixed-methods study, feedback from clinicians was used to inform several strategies to improve content and design of the EHR inbox as well as related clinical workflows and organizational policies.⁴⁵ In another example to reduce inbox-related inefficiencies, one health system applied Pareto diagrams, value stream mapping, and root cause analysis to prioritize administrative burden inefficiencies to develop a new, standardized inbox.⁴⁶ The resulting inbox included clearly defined message categories to reduce ambiguity of the definitions and streamline user understanding and implementation. Initial assessments of this standardized inbox system showed substantial reduction of inbox entry defect and time spent on inbox review.⁴⁶ Findings from this implementation also showed increased clinician efficiency and satisfaction, improved work-life balance, and decreased burden due to fewer administrative tasks.

EHR vendors are in a unique position to address EHR inbox designs and related workflows associated with burnout. Clinicians also need dedicated time for managing the inbox activities and related tasks. Current ratios of clinical practice to administrative time may need to be revisited by policymakers. Additionally, several strategies to address inbox burden need involvement by both EHR vendor and the healthcare system implementing the EHR, invoking a shared responsibility to implement them. Policymakers need to facilitate this shared responsibility.

Future research directions

Research is needed in additional settings to determine the generalizability of the approaches described previously. Multidisciplinary research that involves informatics, human factors, and cognitive scientists should study the impact of important physician-identified factors⁴⁷ such as reducing message processing complexity, simplifying interface design, providing features to reduce physician cognitive load, facilitating care team communication, and streamlining inbox message content. Ultimately, researchers, clinicians, local administrators, health information technology personnel, and EHR developers will need to work together to translate some of the emerging findings and improve the inbox experience.

CONCLUSION

In summary, many opportunities exist to alleviate EHR-related clinician burden through informatics solutions. For these solutions to be widely available, we need not only additional research to strengthen the evidence of improvement interventions, but also involvement of EHR vendors and the broader informatics community in development and implementation efforts. Further, incorporating effective design capabilities as part of certified health technology could help facilitate improvements and signal the importance of these types of solutions for clinicians.

EHR-related burden occurs in a complex sociotechnical system comprising people, processes with accompanying workflows, organizational policies and procedures, and external regulations in addition to the technology. Multiple organizational strategies and leadership engagement would be needed to reduce EHR-related burden.⁷ While organizational investments in EHR training will result in improved user satisfaction and efficiency,⁴⁸ non–value-added functionality should also be eliminated by each health system working closely with its vendor. One organization, for instance, used a crowdsourcing approach as a means to optimize their EHR implementation. To do this, they engaged frontline members of the care team and asked them to identify aspects of the EHR that were non–value-added.⁴⁹ Furthermore, upcoming regulations through the Centers of Medicare and Medicaid Services around “Patients over Paperwork” hold promise to foundationally change documentation and billing requirements including streamlining traditional requirements such as the review of systems.⁵⁰

Our perspective also highlights the need for sharing emerging best practices across both EHR vendors and health systems. EHR vendors are in a unique position to help share best practices because they are already working with different health systems. Thus, they can serve as hubs for collecting, implementing, and diffusing best practices for all 3 areas discussed herein. We thus recommend the development of regional or national consortia led by EHR vendors that can support such collaborative sharing and implementation of best practices to reduce burnout.⁴⁷ Policymakers should support the development of these consortia.

In conclusion, EHR-related burnout requires sociotechnical approaches in which various actors in our healthcare system play a shared role in helping organizations make evidence-based tools and applications available and easy to use. Because the responsibility is shared, actions are harder to operationalize in the absence of specific policy direction in this area. We call on policymakers to help facilitate the implementation of evidence-based solutions to relieve EHR-related clinician burden.

FUNDING

HS is funded in part by the Houston VA Health Services Research and Development Center for Innovations in Quality, Effectiveness and Safety (CIN13-413), the VA Health Services Research and Development Service (CRE17-127 and the Presidential Early Career Award for Scientists and Engineers USA 14-274), the VA National Center for Patient Safety, and the Agency for Healthcare Research and Quality (R01HS27363). Several studies described in this article were supported by Agency for Healthcare Research and Quality Grant Nos. R01 HS022085, R21 HS023602, R21 HS023631, R21 HS024541, K08 HS022901, and R01 HS27363. The findings and conclusions in this manuscript are those of the authors and do not necessarily represent the official position of the Agency for Healthcare Research and Quality or Department of Veterans Affairs.

AUTHOR CONTRIBUTIONS

All authors contributed substantially to this work, contributed to the revision, gave approval for the final version of the manuscript to be published, and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

CONFLICT OF INTEREST STATEMENT

BK worked on this manuscript while he was employed at the Agency for Healthcare Research and Quality. The views expressed are his own and do not necessarily represent the views of the National Institutes of Health, the Department of Health and Human Services, or the United States Government.