Nurse workarounds in the electronic health record: An integrative review

Abstract

Objective

The study sought to synthesize published literature on direct care nurses’ use of workarounds related to the electronic health record.

Materials and Methods

We conducted an integrative review of qualitative and quantitative peer-reviewed research through a structured search of Academic Search Complete, EBSCO Cumulative Index of Nursing and Allied Health Literature (CINAHL), Embase, Engineering Village, Ovid Medline, Scopus, and Web of Science. We systematically applied exclusion rules at the title, abstract, and full article stages and extracted and synthesized their research methods, workaround classifications, and probable causes from articles meeting inclusion criteria.

Results

Our search yielded 5221 results. After removing duplicates and applying rules, 33 results met inclusion criteria. A total of 22 articles used qualitative approaches, 10 used mixed methods, and 1 used quantitative methods. While researchers may classify workarounds differently, they generally fit 1 of 3 broad categories: omission of process steps, steps performed out of sequence, and unauthorized process steps. Each study identified probable causes, which included technology, task, organizational, patient, environmental, and usability factors.

Conclusions

Extensive study of nurse workarounds in acute settings highlights the gap in ambulatory care research. Despite decades of electronic health record development, poor usability remains a key concern for nurses and other members of care team. The widespread use of workarounds by the largest group of healthcare providers subverts quality health care at every level of the healthcare system. Research is needed to explore the gaps in our understanding of and identify strategies to reduce workaround behaviors.

INTRODUCTION

Background and significance

Healthcare errors are the third leading cause of death in the United States, contributing to an estimated 150 000-440 000 mortalities annually.^1,2 Even more pervasive are errors that do not result in mortality, with estimates that 1 in 7 Medicare beneficiaries is subject to an adverse healthcare event.³ In addition to the immense toll of human suffering, economic consequences due to healthcare error are estimated at $17 billion annually.⁴

In order to stimulate proliferation of electronic health record (EHR) systems, reduce error, and enhance care coordination, legislation enacted in 2009 created strong incentives to healthcare organizations to implement EHRs with specified features to enhance safety and quality of care.^5–7 These features include clinical decision support, computerized provider order entry, and drug or allergy interaction checking.^8,9

Despite the widespread implementation of these safety enhanced EHRs, the United States has not fully reduced the scope of healthcare errors.¹⁰ In fact, the implementation of EHRs can lead to new safety problems including: duplicate medication orders,¹¹ errors in dosing,¹² and unexpected order deletion.¹³ Additionally, communication problems¹⁴ and gaps in care coordination remain.¹⁵ Health information technology,¹⁶ EHR medication safety concerns,¹⁷and poor system usability¹⁸ remain, despite years of development.

Compounding these new safety problems is an increase in clinician’s workload related to clinical documentation that is causing widespread dissatisfaction and burnout.¹⁹ Carayon et al²⁰ found that intensive care unit residents and attending physicians (n = 53) spent more time on clinical review and documentation (increased 40% [P < .001] and 55% [P  < .07], respectively) 3-6 months after EHR implementation. Finally, despite many years of use and the goal of improving safety, EHRs continue to provide insufficient cognitive support to clinicians.²¹

In order to cope with increasing demands clinicians find shortcuts, or “workarounds,” to improve job performance.²² Debono et al²³ defined workarounds as “observed or described behaviors that may differ from organizationally prescribed or intended procedures in which workers ‘circumvent’ or temporarily ‘fix’ an evident or perceived workflow hindrance in order to meet a goal or to achieve it more readily.” Workarounds exist in many other complex sociotechnical work settings such as accounting,²⁴ aerospace,²⁵ chemistry,²⁶ and manufacturing.²⁷ Workarounds can be seen in a positive light,²⁵ and they can be used to innovate and problem solve.²⁸

Workarounds are problematic in health care, even if done with the best intentions,^14,29,30 because the adaptations interfere with processes designed to ensure safety.^31,32 Workarounds can also be the result of collaborative efforts among staff³³ and passed on to junior staff.¹⁴ Workarounds also lead to unintended consequences^31,34 that can create documentation gaps in the EHR,³⁵ impair communication,^14,36,37 and harm patients.

EHR systems can also increase the documentation burden for nurses.³⁸ Healthcare professionals, including nurses, use informal techniques to circumvent the EHR to provide care.²² Workarounds performed by nurses are especially concerning as they are the largest group of health care professionals in the United States,³⁹ and together with midwives comprise nearly 20.7 million individuals, or 50% of the world’s healthcare workforce.⁴⁰ Further, for hospitalized patients, nurses spend more time interacting with patients than other clinicians. Additionally, nurses are often the last line of defense in intercepting and preventing healthcare error⁴¹ before patient harm occurs.

Previous reviews recognize the importance of understanding and preventing nursing workarounds.²³ Halbesleben et al’s⁴² review of workarounds in healthcare settings by all clinician types determined that workarounds are poorly measured, influence outcomes due to system impacts, and remain underresearched. Debono et al’s²³ integrative review of workarounds to the EHR for direct and indirect care concluded that workarounds both enable and comprise care, are done cooperatively and individually, and are influenced by organizational and cultural norms.²³

We extend their work with an additional 7 years of research and focus exclusively on nurse workarounds to EHR as part of direct care including medication administration, and we used Koppel et al’s²² categorization and probable causes for bar code medication administration (BCMA) workarounds and extend their classification to the complete EHR.

Objectives

The objective of this review was to synthesize the state of science of nurse workarounds to the EHR in direct care activities. For the purposes of this review, we report study methods, classification of workaround behaviors and probable causes.

MATERIALS AND METHODS

We applied Whittemore and Knafl’s⁴³ integrative literature review method to research nurses’ EHR workarounds, as this method allows for inclusion of quantitative and qualitative research.

Search strategy

In consultation with an academic research librarian to ensure an appropriate search strategy and database specific terminologies, we searched Academic Search Elite, Cumulative Index of Nursing and Allied Health Literature (CINAHL), Embase, Engineering Village, Ovid Medline, Scopus, and Web of Science with specified search terms in December 2019.

Study selection and exclusion criteria

We applied a 3-stage systematic process for article exclusion by: titles, abstracts, and full articles using PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (Figure 1),⁴⁴ and established interrater reliability between 2 authors of ≥85% at each stage.⁴⁵ We assessed the percent of interrater agreement in each of the 3 stages. Beginning with article titles, D.F. and K.D.L. scored 10%-15% of the articles in rounds using the exclusion rules independently and entered the scores on an Excel® spreadsheet. Each round included independent scoring, comparison of differences, and clarification of rules as needed. Rounds for title exclusion continued until ≥85% agreement was reached. This process was repeated using the abstract exclusion rules between D.F. and K.D.L. and concluded using full article exclusion rules between D.F. and J.M.

Figure 1.

Records screened and included.

Information extraction and classification

We developed a data dictionary for all data extracted. We began with the EHR component, methods, theoretical underpinnings, setting, sample and the authoring team. Categorical data (eg, BCMA, computerized provider order entry [CPOE], full EHR) were entered in an Excel® spreadsheet. Similar to the methods used for interrater agreement for study selection, categorical data were extracted between 2 authors (D.F. and K.D.L.) independently for each grouping, responses were compared, and clarification of rules and review continued until ≥85% agreement was reached for each column in Table 1.⁴⁶

Table 1.

Comprehensive list of studies, setting, workarounds, and participants

First Author	MethodCountry	Hours	Summarized	Method	Workarounds	Setting	Sites	Sample Size
Andersen (2009)44	Qualitative	80	Investigated relationship between clinician, role, device selection and clinical care	1, 2, 4, 7 (assessment of hardware)	Nurses wrote paper notes to track information	Hospital	2	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 35; 27; 8; 0
	Australia
Baysari (2018)72	Qualitative	NA	Identified views, perceptions, and changes in behaviors as CPOE system became routine	2	Nurses did not take computers to the bedside, used paper notes	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 114; 83; 31; 0
	Australia
Blaz (2016)48	Qualitative	202	Described nurses’ use of paper as a tool for care with EHR	1, 2, 6 (paper RN notes)	Handwritten notes that were later transcribed, nurses never directly entered the vitals into the EHR	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 13; 13; 0; 0
	United States
Blijleven (2019)56	Qualitative	NR	Described context in which workarounds are created	1, 2	Nurses used a previous database for Hemophilia patients instead of newly implemented EHR	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others (clerks) = 47; 13; 31; 3
	The Netherlands
Bramble (2013)49	Qualitative-	NA	Identified improvements and challenges following EHR implementation in a rural healthcare clinic	2	Nurses printed out and carried documents to providers to assist in the electronic prescribing	Clinic	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 16; 12; 4; 0
	United States
Bristol (2018)61	Qualitative	NA	Analyzed nurses’ perceptions of unintended consequences of her	3	Chart on paper, unstructured data, entered less descriptive data in the EHR than requested	Hospital	NR	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 144; 144; 0; 0
	United States
Carrington (2011)50	Mixed methods	NA	Examined nurse perception of EHR documentation	2	Save without signature, technical solutions, documentation shortcuts	Hospital	2	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 37; 37; 0; 0
	United States
Chao (2016)21	Mixed methods	90	Analyzed collaborative work routines after implementation of a perinatal EHR	1, 2, 3, 5, 6 (clinical forms, patient charts, training materials)	Use of paper as a supplement for shift report or managing tasks, data entered in free text	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 53; 53; NR; NR
	United States
Cresswell (2012)51	Qualitative	38.5	Identified impact of EHR implementation and staff responses to the system	1, 2, 6(field notes, hospital project documents)	Using less descriptive data to enhance patient flow, entering information in other electronic systems which was then used to transcribe information to the EHR, use of paper notes, delayed data entry	Hospital/clinic	3	Total Participants Nurses; Providers (MD, APRN); Others = 87 (doctors, nurses, pharmacists, social workers, technology staff, therapists, training staff, social workers, ward clerks, specific numbers not reported); NR; NR; NR
	England
Early (2011)64	Quantitative	NA	Reviewed medication override data after BCMA implementation	7 (medication override data review)	Medication bar-codes were not scanned prior to administration	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = NR; 0; 0; 0
	United States
Gaudet (2016)52	Qualitative		Described culture of caring for patient nurse interactions and communication with her	1, 2, 7 (audio recording of nurse patient interactions)	Use of paper as a cognitive tool	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 14; 14; 0; 0
	United States
Hardmeier (2014)65	Mixed methods	NR	Measured BCMA errors and types of workarounds after a new system wasimplemented		Failure to visually confirm patient’s identification, failure to compare medication to the EMAR at least twice before administration, charting medication before administration	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = NR; 0; 0; 0
	United States
Holden (2013)31	Qualitative	136.5	Analyzed nurses’ response to a problem and associated workaround	1, 2, 6 (policies, paper MAR)	Paper to track medication schedules, administered medication without scanning the patient, or the medication barcode, scanning medication barcodes not connected to patient, documenting medication prior to administration, data entered in free text	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 141; 141; 0; 0
	United States
Huang (2016)70	Qualitative	NR	Observed nurses’ medication administration process with new health information technology system	1, 2	administered medication without verifying patient, scanning barcodes not connected to pt	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 8; 8; 0; 0
	United States
Koppel (2008)22	Mixed methods	NR	Measured causes and outcomes of BCMA workarounds	1, 2, 7 (failure modes and effects analysis/medication override data)	15 workaround behaviors within 3 broad categories: omission of process steps, steps performed out of sequence, and unauthorized process steps	Hospital	5	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 42; 36; 0; 6 (2 information technology directors, 4 pharmacists)
	United States
Miller (2011)66	Mixed methods	6	Observed nursing workflow and pharmacist workflow reviewed to BCMA alert overrides	1, 7 (medication override reports)	RN did not scan patient armband, RN did not scan medication, scanned medication outside patient room, RN scanned package after medication removed, scanned medications multiple times to reach cumulative dose, documented medications prior to administration, scanned ID band not connected to patient	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others were RNs and pharmacists; no specific sample size reported; NR; NR; NR
	United States
Mount-Campbell (2019)59	Mixed methods	156	Evaluated nurses’ cognitive artifact through analysis of paper notes	1, 6 (paper RN notes)	Use of paper notes, delayed documentation	Hospital	2	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 20; 20; 0; 0
	United States
Niazkhani (2011)36	Qualitative	NA	Identified complications following CPOE implementation and workflow changes	2, 6 (educational materials)	RNs wrote paper orders to assist MD, administered drugs before orders available	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 21; 6; 12; 3 (2 pharmacists,1 pharmacy technician)
	The Netherlands
Ostensen (2019)60	Qualitative	124	Described nursing practice and care coordination with an EHR in community settings	1, 2	Save without signature, use of paper notes, use of personal mobile device	Long-term care, home health	3	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 17; 17; 0; 0
	Norway
Park (2015)32	Qualitative	230	Described EHR adaptation by clinicians and unintended consequence	1, 2	RNs entered less descriptive data than requested	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 52; 23; 29; 0
	United States
Patterson (2006)68	Qualitative	79	Classified BCMA workarounds in acute and long-term care	1	RNs did not scan patient wristband, RN scans wristbands not connected to patients	Hospital/long-term care	3	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 14; 14; 0; 0
	United States
Rack (2012)67	Mixed Methods	NA	Identified BCMA workarounds and associated medication errors	3, 7 (error review)	RNs did not scan patient ID band, did not scan medication barcode, scanned medications after administered, scanned ID band not connected to the patient	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 220; 220; 0; 0
	United States
Rangachari (2019)58	Mixed Methods	NA	Explored issues related to EHR medication reconciliation	2, 3	RNs entered less descriptive data than requested	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 142; 29; 75; 38 (pharmacists)
	United States
Rathert (2019)57	Qualitative	NA	Examined frontline EHR user experiences in care coordination	2	Delayed data entry	Hospital	2	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 30; 15; 15; 0
	United States
Saleem (2011)35	Qualitative	NA	Identified paper tools and workarounds used to compensate for EHR in clinic settings	1, 2, 7 (EHR change requests)	Use of a paper calendar to track clinic, paper lists to manage work, entering order on behalf of MD, use of external software (Microsoft Excel) to manage requests	Clinic	12	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 16; 3; 9; 4 (2 administrators, 2 MA’s)
	United States
Schoville (2009)55	Qualitative	NR	Examined transition from CPOE design errors and care coordination	1, 2, 6 (website review), 7 (email review)	RN discontinued orders instead of MD’s, paper as a cognitive tool, RN administered medication before order	Hospital	2	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 12; 12; 0; 0
	United States
Stevenson (2018)62	Qualitative	62	Examined vital sign documentation in the her	1, 2	Use of paper as a cognitive tool- Post it™ notes, scraps of paper, notebooks, delayed data entry	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 13; 10; 3; 0
	Sweden
Van Der Sijs (2011)71	Qualitative	NR	Studied hospital workarounds after implementation of CPOE	1, 2	Nurses rescheduled medication doses on a paper MAR	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 9; 4; 4; 1 (pharmacist)
	The Netherlands
Van Onzenoort (2008)69	Mixed methods	NA	Identified workarounds to barcode verification by nurses	2, 7 (BCMA data)	RN administered medication without scanning barcode, gave medication before order	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = NR; 0; 0; 0
	The Netherlands
Varpio (2009)14	Qualitative	80	Described communication among nurses and physicians around the her	1, 2	Use of free text documentation	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 76; 62; 14; 0
Vogelsmeier (2008)54	Qualitative	NR	Described workarounds that occurred with an EHR implementation and medication safety impacts	1, 2, 6 (field notes)	Staff called in orders to circumvent faxing, did not check medications prior to administration, documented medications before administration, asked others for information, entered less descriptive data than requested	Long Term Care	5	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 88; 0; 0; 0
	United States
Yeung (2011)53	Qualitative	44.5	Described vital sign documentation and collection to better understand workflow	1	Nurses used paper notes to record vital signs that were later transcribed into the EHR	Hospital	3	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 24; 24; 0; 0
	United States
Watson (2014)63	Mixed methods	10	Described EHR early warning score system and patient assessment process	1, 2, 7 (retrospective data review)	Vitals on paper notes, batching of tasks	Hospital	1	Total Participants; Nurses; Providers (MD, APRN, PA); Others = 15; 15; 0; 0
	United States

Method of study: 1 = observation, 2 = interview or focus group, 3 = survey, 4 = heuristic, 5 = meeting attendance, 6 = artifact Collection (artifact), 7 = other (details).

APRN: advanced practice registered nurse; BCMA: bar code medication administration; CPOE: computerized provider order entry; EHR: electronic health record; EMAR: electronic Medication Administration Record; MAR: Medication Administration Record; NA: not applicable; NR: not reported; PA: physician assistant; RN: registered nurse.

Table 2.

Quality scoring of included studies

First author	Study Type	Sampling	Detail	Analysis	Score
Rangachari (2019)58	5. Mixed methods	3. Random or 100%	1. Methods and tools	3. Inferential	12
Koppel (2008)22	5. Mixed methods	3. Random or 100%	1. Methods and tools	2. Descriptive	11
Miller (2011)66	5. Mixed methods	3. Random or 100%	1. Methods and tools	2. Descriptive	11
Van Onzenoort (2008)69	5. Mixed methods	2. Purposive or case matching	1. Methods and tools	3. Inferential	11
Carrington (2011)50	5. Mixed methods	2. Purposive or case matching	1. Methods and tools	2. Descriptive	10
Watson (2014)63	5. Mixed methods	2. Purposive or case matching	1. Methods and tools	2. Descriptive	10
Andersen (2009)44	5. Mixed methods	2. Purposive or case matching	1. Methods and tools	2. Descriptive	10
Mount-Campbell (2019)59	5. Mixed methods	2. Purposive or case matching	1. Methods and tools	2. Descriptive	10
Early (2011)64	4. Quantitative	2. Purposive or case matching	1. Methods and tools	2. Descriptive	9
Rack (2012)67	5. Mixed methods	2. Purposive or case matching	1. Methods and tools	1. Narrative	9
Bramble (2013)49	3. Qualitative	3. Random or 100%	1. Methods and tools	1. Narrative	8
Blaz (2016)48	3. Qualitative	2. Purposive or case matching	1. Methods and tools	2. Descriptive	8
Hardmeier (2014)65	5. Mixed methods	0. Not explained	1. Methods and tools	2. Descriptive	8
Chao (2016)21	5. Mixed methods	0. Not explained	1. Methods and tools	2. Descriptive	8
Blijleven (2019)56	3. Qualitative	2. Purposive or case matching	1. Methods and tools	1. Narrative	7
Cresswell (2012)51	3. Qualitative	2. Purposive or case matching	1. Methods and tools	1. Narrative	7
Patterson (2006)68	3. Qualitative	2. Purposive or case matching	1. Methods and tools	1. Narrative	7
Rathert (2019)57	3. Qualitative	1. Convenience	1. Methods and tools	2. Descriptive	7
Schoville (2009)55	3. Qualitative	1. Convenience	1. Methods and tools	2. Descriptive	7
Varpio (2009)14	3. Qualitative	2. Purposive or case matching	1. Methods and tools	1. Narrative	7
Baysari (2018)72	3. Qualitative	1. Convenience	1. Methods and tools	1. Narrative	6
Bristol (2018)61	3. Qualitative	0. Not explained	1. Methods and tools	2. Descriptive	6
Gaudet (2016)52	3. Qualitative	1. Convenience	1. Methods and tools	1. Narrative	6
Huang (2016)70	3. Qualitative	1. Convenience	1. Methods and tools	1. Narrative	6
Niazkhani (2011)36	3. Qualitative	1. Convenience	1. Methods and tools	1. Narrative	6
Ostensen (2019)60	3. Qualitative	1. Convenience	1. Methods and tools	1. Narrative	6
Saleem (2011)35	3. Qualitative	1. Convenience	1. Methods and tools	1. Narrative	6
Stevenson (2018)62	3. Qualitative	1. Convenience	1. Methods and tools	1. Narrative	6
Van Der Sijs (2011)71	3. Qualitative	1. Convenience	1. Methods and tools	1. Narrative	6
Yeung (2011)53	3. Qualitative	0. Not explained	1. Methods and tools	2. Descriptive	6
Holden (2013)31	3. Qualitative	0. Not explained	1. Methods and tools	1. Narrative	5
Park (2015)32	3. Qualitative	0. Not explained	1. Methods and tools	1. Narrative	5
Vogelsmeier (2008)54	3. Qualitative	0. Not explained	1. Methods and tools	1. Narrative	5

Study design scores: 3 = qualitative design, 4 = quantitative design; 5 = mixed qualitative and quantitative descriptive. Sampling: 0 = not explained; 1 = convenience; 2 = purposive or case matching/cohort; 3 = random or 100%. Method detail: 1 = methods and tools; 0 = not explained. Analysis (highest level reported): 1 = narrative; 2 = descriptive statistics; 3 = inferential statistics.

Specific workaround behaviors were extracted using each of the included studies author’s narrative description and then grouped with similar meaning into Koppel et al’s²² 3 broad categories of workarounds: omission of process steps, steps performed out of sequence, and unauthorized process steps. All workaround behaviors were then categorized into one of the probable causes identified by Koppel et al (see Figure 2) or were assigned a new probable cause inductively.

Figure 2.

Koppel et al²² framework used in the present review.

RESULTS

Our search yielded 5221 articles, and after removing duplicates (2477) and applying systematic exclusion rules, a total of 33 studies were included: 21 focused on the full EHR,^{14,21,32,35,47–63} 9 on BCMA,^{22,31,64–70} and 3 on the CPOE.^36,71,72 We synthesized study methods, workarounds identified and probable causes. Included studies’ details are presented in Table 1.

Quality appraisal

A quality review tool suitable for qualitative and quantitative studies was used to evaluate research study design with 4 criteria: study type, sampling methodology, data collection, and analysis (Table 2).⁴⁶ Like the exclusion process, articles were independently scored on a spreadsheet between D.F. and K.D.L. for several rounds until we also established ≥85% interrater reliability.⁴⁵

Study methods

A majority of studies (n = 22) used qualitative methods,^{14,31,32,35,36,47–49,51–57,60–62,68,70–72} 10 used mixed methods,^{21,22,50,58,59,63,65–67,69} and 1 used quantitative methods.⁶⁴ The majority of the qualitative studies (n = 16) used 2 or more qualitative techniques to collect and analyze workarounds.^{14,31,32,35,36,48,51,52,54,55,60,62,63,67,70,71} Sixteen studies used observation,^{14,31,32,35,48,51,53–55,60,62,63,66,68,70,71} 17 used interviews,^{14,31,32,35,36,48,51,52,54–56,60,62,63,71,72} and 2 used focus groups.^49,70 Four studies featured artifact analysis,^31,36,48,51 including paper notes or sheets nurses carry to organize patient assignment and tasks,⁴⁸ project initiative or lessons learned documents,⁵¹ paper medication administration record used before CPOE,^31,36 nursing policies,³¹ and educational material.³⁶

Among 10 mixed methods studies, the quantitative component of research included survey,^21,58,67 medication administration data,^{22,63,65,66,69} incident reports,⁶⁷ retrospective statistics,⁶⁵ descriptive statistics of interview content themes,⁵⁰ descriptive statistics of paper nurses’ cognitive tool,⁵⁹ and time spent on admission documentation with EHR.⁵² The qualitative components included interview,^{21,22,47,50,52,58,69} observation,^{22,47,52,59,65} focus group,⁶⁷ analysis of nurse-patient interaction recordings, and artifact review (clinical forms, paper chart documentation, and EHR training materials).²¹

Study setting

Most studies (n = 25) were conducted exclusively in hospital inpatient units.^{14,21,22,31,32,36,47,48,50,52,53,55,56,59,61–67,69–72} Two were in outpatient clinics, and^35,49 4 were in mixed settings: 1 in a hospital with both acute and long-term care units,⁶⁸ 2 in a hospital and outpatient clinic setting,^51,57 and 1 with nurses in both long-term care and a home health setting.⁶⁰ One was conducted in long-term care.⁵⁴ The majority of studies (n = 21) were conducted in the United States,^{21,22,31,32,35,48–50,52,54,55,57–59,63–68,70} with 4 in the Netherlands,^36,56,69,71 3 in Canada,^14,53,55 2 in Australia,^47,72 1 in England,⁵¹ 1 in Sweden,⁶² and 1 in Norway.⁶⁰

Conceptual and theoretical frameworks

The majority of the articles (n = 22) did not report a theoretical basis.^{21,22,35,36,47–49,53–55,57–59,62–66,68–71} The remaining 11 each used different theories or frameworks: information theory,⁵⁰ actor network theory,⁵¹ theory of dynamic nurse-patient relationships,⁵² cognitive systems engineering,³¹ technological and organizational adaptation process model,³² complexity theory,⁶⁷ constructivist grounded theory,¹⁴ extended technology acceptance model,⁷² social constructionist,⁶⁰ sociotechnical framework,⁵⁶ and resilience engineering.⁶¹

Study subjects

Although the aim of this integrative review is to study workarounds used by nurses to the EHR, a number of researchers also included other members of the care team. Half of the studies (n = 18) focused solely on nurses,^{21,31,48,50,52–55,59–61,63–65,67–70} and the remaining articles (n = 15) included non-nurse participants.^{14,22,32,36,47,49,51,56–58,62,66,72}

Workaround classifications

Of the 33 articles, 10 studies^{14,22,31,36,55,56,67–70} explicitly classified workarounds by categories (Table 3). Two^66,67 applied the same broad categories initially proposed by Koppel et al’s²² omission of process steps, steps performed out of sequence, and unauthorized steps. Four articles^14,31,69,70 classified workarounds using different terms with the same or highly similar meaning to the Koppel et al’s original classification, and 4 used unique categories.^36,55,56,68

Table 3.

Workaround categorization

Categorization approach	First author	Categories
Reference category	Koppel (2008)22	Broad Workaround Categories: Omission of process steps Steps performed out of sequence Unauthorized steps
		Probable Causes: Technology related Task related Organizational Patient related Environmental
Same categories as Koppel (2008)22	Miller (2011)66	Omitted Step Incorrect Sequence Unauthorized Step
	Rack (2012)67	Technology Related Task Related Patient Related Organizational Enviornmental
Similar categories to Koppel (2008)22	Holden (2013)31	BCMA creates new problems BCMA permits a new problem-solving behavior BCMA blocks a familiar problem-solving path
	Huang (2016)70	Barcode scanning tasks Computer interface or operating issues Unfamiliar procedures Omitted procedures Logistic issues Others
	Varpio (2009)14	Abandoning workarounds: staff avoided using the EHR because the system impeded their workflow Forcing: staff made the EHR accommodate their professional workflow Submitting: staff compromised their work patterns to adopt the pathways through the EHR
	Van Onzenoort (2008)69	Technological problems Time related
Unique categorization	Blijleven (2019)56	Avoidable: workaround that isn’t required to move forward Unavoidable: workaround that is necessary to move forward Cascading workarounds: creates additional workarounds Noncascading workarounds- do not create additional workarounds Incidental: Used to temporarily overcome constraints that are uncommon situations Routinized: Used to overcome constraints that occur frequently Unanticipated: created within unexpected constraints. Anticipated: used when upcoming workflow constraints are known beforehand
	Niazkhani (2011)36	Prescribing Communication of orders Dispensing Administration Monitoring
	Patterson (2006)68	Patient identification Medication administration
	Schoville (2009)55	Workflow timing of events Communication changes System problems Learning curve

EHR: electronic health record.

Workaround strategies and behaviors

The 33 articles identified 8 workaround strategies: (1) paper as a cognitive tool, (2) bypassing patient identification checks, (3) data entry strategies, (4) bypassing EHR medication safety measures, (5) workarounds to the ordering process, (6) assisting physician’s workflow, (7) bypassing information in the EHR, and (8) scanning violations (see Figure 3). These 8 strategies represent 36 specific workaround behaviors (see Figure 4). Each behavior is categorized subsequently into one of Koppel et al’s²² workaround categories.

Figure 3.

Registered nurse (RN) workaround strategies and behaviors. BCMA: bar code medication administration; EHR: electronic health record; EMAR: electronic Medication Administration Record.

Figure 4.

Workaround strategy frequencies. EHR: electronic health record.

Omission of process steps

Omission of process steps occurred in 13 studies.^{22,31,54,55,60,64–70,72} These strategies included bypassing EHR medication safety measures and bypassing patient identification checks. Bypassing EHR safety measures included the following behaviors: (1) medication barcodes were not scanned prior to medication administration,^{22,31,64,66,67,69} (2) medications were not compared with the electronic Medication Administration Record,^22,54,65,72 (3) medications were administered prior to reviewing relevant information,^22,54 (4) visual confirmation of the package was omitted,^22,54,55 and (5) medication double checks were ignored.^22,55,65 Bypassing of patient identification measures included: (1) patient wristbands were not scanned^{22,31,66–68} and (2) patient identification was not validated verbally by the nurse.^31,70 Nurses consulted other staff members for patient information in lieu of the EHR^22,54,60 and avoided checking the EHR for new orders.²²

Steps performed out of sequence

Steps were performed out of sequence in 13 studies.^22,31,36,5,^{1,53,54,55,57,62,63,65,66,70} In 8 studies,^{22,31,36,54,55,65,66,70} nurses documented in a different sequence than the prescribed safety-focused workflow for medication administration by documenting before administering them, while nurses “batched” task documentation (delayed documentation of several tasks into one time period).^{51,53,57,62,63}

Unauthorized steps

Unauthorized steps was the most common technique in this review: use of paper was identified in 17 studies,^{21,31,35,47,48,51–55,59–63,71,72} identification violations in 6,^{22,31,66–68,70} medication violations in 7,^{22,36,54,66,67,69,71} and workarounds conducted to assist other clinicians in 7,^{31,35,36,49,54,55,71} and 8 manipulated the EHR to accomplish a task.^{31,32,35,36,49,54,55,71}

Use of paper

The most common workaround to the EHR (n = 17) employed was the use of paper.^{21,31,35,47,48,51–55,59–63,71,72} Paper tools assisted nurses with tracking medications,^{31,47,54,71,72} planning patient care,^{48,51,52,55,59,61,63} shift change report,^21,59 long-term care resident information,^54,59 vital signs that were later transcribed into the EHR,^53,59,62,63 and use of calendar to manage clinic schedules.³⁵

Bypassing patient identification

Patient identification workarounds appeared in 6 studies.^{22,31,66–68,70} In these workarounds, nurses scanned patient ID barcodes that were attached to another object,^31,67,68,70 were attached to sheet of paper,^31,66 or were in the nurse’s pocket.²²

Medication violations

Unauthorized process steps during medication administration were identified in 6 studies.^{22,36,54,66,67,69} These include scanning violations (situations when medication packaging was scanned after it was administered and separated from the package),^22,66,67 while the correct process was to scan medications before opening the package. Nurses scanned one medication several times to reach the cumulative dose,²² administered a partial dose, scanned and documented the full dose of medication,²² and scanned medications for multiple patients at the same time.^22,66 Additional steps were that nurses administered medications without the computer screen in view.^22,72 Similarly, staff scanned medications outside of the patient room^22,66 instead of inside the room and in front of the patient.

Nurses bypassed EHR medication safety features when they administered medications before the order was available,^{22,36,54,66,69} documented medications before they were actually administered,^31,65,66 and disabled audio alarms on the scanner units.²² Nurses entered improper medication doses in order to facilitate BCMA,⁵⁴ entered multiple doses in the EHR,⁵⁴ and adjusted medication administration times on a paper Medication Administration Record.⁷¹

Conducting workarounds to assist physician workflow

Six studies^{35,36,49,54,55,71} identified the role nurses played to assist other clinicians, primarily physicians, to work around the EHR. Nurses printed out documents that required provider action^35,36,49 and performed ordering workarounds: prepared written orders for providers,^35,36,54 entered new orders to trigger follow-up actions via the EHR,^35,55,71 discontinued orders,⁵⁵ and called in medication orders to pharmacy.⁵⁴

Data entry strategies

Numerous data entry strategies were used by nurses: data were entered in free text or comment fields, which belong in structured fields,^{14,21,31,54,61} and nurses entered less descriptive information to expedite documentation and patient care.^{32,49,51,54,58,61} Nurses continued to use a legacy database system in lieu of or in addition to a newly implemented EHR⁵⁶; used additional software to track consultations³⁵; used outside software features to document, then copied and pasted text into the EHR from other systems, eg, word processing software, because spellcheck was not always available^35,51; and saved documentation without signing.^50,60 Additionally, nurses assisted their colleagues in the BCMA process by documenting due to other nurses’ discomfort in the system.³¹ In a community setting, nurses used a mobile phone to take photographs to track wound healing, and for medication reference information.⁶⁰

Probable causes

Every study identified and described the probable causes of workarounds. In the following, we categorize the identified causes using Koppel et al’s²² 5 categories of probable workaround causes while adding our own inductively created category of “usability” (see Figures 2 and 5; Table 4).

Figure 5.

Probable causes. BCMA: bar code medication administration; EHR: electronic health record; Pt: patient.

Table 4:

Top 10 probable causes of nurse workarounds to the electronic health record (EHR). BCMA: bar code medication administration.

Technological

Technology-related “probable causes” were identified in most articles (n = 16),^{22,31,32,36,51,54,55,61,63,64,66–70,72} with some studies identifying more than 1 probable cause. These causes include: (1) problems with the hardware or software and (2) perceptions by the clinician using the technology.

The most common technology-related cause, wireless infrastructure problems, appeared in 7 studies.^{22,36,54,55,64,66,70} Other problems included scanner malfunctions, battery failure, or computer freezing,^{22,31,32,47,55,67} and slow processing speeds,^{51,54,61,63,72}and 2 studies identified negative perceptions of the technology.^22,63 In addition, the following technological causes were identified by 1 study each: nurses encountered system downtime,⁶⁴ nurses in a long-term-care setting avoided scanning patient wristbands due to being reportedly too familiar with patients,⁶⁸ and nurses at an organization with a sepsis alert system did not trust the EHR accuracy and used their own methodology to assess sepsis risk.⁶³ User dissatisfaction with BCMA functionality also facilitated workarounds.²²

Usability

A majority of the studies, 25, reported poor usability,^{14,21,22,31,35,36,47,48,50–57,59–62,64,66,67,69,70–72} which included a wide range of problems. Computer hardware that was bulky or difficult to use,^{21,22,31,36,47,54,62,68} unclear audio alerts,^22,70 multiple scan attempts needed to read medication barcodes,^22,64 multiple screens to complete an action,^22,54,55 timeout of BCMA scanners,²² and the EHR.⁵⁷

Nurses also reported that information was difficult to locate,^{14,21,22,31,50,52,55,57,60–62,72} information was located in several screens,^14,35,54 EHR design was not intuitive,^48,50,51,55 the font size small and difficult to read,^47,62,70 the screens were small,^31,47,70 there was a lack of spell check in narrative documentation,⁵¹ there were nonreadable medication barcodes,^{22,31,66,67,69} there were nonreadable patient wristbands,^{22,31,36,67,68,70} there was system functionality that prevented documentation until prior tasks were completed,^22,54 and there were extensive mandatory data.^53,57 Additional causes include nonfixed mouse speeds,⁷⁰ as well as software integration between multiple clinical systems and scheduling³⁵and limitations to medication schedule adjustment.⁷¹

Task

Task-related factors include protocols or situations that nurses were not familiar with or expected to slow performance, and were identified in 14 studies.^{22,31,32,37,51,53,57,58,60,63,67–69,72} These include (1) medications that do not follow the typical processes, such as barcodes that are located inside the medication package instead of their normal location outside the package, or those with multiple barcodes²²; (2) belief that the scanning procedure or EHR was slower than other methods, which increases time on the task^{31,37,51,67–69}; (3) discarded medication packaging²²; (4) undocumented previous doses, which required a workaround to follow the process³¹; (5) emergency situations²²; (6) not enough time to document^{22,32,53,57,58,69}; (7) too busy to review EHR data^60,72; and (8) delayed documentation to communicate with other members of the care team.⁶³

Organizational

Our review identified organizational causes of workarounds in 17 articles.^{22,31,35,36,47,4},^{9,53,55,56,57,60,61,63–65,69,70} Organizational-related factors are instances in which the institutional policy does not align with standard prescribed procedures,²² eg, instances when a patients’ medications from home are administered in the hospital. A number of the factors related to medication dispensing issues, such as preparation or dispensing practices,²² partial dose medications,²² orders that were nonformulary,²² and home medications that were not barcoded despite all hospital-supplied medications having a barcode.^22,64 Additional organizational probable causes include knowledge deficits, such as untrained staff,^{22,35,47,56,57,61,70} nurse unfamiliarity with BCMA safety features,^22,65,69 perceptions certain BCMA processes should be completed by pharmacy instead of nursing,²² and organizational policies that were not updated to accommodate the EHR.^54,70 Resource issues such as insufficient devices,^{31,36,53,54,60,63} inadequate staffing,^22,69 and a culture in which physicians were unwilling to use the EHR^36,51,55 contributed to workarounds. Finally, in one setting lab, nutrition and patient management orders missing from the EHR required clinicians to use paper orders or contact departments by phone.⁵⁵

Patient related

Patient-related factors are special situations in which the nurse does not follow prescribed methods of EHR use due to patient characteristics.²² This includes situations when a barcode or ID band was not accessible for scanning during a sterile procedure, with poorly fitting wristbands, or owing to the interference when the patient does not allow the nurse to use BCMA due to combativeness.^22,67 Nurses also avoided documenting in the EHR in front of the patient.^53,63

Environmental

Environmental factors occurred due to the physical arrangement of the technology, patient, and hospital or healthcare space.^{22,36,60,62,67,70} Limitations to the care area, such as space constraints in radiology,²² the operating room,^21,22,47 limited use of the EHR, and infection prevention concerns, limited use of computers in the patient room.^22,62,67 Remaining causes included medications being far from the scanner (eg, medications that require refrigeration),²² loud ambient noise that interferes with a nurse’s ability to distinguish notifications,²² and fatigue due to eye strain.⁷⁰

DISCUSSION

We found that nurses’ workarounds to the EHR persist despite more than a decade of research in this area. Importantly, nurse workarounds appear to be an international phenomenon, and although studied most frequently in acute inpatient settings,^{14,21,22,31,32,36,47,48,50,51–53,55,56,59,61–67,69–72}. workarounds also occur in outpatient^35,49,51 as well as long-term care⁵⁴ and home care.⁶⁰ This is a large safety risk, as nurses play a critical role in patient care, as the last line in patient protection against error,⁴¹ and are the largest group of healthcare providers in the United States, and with midwives, comprise nearly half of the worldwide healthcare workforce.⁴⁰

Overall, we found that Koppel et al’s²² categorization of workarounds for BCMA remained quite robust over 12 years and also could be applied more broadly to the EHR. However, we found that usability, not previously identified by Koppel et al, was the most frequent cause of workarounds, appearing in 25 studies.^{14,21,22,31,35,36,47,48,50–57,59–62,64,67,69,70–72} Usability, defined by Nielsen to include the 5 attributes of (1) easy to learn and (2) remember, (3) efficient to use, (4) has few errors, and (5) is subjectively pleasing,⁷³ has emerged as a critical problem in health care that increases chance of errors across an array of health information technologies.^18,74,75

Although Koppel et al²² may have categorized certain BCMA workarounds under technological-related causes, we believe these are usability problems. Multiple screens needed to complete an action^22,54,55 difficulty finding information,^{21,22,31,50,52,55,56,60,62} and a need for multiple scans^22,55,64 illustrate a lack of efficiency, or ease of use. Organizational limitations, such as insufficient devices in acute care,^{31,36,53,54,63} and community settings, resulted in a workaround that nurses used personal mobile devices to photograph wounds to track healing and retrieve medication information, which impairs access to information at the point of care.⁶⁰ These actions present unnecessary risk because personal mobile devices may be hacked or stolen along with patient information.

Perhaps we should not be surprised about the number of usability problems, given that the existence of substandard usability testing by some EHR vendors^76,77 and limited federal policies to ensure EHR usability. EHR usability problems are causing increasing clinician dissatisfaction and burnout; however, most of this work focuses on physicians.^78–81 In this review, we also bring attention to the problem of poor usability for nurses and join researchers and advocates who call for stronger federal policies and investment to promote more useable health information technologies.^82,83

We were surprised that paper is a persistent artifact that has not disappeared despite introduction of the EHR.^{21,31,35,47,48,51–55,59–61,63,71,72} We agree with Chao,²¹ Gaudet,⁵² and Keenan et al⁸⁴ that in some instances, the EHR is not seen as an adequate tool to support the nurse in the provision of care^21,59 or easily accessible information.^52,84

Nurses have identified electronic documentation as time-consuming and cumbersome.⁶² Similarly, batching, a process in which nurses wait to document on several different tasks at one time,^{51,53,57,62,63} hints at the perceived limitations by frontline caregivers in using the tools available. These problems indicate the case for innovative solutions necessary for clinicians providing care. These workaround behaviors are rife for error and unintended consequences. Incomplete documentation such as in an emergency room triage³² an incomplete medication history,⁵⁸ or legacy databases⁵⁶ can produce information gaps, while delayed documentation^51,53,62,63 can impair a care team’s decision making.

There are ample opportunities to improve the science of nursing workarounds to the EHR, using quantitative methods to examine workaround practices.⁶⁴ Qualitative research is valuable to understand what, why, and how workarounds occur.⁸⁵ However, the development and use of common quantitative measures in this area would allow comparisons across institutions to catalyze change. Although workaround behaviors and causes have been identified, we found a major gap in research in the ambulatory setting,^35,49,57 little research in long-term care,^54,60,68 and a paucity of literature related to workaround prevention.

Limitations

Although we applied rigorous methods, there are limitations of this review. Exclusion criteria of English-only results and inclusion limited to peer-reviewed published literature may result in a biased sample.

CONCLUSION

This review highlights the many factors that continue to contribute to nurse workarounds that continue to undermine quality health care. The widespread use of workarounds by the largest group of healthcare providers subverts quality health care at every level of the healthcare system. Research is needed to explore the gaps in our understanding of nurse workarounds identified in this review, or risk of patient harm will remain.

AUTHOR CONTRIBUTIONS

All authors contributed substantially to conception, design, and data analysis. The article was drafted by DF, extraction reliability conducted with JM, KDL provided guidance on the methods and analysis. The manuscript was critically revised by DF, with all authors providing final approval of the version to be published.

SUPPLEMENTARY MATERIAL

Supplementary material is available at Journal of the American Medical Informatics Association online.