Keywords
Human factors engineering - usability - human-centered design - health information
technology - patient safety
1 Introduction
In several countries, we are moving beyond issues related to implementation and change
management of health information technology (IT); today, the basic IT infrastructure
is in place and health care organizations are investing in designing and implementing
the next generation of health IT “by-products”, such as clinical dashboard, status
display, clinical decision support, and patient-facing technologies. Usability and
other human factors challenges with health IT continue to be reported, as illustrated
in a New Yorker article by Atul Gawande entitled “Why doctors hate their computers”
[1]. The experience of usability problems with health IT extends to other health care
professionals, including nurses [2]. Staggers et al. [2] gathered input on health IT problems experienced by nurses from a group of nursing
leaders, IT experts, engineers and human factors experts. The consensus among the
group of 27 experts was that “current vendors do not yet understand how to support
nurses’ work, their critical thinking, and decision making” (page 192). For instance,
current EHR designs allow for single patient views whereas nurses care for complex
groups of patients in a non-linear, non-standardized manner. Patients also experience
usability challenges with health information technologies, such as patient portals
[3].
Over time, implementation of health IT may lead to benefits for patients as demonstrated
by the study by Lin et al. [4], which shows positive outcomes (e.g. hospital mortality) associated with health
IT after sufficient time and effort has been spent on implementation, maintenance,
and improvement of the system. However, we cannot rely on “time and effort” and just
hope that positive outcomes of health IT implementation will happen after a while;
in the meantime, negative outcomes may occur, which can endanger patients (e.g., hazards
and medical errors), frustrate healthcare professionals, and impede their performance
(e.g., burnout). We need to be vigilant and proactive in order to achieve benefits
with health IT and avoid negative patient safety outcomes [5]. Human factors and human-centered design play a critical role in ensuring that health
IT is well designed and fits with clinical and patient workflows.
We conducted a review of the literature on human factors and usability of health IT
published between 2016 and 2019 in various databases: PubMed/Medline, Google Scholar,
and PsycINFO. We also searched recent tables of content of major health informatics
journals, such as JAMIA, IJMI, ACI, and Health Informatics. Building on chapters published
in previous IMIA Yearbooks [6]–[8], we continue to see reports of usability problems with health IT. Emerging literature
shows that development and implementation of human-centered design methods to use
existing IT infrastructure can create value for patients and clinicians, for instance,
in implementing usable technologies, e.g., clinical decision support (CDS) for supporting
clinical decision making and care coordination. This chapter also highlights new issues
in human-centered design of health IT, such as design of health IT for teams.
2 Usability Problems of Health IT and Their Negative Impact on Patients and Clinicians
2 Usability Problems of Health IT and Their Negative Impact on Patients and Clinicians
An extensive account of patient safety issues with health IT was recently published
in the US trade literature with a dramatic title: “Death By 1,000 Clicks: Where Electronic
Health Records Went Wrong” [9]. The authors provided examples of poor health IT usability experienced by physicians
and their impact on patient outcomes. Health IT usability problems experienced by
physicians, nurses, and other health care professionals can produce multiple negative
outcomes for both patients (e.g. threats to patient safety) and clinicians (e.g. frustration,
work-arounds, burnout) [2], [10]–[12].
2.1 Impact of Health IT on Patient Safety
We continue to see evidence of the linkage between poor design of health IT and patient
safety consequences. An analysis of medication safety events reported by pediatric
clinicians in three children hospitals across the US showed that more than one-third
(36%) were related to EHR usability issues, in particular the lack of system feedback
and poor visual display of information [13]. The alarming statistic of this study is that 19% of the EHR usability-related events
could have resulted in patient harm. A prospective study of medication safety among
624 patients in two ICUs of a medical center identified a total of1,622 preventable
adverse drug events after EHR implementation [14]. Thirty-four percent of medication events were related to EHR technology; they included
orders with missing information and accepted duplicate medication orders. These findings
echo conclusions by Kushniruk et al. [6], [7] in previous IMIA Yearbooks that the lack of attention to human factors with health
IT and the failure to engage end-users in the design and implementation processes
produce negative consequences for patient safety. Users’ participation and analysis
of their work are key elements of the human-centered design process. Increased recognition
of the benefits of deep analyses of clinical workflow in this human-centered design
process fits with the focus of resilience engineering on the actual work done as opposed
to “work-as-imagined” [15], [16].
2.2 Impact of Health IT on Clinician Quality of Working Life
In addition to the impact of poor health IT usability on patient safety, there is
increasing attention on the negative impact of health IT use on clinicians. Unmanageable
demands for clinical documentation and poor health IT usability and workflow integration
are some of the EHR-related factors that have been suggested as contributing to clinician
burnout. In a survey of more than 5,000 physicians across all specialties in the US,
Shanafelt et al. [11] found that the use of EHR and CPOE technologies by physicians was associated with
lower physician satisfaction with time spent on clerical tasks and a higher likelihood
of professional burnout. A study of EHR event log data showed that primary care physicians
spent almost six hours per day interacting with the EHR both during and after clinic
hours, therefore, indicating potential challenges with workload and interferences
between the professional and private life spheres [17]. A survey of 4,197 physicians confirms the challenges and potential stressful impact
of health IT [18]. Sixty-four percent of the physicians agreed or strongly agreed that the EHR added
to their daily frustration, 38% reported that time spent on the EHR at home was moderately
high or excessive, and 46% indicated that time spent on documentation was “poor or
marginal”. All of these EHR-related stressors were significant predictors of physician
burnout. Clinicians have the feeling that they put a lot of time and effort to support
technology, whereas technology should support them in their clinical activities [1].
2.3 Old and New Challenges for Human Factors and Usability of Health IT
Studies on the impact of poor health IT design (and implementation) for both patients
and clinicians are important; but we need to better understand the linkage between
health IT usability and patient safety. We also need to move beyond analyzing health
IT usability and its impact on patient safety, and create more evidence for actually improving the design of health IT. If we want human factors to be taken seriously into account, we should not be shouting
from the sideline, but get actively involved in the design and implementation of health
IT, and evaluate the impact of our human factors methods and principles on the technology
in practice. This research evidence should include information on how to design better
(more usable) technologies, as well as their positive impact on key outcomes, e.g.,
patient safety. Section 4 provides examples of human-centered design applications
that produce benefits in usability as well as improvements in care process and patient
outcomes. We end the chapter with a discussion of the emerging challenge of designing
health IT for teams.
3 Improved Understanding of Health IT Usability and How It Is Linked to Patient Safety
3 Improved Understanding of Health IT Usability and How It Is Linked to Patient Safety
Two recent viewpoints in JAMA raised the alarm about usability of health IT that “remains
suboptimal” and provided policy recommendations to deepen and broaden the application
of human factors methods and principles in health IT [19], [20]. Deeper understanding of how usability of health IT can actually be improved is
at the core of these recommendations. This improvement will require that we more clearly
identify where and when usability gets created or breaks down, how usability is linked to patient safety, and how usability methods are used (and not used) by health IT designers and implementers.
3.1 Improving Usability of Health IT at Design and Implementation Stages
Recent research is helping us to better understand the problem of health IT usability
and its relation to patient safety. Usability of health IT is created (or influenced)
at both stages of design (e.g., IT vendors) and implementation (i.e. software configuration
by health care organizations). Early research showed the lack of systematic integration
and use of human factors methods by health IT vendors [21], [22]. Recently, Ratwani et al. [23] documented variations in health IT usability at the level of health care organizations.
Using screen capture software, they recorded EHR tasks performed by Emergency Department
(ED) physicians in four different hospitals that used two different EHR technologies
(i.e., Epic and Cerner). At each hospital, between 12 and 15 ED physicians participated
in the study; each physician performed a total of six tasks: two diagnostic imaging
orders, two laboratory orders, and two electronic prescriptions. The researchers extracted
the following usability measures from the videos: total task duration, number of clicks,
and error rate. Results showed large variation in all three usability measures across
the two EHR technologies as well as across the hospitals. For instance, the error
rate varied between 17% and 50% for the ordering of prednisone taper, between 0% and
30% for ordering Tylenol, and was consistently 0% for entering the lab order of lactate.
These data confirm the role of implementation processes in influencing usability.
Health care organizations make decisions about customization, configuration, and other
implementation processes (e.g., training, user input) that can have significant influence
on the final usability of the EHR technology in use. Therefore, improving usability
and other human and organizational aspects of health IT will require action and effort
by both designers and implemented of health IT.
3.2 Developing Better Understanding of Linkage Between Usability and Patient Safety
In addition to understanding that usability is constructed by various actors involved
in the design and implementation of health IT, we are developing a deeper knowledge
of the mechanisms by which usability influences patient safety. Research by the French human factors group at the University of Lille is producing
important knowledge about the impact of usability on patient safety; this is critical
research as it helps to untangle the “causal chain” between the technology and outcomes,
including patient safety [24]
[25]
[26]. Watbled and colleagues [26] provided empirical evidence for the linkage between usability flaws (i.e., violations
of usability principles), usage problems, and negative outcomes in the context of
health IT for discharge summaries. The researchers used three usability evaluation
methods: heuristic evaluation based on Scapin and Bastien’s method [27], user testing in an experimental setting, and in-situ observations of the technologies’
use after implementation. They identified multiple usability flaws in the two technologies,
and documented the chain of evidence between usability flaws, usage problems, and
negative outcomes. For instance, the heuristic evaluation identified the need for
additional steps to insert a dictated discharge summary in the electronic medical
record. This usability flaw led to a usage problem, as additional tasks were needed.
Because of the increased workload, there was a potential negative outcome of inserting
the wrong discharge summary in a given patient’s record. This type of in-depth analysis
helps to identify the specific usability flaws that can have negative consequences on patient safety, therefore, helping to focus
on high-risk design issues that need to be addressed by designers and/or implementers.
3.3 Need to Understand Work System of IT Designers and Implementers
As we are developing a deeper knowledge about health IT usability and its role in
patient safety, we are clearly justifying the need for more and better application
of human factors methods and principles in the design and implementation of health
IT. However, a major open issue remains: our lack of knowledge about the work of health IT designers and implementers. This issue represents an interesting dilemma for the human factors community as
a core principle of our discipline is to first understand the work situation and then
to use this knowledge to develop tools and technologies to support the actual work;
but we have not yet applied our human factors approaches to analyze the work of health
IT designers. We know very little about the work of health IT designers and implementers;
without knowledge on their work context, it remains difficult to elaborate recommendations
on how to integrate human factors in the health IT development life cycle.
Outside of health care, research has been conducted to examine the actual work of
designers and engineers and their experience of constraints, demands, and stress,
which may limit their ability to apply human factors methods and principles systematically
while designing, for instance, off-shore oil platforms [28], [29]. Ironically, Charles Perrow, one of the fathers of safety, and the author of “Normal
Accidents” [30], conducted the first study on the organizational context of human factors engineering
implementation [31], and identified several organizational factors (e.g., lack of support from top management)
that can limit the impact of human factors engineers on the design of technologies
and equipment. We need to do justice to our own human factors discipline and apply
our human factors methods to develop knowledge about the work system of health IT
designers and implementers; this knowledge is a critical open research area that will
contribute to the effective translation of our human factors knowledge into practice
[6]. This research would need to address multiple barriers, such as researchers’ access
to health IT vendors and trust between designers, implementers, and researchers [20].
4 Human-centered Design of Health IT
4 Human-centered Design of Health IT
In previous editions of the IMIA Yearbook, authors emphasized the importance of embedding
human factors methods in the early stage of health IT [7], [32] and they described approaches for implementing human-centered design for health
IT [32], [33]. We continue to see research that shows the positive impact of human-centered design
on usability, as well as important care process and patient outcomes. This research
is necessary to demonstrate the “value” of human factors-based approaches in health
IT design.
4.1 Impact of Human Factors Engineering in Health IT Design on Usability
We continue to see studies that implement various human factors methods and principles
to design health IT applications and demonstrate benefits on various usability dimensions.
For instance, Savoy et al. [34], [35] developed a deep understanding of cognitive requirements related to primary care
physicians or specialists requesting a consultation; this was accomplished using a
combination of semi-structured interviews and direct observations [35]. An interdisciplinary team (including human factors engineers) developed design
guidelines for each cognitive requirement, which were then used to develop a prototype.
A within-subject experiment of the prototype in comparison to the usual EHR consultation
template showed improvement in multiple usability dimensions: perceived satisfaction,
fewer mouse clicks, and faster completion but only for one of the subtasks in the
consultation request process [34]. A team of ED clinicians and human factors engineers applied cognitive systems engineering
methods to develop an ED information system prototype [36]. A usability evaluation of the prototype with 18 ED clinicians (attending physicians,
resident physicians, physician assistants, and nurses) provided strong evidence that
the ED information system prototype supported a range of cognitive activities, e.g.,
patient identification with specific characteristics (e.g., longest waiting time in
the ED) and assessment of resources for current patient demands. The study participants
rated the prototype’s usability and usefulness very high.
Studies are also published on the role of human factors engineering in designing usable
health IT for patients, such as a diabetes dashboard embedded in a patient portal
[37], an app for geriatric heart failure self-care [38], and an app for medication management [39]. The domain of consumer health IT has embraced human factors and usability methods
[40]. Participation of patients and caregivers in health IT design processes can not
only help to produce more usable technologies, but also lead to other benefits and
patient-centered outcomes such as empowerment and shared decision making [41].
Studies demonstrate that critical applications of human factors engineering in the
early stage of health IT design lead to benefits for both clinicians and patients.
However, the question remains about dissemination of human factors knowledge on a
larger scale. Brunner et al. [42] provided a unique perspective on that question when they examined implementation
of user-centered design practices in 170 United States Veteran Affairs primary care
clinics. They assessed whether primary care clinics were using the following practices
when developing computerized clinical reminders and/ or disease-specific templates,
i.e., CDS: (1) pilot testing of CDS before full-scale implementation, (2) assessment
of provider satisfaction post-implementation, (3) formal evaluation of CDS usability,
and (4) analysis of the impact of CDS on performance. The implementation of these
four practices was uneven across the 170 primary care clinics: 74% implemented pilot
testing, 62% carried out the assessment of post-implementation provider satisfaction,
36% performed formal usability assessment, and 79% did the analysis of the CDS impact.
The practice “analysis of CDS impact on performance improvement” was associated with
a greater utility of the CDS as perceived by primary care clinic directors. This research
helps us to understand the organizational context (e.g., primary care clinic) in which
usability and other human factors methods can (should) be implemented.
4.2 Value of Human Factors Engineering in Health IT Design in Improving Care Process
and Patient Outcomes
A few studies are emerging to show the actual value of incorporating human factors
methods and principles in the health IT design process, demonstrating impact on key care process and patient outcomes. This is a critical step forward in demonstrating the value of human factors engineering
as we show that human factors approaches “make a difference” in improving care processes
and patient outcomes, such as medication safety [43] and appropriate medication selection during a code [44]. An interdisciplinary team at Johns Hopkins University in the US (including a human
factors engineer) applied two rounds of usability testing to significantly reduce
the incidence of duplicate medical records from 4% to 2.3% [45]. The improvement was sustained even two years after the new medical record creation
process was implemented. Round 1 of usability testing included the mapping of the
patient search and selection process used by registrars, the direct observation of
registrars to understand their search behaviors, and the identification of barriers
experienced by registrars in searching patient names. Results of this round led to
multiple modifications to the patient name search process, which were tested in the
second round of usability testing. This study confirms the need for an iterative process
with multiple methods and approaches in order to ensure good usability of health IT
[7]. A critical aspect of this study is demonstrating the impact of using human factors
methods on key care processes and outcome measures, such as reduction of the number
of duplicate medical records. We need more research that demonstrates the value of
human factors engineering for the design of health IT in improving care process and patient outcomes.
4.3 Design of Health IT for Teams
As team-based care is increasingly implemented in the continuum of patient care [46], we are seeing an unfortunate mismatch between health IT designed for individual
tasks and what we need, i.e., health IT designed for care processes and teams [47]. To design technologies for teams, we need a better understanding of the critical
information that team members need to share and communicate for shared team awareness
[48], [49]. A couple of research projects focus on the application of human factors engineering
to design health IT for teams. Two research teams in Ottawa, Canada [48], [50]–[52] and at the University of Illinois [53] used various human factors methods that provided input to build shared team displays
for supporting cognitive work and communication during resuscitations. In Canada,
Parush et al., designed a smart display to support cognitive work and team communication
during ED resuscitation [48], [50]–[52]. Using multiple methods (e.g., interview of ED clinicians, observation of live resuscitations),
they identified key information needs, and then implemented an iterative human-centered
design process to develop a prototype for the shared situation display [48], [52]. They tested the display in a simulator environment with three teams made of attending
physicians, residents, nurses, and respiratory therapists. Resuscitation teamwork
was rated using the Clinical Team Scale (CTS). CTS scores were higher when teams used
the shared display than when they did not use it [54]. When teams did not use the shared display, they spent more time exchanging information
on the patient status; when they used the display, they spent more time discussing
the treatment for the patient. This provided some evidence that the shared status
display helped the team in focusing on the patient’s treatment rather than on just
exchanging information. A qualitative follow-up analysis of data gathered in the experiment
debriefs and focus groups studies confirmed how the shared display helped to build
team situation awareness during resuscitations [50].
Wu et al. [53] used the contextual design methodology [55] to develop initial design requirements for a shared display to support cognitive
work (i.e., information assembling, recall, and real-time tracking) in cardiac arrest
resuscitations. They conducted three simulation sessions with ICU physicians and nurses
who were also interviewed post-simulation. This iterative process helped to update
the list of design requirements and provided feedback on the design of the shared
display. The final stage of this human-centered design process consisted of three
scenario-based simulations with 12 ICU physicians and nurses using a mannequin in
a simulator. Different physicians and nurses participated in the three simulations
in order to increase the diversity of perspectives and inputs in the design process.
When using the shared display, physicians and nurses reported lowers levels of mental
workload. During the debriefing interviews, they indicated the reasons why and how
the shared display could improve their performance. For instance, they mentioned that
the system could improve their ability to adhere to best practice guidelines for resuscitation,
as well as their common (and shared) understanding of the treatment progression.
There is an urgent need to develop research and knowledge about how to design health
IT that can support the work of teams, in particular care teams that are distributed
over time and space along the patient journey [56]–[58]. This research requires a shift in focus from the individual to the team, as well
as innovative methods to evaluate the impact of these team-based technologies on clinical
care processes and team functioning. Several studies [43], [52], [53] have shown how scenario-based simulation can be used to evaluate the impact of human
factors-based design of health IT on important clinical outcomes (e.g., prescription
errors). We should extend the evaluation of health IT design beyond individual “subjective”
measures such as end-user satisfaction, and include team outcomes such as shared understanding
and team communication, as well as “objective” measures, e.g., task duration, number
of clicks and scrolls needed to navigate through the technology, and proxy clinical
outcomes such as the number of medication errors, the number of correct diagnoses,
and appropriate clinical decision making. We need further research efforts to develop
innovative human factors and usability methodologies such as those relying on multiple
methods, including sensors and EHR-based methods [59].
5 Conclusion
Recent literature on human factors and organizational aspects of health IT continues
to bring up problems of usability, poor workflow integration, and potential negative
impact on patient safety and other patient and clinician outcomes (e.g., frustration,
burnout). It is encouraging to see emerging and growing literature on the actual role
of human factors engineering in designing more usable and safer health information
technologies. This research needs to grow and address critical aspects of the changing
landscape of health care, including designing health IT for teams. Designing usable
health IT for individual tasks of clinicians or patients is challenging, but designing
health IT for teams with various needs, perspectives, and motivations is even more
challenging. Further understanding how the collaborative health IT design process
can integrate multiple perspectives is paramount [60].
Recent media attention to the problem of health IT usability and its negative impact
on patient safety and clinicians [1], [9] may provide the impetus for increased resources dedicated to health IT usability
and the broad application of human factors methods and principles in the design and
implementation of health IT. In order to respond to the call for increased integration
of human factors in health IT design, we need to better understand the actual work
performed by health IT designers. This can be done by applying human factors method
and principles to the work of health IT design and implementation: applying human
factors approaches to human factors in practice (i.e., health IT design and implementation)
is needed.
