CHAPTER 11: The Human–Technology Interface

Remie V. Tabigne, BSN

Ruelanie Rhoan P. Sarmiento, BSN
Introduction

Poorly designed technology can lead to errors, lower

productivity, or even the removal of the system (Alexander
& Staggers, 2009). Unfortunately, as more and more kinds
of increasingly complex health information technology
applications are integrated, the problem becomes even
worse (Johnson, 2006). However, nurses are very creative
and, if at all possible, will design workarounds that allow
them to circumvent troublesome technology. However,
workarounds are only a Band-Aid; they are not a long- term
solution.
The Psychology of Everyday Things, Norman (1988)

• argued that life would be a lot simpler if people who built

the things that others encounter (such as light switches)
paid more attention to how they would be used.
• At least one everyday thing meets Norman’s criteria for
good design: the scythe.
• The scythe’s design fits perfectly with its intended use
and a human user.
The Human–Technology Interface
• anytime a human uses technology, some type of
hardware or software enables and supports the
interaction.
• hardware and software that defines the interface.
• array of light switches described previously was actually
an interface (although not a great one) between the
lighting technology in the room and the human user.
In today’s healthcare settings
• hospitals may use bar-coded identification cards (to log
their arrival time into a human resources management
system)

• Using the same cards, they might log into their patients’ EHR, access
their patient’s drugs from a drug administration system
• administer the drugs using bar-coding technology.
Other examples of human–technology interfaces
According to Rice and Tahir (2014)

• defibrillator
• a patient- controlled analgesia (PCA) pump
• any number of physiologic monitoring systems
• electronic thermometers
• and telephones and pagers
 Different brands or versions of the same device
• Enter data into an EHR
– one might use a keyboard
– light pen
– a touch screen
– voice

Healthcare technologies may present information via

• computer screen,
• printer
• or smartphone.
 Patient data might be displayed differently;
• in the form of text, images (e.g., the results of a brain
scan), or even sound (an echocardiogram); in addition,
the information may be arrayed or presented differently,
based on roles and preferences
 Human–technology interfaces mimic face-to-
face human encounters
• faculty members are increasingly using
videoconferencing technology to communicate with
their students
• telehealth allows nurses to use telecommunication and
videoconferencing software to communicate more
effectively and more frequently with patients at home by
using the technology to monitor patients’ vital signs,
supervise their wound care, or demonstrate a procedure.
According to Gephart and Effken (2013)

The National eHealth Collaborative Technical Expert Panel

recommends fully integrating patient-generated data (e.g.,
home monitoring of daily weights, blood glucose, or blood
pressure readings) into the clinical workflow of healthcare
providers” (para. 3).
Telehealth technology has fostered other virtual interfaces
• system-wide intensive care units in which intensivists and
specially trained nurses monitor critically ill patients in
intensive care units, some of whom may be in rural
locations
• allow patients to interact with a virtual clinician (actually a
computer program) that asks questions, provides social
support, and tailors education to identify patient needs
based on the answers to screening questions.
• These human– technology interfaces have been remarkably
successful; sometimes patients even prefer them to live
clinicians.
Human–technology interfaces may present information

• using text, numbers, images, icons, or sound

• Auditory, visual, or even tactile alarms may alert users to
important information
• Users may interact with (or control) the technology via
keyboards, digital pens, voice activation, or even touch.
The growing use of large databases for research
(Vincent, Hastings-Tolsma, & Effken, 2010)

• led to the design of novel human–

technology interfaces that help
researchers visualize and understand
patterns in the data that generate new
knowledge or lead to new questions.
• Many of these interfaces now
incorporate multidimensional
visualizations;
• in addition to scatter plots,
histograms, or cluster representations
 Some designers use variations in sound to help researchers hear the patterns in large datasets

• Quinn (the founder of the Design Rhythmics Sonification Research Laboratory at the
University of New Hampshire) and Meeker (2000)
– “climate symphony,” different musical instruments, tones, pitches, and phrases
are mapped onto variables, such as the amounts and relative concentrations of
minerals
– help researchers detect patterns in ice core data covering more than 110,000 years
– Climate patterns take centuries to emerge and can be difficult to detect.
– The music allows the entire 110,000 years to be condensed into just a few minutes,
making detection of patterns and changes much easier.
The human–technology interface is ubiquitous

• in health care
• it takes in many forms
• A look at the quality of these
interfaces follows.
• Be warned: It is not always a
pretty picture
The Human–Technology Interface Problem
• The Human Factor, Vicente (2004)
– safety problems in health care identified by the Institute of Medicine’s
(1999) report and noted how the technology (defined broadly) used often
does not fit well with human characteristics.
– Vicente described his own studies of nurses’ PCA pump errors.
– Nurses made the errors (because of the complexity of the user interface,
which required as many as 27 steps to program the device)
– Vicente and his colleagues developed a PCA in which programming
required no more than 12 steps.
• Doyle (2005)
– reported that when a bar-coding medication system interfered
with their workflow, nurses devised workarounds;
– such as removing the armband from the patient and attaching it
to the bed, because the bar-code reader failed to interpret bar
codes when the bracelet curved tightly around a small arm.
• Koppel et al. (2005)
– reported that a widely used computer-based provider order
entry (CPOE) system
– to decrease medication errors actually facilitated 22 types of
errors because the information needed to order medications
was fragmented across as many as 20 screens
– available medication dosages differed from those the
physicians expected, and allergy alerts were triggered only
after an order was written.
• Han et al. (2005)
– reported increased mortality among children admitted to
Children’s Hospital in Pittsburgh after CPOE implementation.
– Three reasons were cited for this unexpected outcome.
– First, CPOE changed the workflow in the emergency room.
• Before CPOE, orders were written for critical time-sensitive treatment
based on radio communication with the incoming transport team before
the child arrived.
• After CPOE implementation, orders could not be written until the
patient arrived and was registered in the system (a policy that was later
changed).
• Second, entering an order required as many as 10 clicks
and took as long as 2 minutes; moreover, computer
screens sometimes froze or response time was slow.
• Third, when the team changed its workflow to
accommodate CPOE, face-to- face contact among team
members diminished.
In 2005, a Washington Post article
• reported that Cedars-Sinai Medical Center in Los Angeles had
shut down a $34 million system after 3 months because of the
medical staff’s rebellion.
• Reasons for the rebellion included the additional time it took to
complete the structured information forms, failure of the system to
recognize misspellings (as nurses had previously done), and
intrusive and interruptive automated alerts (Connolly, 2005).
• Even though physicians actually responded appropriately to the
alerts, modifying or canceling 35% of the orders that triggered
them, designers had not found the right balance of helpful-to-
interruptive alerts.
• Such unintended consequences (Ash, Berg, & Coiera,
2004) or ;
• unpredictable outcomes (Aarts, Doorewaard, & Berg,
2004) of healthcare information systems
– may be attributed, in part, to a flawed implementation process
– but there were clearly also human– technology interaction
issues.
– technology was not well matched to the users and the context
of care.
– In the pediatric case, a system developed for medical–surgical
units was implemented in a critical care unit.
(Walsh & Beatty, 2002) & (Vicente, 2004).

• Human–technology interface problems are the major

cause of as many as 87% of all patient monitoring
incidents (Walsh & Beatty, 2002).
• It is not always that the technology itself is faulty. In fact,
the technology may perform flawlessly, but the interface
design may lead the human user to make errors (Vicente,
2004).
Rice and Tahir (2014)
• reported on two errors that remind us we still have a long
way to go to ensure patient safety:
• In 2011, a pop-up box on a digital blood glucose reader
was misread and the patient was given too much insulin,
sending her into a diabetic coma; in 2013, a patient did
not receive his psychiatric medicine for almost 3 weeks
because the pharmacy’s computer system was set to
automatically discontinue orders for certain drugs, and
there was no alert built in to notify the team providing care
to this patient that the drug was suspended.
Remember : It is not only a technology or human interface issue

• Many of these problems occur when new technology is

introduced or existing technology is modified.
• we must examine how the technology tools are tested,
how the human users are prepared for their use, and how
the tools are integrated into the care delivery process
(Rice & Tahir, 2014).
Improving the Human– Technology Interface
• Each of these areas of
study is multidisciplinary
and integrates
knowledge from multiple
disciplines (e.g., computer
science, engineering,
cognitive engineering,
psychology, and sociology)

Figure 11-1 Human Factors

and Ergonomics
Figure 11-2 Human Factors and Ergonomics, Continued

• These areas are also

concerned with health
issues arising from
computer and other
technology use. Longo
and Reese (2014)
 Longo and Reese described;

How to prevent computer vision

syndrome • Increased smartphone use, we are
seeing neck issues caused by the tilt of
the head (with the chin on the
• look 20 feet away from your screen chest)while looking down at the
every 20 minutes for a minimum of 20 smartphone or other handheld device.
seconds • We must all be aware of our posture and
• should hold your phone up so that how our work areas are set up when
you are keeping your neck and eyes using our computers, smartphones,
aligned properly with the device’s tablets, and any other devices that
screen for more comfortable viewing consume a great deal of our time during
and interactions our work or personal hours.
Ecological approach to interface design
Effken (2016)

• help us realize a more meaningful EHR

• This approach borrowed from a small field of psychology,
ecological psychology
• “emergedafter the 3-Mile Island nuclear fiasco to allow
complex processes (like nuclear power plants) to be
more easily and safely controlled by operators.
• Ecological displays subsequently have enhanced the
control of airplanes, bottling plants—and even nuclear
power plants.
1990s, the approach began to be extended to the complexities of healthcare” (Effken, para. 2)

• help the user identify deviations from normal physical or

physiological processes.
• It is evident that users and clinicians need the technology and
interfaces necessary to quickly comprehend the multiple discrete
data that are contained in distinct parts of the EHR.
• “Because these are exactly the kind of complex problems that
theywere developed to solve, the analysis and design approaches
derived from ecological psychology are worth examining further as
we attempt to derive a more meaningful EHR” (Effken, 2016, para.
8)
Three axioms have evolved for developing effective human–computer interactions (Staggers, 2003):

(1) Users must be an early and continuous focus during

interface design;
(2) the design process should be iterative, allowing for
evaluation and correction of identified problems; and
(3) formal evaluation should take place using rigorous
experimental or qualitative methods.
Axiom 1: Users Must Be an Early and Continuous Focus During Interface Design

• Rubin (1994) used the term user-centered design to

describe the process of designing products
– users can carry out the tasks needed to achieve their goals
with “minimal effort and maximal efficiency
– user-centered design, the end user is emphasized.
– still a focus of human–technology interface design today
 Vicente (2004)
• argued that technology should fit human requirements at five levels of analysis
(physical, psychological, team, organizational, and political)
– Physical characteristics of the technology (e.g., size, shape, or location)
should conform to the user’s size, grasp, and available space.
– human psychological capabilities (e.g., the number of items that can be
remembered is seven plus or minus two).
– systems should conform to the communication, workflow, and authority
structures of work teams; to organizational factors, such as culture and
staffing levels; and even to political factors, such as budget constraints,
laws, or regulations.
Analysis tools and Techniques
• Task analysis examines how a task must be accomplished.
Generally, analysts describe the task in terms of inputs needed for the
task, outputs (what is achieved by the task), and any constraints on
actors’ choices on carrying out the task.
• Cognitive task analysis usually starts by identifying, through
interviews or questionnaires, the particular task and its typicality and
frequency
– Cognitive task analysis can be used to develop training programs.
– to develop a framework from which a colonoscopy training program
could be designed
 – “Task analysis methods (observation, a think-aloud protocol and cued-recall)
and subsequent expert review were employed to identify the competency
components exhibited by practicing endoscopists with the aim of providing a
basis for future instructional design” (Zupanc et al., p. 10).
– Resulting colonoscopy competency framework consisted of “twenty-seven
competency components grouped into six categories: clinical knowledge;
colonoscope handling; situation awareness; heuristics and strategies; clinical
reasoning; and intra and inter-personal” (Zupanc et al., p. 10)

• Cognitive work analysis was developed specifically for the analysis of

complex, high-technology work domains, such as nuclear power plants,
intensive care units, and emergency departments, where workers need
considerable flexibility in responding to external demands
A complete CWA includes five types of analysis:

(1) work domain ( functions of the system and identifies the information that
users need to accomplish their task goals.)
(2) control tasks (investigates the control structures through which the user
interacts with or controls the system)
(3) strategies (how work is actually done by users to facilitate the design of
appropriate human–computer dialogues.)
(4) social–organizational ( identifies the responsibilities of various users (e.g.,
doctors, nurses, clerks, or therapists) so that the system can support
collaboration, communication, and a viable organizational structure.)
(5) worker competencies ( identifies design constraints related to the users
themselves)
 Analysts typically borrow tools (e.g., ethnography) from the social
sciences for the two remaining types. Hajdukiewicz, Vicente, Doyle,
Milgram, and Burns (2001) used CWA to model an operating room
environment. Effken (2002) and Effken et al. (2001) used CWA to analyze
the information needs for an oxygenation management display for an ICU.
Other examples of the application of CWA in health care are described by
Burns and Hajdukiewicz (2004) in their chapter on medical systems (pp.
201– 238). Ashoon et al. (2014) used team CWA to reveal the interactions of
the healthcare team in the context of work models in a birthing unit. They felt
that team CWA enhances CWA in complex environments, such as health
care, that require effective teamwork because it reveals additional
constraints relevant to the workings of the team. The information gleaned
about the teamwork could be used for systems design applications
Axiom 2: The Design Process Should Be Iterative, Allowing for Evaluation and Correction of Identified Problems

• An excellent place to start is with Norman’s (1988, pp. 188–189) principles:

1. Use both knowledge in the world and knowledge in the head.
In other words, pay attention not only to the environment or to the user, but to both, and
to how they relate. By using both, the problem actually may be simplified.
2. Simplify the structure of tasks.
For example, reduce the number of steps or even computer screens needed to
accomplish the goal.
3. Make things visible:
Bridge the gulf of execution and the gulf of evaluation. Users need to be able to see
how to use the technology to accomplish a goal if they do, then designers have bridged
the gulf of execution. They also need to be able to see the effects of their actions on
the technology . This bridges the gulf of evaluation.
 4. Get the mappings right.
Here, the term mappipping is used to describe how environmental facts (e.g., the order of light
switches or variables in a physiologic monitoring display) are accurately depicted by the information
presentation.

5. Exploit the power of constraints, both natural and artificial.

Because of where the eyes are located in the head, humans have to turn their heads to
see what is happening behind them; however, that is not true of all animals. As the
location of one’s eyes constrains what one can see, so also do physical elements, social
factors, and even organizational policy constrain the way tasks are accomplished. By
taking these constraints into account when designing technology, it can be made easier
for humans use.
6. Design for error. Mistakes happen. Technology should eliminate predictable errors and
be sufficiently flexible to allow humans to identify and recover from unpredictable errors.
7. When all else fails, standardize. To get a feel for this principle, think how difficult it is
to change from a Macintosh to a Windows environment or from the iPhone operating
system to Android.
 Kirlik and Maruyama (2004)
• described a real-world human–technology interface that follows Norman’s
principles.
– authors observed how a busy expert short-order cook strategically managed
to grill many hamburgers at the same time, but each to the customer’s
desired level of doneness. T
– The cook put those burgers that were to be well-done on the back and far
right portion of the grill, those to be medium well-done in the center of the
grill, and those to be rare at the front of the grill, but farther to the left.
– The cook moved all burgers to the left as grilling proceeded and turned them
over during their travel across the grill.
– Everything the cook needed to know was available in this simple interface.
Several useful books have been written about effective interface design

• In addition, a growing body of research is exploring new ways to present

clinical data that might facilitate clinicians’ problem identification and
accurate treatment (Agency for Healthcare Research and Quality, 2010).
• Just as in other industries, health care is learning that big data can provide
big insights if it can be visualized, accessed, and meaningful (Intel IT
Center, 2013).
• Often, designers use graphical objects to show how variables relate. The
first to do so were likely Cole and Stewart (1993), who used changes in the
lengths of the sides and area of a four-sided object to show the relationship
of respiratory rate to tidal volume.
• Other researchers have demonstrated that histograms and polygon
displays are better than numeric displays for detecting changes in
patients’ physiologic variables (Gurushanthaiah, Weinger, & Englund,
1995). When Horn, Popow, and Unterasinger (2001) presented
physiologic data via a single circularobject with 12 sectors (where each
sector represented a different variable), nurses reported that it was easy
to recognize abnormal conditions, but difficult to comprehend the patient’s
overall status.
• This kind of graphical object approach has been most widely used in
anesthesiology, where a number of researchers have shown improved
clinician situational awareness or problem detection time by mapping
physiologic variables onto display objects that have meaningful shapes,
such as using a bellows-like object to represent ventilation.
Axiom 3: Formal Evaluation Should Take Place Using Rigorous Experimental or Qualitative Methods

• Perhaps one of the highest accolades

that any interface can achieve is to
say that it is transparent.
• An interface becomes transparent
when it is so easy to use that users
no longer think about it, but only
about the task at hand.
For example, a transparent clinical interface would
enable clinicians to focus on patient decisions rather
than on how to access or combine patient data from
multiple sources. In Figure 11-3, instead of the nurse Figure 11-3 Nurse–Patient Interaction Framework in
interacting with the computer, the nurse and the patient Which the Technology Supports the Interaction
interact through the technology interface. The more
transparent the interface, the easier the interaction
should be.
• Usability is a term that denotes the ease with which
people can use an interface to achieve a particular
goal.
– Usability of a new human–technology interface needs to be
evaluated early and often throughout its development.
– Typical usability indicators include ease of use, ease of
learning, satisfaction with using, efficiency of use, error
tolerance, and fit of the system to the task (Staggers, 2003).
– Some of the more commonly used approaches to usability
evaluation are discussed next.
Surveys of Potential or Actual Users

Chernecky, Macklin, and Waller (2006) assessed cancer patients’ preferences for
website design. Participants were asked their preferences for a number of design
characteristics, such as display color, menu buttons, text, photo size, icon
metaphor, and layout, by selecting on a computer screen their preferences for
each item from two or three options.
Focus Groups

Typically used at the very start of the design process, focus groups can help the
designer better understand
users’ responses to potential interface designs and to content that might be included in
the interface.

Cognitive Walkthrough

In a cognitive walkthrough, evaluators assess a paper mockup, working prototype, or

completed interface by observing the steps users are likely to take to use the interface to
accomplish typical tasks. This analysis helps designers determine how understandable
and easy to learn the interface is likely to be for these users and the typical tasks .
Heuristic Evaluation

A heuristic evaluation has become the most popular of what are called “discount
usability evaluation” methods. The objective of a heuristic evaluation is to detect
problems early in the design process, when they can be most easily and economically
corrected. The methods are termed “discount” because they typically are easy to do,
involve fewer than 10 experts (often experts in relevant fields such as human–
computer technology or cognitive engineering), hey are called “heuristic” because
evaluators assess the degree to which the design complies with recognized usability
rules of thumb or principles (the heuristics), such as those proposed by Nielsen
(1994).

For example, McDaniel and colleagues (2002) conducted a usability test of an

interactive computer- based program to encourage smoking cessation by low-income
women. As part of the initial evaluation, healthcare professionals familiar with the
intended users reviewed the design and layout of the program. The usability test
revealed several problems with the decision rules used to tailor content to users that
were corrected before implementation
Formal Usability Test

Formal usability tests typically use either experimental or observational studies of

actual users using the interface to accomplish real-world tasks. A number of
researchers use these methods. For example, Staggers, Kobus, and Brown (2007)
conducted a usability study of a prototype electronic medication administration
record. Participants were asked to add, modify, or discontinue medications using
the system. The time they needed to complete the task, their accuracy in the task,
and their satisfaction with the prototype were assessed (the last criterion through a
questionnaire). Although satisfaction was high, the evaluation also revealed design
flaws that could be corrected before implementation.
Field Study

In a field study, end users evaluate a prototype in the actual work setting just
before its general release. For example, Thompson, Lozano, and Christakis (2007)
evaluated the use of touch-screen computer kiosks containing child health–
promoting information in several low-income, urban community settings through an
online questionnaire that could be completed after the kiosk was used. Most users
found the kiosk easy to use and the information it provided easy to understand.
Researchers also gained a better understanding of the characteristics of the likely
users (e.g., 26% had never used the Internet and 48% had less than a high school
education) and the information most often accessed (television and media use,
and smoke exposure)
A Framework for Evaluation

Ammenwerth, Iller, and Mahler (2006) proposed a fit between individuals, tasks, and technology (FITT) model
( suggests each of these factors be considered in designing and evaluating human– technology interfaces)

It is not enough to consider only the user and technology characteristics; the tasks that the technology supports must
be considered as well.

The FITT model builds on DeLone and McLean’s (1992) information success model, Davis’s (1993) technology
acceptance model, and Goodhue andThompson’s (1995) task technology fit model.
 Notable strength of the FITT model
• encourages the evaluator to examine the fit between the various pairs
of components: user and technology, task and technology, and user
and task
• Johnson and Turley (2006) compared how doctors and nurses
describe patient information and found that doctors emphasized
diagnosis, treatment, and management, whereas the nurses
emphasized functional issues.
• Although both physicians and nurses share some patient information,
how they thought about patients differed.
• EHR needs to present information (even the same information) to the
two groups in different ways.
Technology Acceptance model and Task–technology
Fit model

• Hyun, Johnson, Stetson, and Bakken (2009) used a combination of

two models (technology acceptance model and task–technology fit
model)
– to design and evaluate an electronic documentation system for
nurses. To facilitate the design, they employed multiple methods,
including brainstorming of experts, to identify design requirements.
– To evaluate how well the prototype design fit both task and user,
nurses were asked to carry out specific tasks using the prototype in
a laboratory setting, and then complete a questionnaire on ease of
use, usefulness, and fit of the technology with their documentation
tasks.
– Because the researchers engaged nurses at each step of the
designprocess, the result was a more useful and usable system.
Future of the Human– Technology Interface

Increased attention to improving the human– technology interface through human factors
approaches has already led to significant improvements in one area of health care:
anesthesiology.

Anesthesia machines that once had hoses that would fit into any delivery port now have
hoses that can only be plugged into the proper port. Anesthesiologists have also been
actively working with engineers to improve the computer interface through which they monitor
their patients’ status and are among the leaders in investigating the use of audio techniques
as an alternative way to help anesthesiologists maintain their situational awareness.

As a result of these efforts, anesthesia-related deaths dropped from 2 in 20,000 to 1 in

200,000 in less than 10 years (Vicente, 2004). It is hoped that continued emphasis on human
factors (Vicente, 2004) and user-centered design (Rubin, 1994) by informatics professionals
and human–computer interactions experts will have equally successful effects on other parts
of the healthcare system. The increased amount of informatics research in this area is
encouraging, but there is a long way to go.
Technologies that are entering
our lives on a daily basis can
enhance or challenge our ability
to complete both our activities
of daily living and our
professional tasks. As our home
monitoring and patient
technologies increase, the
user’s (patient’s or nurse’s)
ability to use the technology is
paramount.
Figure 11-4 Human Technology Interface and
Task Completion
REMEMBER:

No matter who is using the technology, the human–technology interface addresses the user’s ability and the technology’s functionality to complete the task demands.

THANK YOU!