i50

SAFETY BY DESIGN

Work system design for patient safety: the SEIPS model

P Carayon, A Schoofs Hundt, B-T Karsh, A P Gurses, C J Alvarado, M Smith, P Flatley Brennan
...............................................................................................................................

Qual Saf Health Care 2006;15(Suppl I):i50–i58. doi: 10.1136/qshc.2005.015842

Models and methods of work system design need to be It defines three categories of the environment as
potential contributors to patient safety: physical
developed and implemented to advance research in and (e.g. noise), social (e.g. poor communication),
design for patient safety. In this paper we describe how the and biological (e.g. patient factors). Our SEIPS
Systems Engineering Initiative for Patient Safety (SEIPS) model5–7 goes further by clearly specifying the
system components that can contribute to causes
model of work system and patient safety, which provides a and control of medical errors, incidents and
framework for understanding the structures, processes and adverse events, showing the nature of the
outcomes in health care and their relationships, can be interactions between the components, showing
how the design of the components and their
used toward these ends. An application of the SEIPS model interactions can contribute to acceptable or
in one particular care setting (outpatient surgery) is unacceptable processes, and nesting itself in a
presented and other practical and research applications of model familiar to many health care profes-
sionals—namely, Donabedian’s quality
the model are described. model.11 12
........................................................................... A comparison of the strengths and weaknesses
of the SEIPS model, the Reason/Vincent model,
and Donabedian’s quality model is shown in

M
ost errors and inefficiencies in patient
care arise not from the solitary actions of table 1. The SEIPS model explains how the
individuals but from conflicting, incom- design of the work system can impact not only
plete, or suboptimal systems of which they are a the safety of patients but also employee and
part and with which they interact. To improve organizational outcomes. Employee outcomes
the design of these systems, the US Institute of include safety, health, satisfaction, stress and
Medicine (IOM) has proposed the application of burnout; organizational outcomes include rates
engineering concepts and methods—in particu- of turnover, injuries and illnesses, and organiza-
lar, human factors and systems engineering.1–3 tional health (profitability).
Emphasis on system design was promoted in a In this paper we describe the SEIPS model and
recent report by the National Academy of its research and practical applications, and
Engineering and the IOM: ‘‘… it is time to… propose that this model can be used to help
establish a vigorous new partnership between engineer- address the systemic problems of patient safety.
ing and health care and hasten a transition to a
patient-centered 21st century health care system’’.4 Our SEIPS MODEL OF WORK SYSTEM AND
research program, the Systems Engineering PATIENT SAFETY
Initiative for Patient Safety (SEIPS, http:// Our systems engineering approach to patient
www2.fpm.wisc.edu/seips/), originally funded safety is anchored within the industrial engi-
by the Agency for Healthcare Research and neering subspecialty of human factors. The
Quality, meets this challenge through a novel discipline of human factors emphasizes interac-
integration of human factors and healthcare tions between people and their environment that
quality models and proposes the SEIPS model contribute to performance, safety and health,
of work system5–7 and patient safety. and quality of working life, and the goods or
Patient safety researchers clearly recognize the services produced.13 14 It is important to char-
need for human factors engineering and systems acterize these many interactions between people
approaches to patient safety research, analysis, and their environment in a concise and coherent
and improvement. However, noticeably missing manner to identify points for improvement or
See end of article for from the patient safety literature are models to intervention. In order to achieve this goal, we use
authors’ affiliations guide studies to empirically examine system the work system model developed by Carayon
....................... design in relation to patient safety and medical and Smith (fig 1).5–7 According to the work
Correspondence to: errors. The model described by Reason,8 often system model, a person (the person could be a
Professor P Carayon, referred to as the ‘‘Swiss cheese’’ model, is care provider, another employee of a healthcare
Department of Industrial probably the most well known system model institution such as a biomedical engineer, a unit
and Systems Engineering, used within the patient safety community. clerk, or the patient) performs a range of tasks
Center for Quality and
Productivity Improvement,
Vincent et al9 have expanded Reason’s model using various tools and technologies. The perfor-
University of Wisconsin- and described seven categories of factors that mance of these tasks occurs within a certain
Madison, Madison, WI influence clinical practice, such as organizational physical environment and under specific organiza-
53726, USA; carayon@ and management factors, work environment, tional conditions. The five components of the work
engr.wisc.edu team factors, task factors and patient character- system (person, tasks, tools and technologies,
Accepted istics. The Haddon model, which is used com- physical environment, organizational conditions)
2 September 2006 monly in epidemiology and injury prevention, interact with each other and influence each
....................... has been proposed for use in quality and safety.10 other. The interactions between the various

www.qshc.com
SEIPS model i51

Table 1 Comparison of the SEIPS model, the Reason/Vincent model, and Donabedian’s
quality model
Model Strengths Weaknesses

SEIPS model of work system Focus on system design and its impact Descriptive model; no specific
and patient safety on processes and outcomes guidance as to the critical elements
Broad view of processes
Description of system, its components
and interactions among components
Impact on patient safety and employee/
organizational outcomes

Reason/Vincent model of Focus on etiology of accidents and No discussion of processes

accidents and adverse adverse events No guidance for system redesign and
events Description of contributing factors improvement of patient safety

Donabedian’s model of Description of relationships between Narrow description of ‘‘structure’’

quality (structure-process- structure, processes and outcomes Limited description of processes
outcome, SPO)

components ‘‘produce’’ different outcomes: performance, organizations to health care. Overall, the work system in which
safety and health, and quality of working life. care is provided affects both the work and clinical processes,
Assessing patient safety and designing systems that which in turn influence the patient, employee, and organiza-
produce safe patient care can be accomplished by using the tional outcomes of care. Changes to any aspect of the work
SEIPS model that integrates Donabedian’s structure-process- system will, depending on how the change or improvement is
outcome (SPO) framework and the work system model designed and implemented, either negatively or positively
(fig 1). The structure of an organization (or, more generally, affect the work and clinical processes and the consequent
the work system) affects how safely care is provided (the patient, employee, and organizational outcomes. Table 2
process); and the means of caring for and managing the displays elements of the various SEIPS model components.
patient (the process) affects how safe the patient is (out- This is not an exhaustive list of elements, but should be
come). We suggest that the work system model applied to considered as examples.16 Some of the elements have recently
patient safety complements and expands Donabedian’s been emphasized—for example, teamwork17 18 which is an
framework. Sainfort et al15 have proposed an earlier adapta- element of the organizational component; our SEIPS model
tion of the work system model and the concept of healthy does not highlight any single element of the work system.

Figure 1 SEIPS model of work system and patient safety.

www.qshc.com
i52 Carayon, Hundt, Karsh, et al

Table 2 Components and elements of the SEIPS model

Components Elements (examples)

Work system or Person Education, skills and knowledge

structure Motivation and needs
Physical characteristics
Psychological characteristics
Organization Teamwork
Coordination, collaboration and communication
Organizational culture and patient safety culture
Work schedules
Social relationships
Supervisory and management style
Performance evaluation, rewards and incentives
Technologies and tools Various information technologies: electronic
health record, computerized provider order
entry and bar coding medication administration
Medical devices
Other technologies and tools
Human factors characteristics of technologies
and tools (e.g. usability)
Tasks Variety of tasks
Job content, challenge and utilization of skills
Autonomy, job control and participation
Job demands (e.g. workload, time pressure,
cognitive load, need for attention)
Environment Layout
Noise
Lighting
Temperature, humidity and air quality
Work station design
Process Care processes and other processes Care processes
Other processes: information flow, purchasing,
maintenance, cleaning
Process improvement activities
Outcomes Employee and organizational Job satisfaction and other attitudes
outcomes Job stress and burnout
Employee safety and health
Turnover
Organizational health (e.g. profitability)
Patient outcomes Patient safety
Quality of care

Donabedian’s model The implication is, therefore, that aspects of the structure
Traditional approaches to quality assurance often rely upon may be of lesser consequence. As a result, components of the
SPO measures of quality as conceptualized by Donabedian.19 work system (such as the organization or the environment)
Donabedian’s model has proved valuable in examining the and their interdependencies may be overlooked or under-
clinical processes and outcomes of care, but it is limited in its emphasized. According to Donabedian’s model, poor quality,
recognition of the interactions and interdependencies among in turn, results by not following what is defined as the
system components. Donabedian’s model explicitly links the appropriate or correct means of performing a task (poor
structure and processes of care to subsequent patient process). Likewise, a bad outcome is associated with the poor
outcomes. The SEIPS model builds on this idea by showing performance of an individual (or a group of individuals). As a
how work system design (structure) is linked to patient result, practitioners tend to associate the SPO model with
safety (outcome) through care processes. In Donabedian’s traditionally punitive efforts and reporting—for example, the
model, the structure includes the organizational structure (work credentialing process of medical staff requires reporting
system model component = organization); the material performance measures, generally originating from quality
resources (work system model components = environment, assessment activities. Conversely, patient safety activities
technology/tools); and the human resources (work system place greater emphasis on the system in which practitioners
model components = care provider, tasks). Donabedian’s work and less on individual performance.21
two other means of assessing quality include evaluating the In contrast to the SPO model, the SEIPS model emphasizes
process(es) of care (how provider tasks and clinical processes the structure. It proposes the work system model as an
are both organized and performed) and evaluating the expansion of the structure by addressing elements of the
outcome(s) of care (assessing the clinical results and impacts work system model such as the physical environment,
of and patient satisfaction with the care provided). organizational culture and climate, error reporting and
Donabedian20 concludes that direct relationships may exist analysis, and work design that are so much a part of the
between structure, process and outcome. current patient safety focus. The work system model also
In Donabedian’s model, however, there is a clear statement allows linkage of the various elements of the structure in the
that ‘‘we must begin … with the performance of physicians and SPO model—that is, the organizational structure and
other health care providers’’.19 This reflects the fact that most of material and human resources. The implication is that
the focus of the SPO model centres on the providers and their outcomes (both patient and employee/organizational, as
relationship with the processes and outcome(s)—that is, proposed by the SEIPS model) are related and that they also
quality is assessed by the way in which care is provided by an are associated with structure aspects of the SPO model.
individual or care team as well as the outcome of the care. Battles and colleagues22 23 have made a similar effort at

www.qshc.com
SEIPS model i53

expanding Donabedian’s model and have called for increas- SEIPS model and the professional model
ing attention to the ‘‘structure’’ element. The individual in the work system can be a physician.
Our proposed model enriches Donabedian’s model in four Professionals such as physicians, nurses, pharmacists, and
major ways: other licensed healthcare providers are bound by ethical
mandates to serve in their clients’ best interests.25 The
(1) it adds employee and organizational outcomes to the list concept of the physician as a professional implies both great
of important outcomes to consider; responsibility and great respect. This professionally motivated
(2) it specifies possible relationships between patient out- sense of autonomy and responsibility may, in part, explain
comes and employee/organizational outcomes; physician resistance to organizational strategies to change
(3) it includes other processes besides care processes; and their behavior. However, it has been noted that physician
professionalism also fosters a ‘‘culture of blame’’ when things
(4) it proposes a more comprehensive definition of ‘‘struc-
go wrong because, if physicians are responsible for the entire
ture’’.
medical process, then they are also exclusively to blame for
poor quality care.26 A system redesign approach to changing
The individual in the work system model physician behavior may not only be a more effective
In the SEIPS model the individual is at the centre of the work approach, but may be better accepted by physicians than
system. In turn, the work system should be designed to traditional organizational efforts to improve quality using
enhance and facilitate performance by the individual and to incentives.27 For example, system redesign has been
reduce and minimize the negative consequences on the employed in anesthesiology to develop devices such as a
individual (such as reduced stress) and therefore the system of gas connectors that do not allow a gas hose or
organization (for example, improved organizational perfor- cylinder to attach to the wrong site. This type of technological
mance). The individual at the centre of the work system advance, in addition to an emphasis on a ‘‘culture of safety’’,
could be any healthcare provider performing patient care- has helped to decrease deaths due to anesthesia.28 These
related tasks or a patient receiving care. system redesigns support rather than conflict with the
It is important to recognize that a healthcare work system physician’s role as a professional.
often includes both healthcare providers and the patient Quality improvement efforts that incorporate system rede-
being cared for. For instance, if the focus of the work system sign may therefore be more successful in changing physician
is on the nurse administering a medication, the individual of behavior because they preserve physician professionalism.
the work system would be the nurse. The patient is involved Redesigning a system to make it ‘‘easy to do things right and
in various components of the nurse’s work system. The hard to do things wrong’’ supports this approach to changing
nurse’s task of actually administering the medication physician behavior. Involvement of physicians in the system
involves the patient. The physical environment may involve redesign process is critical to both the success of the system
noise and distractions coming from requests from other and to physician acceptance of the new system.26 The SEIPS
patients. If the focus of the work system is on the patient model, which maintains a sense of job control and participa-
taking a medication, the individual of the work system would tion in the process of organizational change, may improve
be the patient. Nurses and other healthcare providers and physician acceptance of, and adherence to, new systems.
staff are involved in various components of the patient’s Processes in the SEIPS model
work system. The patient performs the task of taking a Donabedian’s model focuses on care process(es)—that is,
medication that has been ordered by a physician and how care is provided, delivered and managed. The SEIPS
administered by a nurse. Any healthcare work system can model expands the concept of process to include not only care
therefore involve multiple individuals such as healthcare processes but also other processes that support the care
providers and patients. The design of the work system must process such as maintenance, housekeeping and supply chain
meet all of their needs—not just the needs of one management. These other processes need to be designed to
individual—for the design to be effective. support the delivery of safe care. For instance, performance
obstacles such as inadequate supplies in isolation rooms may
The patient in the SEIPS model prevent ICU nurses from safely delivering care to critical
It is important to remember that the term ‘‘patient safety’’ patients.29 In this example, the supply chain management
includes the ‘‘patient’’.24 The patient fits in the SEIPS model process does not support the care process.
in different ways. Firstly, if the individual in the work system It is also important to understand that processes are very
is a healthcare provider, tasks performed by the individual much influenced by work system design.30 For instance, a
often involve the patient—for example, surgical tasks. The care process can be considered as a series of steps or tasks
patient is also involved in patient care processes and is the performed by an individual or a team of individuals using
‘‘recipient’’ of good or bad outcomes of the care process. From various technologies and tools; the care process may involve
this viewpoint, the patient is an ‘‘input’’ into the SEIPS various locations, therefore different physical environments.
model. Secondly, the SEIPS model can be applied directly to The care process is also affected by multiple organizational
the patient. The individual in the work system could be the characteristics such as the need for coordination and collabora-
patient who is performing tasks (for example, visiting his/her tion among the healthcare providers involved in the process.
physician, receiving medications from the pharmacist) using In the next section we describe an example of the analysis of
various technologies and tools (for example, email to the outpatient surgery process applying the work system
communicate with the physician, prescription order) in a components.
certain physical environment (such as a clinic or pharmacy)
under certain organizational conditions (for example, waiting Balanced work system
to see the physician, rushing to get the prescription filled). The SEIPS model uses the concept of ‘‘balance’’ proposed by
Using the SEIPS model with the patient at the centre of the Carayon and Smith.5 6 Some negative elements in a work
work system model can help to identify deficiencies in the system that are hard to change may be overcome by focusing
healthcare system that can impair the patient’s capacity to on some other positive elements. For example, in a care unit
receive high quality safe care, and therefore contribute to the where the layout is not optimal for patient care in terms of
design of systems and processes for delivering patient centred the visibility of the patients, assigning patients located next
care. to each other to the same nurse may help with individual

www.qshc.com
i54 Carayon, Hundt, Karsh, et al

nurse outcomes (fatigue, stress) as well as patient safety creates a problem for the nurse work system. As explained
(nurse can monitor the patients continuously). Or the above, the design of the work system needs to consider the
difficulty of not having an adequate number of nursing needs of all the people involved.
assistants in the unit may be overcome by strong teamwork
and collaboration among nurses. Impact of work system and processes on outcomes
One element of the work system that has been extensively Our model emphasizes the linkages between patient out-
focused on in health care is the skills and knowledge of the comes and employee/organizational outcomes. The fact that
individual healthcare provider. The SEIPS model is useful for the SEIPS model explains how the design of a system can
understanding that, although the skills and knowledge of an impact patients, employees, and the organization has its roots
individual healthcare provider are important, it is not in the theory of healthy work organizations (HWOs). HWOs
sufficient by itself to ensure high quality care and patient are organizations that have both good organizational out-
safety. The entire work system needs to be well designed for comes and a healthy and safe workforce.31 32 A healthcare
optimal performance. For example, a nurse who has excellent HWO would also provide high quality safe patient care.15
skills and knowledge may not give the highest quality and Some evidence exists that healthcare organizations are not,
safest of care to the patient if the equipment she/he needs to in fact, healthy organizations. Healthcare workers experience
use is outdated or the medication that she/he gives to the many negative consequences of poor system design such as
patient is not available in the unit at the scheduled time. This job dissatisfaction, burnout, intentions to quit, reduced
is an example of a poor balance in the work system where mental health, and injuries.6
one element (for example, outdated equipment) creates a Others have also emphasized the important relationship
barrier or obstacle to optimal performance.29 The same between patient outcomes and employee/organizational out-
individual healthcare provider practicing in two different comes.15 33 Poor employee/organizational outcomes, such as
work systems may demonstrate different performance. For back injuries experienced by nurses, are likely to be related to
example, a nurse who has a high patient load may provide poor patient outcomes.34 The experience of musculoskeletal
high quality and safe care and may not feel very tired at the pain or discomfort may affect the nurse’s psychological and
end of the shift because he/she may be receiving help from physical resources necessary to perform her job safely. In
other nurses (teamwork and support) and may have the right addition, work system factors are likely to simultaneously
tools and equipment in the unit and the patient room to contribute to negative employee/organizational outcomes and
support her. This is an example of a balanced work system. negative patient outcomes such as medical errors.35
On the other hand, the same nurse who has the same patient Performance obstacles in the work system can not only
load may not perform well in another work system where affect the healthcare provider’s capacity to perform his/her
she/he does not get much support from peers and does not job, but also affect their attitudes toward their organization
have the right tools and equipment in the unit. This example such as job dissatisfaction and frustration.29
shows a lack of balance in the work system. The SEIPS model specifies feedback loops from processes
The challenge of achieving a balanced work system is to work system and from outcomes to work system. These
highlighted in the following example. This example describes feedback loops represent pathways to design or redesign the
a mechanism for reducing physician workload in the work system. Poor processes and outcomes can be triggers for
selection and insertion of central venous catheters. system redesign: the need would then arise to identify
Traditionally, a physician, assisted by a nurse, places negative work system elements that affect processes and the
catheters in the central venous system for patient therapy quality and safety of care, as well as employee and
and monitoring. However, with the introduction of periph- organizational outcomes.
erally inserted central venous catheters (PICCs), this practice
has changed. PICCs are inserted by the nurses and/or APPLICATION OF THE SEIPS MODEL TO
physicians certified in this procedure. By following patient OUTPATIENT SURGERY
care policy and procedures for central venous catheter use, an In collaboration with our SEIPS partners, we identified
ICU physician determines which type of central catheter outpatient surgery services as the first target for testing the
should be placed. A PICC line team composed of certified SEIPS model as a guide for patient safety assessment and
nurses inserts and cares for the patient’s PICC line, thus intervention. We have subsequently applied the SEIPS model
freeing up the physician from this time consuming proce- to our pilot study of five outpatient surgery centres located in
dure. However, the PICC line intervention may not decrease Madison, Wisconsin.36 The goal of the project is to identify
the nurse’s workload but add to the workload as more and elements within outpatient surgery systems and processes
more patients require PICC line insertion rather than central where safety threats may exist, and to plan mediating efforts
venous catheter insertions. in a manner that is congruent with the context.36 The SEIPS
This example shows the need to focus on the entire work model was used for two purposes: (1) to guide the
system relevant to the particular care process (PICC line assessment of systems, processes and outcomes in each
insertion): a narrow focus on the physician work system outpatient surgery centre for the development of system

Table 3 Application of the SEIPS model to the SEIPS outpatient surgery project
Evaluation of system redesign
Phase of assessment and determination of system redesign interventions interventions

Shadowing of Review of floor Assessment of position Employee

Components of SEIPS model Staff survey patients plans descriptions questionnaire Patient survey

Work system ! ! ! ! !
Processes ! ! !
Employee and organizational !
outcomes
Patient outcomes ! ! !

www.qshc.com
SEIPS model i55

redesign interventions, and (2) to guide the evaluation of the

system redesign interventions (table 3). Box 1 SEIPS model for system design

Assessing work systems, processes, and outcomes of N The management of a rural hospital wants to meet the
outpatient surgery guidelines of the Leapfrog Group and decides to invest
In the assessment phase of our study of outpatient surgery in computerized provider order entry (CPOE).
we used a variety of methods to understand the work
systems, care process, and various outcomes including:
N Management soon realizes that, to achieve successful
implementation of CPOE, they need to be proactive
and redesign their systems and processes. Looking at
N completion by staff of an open ended questionnaire aimed
at identifying healthcare professionals’ areas of concern the SEIPS model, they realize that the CPOE technology
regarding quality and patient safety, as well as those will probably interact with existing technologies, tasks,
aspects of their work system that promote patient safety environment, organization, and people (work system),
and a healthy work environment;29 37 so they begin to make an inventory of the possible
interactions.
N ‘‘shadowing’’ of patients undergoing outpatient surgery to
better understand the information flow and any short- N The CPOE technology will have to be integrated with
comings and strengths associated with the system from a the existing computerized decision support system,
patient’s perspective;38 pharmacy information management system, labora-
tory information management system, radiology infor-
N review of floor plans to aid in assessing the physical flow
of patients and staff; and mation system, billing software, and electronic medical
record (technologies and tools). Physicians’ task of
N assessment of position descriptions to clarify roles of those
providing care. ordering medications will change, pharmacy transcrip-
tion may be eliminated, and additional technical
The first data collection method (staff questionnaire) support (organization) will likely be required. An
provides data on components of the work system as well as informational campaign and training materials for
issues related to the work system and processes perceived by physicians and pharmacists need to be developed
the staff as affecting patient outcomes, in particular quality of (organization). Additional computers (technologies
care and patient safety.37 The initial staff questionnaire and tools) will need to be purchased for ordering,
included three questions: which means additional space for the computers. From
an organizational point of view, because order capture
N What do you think are the main issues related to quality of
patient care and patient safety in your outpatient surgery
will be enhanced, billing accuracy should improve as
well. New policies for physician ordering and phar-
unit?
macy verification will have to be developed (organiza-
N Please think of instances in the past year when you feel
your performance was challenged or below par due to
tion). Various processes will therefore have to be
redesigned.
problems in the OSC [Outpatient Surgery Center] ‘‘sys-
tem’’. Please briefly describe any such instance(s) you N The administration creates task forces to work on all of
experienced by explaining the situation and what you these design issues before the CPOE is implemented so
think caused it? that their healthcare system and processes are ready.
N Please think of instances in the past year when you feel
your performance was exceptional. Please briefly describe
any such instance(s) you experienced by explaining the
situation and what you think caused it. observer maintained a two dimensional log: (1) listing the
chronological sequence of steps the patient underwent and
Staff responses to the questionnaire can be associated with (2) recording observations according to the work system
numerous aspects of the work system.29 Tasks, tools, and component(s). For example, a patient whose vital signs are
organization coincide with an issue repeatedly reported by being collected during intake by the preoperative nurse may
staff concerning the process of obtaining clinical information have the following log entries at a given time:
on patients in a timely fashion. Here staff conveyed the
inherent inefficiencies of tracking down information that
should have been previously provided. Likewise, quality and
N Task: patient vitals taken.

safety issues identified were related to insufficient and N Environment: patient door open; noisy and distracting
interactions between staff in hallway.
potentially inaccurate information then collected because of
the last minute nature of these clinical assessments. Staff N Tools/technology: vitals recorded manually in patient’s
chart.
offered suggestions to remedy this problem through redesign
of forms and changes in policy and job design. Staff N Organization: nurse conveys that he will most likely not
follow the patient throughout her stay.
responses to the questionnaire were also associated with
various processes (including care processes) and outcomes.
They commented about the low quality of communication The review of floor plans provides information on the
with patients regarding preoperative preparation (for exam- physical aspects of the work system and their potential impact
ple, understanding of instructions) and postoperative recov- on processes. In this case the plans provided an understanding
ery (for example, inconsistencies in providers’ instructions). of the work flow and offered a greater appreciation for the
They also emphasized coordination issues related to patient confidentiality issues identified during the patient shadow-
information. For instance, unavailability of patient related ing. Likewise, it was easier to understand comments on the
information sometimes can lead to cancellations of surgeries staff questionnaire concerning the work space. For instance,
on the day they are scheduled. concern was expressed by staff regarding the lack of privacy
The second data collection method (patient shadowing) for patients and high noise levels. A review of the floor plans
records the components of the work system over time to collect helped the research team further understand the reasons for
data on the patient care process.38 During the shadowing the these concerns.

www.qshc.com
i56 Carayon, Hundt, Karsh, et al

Box 2 SEIPS model for proactive hazard analysis Box 3 SEIPS model for accident investigation

N A nursing home director, knowing that medication N A major academic teaching hospital has a sentinel
administration errors have become a major concern at event and sets out to conduct a root cause analysis
many nursing homes, decides to launch a hazard (RCA).
analysis of the medication administration system at her
nursing home before her nursing home has a problem.
N In order to make sure that the RCA team does not jump
to conclusions or fall into blaming the individuals at the
N To decide what information to collect, she refers to the sharp end, the hospital patient safety officer shows the
SEIPS model. From the model she understands she will RCA team the SEIPS model. She explains to the team
have to collect data about hazards related to all the that the many potential causes of the event may have
people involved in the medication administration come from the design of the tasks and procedures, the
process as well as the technology, tasks and proce- environment, the technology used, the organizational
dures, organizational policies and culture, and envir- culture or reward system, or most likely, some
onment related to medication administration. This is far interaction among the elements.41
more data than she would have otherwise thought to
collect.
N The RCA team then examines all of those factors in
order to understand how they each might have
N Through the data collection it is discovered that nurses contributed to the sentinel event (patient outcome).
are administering on average six medications per
patient (task), that nursing home policy only allows
them a certain number of minutes for medication
passes (organization), and that to follow all nursing CONCLUSION
home administration protocols would require at least The SEIPS model is useful for providing a view of the whole
double the amount of time allotted (interaction between system instead of focusing on only one aspect of the work
organization and task). Because of that, nurses are system and treating that aspect in isolation. It is descriptive,
circumventing the supposed ‘‘safety’’ protocols for not prescriptive. It does not tell if a change in one factor in
administration. the work system leads to any specific employee, organiza-
tional or patient outcome. However, it provides a framework
on how to think about the different aspects of a work system,
their interactions, and possible outcomes. This can be
Finally, the assessment of position descriptions provides considered as a limitation of the model because it does not
general information on the work system of the various provide specific guidance as to the critical elements; but it
healthcare professionals. This gave us greater insight can also be a strength because the model is generic and
concerning the expectations for the various positions as well adaptable to the particular context or situation.
as how each centre was organized according to the various It is critical to understand how resistance to a systems
tasks. approach may limit effective implementation of the SEIPS
model in practice. In particular, provider resistance may
Evaluation of system redesign interventions
Two data collection methods were used to evaluate the
various system redesign interventions implemented in the Box 4 SEIPS model for patient safety research
five outpatient surgery centres: a structured employee
questionnaire and a patient survey. The employee question-
naire includes questions on several components of the work
N Research on patient safety is generally concerned with
understanding (a) the predictors of safe or unsafe care
system, particularly the issues targeted by the redesign practices, (b) the predictors of potential or actual
interventions (such as communication and coordination).37 39 patient harm, and (c) testing interventions to improve
It also includes questions on employee outcomes (for patient safety.
example, quality of working life such as job satisfaction
and stress), as well as staff perceptions of quality and safety N The SEIPS model can guide researchers in identifying
of care provided by their outpatient surgery centres. The
potential predictors by helping them think about all of
patient survey focuses on processes (such as medication the relevant factors in the system, as opposed to just
related information) and patient outcomes (such as symp- focusing on what seems to be relevant (for example,
toms and complications from surgery).40 caregiver characteristics or workload).
N The SEIPS model can also help intervention research-
ers. To successfully study an intervention, the interven-
USING THE SEIPS MODEL OF WORK SYSTEM AND
tion must be designed appropriately and the correct
PATIENT SAFETY: DESIGN AND RESEARCH
indicators of success or failure must be measured. The
APPLICATIONS SEIPS model can help guide the design of the
Because of its emphasis on a systems approach, the SEIPS
intervention to make sure that the relevant technologi-
model can be used both proactively and reactively to improve
cal, organizational, job, environmental and personnel
patient safety by focusing on the design of work systems. It
factors are being considered. Furthermore, the SEIPS
can be used proactively to guide system design and hazard
model can guide the measurement of success by
analysis, or reactively to guide patient or employee injury
investigations. It can also guide patient safety researchers pointing to changes in the technology, organization,
toward developing research questions and understanding jobs, environment, or personnel that might indicate
what variables to measure for a particular study. Examples success or failure. According to the SEIPS model,
are used to illustrate the case for each in boxes 1–4. For each success or failure needs to be evaluated on both patient
example, components of the SEIPS model are highlighted in outcomes and employee/organizational outcomes.
italics.

www.qshc.com
SEIPS model i57

A P Gurses, Department of Anesthesiology, University of Maryland,

Key messages Baltimore, Maryland, USA
Funding for this research was provided by the Agency for Healthcare
N The design of work systems affects patient safety as well Research and Quality Grant # P20 HS11561-01.
as employee and organizational outcomes. Competing interests: none.
N The design of work systems influences processes which, An earlier version of this paper was presented at the Organizational
in turn, affect outcomes. Design and Management Conference, 2003 (Carayon P, Alvarado CJ,
Brennan P, et al. Work system and patient safety. In: Luczak H, Zink KJ,
N The individual is at the centre of the work system and
eds. Human factors in organizational design and management—VII.
can be a healthcare provider, a healthcare team, or the Santa Monica, CA: IEA Press, 2003:583–9).
patient.
N The work system needs to be balanced and to consider
REFERENCES
the requirements of the various stakeholders involved in
the process(es). 1 Institute of Medicine. Crossing the quality chasm: a new health system for the
21st century. Washington, DC: National Academy Press, 2001.
N Patient outcomes and employee/organizational out- 2 Institute of Medicine. To err is human: building a safer health system.
comes are influenced by the work system design. They Washington, DC: National Academy Press, 1999.
3 Institute of Medicine. Keeping patients safe: transforming the work
also influence each other. environment of nurses. Washington, DC: National Academies Press, 2004.
4 Reid PR, Compton WD, Grossman JH, et al. Building a better delivery
system. A new engineering/health care partnership. Washington, DC:
National Academies Press, 2005:15.
5 Carayon P, Smith MJ. Work organization and ergonomics. Appl Ergon
derive from a mismatch between the implications of the work 2000;31:649–62.
system model and the degree of autonomy expected by 6 Smith MJ, Carayon-Sainfort P. A balance theory of job design for stress
reduction. Int J Ind Ergon 1989;4:67–79.
providers in their work environment. The professional model 7 Smith MJ, Carayon P. Balance theory of job design. In: Karwowski W, ed.
assumes that providers have the ultimate authority to define International encyclopedia of ergonomics and human factors. London: Taylor
and carry out the core tasks of providing medical care.42 & Francis, 2000:1181–4.
Because the professional status of providers is linked so 8 Reason J. Human error. Cambridge: Cambridge University Press, 1990.
9 Vincent C, Taylor-Adams S, Stanhope N. Framework for analysing risk and
strongly to their autonomy, they may resist system efforts to safety in clinical medicine. BMJ 1998;316:1154–7.
manage the process of medical care as infringing on their 10 Brasel KJ, Layde PM, Hargarten S. Evaluation of error in medicine:
autonomy and degrading their professional status. application of a public health model. Acad Emerg Med 2000;7:1298–302.
11 Donabedian A. The quality of medical care. Science 1978;200:856–64.
Recognizing potential resistance and obtaining strong sup- 12 Donabedian A. Evaluating the quality of medical care. Milbank Memorial
port from key opinion leaders43 is essential to effective Fund Q 1966;40:166–206.
implementation of the SEIPS model. 13 Wilson JR. Fundamentals of ergonomics in theory and practice. Appl Ergon
2000;31:557–67.
We have shown that the SEIPS model can be applied in 14 Grandjean E. Fitting the task to the man: an ergonomic approach. London:
many different ways to patient safety challenges. Healthcare Taylor & Francis, 1980.
institutions interested in understanding the relationship 15 Sainfort F, Karsh B, Booske BC, et al. Applying quality improvement principles
to achieve healthy work organizations. J Qual Improve 2001;27:469–83.
between the work system and patient safety may start by 16 Carayon P, Alvarado CJ, Hundt AS. Work system design in healthcare. In:
answering the following questions: Carayon P, ed. Handbook of human factors and ergonomics in healthcare
and patient safety. Mahwah, New Jersey: Lawrence Erlbaum Associates,
N What are the characteristics of the person(s) performing
the tasks or involved in the work?
2007:61–78.
17 Sexton JB, Thomas EJ, Helmreich RL. Error, stress, and teamwork in medicine
and aviation: cross sectional surveys. BMJ 2000;320:745–9.
N What tasks are being performed and what are the
characteristics of the tasks that may contribute to safe or
18 Morey JC, Simon R, Jay GD, et al. Error reduction and performance
improvement in the emergency department through formal teamwork training:
evaluation results of the MedTeams project. Health Serv Res
unsafe patient care? 2002;37:1553–81.
N What in the physical environment can be sources of error
or promote safety?
19 Donabedian A. The quality of care. How can it be assessed? JAMA
1988;260:1743–8.
20 Donabedian A. The definition of quality and approaches to its assessment.
N What tools and technologies are being used to perform the
tasks and do they increase or decrease the likelihood of
Ann Arbor, MI: Health Administration Press, 1980.
21 Leape LL, Bates DW, Cullen DJ, et al. Systems analysis of adverse drug events.
JAMA 1995;274:35–43.
untoward events? 22 Battles JB, Lilford R. Organizing patient safety research to identify risks and
N What in the organization prevents or allows exposure to
hazard? What in the organization promotes or hinders
hazards. Qual Saf Health Care 2003;12(Suppl II):ii2–7.
23 Coyle YM, Battles JB. Using antecedents of medical care to develop valid
quality of care measures. Int J Qual Health Care 1999;11:5–12.
patient safety? 24 Brennan PF, Safran C. Patient safety. Remember who it’s really for. Int J Med
N How do the tasks, physical environment, tools/technolo-
gies, and organization facilitate or hinder the performance
Informatics 2004;73:547–50.
25 ABIM Foundation. ACP-ASIM Foundation, European Federation of Internal
Medicine. Medical professionalism in the new millennium: a physician charter.
of the individuals involved in the work system? Ann Intern Med 2002;136:243–6.
26 Leape LL. Error in medicine. JAMA 1994;272:1851–7.
27 Flynn EA, Barker KN, Pepper GA, et al. Comparison of methods for detecting
..................... medication errors in 36 hospitals and skilled-nursing facilities. Am J Health-
System Pharm 2002;59:436–46.
Authors’ affiliations 28 Gaba DM. Anaesthesiology as a model for patient safety in health care. BMJ
P Carayon, B-T Karsh, P Flatley Brennan, Department of Industrial and 2000;320:785–8.
Systems Engineering, College of Engineering, University of Wisconsin- 29 Carayon P, Gurses AP, Hundt AS, et al. Performance obstacles and facilitators
Madison, Madison, Wisconsin, USA of healthcare providers. In: Korunka C, Hoffmann P, eds. Change and quality
in human service work. Munich: Hampp Publishers, 2005:257–76.
P Carayon, A Schoofs Hundt, B-T Karsh, C J Alvarado, M Smith, 30 Carayon P, Schultz K, Hundt AS. Righting wrong site surgery. Jt
P Flatley Brennan, Center for Quality and Productivity Improvement, Comm J Quality Saf 2004;30:405–10.
College of Engineering, University of Wisconsin-Madison, Madison, 31 Sauter S, Lim SY, Murphy LR. Organizational health: a new paradigm for
Wisconsin, USA occupational stress at NIOSH (in Japanese). Japanese J Occup Mental Health
P Flatley Brennan, School of Nursing, University of Wisconsin-Madison, 1996;4:248–54.
32 Murphy LR, Cooper CL, eds. Healthy and productive work: an international
Madison, Wisconsin, USA
perspective. London: Taylor and Francis, 2000.
M Smith, Department of Population Health Sciences, School of Medicine 33 Kovner C. The impact of staffing and the organization of work on patient
and Public Health, University of Wisconsin-Madison, Madison, outcomes and health care workers in health care organizations. Jt
Wisconsin, USA Comm J Qual Improve 2001;27:458–68.

www.qshc.com
i58 Carayon, Hundt, Karsh, et al

34 Nelson A, Baptiste A. Evidence-based practices for safe patient handling and 39 Carayon P, Alvarado CJ, Hundt AS, et al. Employee questionnaire survey for
movement. Online Journal of Issues in Nursing 2004;9. assessing patient safety in outpatient surgery. In: Henriksen K, Battles JB, Marks E,
35 Keijsers GJ, Schaufeli WB, LeBlanc PM, et al. Performance and burnout in et al, eds. Advances in patient safety: from research to implementation. Rockville,
intensive care units. Work & Stress 1995;9:513–27. MD: Agency for Healthcare Research and Quality, 2005:461–73.
36 Carayon P, Hundt AS, Alvarado CJ, et al. Implementing a systems 40 Hundt AS, Carayon P, Springman SR, et al. Outpatient surgery and patient
engineering intervention for improved patient safety—example in outpatient safety—the voice of the patient. In: Henriksen K, Battles JB, Marks E, et al, eds.
surgery. In: Henriksen K, Battles JB, Marks E, et al, eds. Advances in patient Advances in patient safety: from research to implementation. Rockville, MD:
safety: from research to implementation. Rockville, MD: Agency for Agency for Healthcare Research and Quality, 2005:445–59.
Healthcare Research and Quality, 2005:305–21. 41 Carayon P, Alvarado C, Hundt AS. Reducing workload and increasing patient
37 Carayon P, Alvarado CJ, Hundt AS, et al. Patient safety in outpatient safety through work and workspace design. Washington, DC: Institute of
surgery: the viewpoint of the healthcare providers. Ergonomics Medicine, 2003.
2006;49:470–85. 42 Freidson E. Profession of medicine: a study of the sociology of applied
38 Carayon P, Alvarado C, Hsich Y, et al. A macroergonomic approach to knowledge. New York: Harper & Row, 1970.
patient process analysis: application in outpatient surgery, XVth Triennial 43 Flynn KE, Smith MA, Davis MK. From physician to consumer: the effectiveness
Congress of the International Ergonomics Association.Seoul, South Korea, of strategies to manage health care utilization. Med Care Res Rev
2003. 2002;59:455–81.

www.qshc.com