Patient Navigation: Development of a
Protocol for Describing What
Victoria A. Parker, Jack A. Clark, Jenniffer Leyson,
Elizabeth Calhoun, Jennifer K. Carroll, Karen M. Freund, and
Tracy A. Battaglia
Objective. To develop a structured protocol for observing patient navigators at work,
describing and characterizing specific activities related to their goals.
Data Sources/Setting. Fourteen extended observations of navigators at three pro-
grams within a national trial of patient navigation.
Study Design. Preliminary observations were guided by a conceptual model derived
from the literature and expert consensus, then coded to develop and refine observation
categories. These findings were then used to develop the protocol.
based on emergent themes. Using these codes, the team refined the model and con-
of navigator actions.
Findings. Navigator actions across a wide variety of settings can be categorized in a
matrix with two dimensions. One dimension categorizes the individuals and organi-
zational entities with whom the navigator interacts; the other characterizes the types of
tasks carried out by the navigators in support of their patients.
Conclusions. Use of this protocol will enable researchers to systematically character-
ize and compare navigator activities within and across programs.
Key Words. Continuity of patient care, social support, health services needs and
demands, case management, patient navigation
Challenges arise when people contemplate cancer screening. They multiply
when tests suggest a threatening disease and indicate the need for follow-up
investigations. And the challenges expand dramatically when such investiga-
tions reveal cancer, as people become ‘‘patients’’ in the complex realm of
referrals, consultants, examinations, decisions, and often arduous treatment
rHealth Research and Educational Trust
Health Services Research
Individuals who are socially and economically disadvantaged may find
cancer care all the more problematic (Baquet et al. 2005). These people are at
substantial risk of receiving inadequate care at each step of the cancer care
continuum: screening, diagnostic follow-up of suspicious results, treatment
when cancer is diagnosed, and survivorship surveillance. Moreover, the sys-
tems of care available to them, such as ‘‘safety net’’ institutions, are often beset
Patient navigation has emerged in the past decade in response to these
widely recognized disparities in cancer care. Health care advocates, policy
makers, and innovative health care organizations have called for the adoption
of patient navigation to assist patients and remedy inefficiencies in the pro-
vision of timely care (Freeman 2006; Jandorf et al. 2006). As a result, patient
navigation services have proliferated rapidly in recent years.
cancer disparities (Gabram et al. 2008). Currently, a large, multisite cooper-
ative study of patient navigation is being conducted. The National
Cancer Institute’s Patient Navigation Research Program (PNRP) is designed
to evaluate the effectiveness of navigation in improving timeliness of care
(i.e., time to follow-up of abnormal screening results and to completion of
treatment when cancer is diagnosed) and patient satisfaction (Freund et al.
2008). The nine cooperating studies of the PNRP provide a laboratory
their activities to patient outcomes. The first step in such an effort is to design
AddresscorrespondencetoVictoria A.Parker, D.B.A.,Assistant Professor, DepartmentofHealth
the Center for Organization, Leadership and Management Research, VA Boston Healthcare
System, Boston, MA. Jack A. Clark, Ph.D., Associate Professor, is with the Center for Health
Quality, Outcomes, and Economic Research— —VA HSR&D, Edith Nourse Rogers Memorial
Veterans Hospital, Bedford, MA and the Department of Health Policy and Management, Boston
University School of Public Health, Boston, MA. Jenniffer Leyson, M.A., Senior Qualitative Data
Analyst, is with the Department of Health Policy and Management, Boston University School of
Public Health, Boston, MA. Elizabeth Calhoun, Ph.D., Associate Professor and Senior Research
Scientist, is with the Division of Health Policy and Administration, University of Illinois School of
Public Health, Chicago, IL. Jennifer K. Carroll, M.D., M.P.H., Assistant Professor, is with the
Department of Family Medicine and Oncology, University of Rochester Medical Center, Family
Medicine Research Programs, Rochester, NY. Karen M. Freund, M.D., M.P.H., Professor of
Medicine and Epidemiology, and Tracy A. Battaglia, M.D., M.P.H., Assistant Professor of Med-
icine and Epidemiology, are with Women’s Health Unit, Section of General Internal Medicine,
Department of Medicine and Women’s Health Interdisciplinary Research Center, Boston Uni-
versity School of Medicine, Boston, MA.
Observing Patient Navigation 515
procedures for systematically observing navigators’ activities. This paper
reports on the development of a protocol for observing what navigators
Despite continued advances across the spectrum of cancer care (Brenner,
Gondos, and Arndt 2007), the distribution of these advances remains uneven.
They are less likely to be enjoyed by those segments of our society defined by
minority racial and ethnic status, low income, and limited health insurance
(Weir et al. 2003; Shavers, Fagan, and McDonald 2007). Disparities in cancer
care are persistent and may in some instances actually be widening (Ries et al.
2007). Inequitable outcomes may result from, among other factors, well-
at-risk populations (Chang et al. 1996; Peterson, Han, and Freund 2003;
Battaglia et al. 2007).
Several seminal reports (Smedley et al. 2003; Weir et al. 2003) have
highlighted the barriers to cancer care inherent to socioeconomic disadvan-
tage. Patient navigation is a community-based approach to reducing these
barriers (Dohan and Schrag 2005; Battaglia et al. 2007; Ell et al. 2007;
Ferrante, Chen, and Kim 2008). Guided both by principles of disease
management and by cultural sensitivity, navigators are responsible for iden-
to their receiptof that care(Vargaset al.2008). Navigation programs also seek
to remedy systemic barriers to care within organizations delivering care.
Patient navigation services address barriers by assigning trained supportive
staff who track patients and assist them in completing their diagnostic and
treatment care, while also advocating for solutions to systemic causes of those
Navigation programs are usually funded through local resources
or foundation support because insurers do not reimburse this care. Local
innovation results in tremendous variability in program structures and activ-
ities. Trailing behind these developments is a small, but growing body of
research documentingtheefficacy ofnavigation (Battagliaetal.2007; Ellet al.
provide evidence that navigation is effective, the key components of a suc-
cessful navigation program are not well understood.
516 HSR: Health Services Research 45:2 (April 2010)
Problems often arise in the evaluation of complex innovations such as patient
navigation because the interventions have not been fully defined and devel-
oped (Campbell et al. 2000). Early evaluations of innovative programs often
simply assume either that the intervention is in place as planned or that some
variation is acceptable (Eccles et al. 2003). However, such assumptions lead
evaluators to overlook the effects of variation. And in the case of patient
navigation,alackofinformationabout variationindefinition, style,and scope
could lead to inaccurate conclusions about its effectiveness.
There is no generally accepted definition of patient navigation. Review-
ing 56 articles published before early 2004, Dohan and Schrag (2005) identify
two types of definitions: ‘‘service focused’’ and ‘‘barrier focused.’’ Service-
focused definitions attend to activities such as connecting individuals to re-
(2005) criticize these definitions as nonspecific: such activities could be, and
often were, performed by other providers as part of their duties. Barrier-
focused definitions, they argue, attend to activities that identify and remove
impediments preventing patients from moving through screening, diagnostic
follow-up, and treatment. Furthermore, these responsibilities were distin-
guishable from those usually assigned to social workers, case managers, com-
munity outreach workers, and health advocates. The latter roles, they argue,
were typically proactive, providing education and counseling services, while
navigation was essentially reactive to emergent impediments to care.
However, defining patient navigation in terms of resolving barriers for
individual patients may be too constricting. While service-oriented definitions
could blur distinctions between navigators and other service providers, fo-
cusing only on what navigators do in relation to barriers facing individual
patients may obscure a variety of related activities they perform. They may
tweak organizational practices to expedite patient care, develop local re-
sources for multiple patients, and build cooperative relationships with and
among clinic staff that facilitate more efficient movement of patients through
systems. Thus, while staff other than navigators may facilitate patient access to
services, to define patient navigation work solely in terms of barrier reduction
risks artificially excluding other patient navigation functions.
We sought to avoid such exclusions. Informed by research to date, we
sometimes externally——to facilitate patients’ receipt of care from providers.
This framing gives attention to the fact that navigators often must involve
Observing Patient Navigation517
others in their work, which, in turn, suggests that navigators’ networks of
relationships with these ‘‘others’’ might be essential to achieving their objec-
in terms of tasks and networks: navigators do things for patients by working with
patients and other actors in both the social network of the organization itself and the
community in which the organization resides.
Social networks have been described as ‘‘patterns of relations joining
actors’’ (Marsden 1990). Keating et al. (2007) use the social network approach
to understand patterns of advice-giving and -following within a primary care
practice. Earlier navigation research emphasized the pivotal role of the nav-
igator in helping patients access necessary services, which suggests that partof
navigation is knowingto whom to go for specific support. Thus, to understand
what navigators do, we must understand their patterns of relations with others
who provide services that facilitate screening, diagnosis, and/or treatment.
Underscoring the importance of understanding these networks is a recent
analysisofearlypatientnavigation programsthatdefinespatientnavigation as
a system, rather than a person (Vargas et al. 2008). The network framework is
useful in bridging the service-focused/barrier-focused dichotomy: it usefully
attends to the question of how navigators achieve their efforts on behalf of
patients. Whether obtaining services proactively or in response to a specific
barrier, navigators engage others in their networks to find, arrange, and seek
reimbursement for those services.
The social network concept illuminates part, but not all, of the scope of
patient navigation. We need also to describe the activities of navigators. Task
analysis, with its emphasis on interaction of persons and environment, offers a
useful complement. Task analysis is concerned with identifying the goal of a
task, the criteria for reaching that goal, and the relevant resources and con-
straints. It also emphasizes that task-directed actions are determined by both
the person carrying them out and the relevant environment (which presum-
ably includes the person’s social network), and that people develop personal
expertise in how to accomplish their tasks (Norros and Nuutinen 2002). Nav-
igators need to build a working knowledge of the tasks they must perform and
a network of contacts to support their actions.
Development of the observation protocol was grounded in a qualitative study
of PNRP navigators at work. We began by surveying the nine PNRP sites to
518 HSR: Health Services Research 45:2 (April 2010)
characterize structural attributes of each program that would define the con-
site (facilities, geographic location, and populations served), its size, and the
spectrum ofnavigationservicesoffered.Basedon thesefindings, weselecteda
convenience sample of three pilot programs that represented the diversity of
settings and approaches to navigation implemented at the nine independent
We also developed a preliminary guide for observing navigators at
work. This guide was designed to enable the collection of comparable field
observations of what navigators do (tasks) and the people and entities with
whom they interact (networks) in accomplishing those tasks. The resulting
data were analyzed to develop a comprehensive, yet simple observation pro-
tocol for observation of navigators across all PNRP sites and elsewhere.
The nine programs in the PNRP were selected through a competitive national
process and designed according to general program criteria set by the NCI
(Freund et al. 2008). The NCI sought to maximize diversity within this group
to assess the usefulness of patient navigation across a range of settings. Thus,
the programs differed in many respects, but they met common requirements
regarding navigator training, patient population, patient criteria for inclusion,
and collection of data. The programs address different combinations of four
cancers——breast, cervical, prostate, and colorectal——where navigation would
likely have a detectable effect in facilitating follow-up of suspicious screening
resultsand completionoftreatment(Freund etal.2008).Eachofthesecancers
is associated with a distinct patient population and pattern of care.
The locations where navigators work and interact with patients, medical
providers, and others vary: hospital evaluation clinics, inpatient wards, and
treatment units; primary care clinics; and community health centers. The
scope of navigator involvement relative to the cancer care continuum also
varies; some navigate in all phases, while others navigate only from diagnosis
through treatment. At some sites, navigators focus on case finding, while at
others, the focus is on supporting patients through treatment. The number of
navigators employed varies, as does navigators’ involvement in competing
clinical or administrative responsibilities. Some navigators were hired and are
supervised directly by the research program; others were hired by the clinical
care sites themselves as a subcontract to the research program. Finally, the
programs vary with respect to the professional background and training of
Observing Patient Navigation519
navigators: some use navigators with clinical training and credentials, while
others use ‘‘lay’’ navigators selected for congruence with the target patient
In selecting the pilot sites, we excluded one program due to restrictive
local access requirements. Site 2 was selected because it is the research team’s
home site and had a small, longstanding navigation program that predates the
along the dimensions outlined above. Information about all eight programs is
presented in Table 1.
Preliminary Data Collection
We enlisted other PNRP investigators to collaborate at their respective sites.
Multisite, multi-investigator qualitative research required developing a com-
mon protocol. We designed a semistructured observation guide to support
collection of comparable data across all sites. During a 1-day training, ob-
servers from each site were directed to describe specific actions of navigators
and note the following: (1) the approximate duration of actions; (2) the parties
with whom navigators interacted; (3) whether interactions were in person or
via phone/email; (4) the relevance of the action to navigation.
Observers also were directed to query navigators about their actions at
moments that would not interrupt observed activities. Navigators were asked
to explain the relevance of observed activities to particular navigation issues
and challenges, including (1) actions navigators took to develop a relationship
with the patient and others relevant to the patient’s case; (2) the role of others
who were consulted for advice, direction, or assistance; (3) the initiation and
extent of an interaction with a person; (4) the nature of the problems being ad-
dressed. Thus, while we focused on directly observable behaviors, we also
explored navigators’ reflections on the scope of and rationale for their actions.
Each navigator at each site was observed at least twice. Observations
were scheduled in consultation with navigators to bestcapture the variation in
their workflow, so the lengths of the sessions were defined by the ways
navigators scheduled their own activities. For example, if a navigator worked
with a specific provider seeing patients for screening follow-up, the observa-
tion ran the length of the provider’s clinic. Most observations lasted about 4
At the end of each observation, navigators were asked further questions
to characterize the representativeness of the actions just observed:
? How did you decide what to spend time on today?
520HSR: Health Services Research 45:2 (April 2010)
Type of Site(s)
Where Navigator Is Based
(Set at Site)
Navigators Have Other
health centers (CHCs)
High school diploma
High school diploma
CHCs, VA outpatient
High school diploma
BS/BA or LPN
Specialty clinics at
High school diploma,
2 years of work
experience in medical
clinics and treatment
units, primary care
Completion of local
and in-house training
BA/BS and relevant
PNRP and local training
Observing Patient Navigation521
? Howdidthisactionrelatetoprioractivitiesfora particularpatientor
patients in general?
? Does this action reflect any specific strategies for assessing/obtaining
what the patient needs?
? What will you do next?
? Did the problematic situation/barrier get resolved?
? How typical a day was this for you?
The observation guide was refined in a 1-day meeting of the observers,
followed by regular conference calls. Issues of reliability were addressed by
comparing observations within and between the three sites. Apparent dis-
crepancies in the quality of observations were discussed and resolved by de-
veloping a consensus among all observers.
Using the observation guide, investigators conducted a total of 18 comparable
observations of nine navigators working in three program sites of the PNRP.
The analysis focused on refining the observation protocol. Data from all three
sites were compiled by the project PIs. Fieldnotesfrom each observation were
imported into software supporting text-based analysis, HyperRESEARCH
(ResearchWare Inc. 2008).
The analysis was informed by the general approach of grounded theory
methodology (Glaser and Strauss 1967). The first three authors each reviewed
separately one set of fieldnotes from each of the three sites, identifying themes
that characterized the activities reported. The team then met to compare
descriptive codes and reach consensus on code definitions before coding the
rest of the fieldnotes. The primary aim of this analysis was to define a com-
prehensive, yet parsimonious set of categories that would enable other ob-
servers to reliably categorize navigator behavior.
Once these categories were developed, we also drafted observation in-
structions to guide subsequent field observations of the work of navigators.
Drafts of the analytic categories were developed by the Site 2 researchers and
shared with the Sites 4 and 7 observers for substantive critique and revision in
a day-long research team meeting, which yielded the final protocol reported
522 HSR: Health Services Research 45:2 (April 2010)
The three sites provided a wide array of contexts for observing naviga-
tion. They differed with respect to the scope of the navigation program, the
phases of cancer care addressed (i.e., screening, diagnosis, or treatment),
the history/longevity of the program, the emphasis placed on various nav-
igator responsibilities, and the background (e.g., clinical, survivor, cultural/
ethnic) of the navigators, as well as their physical and organizational loca-
care, or treatment clinics). These contextual differences appeared to influ-
ence, to an undetermined extent, what navigators do. Thus, we sought an
observation protocol that would reliably capture activities in these diverse
Domains of Navigator Behavior
Guided by the concepts of task and network, we defined five categories each of
navigator tasks and social networks. The task categories include navigating
with a patient, facilitating for a patient, maintaining systems for all patients,
documenting/reviewing actions, and other tasks. The five network categories
include patient(s), clinical provider(s), nonclinical staff, formal and informal
support, and medical record systems. Each of these categories is defined,
described, and illustrated below.
Task Categories. Navigating tasks consist of identifying and mitigating barriers
with patients. They include telling (explaining when and where biopsy will be
done, describing what it will be like); inquiring (asking about barriers to
attending the appointment, exploring the patient’s concerns); supporting
(listening to fears about treatment); and coaching (discussing questions that
need to be asked at next appointment and how to ask them).
Facilitating tasks are performed for a specific patient. They include
finding (locating current patients and ensuring that they will come to
appointments); coordinating team communication (ensuring the entire care
team is aware of the next steps); integrating information (ensuring that
different types of patient data are documented and shared as needed); and
seeking collaboration (enlisting other providers in addressing the patient’s
Maintaining systems tasks support all patients. They include identifying
potential patients (reviewing lab results to note patients who need follow-up);
building networks and referral routines (meeting with clinicians to explain
Observing Patient Navigation523
navigator role and discuss referral criteria); and reviewing cases (checking on
ticklers and open issues).
Documenting activities and reviewing information constitute another
major navigator task. They include recording navigator actions (recording
steps taken with or on behalf of the patient in the patient’s medical record
or a separate navigation file); handling test results (retrieving and entering
patient data from labs, radiology, or other sources); and processing other
necessary information (recording information or activities relevant to
Other activities are those apparently unrelated to navigation. It was
important to capture all network interactions, even when their relevance to
navigation was not apparent. For example, many navigators have other
distinct roles unrelated to navigation; documenting these other activities will
help in understanding how the navigator role fits in with other roles, both
formally and informally. This category includes research-related activities,
such as consenting patients, providing clinical back-up, activities unrelated to
navigation (interpreting for nonnavigated patients), and socializing (having
informal conversation with co-workers).
Network Categories. Navigators may interact with a specific patient, such as
when phoning the patient with information about an upcoming diagnostic
Navigators may also interact with providers, both within and
outside their immediate location. For example, s/he might speak with the
physician to confirm the meaning of a test result before discussing it with
Nonclinical staff, such as receptionists or administrators coordinating
insurance, represent another group with whom the navigator may interact.
People who provide supportive services, either formally (social workers,
translators, transportation staff) or informally (friends, family) within or
outside the facility are another group with whom navigators interact.
The final category——paper or electronic medical record systems——could be
perceived as merely a means to communicate with members of the other four
network categories, and it does function in that way. However, our
preliminary observations indicated that, in the eyes of the navigator, the
medical record itself takes on some of the qualities of a person, in that it needs
to be informed and/or consulted before other actions are taken. This
observation is consistent with those of many studies of human–computer
interaction (Turkle 2003).
524HSR: Health Services Research 45:2 (April 2010)
Observation Protocol Refinement
The current observation protocol incorporates solutions to problems encoun-
tered in the field using the preliminary observation guide. Initially, observers
were required to take continuous notes, recording the duration and mode of
patient on whose behalf the activity wastaken, plus descriptive narrative. This
recording burden proved too onerous in the field: recording all observed
activities not only interfered with the primary goal of noting tasks and social
networks used by the navigators, but it did not produce more useful data.
Based on this early finding, two important changes were made to the
observation protocol: activities were observed in 15-minute intervals, and
coding focused on the primary activity of each interval. This time sampling
methodology facilitates detailed reporting of navigator activities without at-
tempting to capture everything that occurs during an observation. Observers
start a new form every 15 minutes, focusing notes and coding on the nav-
igator’s primary activity during that period. Thus, each hour of observation
time yields four distinct chunks of description and activity coding. This sam-
pling interval provides some sense of the relative proportion of a navigator’s
time spent on different activities, while allowing observers to record more
detailed notes about the main activity. This approach necessarily involves
some observer judgment: sometimes a navigator tackles multiple short tasks
during a single 15-minute interval. In such instances, observers were in-
structed to either group-like tasks into a single entry (making appointment
or focus on the first activity during the time period.
Through discussion, a five-by-five matrix emerged, with tasks on the
vertical axis and social networks on the horizontal axis. The observation form
itself was redesigned to incorporate on a single page both this simple matrix
and an open area for handwritten fieldnotes (see Figure 1).1
After field testing, several additional refinements were made. Certain
combinations of tasks and networks cannot occur. For example, the task of
navigating can be performed only with a patient, while reviewing a patient’s
file can be done only with the medical record. To further simplify the form,
matrix cells representing combinations that cannot occur are blacked out.
Also, observers at some sites reported that a significant amount of nav-
igation is carried out by telephone, leaving and returning voicemail messages.
Therefore, for each observation of a navigator action that involves contact
with a patient, the observer also notes whether the interaction is synchronous
Observing Patient Navigation525
by recordingeither ‘‘S’’or ‘‘A’’ in the appropriatecell. For all other cells in the
matrix, a checkmark is used.
Finally, because some observed activities may involve more than one
person or task, observers are encouraged to mark more than one cell if that
* Begin a new form every 15 minutes *
For each task observed, mark the appropriate empty cell. (Darkened cells are not in use.)
(S or A)
Med rec /
If you cannot decide what box to mark, check here: _____
Copyright 2009 Trustees of Boston University. Reproduced here with permission.
Figure1:Protocol Data Collection Form
526HSR: Health Services Research 45:2 (April 2010)
bestreflects what they areseeing.Forexample,if the navigatoraccompaniesa
patient to a physician visit, the observer puts an ‘‘S’’ in the cell representing
‘‘navigate/patient’’ and a checkmark in the box representing ‘‘facilitate/
The matrix supports coding of real-time activities as they occur, but we
realized the need for simultaneous, structured, narrative fieldnotes, as well.
Hence, we developed observation guidelines directing the observer to note
relevant contextual factors, such as the location of navigation activity, the
navigator if known, and the navigator’s other roles (if any) in the organization.
This matrix facilitates rapid categorization of tasks and networks, allow-
ing the observer to concentrate on writing narrative description that will doc-
ument important information about context and content that cannot be fully
captured by the matrix. The observer is encouraged to ask the navigator
why.2Observers also are asked to record their impressions about interactions,
clearly identifying these notes as their perceptions. For example, the observer
might write ‘‘navigator and patient embraced warmly and seem to know each
other well.’’ While these impressions are particular to specific observers, they
nevertheless add richness to the description.
Patient navigation represents an emerging innovation in care adapted to a
variety of specific local contexts. Capturing local adaptations is crucial to
meaningfully assess the efficacy of navigation across different sites. The pro-
tocol we have developed reflects a plausible, generic definition of navigation
that has been found thus far to be applicable in multiple contexts. We expect
the protocol to be useful in capturing the existing variation in navigation
programs, and we will use it for this purpose as our research program goes
forward. This type of data is essential to inform both research and practice.
While this protocol represents an important step, it certainly does not
capture every detail of navigators’ actions. For example, while interactions
with other providers are categorized and noted, the protocol does not note
how extensive or collaborative they might appear to be. Data from narrative
fieldnotes will compensate for this limitation, while also providing the poten-
tial for further revisions to the protocol based on emerging patterns in these
Observing Patient Navigation 527
This protocol is grounded in a qualitative study of navigation in three Download full-text
sites of the PNRP. As such, it has a measure of validity, yet its validity requires
further examination through application. The current definitions of the task/
network categories may reflect the particular realities of the sites we have
Likewise, variation in organizational, political, and community contexts in
which operational navigation programs are developed may require modifi-
cations to this protocol. Moreover, while this protocol enables the systematic
observation of what navigators do, it is only part of a comprehensive method
for evaluating the processes and outcomes of patient navigation. The mix of
patients served, their resources and ability to access care; the types of health
problems for which access is needed; and the specific array of health services
and providers for which navigation is needed must be taken into account in
evaluating the effectiveness of what navigators do, as captured by this, or any
of this writing, 89 observations have been completed, and none has presented
activities that fall outside the task/network categories described above. Quan-
titative and qualitative analyses of the data collected will enable the research
team to characterize variation, both within and across sites, in navigator tasks,
networks, and emphasis.
Protocol development thus far has illuminated important dimensions
along which navigation programs may vary. By accurately characterizing this
variation, researchers should be better able to interpret variation in patient
outcomes associated with different navigation programs. While this protocol
does not provide information on program effectiveness, it provides important
process information that may help explain the connections between naviga-
tion context and outcomes. Of course, outcomes also may be affected by the
actions of other health care providers and advocates, which may overlap with
the actions of navigators. It is important not to over-attribute outcomes to the
actions of specific navigators, but rather to keep analysis at the programmatic
level. In other words, while patient navigator actions may often be directed at
specific patients, navigation is properly conceptualized as a systemic inter-
vention that changes how care is delivered. Thus, any change in outcomes
navigator, whose actions have been influenced by the presence of the nav-
igation program. The information obtained from further use of this protocol
528 HSR: Health Services Research 45:2 (April 2010)