Finding the answers in primary care: information seeking by rural and nonrural clinicians.

Abstract

Information systems for rural practice may assume that rural clinicians have different information seeking, but studies have not directly compared rural and nonrural information needs using common methodology.
Compare rural and non-rural: 1) information needs; 2) information seeking; 3) effectiveness of information seeking; and 4) use of information resources.
Observation and interviews during one half-day of office practice; telephone follow-up 2-10 days later.
Primary care physicians (39), nurse practitioners (42), and physician assistants (22) in ambulatory practices in rural and nonrural Oregon.
1) number of questions asked, 2) number of questions pursued, 3) number of questions answered), and 4) use of knowledge resources.
Rural clinicians practiced in smaller groups, but were otherwise similar to nonrural clinicians. During half-day interviews, clinicians cared for an average of 8.2 patients (95% CI 7.5 - 8.8) and asked an average of 0.83 questions per patient seen (95% CI 0.73 - 0.92). At follow up, they had pursued an average of 47% of their questions (95% CI 40 - 53%), and reported being successful in finding an answer to 77% of those they pursued (95% CI 70 - 84%). There were no statistically significant differences between rural and nonrural clinicians for any of these variables.
Rural and nonrural clinicians had similar information needs, information seeking, knowledge resource use, and effectiveness at finding answers to their questions. Human consultants, digital resources, and library-based resources were less available, but these differences in availability had little impact on their use.

Full text

Page 1

1133
MEDINFO 2004
M. Fieschi et al. (Eds)
Amsterdam: IOS Press
© 2004 IMIA. All rights reserved
Finding the answers in primary care: Information seeking by rural and nonrural
clinicians
Paul N. Gorman, Patricia Yao, Veena Seshadri
Department of Medical Informatics and Clinical Epidemiology, Oregon Health & Science University, Portland, Oregon, USA
Paul N. Gorman, Patricia Yao, Veena Seshadri
Abstract
Background: Information systems for rural practice may assume
that rural clinicians have different information needs, but stud-
ies have not directly compared rural and nonrural information
needs using common methodology. Objective: Compare rural
and non-rural: 1) information needs; 2) information seeking; 3)
effectiveness of information seeking; and 4) use of information
resources. Design: Observation and interviews during one half-
day of office practice; telephone follow-up 2-10 days later. Par-
ticipants & Setting: Primary care physicians (39), nurse practi-
tioners (42), and physician assistants (22) in ambulatory
practices in rural and nonrural Oregon. Measures: 1) number of
questions asked, 2) number of questions pursued, 3) number of
questions answered), and 4) use of knowledge resources. Re-
sults: Rural clinicians practiced in smaller groups, but were oth-
erwise similar to nonrural clinicians. During half-day
interviews, clinicians cared for an average of 8.2 patients (95%
CI 7.5 - 8.8) and asked an average of 0.83 questions per patient
seen (95%CI 0.73 - 0.92). At follow up, they had pursued an av-
erage of 47% of their questions (95% CI 40 - 53%), and reported
being successful in finding an answer to 77% of those they pur-
sued (95% CI 70 - 84%). There were no statistically significant
differences between rural and nonrural clinicians for any of
these variables. Conclusions: Rural and nonrural clinicians had
similar information needs, information seeking, knowledge re-
source use, and effectiveness at finding answers to their ques-
tions. Human consultants, digital resources, and library-based
resources were less available, but these differences in availabil-
ity had little impact on their use.
Keywords:
Information needs, information seeking, rural primary care
Introduction
Information systems or services are often developed specifically
to address the needs of rural medical practice. Examples include
traditional outreach services such as circuit librarians, as well as
technology-based programs such as online literature searching.
The rationale for these systems has been the expectation that ru-
ral clinicians have different information needs and patterns of in-
formation seeking than their nonrural colleagues.
The intended user of these systems appears to be based on com-
mon assumptions about rural practice. For example, the "typi-
cal" rural physician is expected to be older, busier, and more
familiar with their patients and community, based on years of ex-
perience in a stable practice setting. Rural physicians are thought
to have fewer questions about patient care, less access to infor-
mation sources, and less time or inclination to use the sources
they have. The result is a physician who, compared to his urban
counterparts, has greater unmet information needs and may be
isolated and out of date with respect to recent advances in med-
ical knowledge.
Empirical data to support these ideas about rural clinicians are
limited, and there are important shortcomings to the data that do
exist. First, most studies of rural physicians’ information seeking
do not directly compare them to urban physicians. (1) Second,
most information seeking studies depend on self-report, but the
landmark Covell study showed the inadequacy of self-report:
physicians underestimated their information needs and misper-
ceived their information seeking and use of knowledge resources
(2). Third, studies of physicians' information needs are difficult
to compare because of differing definitions of key variables, se-
lection of subjects and settings, and data collection methods (3,
4). Fourth, most such studies focus on physicians, but nurse
practitioners and physician assistants are an increasingly impor-
tant source of primary care in rural areas, and their needs also
must be considered.
We analyzed data from a study of information seeking by Ore-
gon primary care clinicians (PCCs) to answer three questions
about rural clinicians when compared to nonrural clinicians: 1)
Are their information needs different? 2) Is their access to infor-
mation resources different? and 3) Is their information-seeking
behavior different?
Materials and Methods
Operational Definitions
For purposes of this study, rural means more than ten miles from
a town of 30,000 or greater, as defined by the Oregon Office of
Rural Health. Primary care clinician (PCC) refers to physicians,
nurse practitioners, and physician assistants engaged in family
practice, internal medicine, pediatrics, or equivalent specialties,
practicing at least half time in direct care and no more than half
time in a special interest area. While there are important reasons
to consider a broader view of information needs, for purposes of
this study information need refers to clinical questions articulat-
ed by a PCC in the course of patient care. Information seeking

Page 2

PN. Gorman et al.
1134
refers to activity undertaken to satisfy a perceived need for infor-
mation. (5)
Subjects, Setting, and Sampling
A database of clinicians in primary care was compiled from Or-
egon Board of Nursing and Board of Medical Examiners data,
and classified as rural or nonrural based on a postal code list pro-
vided by the Oregon Office of Rural Health. The sampling strat-
egy was to recruit 100 participants, with sufficient numbers in
each group to permit meaningful comparison, while minimizing
unnecessary bias. Letters of invitation, and an initial survey were
mailed to all primary care nurse practitioners (591), all primary
care physician assistants (132), all rural primary care physicians
(806), and a random sample of 800 nonrural primary care physi-
cians. Respondents who were eligible and willing to participate
(529) were stratified by location and by professional degree,
sorted in random order, and contacted sequentially by telephone
until roughly equal numbers of in each professional group were
included.
Data Collection
Data was collected in a postal questionnaire and paired inter-
views modified from the method of Covell. (2) The question-
naire covered demographic and practice setting data including
knowledge resource data availability. Interviews were conduct-
ed by medical librarians, graduate students in qualitative meth-
ods, and nurses or nurses in training, following two half days of
training to minimize variation in data collection.
Interviews during one half day of typical office practice, began
with a pre-session briefing.. After each patient was seen, the cli-
nician was asked, "Do you have any questions about how best to
care for this patient?" Each question was recorded, along with its
clinical context. Interviewers also recorded any observed infor-
mation seeking, such as phoning consultants or looking in text-
books. After hours, clinicians verified each question, and were
told to expect a follow-up call but were not informed of its pur-
pose.
Telephone interviews took place two to ten days after office in-
terviews. The clinician was asked whether he or she had had a
chance to pursue each question and if so, whether an answer was
found, the source of the answer, their estimate of the time spent
pursuing it, and whether the answer had, in their judgment, had
an impact on the patient's care.
Data Analysis
Rural-nonrural comparisons were made using Student's t test for
normally distributed data, the rank-sum test for non-normally
distributed data, and Fisher's exact test for categorical data.
Analyses were performed using JMP and SAS software (SAS In-
stitute, Cary, North Carolina, USA).
Results
Clinicians and Practice Settings
Demographic variables are summarized in Table 1. One hundred
three clinicians were interviewed, with a slightly larger propor-
tion of physicians in the nonrural group. The age distribution of
rural and nonrural clinicians was similar, with a mean age of 44
years and a slight but non-significant trend toward younger age
among rural clinicians. There were approximately equal num-
bers of women both groups. This is in part a reflection of the
stratification by professional degree, with roughly equal num-
bers of nurse practitioners, who are overwhelmingly women,
and physicians and physician assistants, who are predominantly
men.
Rural and nonrural clinicians were similar with respect to their
years of experience and professional activities. Overall, the me-
dian years since completing training was 12, with no significant
difference between rural and nonrural groups. Despite having
over a decade of practice experience, clinicians had been in their
current practice settings a median of only four years, again with
no significant difference between the rural and nonrural groups.
Participation in other professional activities that might influence
information needs or information seeking, such as research, pub-
lications, teaching, and administrative duties, was not different
between the two groups.
Table 1: Clinician and Practice Setting Characteristics.
As indicated in Table 1, rural clinicians were more often in solo
or small group practice than nonrural clinicians. Regardless of
group size, both rural and nonrural clinicians tended to practice
in groups with others in the same specialty. Thus although rural
PCCs had fewer clinicians with them in their practice sites, there
was no difference in the spectrum of specialties, and presumably
specialized knowledge, on hand to meet their information needs.
not
rural
55
rural
48
total
103
p
clinicians inter-
viewed
mean age 44.3 43.6 440.63
number of women 2824520.93
profes-
sional
degree
MD 2316 39 0.59
NP 222042
PA 1012 22
median yrs experi-
ence
median yrs onsite
13 10 120.09
5340.25
any “teaching” (n)31 28590.84
any “research” (n)437 0.84
any publications (n)96150.58
any administration
(n)
clinicians
at practice
site
24 25 490.39
Solo279
0.0002
2-415 26 41
5-717 1330
8-14141 15
15-40718
specialties (median)
patients per half day
2 2 2 0.349
0.4178.437.888.17

Page 3

PN. Gorman et al.
1135
Rural and nonrural PCCs were equally busy: caring for a mean
of 8.43 patients for nonrural clinicians (95% CI 7.96-8.9) and
7.88 (95% CI 7.38-8.38) for rural clinicians, during their half-
day interviews.
Table 2: Information Needs and Information Seeking
NR = nonrural NR
means for entire group
questions asked6.43
questions pursued2.4
questions answered1.9
answers with impact1.64
frequencies per clinician
questions per patient0.82
proportion pursued0.43
proportion answered0.77
proportion w/ impact0.63
information seeking time and effort (medians)
hours till pursuit .75
search time 8.5
sources used
Information Need and Information Seeking
Table 2 summarizes the information needs and information
seeking of study clinicians. On average they asked 6.2 clinical
questions per half day (95% CI 5.4 – 7.0), an average rate of 0.83
questions per patient seen (95%CI 0.73 - 0.92). By the time of
telephone follow up, they had pursued on average 2.66 questions
(95% CI 2.43 - 2.91), or 47% of those they had asked (95% CI
40 - 53%). They reported success in finding an answer to an av-
erage of 2.09 questions (95% CI 1.7 – 2.5) for an average of 77%
of those they had pursued (95% CI 70 - 84%), and they expected
or perceived that the information had an impact on their patient
for 68% of the questions they pursued (95% CI 61 - 75%). There
was no statistically significant difference between rural and non-
rural clinicians for any of these variables. Of 275 questions pur-
sued, 42.4% were pursued immediately or within 15 minutes,
66.7% were pursued within 12 hours, and 80% were pursued
within 24 hours, with no significant difference between rural and
nonrural clinicians. Clinicians usually consulted one or at most
two information sources to find answers to their questions, and
spent 10 or at most 15 minutes doing so. Rural clinicians more
often reported using 2 or more information sources and reported
taking longer, on average, to find answers to their questions than
did nonrural clinicians. These differences were statistically sig-
nificant, but small, and it is uncertain whether they are due to dif-
ferences in the accessibility of knowledge resources or to
differences in the nature of the questions being pursued.
Use of Knowledge Resources
Each clinician reported the source used to answer each questions
he or she had pursued, and these sources were grouped by the au-
thors into the categories listed in the Table 3. Both rural and non-
rural PCCs turned most often to human sources of information
and to local print resources to answer their questions. Clinicians
in both groups sometimes obtained answers from organizational
and institutional sources and from patient specific data sources.
Clinicians in both groups made use of digital and library resourc-
es infrequently, if at all.
Differences in access to knowledge resources were not reflected
in differences in their use. Rural clinicians had significantly less
access to human sources of information, especially consultants
in other specialties (6), but they relied on them more often to ob-
tain answers to their questions. Similarly, although rural clini-
cians had less access to digital and library resources, they used
them with equally low frequency. Although the numbers are too
small for meaningful comparisons of the use of most of the re-
sources individually, some of the differences are interesting, if
inconclusive. For example, nonrural clinicians more often
sought insurance information, perhaps reflecting a greater vari-
ety of insurers in metropolitan areas, where the need to deter-
mine whether a test or treatment is covered by the patient's health
plan may be as important as whether the it is recommended by
experts.
Discussion
Findings
This study addressed three questions about nonrural clinicians
compared to rural clinicians: 1) Are their information needs dif-
ferent? 2) Is their access to information resources different? 3) Is
their information seeking behavior different? Overall, we found
that the similarities between rural and nonrural PCCs out-
weighed the differences. Rural clinicians had less access to cer-
tain information resources, but we found little difference in their
information needs, information seeking, knowledge resources
use, or effectiveness at finding answers. Both rural and nonrural
PCCs had about eight questions for every ten patients they cared
for, pursued about 47% of these questions, found answers to
77% of the questions they pursued, and reported that for 68% of
the questions they pursued the information retrieved would have
an impact on patient care.
Differences in availability of knowledge resources did not trans-
late, for the most part, into differences in their use. Both rural
and nonrural clinicians relied most heavily on local, immediately
available knowledge resources and on human sources of infor-
mation. Neither group made use of library-based resources or
digital resources with any frequency. Rural clinicians had less
immediate access to human sources such as consultants, but they
obtained answers from them more often, the difference in avail-
ability being mitigated by an older information technology: the
telephone. Rural clinicians had less access to library and digital
resources, but neither group used these resources with any fre-
quency.
Curley, et al. reported a strong preference among family physi-
cians for human information sources and sources with high ac-
cessibility and availability (7). The present study, and other
observational studies of clinician information seeking (8-10), are
consistent with their model, with two important qualifications.
First, as they noted, accessibility means more than physical dis-
tance, and includes elements of familiarity and usability:. The
fact that an resource is physically present does necessarily mean
it is more accessible, and the proximity of a resource (such as
digital media or library resources in nonrural sites) or its absence
(such as specialty consultants in rural sites) is not by itself suffi-
cient to predict resource use. A busy urban clinician may have
Rtotalp
5.97
2.97
2.3
2.14
6.20
2.7
2.09
1.87
0.58
0.36
0.53
0.19
0.84
0.52
0.76
0.73
0.83
0.47
0.77
0.68
0.99
0.16
0.93
0.17
2
12
1
10.65
0.02
0.04
10
11

Page 4

PN. Gorman et al.
1136
no more access to associates in the next office suite than does a
rural clinician whose associates are in the next county.
Table 3: Use of Knowledge Resource
NR = nonrural
Human Sources
Consultant (other specialty)
Colleague (same specialty)
Self
Pharmacist
Other human source
Referred
Continuing medical educa-
tion
Local Print Resources
Textbooks
Drug compendium
Personal notes
Articles from journals
Organizational & Institutional Information
Insurance requirements, etc.
Local service information
Lab/radiology (general info)
Professional organization
Patient Data/Patient Record
Patient
Lab/radiology (patient data)
Medical record
Digital Sources
Lab/radiology (patient data)
CDROM
World Wide Web
Electronic mail
Personal digital assistant
MEDLINE literature search
Library Sources
Library
The second qualification to the Curley model is that resource ac-
cessibility is but one element in determining the value of using
it. Huth proposed an economic model in which the overall value
of a resource is a function of both the benefit and the cost of its
use (11). This rational choice model may help to explain the in-
formation-seeking behavior of PCCs. In a previous study, we
found that a physician's expectation that a definitive answer ex-
ists was a better predictor of information seeking than the ex-
pected difficulty of finding it, (9) suggesting that the value of
having an answer may outweigh the access cost of obtaining it.
Findings of the present study are also consistent with this view:
clinicians’ access to knowledge resources did not predict their
use of those resources, suggesting that other factors, such as the
value of the information obtained play an important role.
The present study was not a comprehensive examination of rural
practice, and does not refute reports by others on differences in
the nature of rural medical practice and the nature of rural prac-
titioners. (12) But for those variables we did measure, our com-
parison of rural and nonrural PCCs and their information seeking
found more similarities than differences. Rural PCCs were more
often in solo or small group practices, but were otherwise not
significantly different from nonrural PCCs in demographic char-
acteristics, or patterns of information needs, information seek-
ing, and knowledge resource use.
These findings do not suggest that geographic, social, and eco-
nomic factors do not have profound influences on rural health
care and rural clinicians. Rather, our data suggest that, in terms
of information seeking, a clinicians in large urban practices may
be every bit as isolated from the medical knowledge needed to
support optimal care as their rural counterparts. The time pres-
sures and other distractions of the current practice environment
may have as much influence on information seeking as geo-
graphic isolation or other factors affecting the rural clinicians.
Furthermore, from the point of view of designing and deploying
information systems to benefit clinicians and their patients, these
findings suggest the importance of looking beyond the usual as-
sumptions about a system's intended users by obtaining empiric,
observational data about their actual information needs and be-
haviors.
This study provides baseline data from primary care practice on
the extent of information need, the rate of information seeking,
the use of information resources, and the relative success of in-
formation seeking by PCC. Others who evaluate the impact of
new information systems using similar methods may be able to
use our data as a baseline for comparison.
Limitations
Several limitations that may affect the validity or reliability of
these findings must be recognized. These relate mainly to the
population studied and the data collection method employed.
First, it is well known in other areas of research that volunteers
who choose to be studied maybe systematically different from
non-volunteers who do not choose to participate. These differ-
ences have not been examined in information needs studies, so
the impact of this bias is unknown. Second, inclusion of nurse
practitioners and physician assistants was intended to improve
the generalizability of our findings to the population of clinicians
actually providing care in our state, but these practitioners may
be different from physicians with respect to their information
management, so it may not be appropriate to compare our find-
ings to studies which focus on only one of these professional
groups. Third, we used a definition of rural developed by the Of-
fice of Rural Health in Oregon which makes sense for this large-
ly rural state, but may make it difficult to compare our findings
with those of other studies which use other definitions of rural,
such as the county-based standard metropolitan statistical area
(SMSA). Because most nonrural areas in Oregon would hardly
qualify as urban in other areas of the country, we have avoided
the use of the term "urban" to refer to these areas.
With respect to data collection, our model of information seek-
ing influenced the data we collected. (13) The late Diana For-
sythe argued strongly for a broader view of information needs
than that adopted here. We chose the clinical question approach
because it has been employed widely (2, 3, 10, 14, 15) and be-
cause it appears to be quantifiable and reproducible. Barrie has
suggested that the office interview method may produce a biased
estimate of information need by stimulating subjects to think of
questions for the interviewer which would not occur if the clini-
cian were alone. (15) Others have suggested the opposite effect,
where fewer questions would be collected than naturally occur
because of the clinicians' reluctance to reveal their ignorance or
uncertainty, (16) producing an estimate of information need that
is biased in the other direction. Until these methods are directly
NRRtotal
19
13
3
6
2
2
0
32
12
9
3
5
2
1
51
25
12
9
7
4
1
28
11
9
5
25
7
4
6
53
18
13
11
9
5
1
0
2
8
3
1
11
13
4
1
6
8
2
4
8
1
10
16
3
2
1
0
1
0
1
0
1
1
0
1
0
2
2
1
1
1
1
044

Page 5

PN. Gorman et al.
1137
compared, the magnitude, direction, or importance of this poten-
tial bias will remain unknown.
Summary
In summary, we compared the information needs and the infor-
mation seeking activity of rural and nonrural PCCs in Oregon.
We found that overall there were more similarities than differ-
ences. Rural PCCs had less direct access to consultants, digital,
and library-based information resources, but these differences in
access had little impact. Rural and nonrural clinicians had equiv-
alent rates of information need and information seeking, had
similar patterns of knowledge resource use, and were equally ef-
fective at finding answers to their questions. These findings have
implications for the design and deployment of information sys-
tems and services for primary care. Whether rural or nonrural,
factors other than physical availability of information resources,
such as time pressures and the organization of care, may be more
important in limiting access to timely information to support op-
timal patient care for rural and nonrural clinicians alike.
Acknowledgments
Thanks to the primary care clinicians of Oregon for sharing their time
and expertise; to Susan Poulsen, PhD for her valuable assistance with
qualitative methodology Supported by the National Library of Medicine
(R29LM05663).
References
[1] Dee C, Blazek R. Information needs of the rural physician: a
descriptive study. B Med Lib Assn 1993;81(3):259-64.
[2] Covell DG, Uman GC, Manning PR. Information needs in
office practice: are they being met? Ann Intern Med
1985;103(4):596-9.
[3] Gorman PN. Information needs of physicians. J Amer Soc
Information Sci 1995;46(10):729-736.
[4] Smith R. What clinical information do doctors need? [see
comments]. BMJ 1996;313(7064):1062-8.
[5] Krikelas. Information seeking behavior. Patterns and con-
cepts. Drexel Libr Quart 1983;19:5-20.
[6] Gorman PN. Information needs in primary care: A survey of
rural and nonrural physicians. MEDINFO 2001;10(Pt
1):338-342.
[7] Curley S, Connelly D, Rich E. Physicians' use of medical
knowledge resources: Preliminary theoretical framework
and findings. Med Dec Making 1990;10:231-241.
[8] Ramos K, Linscheid R, Schafer S. Real-time information-
seeking behavior of residency physicians. Family Medicine.
2003;35(4):257-60.
[9] Gorman PN, Helfand M. Information seeking in primary
care: How physicians choose which clinical questions to
pursue and which to leave unanswered. Med Dec Making
1995;15(2):113-9.
[10]Ely J, Burch R, Vinson D. The information needs of Family
Physicians: Case-specific clinical questions. J Fam Pract
1992;35:265-269.
[11]Huth EJ. Needed: an economics approach to systems for
medical information. Ann Intern Med 1985;103(4):617-9.
[12]Yawn BP. Rural Medical Practice: Present and Future. In:
Yawn BP, Bushy A, Yawn R, editors. Exploring Rural Med-
icine: Current Issues and Concepts. Thousand Oaks, Cali-
fornia: Sage; 1994. p. 1-16.
[13]Gorman PN. Information seeking in primary care: Concep-
tual models and empirical studies. In: Wilson T, editor.
Information Seeking in Context: Proceedings of the 2nd
ISIC conference. Sheffield, UK: Taylor Graham; 1998. p.
226-240.
[14]Osheroff J, Forsythe D, Buchanon B, Bankkowitz R, Blu-
menfeld B, Miller R. Physicians' information needs: analy-
sis of questions posed during clinical teaching. Ann Intern
Med 1991;114:576-581.
[15]Barrie AR, Ward AM. Questioning behaviour in general
practice: a pragmatic study. BMJ 1997;315(7121):1512-5.
[16]Gerrity MS, DeVellis RF, Earp JA. Physicians' reactions to
uncertainty in patient care. A new measure and new
insights. Medical Care 1990;28(8):724-36.
Address for correspondence
Paul Gorman, MD; Associate Professor; Department of Medical Infor-
matics & Clinical Epidemiology; Oregon Health and Sciences Univer-
sity; 3181 SW Sam Jackson Park Road; Portland, Oregon, 97239;
USA; gormanp@ohsu.edu