PrimeAnswers:
Knowledge at the Point of Care
NLM
Information Systems Grant LM06758
Y2
Progress Report Summary
Specific Aims
A strategic decision was made to use contextual design as the development strategy for the PrimeAnswers project. By targeting the users’ environment and workflow processes, contextual design informs development of a coherent information technology solution based on the reality of how users work (1). Friedman (2) states that an information technology intervention without cultural context is doomed to failure and that post-evaluation is fruitless. Brown (3) states it as: text + context = meaning. This work has focused our effort on building a rich set of core sources on which we build and integrate best available evidence. Success in our target user community requires inclusion of factual information currently used throughout the day that will be enhanced iteratively to become “the” place to go for answers.
Studies and Results
To
implement the contextual design strategy, we have: 1) recruited an evaluator
with experience in ethnographic and contextual techniques; 2) broadened our
in-reach to the pilot clinics and outreach to three mirror clinics in the UW
system to verify evaluation tools; 3) established a collaborative research
initiative with the UW Family Medicine Residency Network including
participation of the Network’s EBM Coordinator in the PrimeAnswers project team
and affiliation with the Family Practice Information Network project in
Missouri; and 4) developed relationships with producers of new content relevant
to identified needs in primary care (e.g., ePocrates
for Palm, BMJ’s Clinical Evidence).
·
A five-part contextual design was completed. 1.
Pilot clinic technology and information inventory to identify potential
disconnects with design in the current environment. (Sample form in Appendix
1.) 2. Baseline survey of providers in the targeted primary care clinics using
a web-based tool. (Sample survey and results tables in Appendix 1.) 3. Observations of information seeking
strategies used by the three project providers to answer questions that arose
during the day. 4. Observations of 10 primary care providers during a typical
half-day clinic. (Methods and results in Appendix 1.) 5. Monthly discussions
between the three primary care providers and the PI, Project Coordinator and
Evaluator. (New human subjects documentation in
Appendix 2.)
·
Time and ability to articulate questions and search potential
sources is the issue to solve at UW. PrimeAnswers users are: 1) 54/46
male/female; 2) the majority have practiced less than 6 years, work 7-10 half
day clinics per week, and treat 6-15 patients per half day clinic; 3)
information seeking occurs most frequently around the patient encounter or
after seeing patients for the day; and 4) clinicians who use a PDA for clinical
reference either have less than 6 or 20+ years of experience. Most information use is focused on: drug
dosage, cost, and availability in the formulary; patient education;
immunizations and travel health; dermatology images and therapy; diagnostic
rules for injuries; dietary counseling; and uncommon presentations of commonly
seen conditions. The most frequently used digital sources are Micromedex and
the MD Consult textbook search. The top print resource is the Pocket
Pharmacopoeia. Our contextual design results indicate that simplification to a
narrow set of best resources and presentation of a few options to reduce the
steps required to find the answer is necessary. The long-term goal is a
seamless integrated system that assists formulation of an answerable question
in a format that looks two years to the future. “Pocket” PrimeAnswers is
critical to this seamless system for routine factual answers (e.g., drug dosage
and cost), which must be configurable with local information, such as “on
formulary.” Several prototypes will be built that include a portal with
personalization and portability for nomadic devices.
·
Availability of digested best evidence changed dramatically
between grant submission and project start date, and again by the end of Year
1. Content collaborations developed were: 1) Dynamic backend query of MD
Consult and Micromedex; 2) Two new herbal products, Natural Medicines
Comprehensive Database and AltMedDex; 3) Negotiated use of selected chapters of
BMJ’s Clinical Evidence for prototyping the Top 20
Presentations; 4) Trials of UpToDate, BMJ Clinical Evidence, Kluwer’s Clineguide and Evidence
Finder as best evidence sources; 5) Discussion with ePocrates
for local UW drug formulary tagging; 6) Compilation of clinical calculators and
decision support rules. Curbside Consult (question and answer) will be released
in December 2000. Questions will be submitted by email, web form or voice mail
and best answers posted on the site. Content inclusion for Top 20 Presentations
is in prototype for asthma, depression and back pain. A beta version of the
PrimeAnswers portal will be released in December. The project site is available
at: http://healthlinks.washington.edu/primeanswers/.
Working prototypes, produced using DreamWeaver, are linked to this site.
·
The HealthLinks team provided systems support. PrimeAnswers
will be based in large part on HealthLinks, which underwent a major conversion
this year. The Microsoft SQL database was moved to a Windows 2000 Server. The
homegrown Java code was converted to Cold Fusion on Windows 2000 server for
long-term sustainability. Cold Fusion’s dynamic query, streaming content and
personalization features are needed for PrimeAnswers. Google or Ultraseek will be used for full text site searching outside
of SQL database queries. Excalibur and XML technologies will be reviewed for
the project in Year 2. To insure availability throughout the project, the “primeanswers.org”
domain was registered for 3 years by the project, and trade marking of the name
“PrimeAnswers” is being pursued at the recommendation of UW Office of
Technology Transfer.
· Personnel Changes: Appointment of three primary care provider advisors for the pilot clinics (Barak Gaster, David Kauff, Diane Timberlake) was completed in June 2000. Previously identified providers were no longer available resulting in a 3-month startup delay. Provider appointments were made in concert with clinic directors. A librarian (Debra Revere) with extensive ethnographic and contextual design experience was recruited as project evaluator.
Long-term
impact of the project to date are: 1) completion of a information needs
assessment of primary care providers using ethnographic and contextual design
techniques that will be published; 2) expansion of the project at UW to the
“mirror” clinics providing a broader investment and long-term integration into
the EMR and work processes; 3) development of collaborations with
evidence-based medicine and primary care targeted content (e.g., eProcrates, BMJ’s Clinical
Evidence, Evidence Finder) to experiment with new formats of information for
primary care providers; and 4) development of a research collaboration with the
UW Family Practice Residency Network and the Family Practice Information
Network (FPIN) to pool question bank results and best answer sources, as well
as the use of information on a pocket-sized device at the point of care.
Plans
The project timeline is on target. Key
outcomes for Year 2 are: recruit student programmer (underway) and
database/web programmer; integrate Information School independent study
students; develop data schema based on a combination of SQL and XML
technologies; implement dynamic query model for remote content and search
engine for local content; evaluate portal and implement personalization
features; create a “pocket” PrimeAnswers on the
Palm platform for “quick tap” answers to factual questions (e.g., ePocrates, 5-Minute Clinical Consult, Guide to
Antimicrobial Therapy); implement Top 20 Presentations; evaluate questions and
sources of answers for Curbside Consult with goal to create knowledge base to
answer 80% of core primary care questions; continue research collaboration with
family practice partners; and publish results of information needs assessment.
Publications.
PrimeAnswers: Knowledge at the Point of Care. Panel
presentation by Debra Revere of the preliminary results of ethnographic and
survey needs assessment, American Society for Information Science Annual
Meeting,
1
Beyer H, Holtzblatt
K. Contextual Design.
2 Friedman C.
Keynote speaker, A Vision For Informatics: Biology,
Health, Education, & Computing, Division of Biomedcical
and Health Informatics,
3 Brown JS, Duguid
P. Social life of information.