Implementing Clinical Decision
Support in an Electronic
Medical Record System
2010 AcademyHealth Annual Research Meeting
Boston, MA
June 26,
26 2010
Mark C. Hornbrook, PhD
Chief Scientist
Kaiser Permanente
The Center for Health Research
3800 North Interstate Avenue
Portland, OR 97227
97227-1110
1110
503-335-6746
503-335-2428
mark.c.hornbrook@kpchr.org
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
Clinical Knowledge Management
 Validated on-line searchable resources
 Inside the firewall
 External
 On-line interactive logic
 Clinical decision support
 What has happened so far?
 Views of patient’s history, medications, and orders
 What should I do next?
 Pop-up prompts for next decision step
 Are you sure you want to do this?
 Pop-up soft prompts regarding the previous entry
 This is not an acceptable action!
 Pop-up hard stops regarding the previous entry
 Cross-patient priority-setting support
 Which patient who did not arrive today deserves my attention next?
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
Knowledge Connection
 Desktop connectivity to internal resources…






Local clinical practice guidelines
D
Drug
iinformation
f
ti di
directory
t
Formulary
Referral resource directory
Fee schedules
Coding definitions
 Desktop connectivity to external resources…






Published clinical practice guidelines
PUBMED
NIH
FDA
Specialty societies
Patient advocacy organizations
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
Computerized Physician Order Entry
 Domain
 Functions
 Referrals
 Electronic order transmission
 Surgical procedures
 Autocoders
 Medications
 Order validation and error checks
 Laboratory tests
 Formularies
 Imaging procedures
 Prescribing/ordering guidelines
 Rehabilitation (PT, OT, ST, RT)
 Patient instructions
 Hospital admission
 Patient copayments
 Skilled Nursing Facility admission
 Scheduling
 Durable medical equipment
 Special requirements
 Home health
 Inventory control
 Hospice
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
Getting Data In
 “This is part of the EMR that many
clinicians dread until the EMR champion
shows them tips
p and tricks to q
quickly
yg
get
data into the EMR. Coding has
improved after our EMR for various
reasons including reminders to make
sure coding is done correctly
correctly.”
Richard Dell, MD, KPNC Orthopedist
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
Order Entry
 “Order entry is actually very easy to do
with an EMR and makes errors
significantly
g
y less if implemented
p
correctly. Order Entry is also faster now
and improves medication reconciliation
when patients are admitted.”
Richard Dell, MD, KPNC Orthopedist
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
EMR Smart Sets
 Linked support that integrates clinician
note generation, orders, after-visit
f
summary,
y etc.
 Activating a smart set function opens and fills in
templates
p
for notes,, order sets,, etc.
 Support tailoring of diagnostic work-ups
t optimize
to
ti i accuracy-costt trade-offs
t d ff
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
Medication Warnings
 Prescribing error alerts
 Drug allergy alerts
 Drug-drug interaction alerts
 Safety surveillance alert
 Cue to order lab test(s) to detect possible adverse
d
drug
effects
ff t
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
After Visit Summaries
 Tailored and personalized instructions
f patients to take with them for
for
f
reference at home
 Specialty-specific templates
 Standard phrase libraries for ease of
data entry
 Personal messages from clinicians
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
Clinical Prevention Services
 Tobacco use and control
 Immunization tracking
 Disease screening
 Pap
p smears
 Mammography
 PSA
 FOBT/Colonoscopy
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
Getting Data Out
 “Everyone loves the ability to quickly get
t th
to
the data
d t needed
d d for
f patient
ti t care,
such as notes, lab tests, and digital
images. The EMR gives you the ability
to use filters to quickly get to data and
summary reports to put the data
together for ease of use.
use ”
Richard Dell, MD, KPNC Orthopedist
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
Osteoporosis Management
 “When a patient has a fragility fracture and
does not quickly get a DXA, and/or begin an
anti-osteoporosis medication, our Healthy
Bones Program uses HealthConnect to
identify that patient and we have our Healthy
Bones Care Managers address the care gap.
This has led to a drastic increase in getting
our members DXA scans and/or antiosteoporosis treatment when indicated and a
steady decrease in our hip fracture rate at
Kaiser Permanente.”
Richard Dell, MD, KPNC Orthopedist
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
Panel Support Tool
 Prioritizes clinical prevention services across PCP’s patient
panel
 Immunizations—who has which immunizations due and how long past due?
 Disease screening—who have the longest noncompliance intervals?
 Adverse event surveillance—who is past-due for scheduled lab
tests and
d ffor h
how llong?
?
 Liver function tests for Statin users
 Alerts for return-to-clinic visits—who is p
past due for how long?
g
 Alerts for referral follow-ups—did patient comply?
 Chronic disease management—who have the longest noncompliance intervals for check
check-ups?
ups?
 Diabetes, hypertension, hyperlipidemia, mental illness, etc.
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
EMR $aving$
 Huge cost savings from eliminating paper records
 Storage,
Storage transportation
transportation, handling
handling, faxing
 Reduced duplication of laboratory testing
 Lab test results are viewable from any work station
 Increased pharmacy efficiency and reduced overall
Rx error is EMR’s shining
g star
 Electronic drug orders
 On-line refills
 Robotic pharmacy
 Automated refill reminders to patients
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
EMR Evolution
 Phase I
 Emulation of hard-copy charts and existing sequential care
processes
 Ease transition for users
 Ease specification of EMR functions to programmers
 Phase II
 Redefine care management to take advantage of medical
informatics
 Phase III
 On-line multi-disciplinary team interactive clinical practice
 Medical e-home and e-CDS for patients and clinicians
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
EMR Challenges
 10% reduction in clinician efficiency that
nobody has figured out how to fix
 Structured user interface is more restrictive and demanding
than pure free text charting
 New error potential = click fatigue
 Wrong patient, wrong dose, wrong drug click
 “Certainly as a practitioner, I can cite many examples of how
well-intended
ll i t d d EMR b
builds
ild llead
d tto iinefficiency,
ffi i
di
disruptions
ti
iin
care, and, at worst, outright errors.”
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
EMR Challenges, continued
 Who enters data into an EMR?
 D
Decentralizing
t li i b
burden
d off EMR d
documentation
t ti ffrom
physicians = increased face-to-face time with patients
 E-mail tsunami
 Patients love direct E-mail connectivity to their clinicians, but
this is creating extraordinary workloads
 Managing clinical practice variations
 How much variation is observed?
 How much variation is “too much”?
practice variation?
 What are the correlates of p
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH
Abstract
Research Objective: Kaiser Permanente (KP) has installed a national electronic medical record (EMR) system which covers all
ambulatory care facilities. KP is in the process of installing the inpatient EMR modules in all its hospitals. KP’s informatics
strategy has included total conversion from paper charts to EMRs since 2003. EMRs have several objectives: 1) establish a
medico legal record of care provided to patients; 2) facilitate physician order entry and forward orders to appropriate departments;
medico-legal
3) document the patient’s health problems, diseases, and clinical parameters; 4) provide immediate access to a patient’s medical
record at every medical facility and at every clinical services department; 5) provide clinical decision support systems to improve
quality and cost-effectiveness of care; and 6) provide population aggregate data on case mix, utilization, costs, and outcomes. The
aggregated data files represent the raw resources for conducting retrospective comparative effectiveness studies, for estimating
predictive models in support of clinical decision making, and for benchmarking risk-adjusted costs and outcomes. EMRs offer a
l i l ffoundation
logical
d ti ffor clinical
li i l d
decision
i i supportt systems
t
(CDS) iin real-time
l ti
att the
th point
i t off order
d entry.
t
CDS can b
be purely
l elective
l ti
and separable, providing searchable information resources for clinicians, and they can be integrated into the physician order entry
(POE) system to provide real-time feedback to clinicians. CDS can be designed as a patient safety net, as a cost-management
strategy, and as a guide to optimizing quality of care in both retrospective and prospective in applications.
Study Design: Qualitative analysis of KP’s CDS initiatives.
Population Studied: Kaiser Permanente (KP), an integrated healthcare delivery system covering eight geographic regions with
nearly nine million members. KP is a closed panel HMO. The Permanente Medical Groups are tied exclusively to the Kaiser
Foundation Health Plan. All monies for physicians’ services are allotted to the medical groups for purposes of producing or
purchasing these services.
Principal Findings: CDS systems should be evidence based. Retrospective analyses of practice patterns and estimation of
predictive risk models require data from large patient samples
samples. EMR data storage is optimized for rapid on
on-line
line access and
recording for individual patients. EMRs are characterized by thousands of tables for data storage, which is completely opposite of
the large rectangular files optimized for statistical analysis. Hence, the utility of EMR data for CER and epidemiology requires
rigorous and complex data extraction. Real-time CDSs interacting with POE raise the risk of alert overload. KP went through a
process of building alerts into the EMR until physicians complained that they were getting so many alerts that it was imposing
significant delays in completing their orders during patient visits, thereby creating delays for patients.
Conclusions: Concerns for patient safety, quality of care, duplication of services, and care coordination motivate development of
CDS applications within EMRs. POE-CDS systems offer the potential for patient safety nets, as well as for detecting orders that
appear to be out of alignment with established practice guidelines. The countervailing pressure is loss of productivity associated
with “dead stops” in a CDS system that requires additional action by the ordering physician.
© 2010, KAISER PERMANENTE CENTER FOR HEALTH RESEARCH