Guidance for the Development of Patient Safety Data Standards

1
Guidance for the Development of Patient Safety
Data Standards

• Institute of Medicine / NAS
• Blackford Middleton, MD, MPH, MSc
• Partners Healthcare System
• Brigham Womens Hospital
• Harvard Medical School

2
Motivation

• To Err is Human
• called for patient safety reporting systems,
  that focus on serious adverse events and
  reporting systems for quality improvement
  purposes that focus on near misses and no-harm
  events.
• develop data standards to maximize the
  usefulness of information derived from the data
  collected, to minimize the reporting burden, and
  to allow for comparisons across reporting systems
  and over time.

3
Committees Questions

• (1) What are likely to be the key IT developments
  for improving and auditing patient safety?
• (2) How will developments in health care IT
  affect the type of patient safety data standards
  needed?
• (3) What is the process/timeframe by which data
  standards get implemented by health care system
  vendors and users?
• (4) What are the barriers to standardization
  of health care data systems?

4
Tenets for Patient Safety Data

• Patient Safety Data Clinical Data
• GIGO
• Free text vs. Structured Data Capture
• Reporting as a by-product of care

5
Information Space (from NCQA)
6
Free Text GIGO
S CC Feeling tired, runny nose HPI Patient
presents with complaint of runny nose and fatigue
for 3 days duration. She states that when the
symptoms began she first noted headache and
listlessness. Over the next day or two she
developed a runny nose with frequent sneezing and
a cough productive of a yellow-green phlegm. She
denies dysuria, diarrhea, muscle aches or cramps,
nausea, vomiting, hematuria, hemoptysis, rash, or
joint pains. She works as a sales clerk at a
local department store with lots of exposure to
the public. ROS as per HPI. O BP 145/90 P
88 regular RR 14 T 99.7 oral HEENT PEERL,
EOMI, Sclerae anicteric. Oral pharynx with
bilateral erythema in posterior pharynx without
hemorrhage or exudate Neck supple with shoddy
lympadenopathy right left Chest clear to
auscultation and percussion CV RRR without
murmur, rub, or gallop A URI, no evidence of
bacterial process. P Chest x-ray to R/O
pulmonary involvement. Will treat conservatively
with OTC cold preps, NS gargle. Advise for Fluvax
when well. RTC prn.
Data?
7
Structured vs. Unstructured Data
8
Structured vs. Unstructured Data
Optimum Mix
Usefulness of Data
Impact on Usability
100Free Text
100 Structured Coded
Middleton B, Renner K, Leavitt MK. Ambulatory
Practice Clinical Information Management
Problems and Prospects J Hlth Info Mgmt,
11497-112, 1997
9
A Framework for Knowledge Engineering for the EMR
Practical Knowledge Engineering for the EMR From
Vocabulary, to Encounter Form, to Report.
Middleton B., Masarie F, Betts C. AMIA Tutorial
1995-8
10
Key Issues for Clinical Information Management

• Data collection as a by-product of routine care
  delivery
• Who obtained the data?
• Why was the data obtained?
• How was it assessed?
• How was it used in post-hoc analysis?

Henry SB, Lenert L, Middleton B. Linking Process
and Outcome with an Integrated Clinical
Information Management System Proc HIMSS 1993
57-81.
11
Quality Integration Cycle
Measurement
Vocabulary
Data
EMR
Analysis
12
Quality Integration Cycle Henry SB, Lenert L,
Middleton B, HIMSS Proc, pp. 57-81, 1993.
Measurement
Clinimetrics Reliability, Bias, Error
Data
Vocabulary
Message HL-7 3.0, ASC X12N, NCPDP, MIB Model
RIM Knowledge Arden, GLIF, GEODE,
SNOMED-CT, ICD-9/10CM, CPT, DSM, LOINC,
NANDA, NIC, NOC, CDISC Sequence data
Analysis
EMR
Cadeuceus, dr. Quality, TSI, APACHE, CSI,PMC
13
Challenges

• Balance between information utility and system
  usability
• Matching concepts term definitions, granularity,
  intended use
• Lacking reference information model
• No defined set of pre-coordinated terms for user
  interface terminology

B. Middleton NCVHS Testimony Oct 14,
1999 http//ncvhs.hhs.gov/991014tr.htm
14
Suggestions

• Establish minimal set of critical quality
  measures (outcome measures)
• Establish minimal required data set to support
  derivation of critical quality measures
• Establish acceptable standard controlled
  vocabulary and codes for minimum data set
• Establish acceptable methods to gather minimum
  data set at the point of care

B. Middleton NCVHS Testimony Oct 14,
1999 http//ncvhs.hhs.gov/991014tr.htm
15
Phased Interoperability

• Basic interoperability
• Point-to-point, pre-HL7
• Structured interoperability
• HL7Reference Information Model
• Semantic interoperability
• HL7 RIM
• Controlled Medical Terminology (SNOMED CT)

B. Middleton NCVHS Testimony Oct 14,
1999 http//ncvhs.hhs.gov/991014tr.htm
16
(1) What are likely to be the key IT developments
for improving and auditing patient safety?

• Improving
• Information access
• CPOE
• CDSS
• Auditing
• CDRs
• Self report
• surveillance

17
(2) How will developments in health care IT
affect the type of patient safety data standards
needed?

• They wont
• Expect PSDS to drive IT adoption
• Facilitate
• Value proposition
• Structured data capture
• Aggregate data

18
(3) What is the process/timeframe by which data
standards get implemented by health care system
vendors and users?

• Coding standards are driven by finance
• E.g. billing requires ICD, CPT
• Real information stds like the RIM, and SNOMED-CT
  take time
• Weave into backend representation
• Evolve front end structured data capture

19
(4) What are the barriers to standardization of
health care data systems?

• No financial motive
• User resistance
• Scale start small, focused
• What form could incentives take?
• Billing data chart attachments
• Payers pay for quality reports data

20
Thank you!

• Blackford Middleton, MD
• bmiddleton1_at_partners.org a