Transcript Slide 1

Physician Clinical Performance
Measurement
Presentation to the Consumer-Purchaser
Disclosure Project
September 29, 2006
Earl Steinberg, MD, MPP
President and CEO, Resolution Health Inc.
Adjunct Professor of Medicine and of
Health Policy & Management,
Johns Hopkins University
© Resolution Health Inc., 2006
All Rights Reserved
Outline
Concerns about the analysis of electronic “administrative”
data
II. Key issues in physician clinical performance measurement
A. Types of data that are available
B. Measures
C. Physician attribution
D. Measurement Period
E. Scoring
F. Reporting
G. How results are used
H. Privacy
III. Q&A
I.
Better Health Care Through Better Information
2
Physician Measurement: Key Topics
• Data
•
•
•
•
•
•
•
Measures
Attribution
Measurement period
Scoring
Reporting
Use of results
Privacy
Better Health Care Through Better Information
3
Potential Data Sources
• Electronic Medical Record
• Claims
• Claims + Other Electronic “Administrative” Data (e.g.
enrollment, benefits, lab results)
• Chart Review
• Self-reported (by MD)
• Survey
Better Health Care Through Better Information
4
Inpatient Services Data?
• Discharge abstract vs UB92 vs Detailed bill?
• Detailed bill required to ID:
▪ Tests/Certain Procedures
- e.g. drug level; oxygen saturation; telemetry
▪ Medications (e.g. systemic steroids)
▪ Timing of provision of particular services relative to:
- date/time of admission
- initiation of treatment (e.g. was CT/MRI done
before anti-coagulation was started?)
Better Health Care Through Better Information
5
Outpatient Services Data?
• Behavioral health services
• DME
• Does each claim (e.g., each Rx claim) have a MD
identifier? How specific is it?
Better Health Care Through Better Information
6
Claims Data vs Chart Abstraction
Claims Data
Chart Abstraction
Missing claims (e.g., OON utilization)
Info. re care provided by other MDs
often missing
Incomplete coding
Incomplete documentation
Errors in coding (both inadvertent and
deliberate)
NA
Variations in coding
Variation in terminology, abbreviations,
symbols and criteria
Limited breadth and depth of codes
NA
NA
Problems with legibility
Low cost
High cost
Better Health Care Through Better Information
7
Pros/Cons of Various Data Sources
PROs
CONs
Claims Data
Widespread Availability
Low Burden
Low Cost
Claim Lag (timeliness)
Not a Problem for CPM
Limited Clinical Detail
Incomplete
Inaccuracies
Chart Abstraction
Clinical Detail
Incomplete
Variability
High Burden
High Cost
Patient Survey Data
Can Address Non-Clinical
Issues
High Burden/Cost
Uncertain Accuracy
Little Clinical Detail
Low Response Rates
Better Health Care Through Better Information
8
Things Not Detected in Claims Data
•
•
•
•
•
•
Patient symptoms
Performance or results of a physical
exam
Family history
Personal habits (e.g., smoking)
Diseases not reliably inferred using
claims (e.g. obesity)
Procedures not reliably captured by
claims data (e.g. influenza vaccine)
•
•
•
•
•
•
Use of OTC meds (e.g. aspirin)
Procedures/ tests performed on
emergent v. elective basis
Clinical disease stage or severity
Documentation in patient chart
Test results
Alternative/ complimentary medical
care
Better Health Care Through Better Information
9
Ways to Deal with Shortcomings
of Claims Data
Shortcoming
How to Address the Shortcoming
Certain tests and services are (may)
not be reflected reliably in claims
data
Do not use measures that focus on use of
those tests and services
(e.g. spirometry; influenza vaccine;
smoking cessation programs)
Use of OTC drugs is not captured in
claims data
Do not use measures that focus on use of
meds that could be purchased OTC (e.g.
Use of aspirin post-MI)
Inaccuracies in diagnoses that are
listed on claims
Require more than one indication that a
member has a disease/condition
Better Health Care Through Better Information
10
Other Ways to Deal With
Limitations of Claims Data
• Use a very conservative approach to identification of “newly
diagnosed Disease X”
Example: Require a long “clean period” without evidence
of the diagnosis
• Make sure patients had insurance coverage for services that
were being looked for
Example: If patient was supposed to be on a drug over a
particular time period, we only look at people with Rx
benefits throughout entire time period
Better Health Care Through Better Information
11
Physician Measurement: Key Topics
• Data
• Measures
•
•
•
•
•
•
Attribution
Measurement period
Scoring
Reporting
Use of results
Privacy
Better Health Care Through Better Information
12
The “Geology” of Clinical Performance
Measurement
A Performance Assessment
Performance Measurement
Performance
Measures
Cost of Care
• Quality of Care
•
Standards
Guidelines
Content of Care
Outcomes of Care
•
•
Expert Opinion
Better Health Care Through Better Information
Evidence
13
Components of Measures: 1. Denominator
• Defines the target population to whom the CPM applies
• Things that should be considered for a process of care measure are
factors that should influence the care that should be delivered to a
patient
• Things that should be considered for an outcome of care measure
are factors that could influence the outcome of care
• Goal is to assure that “apples” are being compared to “apples”
• Goal is typically accomplished through “patient sub-setting”
• Sensitive to the technical specifications used to define the
denominator
Better Health Care Through Better Information
14
Components of Measures: 2. Numerator
• Quantifies whether a particular “event” of interest has occurred
- care consistent with a guideline or standard
- achievement of a normative outcome
• Sensitive to the technical specifications used to define the
numerator
Better Health Care Through Better Information
15
Practice Guidelines Don’t = Quality Measures
• Relationship between a practice guideline and a valid
algorithmic representation of the guideline (i.e. a valid measure
based on the guideline) is “1 to many”, not “1 to 1”
• Potential measures vary in terms of false negative vs false
positive rate
• The devil’s in the details!
Better Health Care Through Better Information
16
Criteria for Assessing Measures
•
Importance of the focus of the measure
•
Variation in performance
•
Potential for performance improvement
•
Validity of the measure
•
Reproducibility of the measure
•
Feasibility of the measure
- availability of data required for the measure
- difficulty of collecting the data
- cost of collecting the data
•
Relevance to consumers and purchasers
Better Health Care Through Better Information
17
Measure Description: Complete Transparency
Rule 114: Diabetics with HbA1c test in past 6 months
Description
This rule identifies the percentage of members over 18 years old with a history of diabetes who have had a
HbA1c test over the past 6 months
Technical Specification
Numerator =
•
Age over 18 years old AND
•
Meet RHI criteria for diabetes before measurement year (based on ICD-9 and NDC codes) AND
•
Eligible for benefits over the most recent 7 months during measurement year
Denominator =
•
At least 1 Procedure code (e.g. CPT-4) for HbA1c test over the most recent 7 months during
measurement year
Applicable Specialties
Internal Medicine
Family Practice
Endocrinology
References
American Diabetes Association. Diabetes Care. January 2006. 29 (suppl 1) s8-s11
Edelman D, Maren K, Olsen K, et al. Utility of Hemoglobin A1c in Predicting Diabetes Risk. Gen Intern Med
2004;19(12):1175-1180
Better Health Care Through Better Information
18
Physician Measurement: Key Topics
• Data
• Measures
• Attribution
•
•
•
•
•
Measurement period
Scoring
Reporting
Use of results
Privacy
Better Health Care Through Better Information
19
Attribution of Care to a Particular MD
• Should a patient or service be associated with only a
single MD or with a group of MDs?
• Can a unique MD (group) ID be constructed?
• What information is available regarding MD
specialty(ies)? How accurate is it?
• Does each claim (e.g., each Rx claim) have a MD
identifier? How specific is it?
Better Health Care Through Better Information
20
Physician Measurement: Key Topics
• Data
• Measures
• Attribution
• Measurement period
•
•
•
•
Scoring
Reporting
Use of results
Privacy
Better Health Care Through Better Information
21
Measurement Period
• Fixed
- 2 yrs of historical data to ‘prime the pump’
- Single (consistent) measurement period
- Easy to understand
- Don’t use some potential observations
• Rolling, Variable
- Once criteria for a denominator are satisfied, the following
12 mos. are used as the measurement period
- Measurement year varies across MDs and across pts for a
given MD
- Could have >1 observation for a given pt for a given MD
- Harder to do and harder to explain
- Maximizes observations
Better Health Care Through Better Information
22
Physician Measurement: Key Topics
•
•
•
•
Data
Measures
Attribution
Measurement period
• Scoring
• Reporting
• Use of results
• Privacy
Better Health Care Through Better Information
23
Construction of a Quality Score
•
Types of scores
- Single (aggregated) score per MD (or MD group)
- A score for certain types of care (e.g. management of diabetes; cardiac
care)
- Measure-specific scores
•
Weighting
- unweighted sum of numerators and denominators across all pt-measures
applicable to a given MD (implies all measures are equally important)
- weighted sum (implies some measures are more important than others)
•
Sample size
- Minimum number of observations for a given MD for a given measure for
that
measure to be considered?
- Minimum number of observations for a given MD for that MD to
receive a score?
- Confidence intervals around point estimates
•
Distinguishing between MDs
- Numerical score vs %-ile vs quartile (or other group)
- How big a difference in score is clinically meaningful?
Better Health Care Through Better Information
24
Physician Measurement: Key Topics
•
•
•
•
•
Data
Measures
Attribution
Measurement period
Scoring
• Reporting
• Use of results
• Privacy
Better Health Care Through Better Information
25
Making Reports Relevant
to Consumers and Purchasers
• Address issues that are important
• Address a spectrum of services
- within a specialty
- across specialties
• Provide information that is understandable/easy to
interpret
• Provide information that is actionable
Better Health Care Through Better Information
26
Physician Measurement: Key Topics
•
•
•
•
•
•
Data
Measures
Attribution
Measurement period
Scoring
Reporting
• Use of results
• Privacy
Better Health Care Through Better Information
27
Potential Uses of Physician Measures
• Inform patients and purchasers
• Quality improvement
• Payment adjustments (e.g. P4P)
• Financial incentives to patients (e.g., high performance
networks/tiering
Better Health Care Through Better Information
28
Physician Measurement: Key Topics
•
•
•
•
•
•
•
Data
Measures
Attribution
Measurement period
Scoring
Reporting
Use of results
• Privacy
Better Health Care Through Better Information
29
Privacy Issues
• Importance of security safeguards in connection with
transfer, storage and use of data
- should have a security policy
- use of passwords and encryption
- de-identification of data (removal of personal
identifiers) during transfer, storage and analysis
- firewalls
- monitor access
- identification of patients to MDs?
• Do physicians have a right to privacy re their clinical
performance?
Better Health Care Through Better Information
30
Data to Results Overview
Data Management and Analysis
Enrollment
Benefits
Med Claims
Rx Claims
Lab Results
Providers
Data Standardization and Integration
Evidence-Based Rules Engine
Identify Apparent Deficiencies in Care
Attribution of Patient-Issue to Provider(s)
Provider-specific Clinical
Performance Assessment
31
Questions?
Better Health Care Through Better Information
32