Using Medicare Files as Sampling
frames
for a Survey of Physicians
Vasudha Narayanan
Presented at
Third International Conference on Establishment Surveys
June 20, 2007
1
Acknowledgements
Co-Authors
Pamela Giambo & Sherm Edwards from
Westat
Project Officers
Gladys Valentin & David Clark from
Centers for Medicare & Medicaid Services
This material was prepared by Westat Inc., under contract with the Centers for
Medicare & Medicaid Services (CMS), an agency of the U.S. Department of Health
and Human Services. The contents presented do not necessarily reflect CMS policy.
2
Issues with Establishment Survey
Sampling Frames
• Cox (1997); Konschnik (1988)
• Complex ownership and management
structures
• Frames clogged with inactive and
duplicate records
• Coverage error
• Unreliable contact information
3
About MCPSS
• Measures satisfaction of physicians and other
medical providers with the services offered by
Medicare Contractors
• Purpose of MCPSS
• Satisfy Medicare Modernization Act (2003)
requirements
• Establish a uniform measure of provider
satisfaction with Contractors
• Provide feedback to Contractors for QI
• Annual Survey
4
About MCPSS
• Samples customers of 51 Medicare
Contractors
• 19 “carriers” handle most physician claims
• Within Contractor, sample stratified by provider
type and state
• Screening for a “knowledgeable” respondent
• Survey can be completed over the web,
telephone or paper
• Response rate: 65%
5
Who is the Customer/Analysis Unit
• Individual(s) who interact with Contractor
• Typically not the physician
• Could have multiple customers within a
practice
• Could represent a single physician, a
practice, multiple solo practices, or
multiple groups
• Customer may interact with more than one
contractor
6
Sample Design Utopia
Sampling Unit
Responding Unit
Analysis Unit
7
Variables Required for MCPSS
Sampling
• Contractor ID
• Tax ID
• Number of claims in reference year
• Provider type
• State
• Contact information
8
Files Available for MCPSS
Physician Frame
• CMS files
• Claims History
• Unique Physician Identification Number
(UPIN)
• Provider Enrollment, Chain, and Ownership
System (PECOS)
• Non-CMS frames
• AMA
• Commercial Vendors
9
Sources of Contact Information
• 2006 sample [2%]
• PECOS [41%]
• [Did not use UPIN]
• Commercial vendor [39%]
• Contractors [as a source of last resort]
[16%]
• Other files [1%]
[Information missing for 1% of the sample]
10
Paradise Lost
• A disconnect between the sampling unit and
responding unit
• Sampling unit: a Tax ID with at least one claim
in the last 12 months
• Responding Unit (s): a physician
representative who has experience with a
Medicare Contractor
• As mentioned earlier
• One respondent for multiple sampled units
• Multiple respondents in one sampled unit
11
Screening Sampled Providers
• To locate the most knowledgeable
respondent(s)
• Hence determine mapping between sampled
unit and responding unit(s)
• To determine if the responding unit is a
third-party billing service
12
2007 Sample Performance
Physician Sample
Rest of Sample
Sample Size
7,895
28,464
1. Completes
51.7%
48.0%
2. Partial completes
2.6%
2.1%
3. Ineligible
6.1%
3.5%
4. Ineligible-Chains
1.4%
23.3%
5. Unknown Eligibility
12.8%
6.6%
6. Refusals
4.4%
2.1%
7. Other Non-response
21.1%
2.1%
RR
56.5%
67.3%
Eligibility rate
91.4%
71.3%
Response rate Calculation (AAPOR-RR3)
1 / ( (1 + 2 + 6 + 7) + (5*e) )
Where e = (1 + 2 + 6 + 7) / (1 + 2 + 6 + 7 +3 + 4
13
Paradise Regained?
• Utopia is inherently unrealistic
• The frame allows for a pretty good
mapping of sampling and analysis units
• Nothing can be done about the ineligibles
• Problem of unknown eligibility
• Poor contact information
14
Contacting the Authors
Vasudha Narayanan
([email protected])
Pamela Giambo ([email protected])
Sherm Edwards ([email protected])
15
Thank you
16
Descargar

Using Medicare Files as Sampling frames for a Survey of