Patterns of Care in Breast Cancer: On Care Patterns of Care in Breast Cancer: On Care Coordination in Underserved Populations and Coordination in Underserved Populations and the Use of Health Claims Data. the Use of Health Claims Data. Roger Anderson, Ph.D. Roger Anderson, Ph.D. Professor Professor Health Services Research Core Health Services Research Core Penn State Cancer Institute Penn State Cancer Institute Penn State Hershey Medical College Penn State Hershey Medical College Research Roundtable Appalachia Community Cancer Network Research Roundtable Appalachia Community Cancer Network September 28, 2007 September 28, 2007 Hershey, PA Hershey, PA
22
Embed
Roger Anderson, Ph.D. Professor Health Services Research Core Penn State Cancer Institute
Patterns of Care in Breast Cancer: On Care Coordination in Underserved Populations and the Use of Health Claims Data. Roger Anderson, Ph.D. Professor Health Services Research Core Penn State Cancer Institute Penn State Hershey Medical College - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Patterns of Care in Breast Cancer: On Care Patterns of Care in Breast Cancer: On Care Coordination in Underserved Populations and the Use Coordination in Underserved Populations and the Use
of Health Claims Data.of Health Claims Data.
Roger Anderson, Ph.D.Roger Anderson, Ph.D.ProfessorProfessor
Health Services Research Core Health Services Research Core Penn State Cancer InstitutePenn State Cancer Institute
Penn State Hershey Medical CollegePenn State Hershey Medical College
Research Roundtable Appalachia Community Cancer NetworkResearch Roundtable Appalachia Community Cancer NetworkSeptember 28, 2007 September 28, 2007 Hershey, PAHershey, PA
Source, North Carolina Cancer Registry, 2005,; SEER, National Cancer Institute, 2005
Age Age adjusted adjusted Breast Breast CancerCancerDeath Death
rate per rate per 100,000100,000
ObjectivesObjectives
Describe assembly of linked-Medicaid- North Describe assembly of linked-Medicaid- North Carolina cancer registry data. Carolina cancer registry data.
Describe the use of this dataset to identify unmet Describe the use of this dataset to identify unmet needs in cancer prevention and control needs in cancer prevention and control
Discuss applicability to ACCN Discuss applicability to ACCN
Discuss health services research implicationsDiscuss health services research implications
Medicaid and Rural HealthMedicaid and Rural Health
Medicaid is an important source of health insurance Medicaid is an important source of health insurance coverage for both rural residents and rural providers.coverage for both rural residents and rural providers.
Rural residents are more likely to live in poverty than Rural residents are more likely to live in poverty than urban residents, and are less likely to have employer urban residents, and are less likely to have employer sponsored health insurance coverage. sponsored health insurance coverage.
Rural residents are more likely to be covered by Rural residents are more likely to be covered by Medicaid than are urban residents. Medicaid than are urban residents.
Percent Residents with Medicaid Coverage During thePercent Residents with Medicaid Coverage During the Past Year in Urban and Rural Counties. Past Year in Urban and Rural Counties. CPS, 2004-2005 CPS, 2004-2005
StateState RuralRural UrbanUrban
KentuckyKentucky 17.417.4 12.112.1
MarylandMaryland 16.816.8 8.68.6
New YorkNew York 17.417.4 18.318.3
OhioOhio 11.211.2 12.312.3
Pennsylvania Pennsylvania 12.012.0 11.311.3
VirginiaVirginia 11.511.5 6.96.9
West VirginiaWest Virginia 18.218.2 11.311.3
North Carolina Project North Carolina Project
NC Tumor Registry NC Tumor Registry + + Incident casesIncident cases Clinical dataClinical data 11stst course of Tx course of Tx
(FORDS)(FORDS) Hospital RegistrarsHospital Registrars Other facilitiesOther facilities Merged dataMerged data CDC proficiency CDC proficiency
testingtesting
NC Medicaid ClaimsNC Medicaid Claims All medical services* All medical services*
1998-99 Medicaid Claims for North Carolina.1998-99 Medicaid Claims for North Carolina. N=1,401 female breast cancer (single primary, all N=1,401 female breast cancer (single primary, all
stages, 20% all cases) stages, 20% all cases)
Caveats of Medicaid claimsCaveats of Medicaid claims: : • Managed care organizations omitted Managed care organizations omitted • Dual eligibility - MedicareDual eligibility - Medicare• Continuous enrollment - pre-and-post diagnosis.Continuous enrollment - pre-and-post diagnosis.• Bundling of claims (date/services)Bundling of claims (date/services)• Completeness of claims (unbilled services?) Completeness of claims (unbilled services?)
Data SourcesData Sources
Caveats of Hospital Registry dataCaveats of Hospital Registry data: : • First course of treatment may be defined as 4-months First course of treatment may be defined as 4-months
post diagnosis.post diagnosis.• Treatment in physician offices may go unreported.Treatment in physician offices may go unreported.• Out-of state services often missingOut-of state services often missing• May exclude VHA casesMay exclude VHA cases• Comorbidity – added in 2003Comorbidity – added in 2003• Non-registry (mid-size) hospitals lower quality data Non-registry (mid-size) hospitals lower quality data
Data SourcesData Sources
Caveats of Medicaid dataCaveats of Medicaid data: :
• 9- 12 months of continuous eligibility is generally 9- 12 months of continuous eligibility is generally needed. needed.
• Medicare files m,ay be needed for dually insured.Medicare files m,ay be needed for dually insured.• Policies on covered services may vary by state. Policies on covered services may vary by state.
MethodsMethods
Test Population (1998-99 cases)Test Population (1998-99 cases): 1,401 cases : 1,401 cases single primary breast cancer in NC registry years single primary breast cancer in NC registry years 1998 - 1999.1998 - 1999.
Test sampleTest sample: 845 (60%) cases enrolled in : 845 (60%) cases enrolled in Medicaid 1 month prior and 12 months post Medicaid 1 month prior and 12 months post registry date of diagnosisregistry date of diagnosis
ApproachApproach: : • 1) Assume registry data is accurate if treatment is listed 1) Assume registry data is accurate if treatment is listed
as provided (not missing or indeterminate). as provided (not missing or indeterminate). • 2) Else, replace data with Medicaid claims (if 2) Else, replace data with Medicaid claims (if
discordant).discordant).• 3) Validate by performing record review on sample of 3) Validate by performing record review on sample of
cases. cases.
Accuracy of Radiation data in BCS sampleAccuracy of Radiation data in BCS sample
Days to first claims since dx Days to first claims since dx (Rad column, Chem column)(Rad column, Chem column) 0.989 (0.985,0.993)0.989 (0.985,0.993) 0.994 (0.988, 0.998)0.994 (0.988, 0.998)
Tum
or S
ize
Tum
or S
ize
CCR Agreement on RadiationCCR Agreement on
Chemotherapy
Examples of Application to Answer Patterns of CareExamples of Application to Answer Patterns of Care And Outcomes Research Questions And Outcomes Research Questions
Total Radiation No Radiation p
N= 344 N=242(70.4%)
N=102(30.0%)
Age group at time of diagnosis
65+ years 159 (46.2%) 58.49 41.51 <.0001 <65 years 185 (53.8%) 80.54 19.46
Race/ethnicity White 175 (50.9%) 69.71 30.29 0.7931 Other 169 (49.1%) 71.01 28.99
1. Based on presence of any paid claims from nursing home (location of service=T), home healthcare services (Q) or skilled nursing facility (cos = 35,36). 2. Tertile distribution of number of beds reported by American Hospital Directory. 3. Based on patient discharges reported by American Hospital Directory 4. Median split of all breast cancer cases with Medicaid enrollment.
Correlates of Under Use of Radiation Treatment with BCS in North Carolina Medicaid
Table 3. Interaction Graph showing unadjusted proportions of Radiation Treatment in Table 3. Interaction Graph showing unadjusted proportions of Radiation Treatment in BCS patients by Metropolitan Status and Hospital SizeBCS patients by Metropolitan Status and Hospital Size
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
L/M SHospital Size
% BCS who received radiation
Metro NonMetro
Kaplan-Meier Survival Curves of All-Cause Mortality by Radiotherapy Treatment
0.00
0.25
0.50
0.75
1.00
0 500 1000 1500 2000 2500
Days since diagnosis
No Radiation Radiation
Cormorbidity among Women with Breast Cancer in NC MedicaidCormorbidity among Women with Breast Cancer in NC Medicaid
N= 1,401N= 1,401
● ● 55 % had at least one other comorbid condition defined in Charlson 55 % had at least one other comorbid condition defined in Charlson comorbidity index comorbidity index
● ● Among those with comorbidity, > 50% had multiple conditions.Among those with comorbidity, > 50% had multiple conditions. ● ● The top three comorbid conditions were: The top three comorbid conditions were: Diabetes (26%) Diabetes (26%) Congestive heart failure (18%), Congestive heart failure (18%), Chronic pulmonary disease (11%). Chronic pulmonary disease (11%).
North Carolina Medicaid enrollees with a diagnosis of diabetes with and without breast cancer:
mean days (SD) ____________________________________________________
Women without cancer 300 (19) With breast cancer 205 (95)
Hormone Medication Persistence. NC Medicaid Breast CancerHormone Medication Persistence. NC Medicaid Breast Cancer
Monali Bhosle OSUMonali Bhosle OSU
Cumulative nonpersistence rate based on number with + ER status who started therapy. No significant association between race and medication persistence adjusting for a type of index therapy and other confounders (hazard ratio (SE) [95%CI]: 1.13 (0.30) [0.68-1.89].
0.00
0.25
0.50
0.75
1.00
Cum
ulat
ive
pers
iste
nce
rate
to in
dex
med
icat
ion
0 50 100 150 200Days on Therapy
White Black
Kaplan-Meier Estimates for Matched Cohort
Current Multi-Site StudyCurrent Multi-Site Study
● ● CDC Patterns of Care Study (Breast and CDC Patterns of Care Study (Breast and Prostate CA)Prostate CA)
7 States + PSU7 States + PSURegistry data + CMS Registry data + CMS Centralized data processing (5 sites).Centralized data processing (5 sites).Pooled data analysisPooled data analysis
ModelModelNCCN guideline concordant careNCCN guideline concordant careHealth system and patient characteristicsHealth system and patient characteristicsCare coordinationCare coordination
ProposalProposal
● ● Develop set of cancer prevention and control outcomes consistent Develop set of cancer prevention and control outcomes consistent with CDC pattern of care studies.with CDC pattern of care studies. ● ● Seek funding to link 7 ACCN States Registry data to Seek funding to link 7 ACCN States Registry data to Medicaid/Medicare.Medicaid/Medicare. ● ● PSCI Health Services Research Core serve as Data and Support PSCI Health Services Research Core serve as Data and Support Center to provide:Center to provide:
IRB templatesIRB templatesData acquisition and linkage Data acquisition and linkage ArchiveArchiveAnalysis supportAnalysis support