Using SEERMedicare Data to Enhance Registry Data to Assess Quality of Care

1
Using SEER-Medicare Data to Enhance Registry
Data to Assess Quality of Care

Joan Warren Applied Research Program National
Cancer Institute
NAACCR June 6, 2007
2
What are the SEER - Medicare data?

• The SEER-Medicare data are the result of the
  linkage of two large population-based sources of
  data cancer registry data from NCIs sponsored
  cancer registries (SEER program) and Medicare
  claims from CMS
• The SEER registries collect detailed clinical,
  demographic and cause of death information for
  persons with cancer
• Medicare data are longitudinal, with claims for
  all covered fee-for-service health care from the
  time of eligibility to death
• There are currently over 1.8 million cases in the
  data

3
Why link the SEER-Medicare Data?  
The linked data can be used for a number of
analyses that span the course of cancer control
activities Diagnosis/ Tx ? Survivorship
? Second Occurrence ? Terminal Care

Rates of second primaries Relationship of
second events to initial treatment and ongoing
surveillance
Patterns of care Peri-operative
complications Volume outcomes studies Extent
of staging Comorbidities
Late effects of treatment Post-diagnostic
surveillance Treatment of prevalent
cancers Survival
Use of hospice services Patterns of care during
the last year of life
Quality of care, health disparities, and cost of
treatment
4
Persons included in the SEER-Medicare Data

• 100 of patients in the SEER data who are found
  to be Medicare eligible
• 5 random sample of persons residing in the SEER
  areas who have not been diagnosed with cancer
• These people can be used to create comparison
  groups as well as to create estimates of
  diagnostic testing and treatment practices in the
  entire population
• Medicare files available for the non-cancer cases
  are the same as for the cancer cases

5
What is included in the SEER-Medicare Data

• SEER Data including
• Incidence, site, stage, initial tx, demographics
  and vital status
• Medicare claims for
• Short stay hospitals
• Physician and lab services
• Hospital outpatient claims
• Home health and hospice bills

6
Other variables available in the SEER-Medicare
data

• 1990 and 2000 Census data at the census tract and
  zip code level for ecological SES measures
• Health Care Service Area from Area Resource File
• Hospital and physician characteristics- ex.
  bedsize, hospital ownership, physician specialty

7
Years of SEER-Medicare Data Available

• SEER data are available for the entire time a
  registry has participated in the SEER program
  some registries go back to 1973
• Medicare claims are available from 1991-2005,
  except for hospital data that are available back
  to 1986
• Cases reported through 2002
• Update of the linkage is underway. It will
  include cases through 2005 with Medicare claims
  through 2006.

8
Limitations of the SEER-Medicare Data

• Observational data- pts are not randomly assigned
  to treatment
• Non-covered services excluded prescription
  drugs, long-term care, free screenings
• Reasons for tests are not known this raises
  challenges w/measuring screening
• Results of tests not available
• Does not include claims for care provided to
  persons in HMOs (about 22 in SEER areas)
• Under 65 population includes only the
  disabled/ESRD

9
Using the SEER-Medicare Data to Assess Quality of
Cancer Care

• The SEER-Medicare data are a good resource to
  measure quality of cancer care
• Data are longitudinal
• Can look at claims prior to diagnosis to adjust
  for pre-existing conditions
• Cross most components of the health care system
• Challenges of using these data to assess quality
  of care
• Secondary data do not capture factors that may
  influence treatment choices especially an issue
  in the elderly
• There are a limited number of treatments for
  which there is consensus regarding treatment

10
Examples of Quality of Care Studies Using
SEER-Medicare Data

• Investigators have used SEER-Medicare data to
• Assess if patients received routinely provided
  care-
• Surgery
• Adjuvant therapy (RT/Chemo)
• Post-diagnostic surveillance
• Examine health system factors related to outcomes
• Hospital and physician characteristics
• Volume outcomes

11
Are All Medicare Beneficiaries with Early-Stage
Non-Small Cell Lung Cancer Receiving Potentially
Curative Surgery?

• Black persons with early stage non-small cell
  lung cancer have poorer survival than do
  comparable white persons
• Early-stage non-small cell lung cancer is
  potentially curable by surgical resection
• Investigators used SEER-Medicare data to estimate
  the rates of surgical treatment between blacks
  and whites and to determine if disparities in
  survival could be explained by differences in use
  of surgery

12
Survival of Medicare beneficiaries aged 65 with
Stage I/II non-small cell lung cancer, by
treatment and race, 1985-1993
13
CONCLUSIONS

• The lower survival rate among black patients
  with early-stage, non-small-cell lung cancer, as
  compared with white patients, is largely
  explained by the lower rate of surgical treatment
  among blacks.

14
Use of Adjuvant Chemotherapy for Medicare
Beneficiaries with Stage III Colon Cancer

• Use of adjuvant chemotherapy following a
  diagnosis of Stage III colon cancer has been
  guideline treatment for many years
• There are concerns that some patients are not
  receiving adjuvant treatment because of their age
  or race
• Investigators have used the SEER-Medicare data to
  assess use of adjuvant chemo in Medicare
  beneficiaries with Stage III colon cancer

15
Receipt of Adjuvant Chemotherapy for Medicare
Beneficiaries with Stage III Colon Cancer by Age
Group
Schrag et al, JNCI 2001
16
Referral to Medical Oncologist and Receipt of
Chemotherapy Among Those Who Saw an Oncologist
Among Medicare Beneficiaries with Stage III Colon
Cancer
Percent
Saw a Medical Oncologist
Received Chemotherapy
Baldwin LM, et al. JNCI Aug 2005.
17
Assessment of Post-diagnostic Surveillance

• SEER-Medicare data have been used to evaluate
  whether
• patients are receiving the recommended
  surveillance
• following a cancer diagnosis
• Persons with superficial bladder cancer who have
  not undergone total cystectomy should undergo
  bladder surveillance with cystoscopy every 3-6
  months
• Men with prostate cancer who opt for expectant
  management should have a PSA test every 6 months

18
Surveillance among Medicare Eligible Patients
with Superficial Bladder Cancer over a 30-month
interval following diagnosis, by Age Group
Source Schrag D et al. J Natl Cancer Inst. 2003
Apr 16.
19
Receipt of PSA Testing 7-24 Months Following a
Diagnosis of Prostate Cancer for Men Choosing
Expectant Management
Shavers, et al., Medical Care 2004
20
Conclusions

• Bladder surveillance Only 40 of the cohort
  received the recommended surveillance
• PSA tests African Americans and Hispanics were
  significantly less likely to receive a PSA test.
  Black men are more likely to be treated with
  expectant management.

21
Does Provider Specialty or Provider Volume Impact
on Patient Outcomes?

• Earlier studies have suggested that provider
  specialty and/or volume may improve patient
  outcomes
• Investigators used the SEER-Medicare data to
  compare outcomes for women following surgery for
  ovarian cancer
• Two studies were done
• Does the specialty of the physician performing
  the surgeon impact on overall survival ?
• Is there a volume-outcome effect?

22
Adjusted Cox proportional hazards model for
death from any cause for Medicare women with
ovarian cancer
P Earle CC et al. JNCI Feb 1 2006
23
Percent of Patients with Stage III/IV Ovarian
Cancer Surviving 48 Months After Surgery by
Hospital and Surgeon Volume
Schrag D. et al. J Natl Cancer Inst. 2006 Feb 1.
24
Conclusions About Ovarian Cancer Treatment

• These data show that the volume of procedures is
  not a significant factor in patient survival
• It appears that physician training is associated
  with improved outcomes

25
Final Thoughts About Using SEER-Medicare Data to
Assess Quality of Care

• Secondary data sources such as SEER-Medicare can
  be a powerful source of information because of
  their size and breadth
• However, these types of data do not offer
  definitive information about quality- why was
  treatment not given, what other factors
  influenced outcomes
• These data should be used to determine where more
  in-depth research should be focused.

26
More Details on the SEER-Medicare data

• SEER-Medicare WEB site http//appliedresearch.canc
  er.gov/seermedicare