Clinical trials of cancer treatments have established approaches with proven ability to cure a large proportion of women diagnosed at an early stage of breast and cervical cancer. Treatment effectiveness is less clear in non-trial settings. Observational studies tracking patterns of treatment and outcomes can clarify these issues. The primary data source for these studies has been cancer registries . a source with limited information on treatment.

HMOs routinely collect data on patient encounters for administrative tracking, quality control, billing and reimbursement purposes. These electronic data systems can allow identification of cancer patients and offer extensive information on treatment patterns and outcomes. If these electronic data sources are found to be complete and accurate, they would provide the opportunity to broaden the population of patients that can be included in studies of patterns of care and outcomes.

The IMPACT study, a collaboration among four members of the HMO Cancer Research Network: Fallon/Meyers Primary Care Institute, Kaiser Permanente Northern California, HealthPartners Research Foundation and Henry Ford Health System, was funded by NCI to determine the completeness and accuracy of HMO electronic data systems to identify breast and cervical cancer patients, treatment received, and delivery system
factors associated with differences in outcomes. Feifei Wei, from HPRF, is the Principal Investigator. Other investigators include Terry Field, Charles Quesenberry, Cheri Rolnick, and Marianne Ulcickas- Yood.

The specific aims of the IMPACT project are to:

• Determine the completeness and accuracy of HMO electronic data for identifying cancer patients, disease stage, and their treatment and outcomes among women age 55 or older with breast cancer and all women with cervical cancer.
  
  
• Determine the completeness and accuracy of claims-type encounter data for tracking treatment and outcomes among women age 55 or older with breast cancer and and all women with cervical cancer.
  
  
• Analyze variations in completeness and accuracy of these data by patient characteristics and among different HMO's
  
  
• Identify the biases associated with claims-type electronic data when they are used to characterize patterns of care and analyze the relationship between treatment and outcomes for women age with breast cancer or cervical cancer.

IMPACT developed very broad patient selection criteria to identify all women who might have developed breast or cervical cancers during specific time periods. Criteria included codes from a variety of sources, including: ICD-9 diagnosis and procedure codes, CPT codes, ICD-O codes and pathologic topography codes. The women who met these initial criteria were randomized for chart abstraction. Chart abstracted data included details of diagnosis and initial treatment, outcomes and comorbidity. The study also pulled electronic administrative data on enrollment

history, diagnoses, procedures, and dispenses of cancer-related drug.

The IMPACT investigators are in the midst of analyses to produce a series of manuscripts. Among the planned topics are the following: development and testing of algorithms for identifying breast and cervical cancer patients in electronic administrative data; characterization of patterns of care and relationships between treatment and outcomes among women age 55 or older with breast cancer and among women of any age with cervical cancer; patterns of use of tamoxifen among women with breast cancer; ability of various comorbidity indices to predict treatment patterns and to modify the relationship between treatment and outcomes; development and testing of algorithms for chemotherapy and radiation therapy in women with breast cancer in electronic administrative data; development and testing of algorithms for identifying recurrence in women with breast or cervical cancer in electronic administrative data; patient and physician characteristics associated with treatment patterns follow-up care for women age 55 or older with breast cancer.

Data collected for the IMPACT study are also serving as the base for an additional study by Dr. Chyke Doubeni, the recent recipient of a minority investigator research supplement from the National Cancer Institute.

-Terry Field, Meyers Primary Care Institute

<< Page 3