Defining care provided for breast cancer based on medical record review or Medicare claims: information from the Centers for Disease Control and Prevention Patterns of Care Study
Introduction
Treatment guidelines, taking into account tumor and patient characteristics, have been promulgated for various types of cancer [1]. Despite this strive toward consistency, presumably within the bounds of evidence-based practice, there is room for different strategies of cancer care, based for example on sequencing of surgery, radiation (RT), and chemotherapy (CTX), and the choice/mix of chemotherapeutic agent(s). It becomes ever more important, therefore, to document treatment patterns and to study the efficacy of different patterns on clinical outcomes such as complications, recurrence, and survival. Medical claims and record review are two sources of information used to analyze treatment patterns; both have their limitations. Given the greater convenience and lower costs of claims data, ensuring their comparability with medical records is important.
In describing patterns of care for breast cancer, there are known discrepancies in medical claims compared with chart review or tumor registry data with respect to type of surgery and receipt of CTX and RT. Comparing Medicare claims to Surveillance, Epidemiology, and End Results (SEER) data, the level of agreement for type of surgery was 88.0% for breast-conserving surgery (BCS) and 94.5% for mastectomy and the sensitivities, claims to SEER, were 91.3% and 87.9% for BCS and 96.2% and 96.0% for mastectomy using either inpatient or physician claims, respectively [2]. For determining use of CTX, the sensitivity of Medicare claims to chart review was 91% [3]. In another study [4] there was a high level of agreement between SEER-Medicare claims and National Cancer Institute-supported Patterns of Care Studies (POC) medical chart review data for use of CTX (kappa 0.73), and a sensitivity of Medicare claims to re-abstracted data of 88% for breast cancer. With respect to RT treatment, comparison of Medicare claims with SEER data finds a sensitivity of 93%, claims to SEER data [5] and comparison of SEER-Medicare linked data to claims data report high levels of agreement in lung (88%), prostate (94%), rectal (94%), breast (94%), and endometrial cancers (95%) [6].
The level of agreement between claims and records according to most of these studies seems to be reasonably high. However, there is little evidence in the literature regarding whether strategies of care with multiple components (surgery, RT, CTX) or specific multiple agent chemotherapeutic regimens have as high a level of agreement. The purpose of this study was to examine the level of agreement between the two data sources regarding various treatment patterns for a sample of breast cancer patients. Our study uses cancer registry data augmented by medical record re-abstraction as the gold standard to which claims data are compared. This study advances the literature by comparing claims with records for both single and multiple components of cancer therapy, including examination of specific chemotherapeutic agents.
Section snippets
Methods
We used data from the U.S. Centers for Disease Control and Prevention's National Program of Cancer Registries Breast and Prostate Cancer Data Quality and Patterns of Care study, a cross-sectional study of patterns of care for breast and prostate cancer. In this study, data pertaining to cancer and its treatment from cancer registries in seven states (California, Georgia, Kentucky, Louisiana, Minnesota, North Carolina, and Wisconsin) were augmented by abstracting from various medical record
Medical records
Abstractors from each of the states re-abstracted treatment information from in- and out-patient medical records and verified it with treating physicians when medical records were insufficient for confirming treatment or there was an indication that information on therapy was incomplete. Comparison of re-abstracted to original frozen registry data is reported elsewhere [8]. Thus, each patient was assigned one of the following physician verification codes: (1) No physician verification; (2)
Statistical analysis
Our primary analysis compares the elements of treatment (BCS ± nodes, mastectomy ± nodes, RT, CTX) from claims with the same elements derived from the re-abstraction of medical records. We report concordant and discordant cases, sensitivity, and kappa statistics for each treatment component (e.g., surgery, RT, CTX) and all possible combinations of these components, and interpret the kappa statistics using the following acceptable rules [9]: excellent (0.81–1.00), substantial (0.61–0.80),
Results
Table 1 reports the distribution of patient characteristics for the 1762 women included in our sample and the 653 women also older than 65 who were excluded because they were lacking complete Medicare Part A and B coverage, were enrolled in an HMO, or because of the absence of any postdiagnosis claims. Excluded women tended to be younger, have more advanced stage, have a “private” source of payment, be black or Hispanic, and come from urban areas, a finding consistent with the exclusion of HMO
Discussion
Our study compared Medicare claims with medical record review as sources for information on treatment strategies for breast cancer. The results showed that claims are a highly sensitive source of information on BCS or mastectomy with nodal assessment, RT therapy, CTX, and various multimodality treatment combinations. Furthermore, agreement of Medicare claims with medical records was excellent for multi-agent CTX regimens overall and substantial for several specific regimens, including agent
Acknowledgments
The Breast and Prostate Cancer Data Quality and Patterns of Care Study was supported by the Centers for Disease Control and Prevention through cooperative agreements with the California Cancer Registry (Public Health Institute) (1-U01-DP000260), Emory University (1-U01-DP000258), Louisiana State University Health Sciences Center (1-U01-DP000253), Minnesota Cancer Surveillance System (Minnesota Department of Health) (1-U01-DP000259), Medical College of Wisconsin (1-U01-DP000261), University
References (24)
- et al.
1999. Information on radiation treatment in patients with breast cancer: the advantages of the linked Medicare and SEER data
J Clin Epidemiol
(1999) - et al.
Medicaid data as a resource for epidemiologic studies: strengths and weaknesses
J Clin Epidemiol
(1989) Using administrative data to identify associations between implanted medical devices and chronic diseases
Ann Epidemiol
(2000)- National Comprehensive Cancer Network. Clinical practice guidelines in oncology....
- et al.
Developing a claims-based version of the ACE-27 Comorbidity Index: a comparison with medical record review
Med Care
(2011) - et al.
External validation of Medicare claims for breast cancer chemotherapy compared with medical chart review
Med Care
(2006) - et al.
Utility of the SEER-Medicare data to identify chemotherapy use
Med Care
(2002) - et al.
Studying radiation therapy using SEER-Medicare-Linked data
Med Care
(2002) - et al.
Epidemiologic methods
(2003) - et al.
Quality of cancer registry data: findings from CDC-NPCR's Breast and Prostate Cancer Data Quality and Patterns of Care Study
J Registry Manag
(2011)
The reliability of medical record review for estimating adverse event rates
Ann Intern Med
Claims data linked to hospital registry data enhance evaluation of the quality of care of breast cancer
J Surg Oncol
Cited by (6)
Researching the Appropriateness of Care in the Complementary and Integrative Health Professions Part 5: Using Patient Records: Selection, Protection, and Abstraction
2019, Journal of Manipulative and Physiological TherapeuticsCitation Excerpt :However, other sources of data (eg, patient interviews, administrative documents) may be more feasible. Comparisons of administrative and medical record data have revealed that certain aspects of care (eg, surgery, radiation treatments) are reliably accessed from either source.2,3 But other data elements (eg, time of symptom onset, contraindications, patient refusals of offered care) are less likely found in the medical record.4,5,6
Validity of Administrative Databases in Comparison to Medical Charts for Breast Cancer Treatment Data
2018, Journal of Cancer EpidemiologyCharacteristics of long-term survivors of epithelial ovarian cancer
2015, Obstetrics and GynecologyT<inf>H</inf>2-polarized CD4<sup>+</sup> T Cells and macrophages limit efficacy of radiotherapy
2015, Cancer Immunology ResearchA review of the use of medicare claims data in plastic surgery outcomes research
2015, Plastic and Reconstructive Surgery - Global Open