BACKGROUND/RATIONALE:
Approximately 35,000 new cases of cancer occur annually and cancer is the second leading cause of death among veterans. The quality of supportive cancer care is widely recognized as inadequate, yet national efforts underway to evaluate and monitor the quality of cancer care do not address supportive care.

OBJECTIVE(S):
The Cancer Quality-ASSIST (Assessing Symptoms Side Effects and Indicators of Supportive Treatment) indicators address a critical gap in quality measurement, and they are being piloted by the National Cancer Institute (NCI) and the Agency for Healthcare Research and Quality (AHRQ) at Johns Hopkins University. ASSIST quality indicators address practice-level performance, multi-disciplinary care, and span care settings and represent the most comprehensive toolkit ever developed for evaluating supportive cancer practice. The objectives of this study were to further develop this toolkit so that it could be used in a national study of the quality of supportive care of veterans with cancer by:
-Pilot testing an abstraction tool for obtaining data on the quality of supportive and end of life care of veterans with cancer
-Identifying a robust set of quality indicators for supportive cancer care that can be used with VA electronic medical records and developing the technical basis for implementing them, and
-Evaluating an administrative data-based strategy for cohort identification.

METHODS:
We used the VA GLAHS cancer registry to identify patients veterans diagnosed with stage IV solid tumors in 2006. Experienced abstractors abstracted data from patients and data elements were aggregated to form the specified quality indicators. Reliability of the quality indicators was assessed by estimating percent agreement and kappa for a sample of charts abstracted by all abstractors. Validity was evaluated by implicit review. As a secondary aim, VA GLA cancer registry data was merged with visit-level and discharge diagnoses codes from the VISN22 datawarehouse in order to evaluate the validity of an administrative-data algorithm to identify patients with cancer, using the cancer registry as the gold standard.

FINDINGS/RESULTS:
Overall, 61 indicators appeared promising measures of the quality of supportive care; 14 of these are undergoing additional evaluation using different specification of the data elements to improve reliability. An additional 25 indicators will need additional evaluation of inter-rater reliability in a larger sample as reliability was not evaluable due to low prevalence of the condition. A Computerized Abstraction Tool is now available for future efforts to evaluate the quality of supportive and end of life care of veterans with cancer, along with detailed guidelines and training materials for data abstraction from patient medical records and analyses plans for the quality indicators. VISN administrative data did not have adequate sensitivity or specificity for cohort identification.

IMPACT:
A robust toolkit now exists to evaluate the quality of supportive cancer care across the VA Healthcare System.

---

External Links for this Project

---

PUBLICATIONS:

Journal Articles

1. Dy SM, Lorenz KA, O'Neill SM, Asch SM, Walling AM, Tisnado D, Antonio AL, Malin JL. Cancer Quality-ASSIST supportive oncology quality indicator set: feasibility, reliability, and validity testing. Cancer. 2010 Jul 1; 116(13):3267-75. [view]

Journal Other

1. Lorenz KA, Malin JL. How might VA-Medicare differences inform the delivery of end-of-life cancer care? Cancer. 2010 Aug 1; 116(15):3533-6. [view]

DRA: Health Systems
DRE: Treatment - Observational
Keywords: Cancer, End-of-life, Quality Measure
MeSH Terms: none