Validation of Claims Algorithms for Progression to Metastatic Cancer in Patients with Breast, Non-small Cell Lung, and Colorectal Cancer
Open Access
- 1 February 2016
- journal article
- research article
- Published by Frontiers Media SA in Frontiers in Oncology
- Vol. 6, 18
Abstract
Background: Validated algorithms for identifying progression to metastatic cancer could permit the use of administrative claims databases for research in this area. Objective: To identify simple algorithms that could accurately detect cancer progression to metastatic breast, non-small cell lung, and colorectal cancer using medical and pharmacy claims data. Methods: Adults with stage I–III breast, non-small cell lung cancer (NSCLC), or colorectal cancer (CRC) in the Geisinger Health System from 2004–2011 were selected. Evidence of progression was extracted via manual chart review as the reference standard. In addition to secondary malignancy diagnosis (ICD-9 code for metastases), diagnoses, procedures, and treatments were selected with clinician input as indicators of cancer progression. Random forests models provided variable importance scores. In addition to codes for secondary malignancy, several more complex algorithms were constructed and performance measures calculated. Results: Among those with breast cancer (17/502 [3.4%] progressed), the performance of a secondary malignancy code was suboptimal (sensitivity: 64.7%; specificity: 86.0%; positive predictive value [PPV]: 13.9; negative predictive value [NPV]: 98.6%); requiring malignancy at another site or initiation of immunotherapy increased PPV and specificity but decreased sensitivity. For NSCLC (61/236 [25.8%] progressed) codes for secondary malignancy alone (PPV: 47.4%; NPV: 84.8%; sensitivity: 60.7%; specificity: 76.6%) performed similarly or better than more complex algorithms. For CRC (33/276 [12.0%] progressed), secondary malignancy codes had good specificity (92.7%) and NPV (92.3%) but low sensitivity (42.4%) and PPV (43.8%); an algorithm with change in chemotherapy increased sensitivity but decreased other metrics. Conclusions: Selected algorithms performed similarly to the presence of a secondary tumor diagnosis code, with low sensitivity/PPV and higher specificity/NPV. Accurate identification of cancer progression likely requires verification through chart review.Keywords
Funding Information
- Amgen
This publication has 18 references indexed in Scilit:
- Identification of metastatic cancer in claims dataPharmacoepidemiology and Drug Safety, 2012
- A method to predict breast cancer stage using Medicare claimsEpidemiologic Perspectives & Innovations, 2010
- Evaluation of Three Algorithms to Identify Incident Breast Cancer in Medicare Claims DataHealth Services Research, 2007
- The Added Value of Claims for Cancer SurveillanceMedical Care, 2005
- An Algorithm for the Use of Medicare Claims Data to Identify Women with Incident Breast CancerHealth Services Research, 2004
- Can Administrative Data Identify Incident Cases of Colorectal Cancer? A Comparison of Two Health PlansHealth Services and Outcomes Research Methodology, 2004
- An approach to identifying incident breast cancer cases using Medicare claims dataJournal of Clinical Epidemiology, 2000
- Patients with Newly Diagnosed Carcinoma of the Breast Validation of a Claim-based Identification AlgorithmJournal of Clinical Epidemiology, 1999
- Evaluation of an Algorithm To Identify Women with Carcinoma of the BreastJournal of Medical Systems, 1997
- Analysis of a claims database for the identification of patients with carcinoma of the breastJournal of Medical Systems, 1994