Patient Performance Status: Prediction Model Development and Validation (Davidoff, PI) Project Summary Cancer is a leading cause of morbidity and mortality, particularly among the elderly in the U.S., and treatment of cancer accounts for a substantial portion of health care spending. Clinical trials are the key mechanism for studying efficacy of new therapies, but retrospective data analyses using administrative (e.g. insurance enrollment and claims) data provide an important complement in examining treatment patterns, disparities in treatment, and effects on survival and costs of care for cancer patients. Although these latter studies commonly control for co-morbid conditions present in patients with cancer, one of the significant limitations is that they cannot control for patient performance status (PS). PS, a measure of functional status and physical performance, is used by clinicians to determine whether a cancer patient receives any treatment, and type and intensity of treatment. When the effects of cancer treatment on survival and costs are analyzed without controls for PS, the resulting estimates may be biased. The objective of this application is to develop a method to incorporate information on PS into analyses of cancer treatment that rely on administrative data. The approach will be to develop a multivariate regression model for PS using a dataset that has PS, enrollment and claims data. The model will be designed so that the independent variables are derived solely from data elements available in enrollment and claims files. The estimated coefficients from the model can be used to predict PS in a dataset that has enrollment and claims data but no direct measure of PS;the predicted PS can be incorporated into analyses using those data.
The specific aims for this project are to: 1) develop a prediction model for PS based on demographic characteristics and measures generated from diagnostic and procedure codes in insurance claims data;and 2) examine predictive validity of the PS model. The source of data for model development will be the 1997-2006 Medicare Current Beneficiary Survey (MCBS), a nationally representative survey of Medicare beneficiaries linked to Medicare Part A and B claims data. The MCBS includes detailed information on functional status and physical performance that will be used to measure PS. Enrollment and claims data will be used to identify and test candidate predictor variables for the model. The second source of data, used in the validation step, will be Surveillance, Epidemiology, and End Results (SEER) cancer registry data from 1997-2002, linked to Medicare claims. The project team recently completed analyses of treatment patterns and survival for elderly Stage III colon cancer and locally advanced non-small-cell lung cancer. The PS model validation will incorporate predicted PS into those previously estimated models to test explanatory power, and to assess the impact of including PS on estimated effects of comorbidity and age on treatment, and effects of treatment on survival. Further model refinement and comparative effectiveness research using this model will be proposed as part of a subsequent R01 grant submission.
Retrospective data analyses using enrollment and claims data provide an important complement to clinical trials in examining use of chemotherapy, disparities in use, and effects on survival of cancer patients. A significant limitation to these analyses is that they cannot control for patient performance status (PS), a measure of functional status used by oncologists to determine treatment, and that has independent effects on survival. The objective of this application is to develop and validate a multivariate prediction model to incorporate information on PS into claims-based analyses. The results of this work will address a key methodologic limitation to many current studies, thus enhancing our understanding of factors that affect treatment for elderly cancer patients, and the effects of those treatments on survival.
| Deepak, Janaki A; Ng, Xinyi; Feliciano, Josephine et al. (2015) Pulmonologist involvement, stage-specific treatment, and survival in adults with non-small cell lung cancer and chronic obstructive pulmonary disease. Ann Am Thorac Soc 12:742-51 |
| Feliciano, Josephine; Gardner, Lisa; Hendrick, Franklin et al. (2015) Assessing functional status and the survival benefit of chemotherapy for advanced non-small cell lung cancer using administrative claims data. Lung Cancer 87:59-64 |
| Davidoff, Amy J; Gardner, Lisa D; Zuckerman, Ilene H et al. (2014) Validation of disability status, a claims-based measure of functional status for cancer treatment and outcomes studies. Med Care 52:500-10 |
| Davidoff, Amy J; Zuckerman, Ilene H; Pandya, Naimish et al. (2013) A novel approach to improve health status measurement in observational claims-based studies of cancer treatment and outcomes. J Geriatr Oncol 4:157-65 |
