Spending Among Cancer Patients between ACO and Non-ACO Participants

Abstract

Of the many initiatives aimed at improving the value of healthcare services in the United States, Accountable Care Organizations (ACOs) in particular are receiving substantial policy attention from both sides of the political aisle. Cancer is a common and costly condition with considerable variation in intensity and quality of care delivered. While early evidence from ACOs shows modest reductions in healthcare costs for Medicare beneficiaries overall, we are unaware of any data looking specifically at the impact of ACOs on spending for patients with cancer. We studied the 2014 claims of a 20% sample of Medicare fee-for-service (FFS) beneficiaries, age 65 and over who were continuously enrolled, and were age 65 or older. We used ACO assignments of patients by CMS to identify those in an ACO vs not in an ACO. We identified patients with breast, central nervous system, gastrointestinal, gynecologic, head and neck, lung, lymphoma and sarcoma malignancies using ICD9 codes. We identified beneficiary age, race, sex, Medicaid eligibility as an indicator of poverty, and chronic conditions as covariates. We calculated mean annual standardized costs overall as well as stratified by the following categories of spending: inpatient, outpatient, physician, Part D, skilled nursing facility, home health, and hospice. Physician spending was further categorized as follows: evaluation and management, procedures, imaging, durable medical equipment and other. Spending was adjusted for patient characteristics and by chronic conditions, using the Chronic Conditions Warehouse. We repeated our analyses stratified by individual tumor types. Our sample consisted of 913,942 beneficiaries with solid tumors, of whom 21% were in an ACO. ACO beneficiaries had modest but significantly higher total annual standardized costs compared to beneficiaries not in an ACO ($22,576 vs. $21, 484; P < 0.001). The ACO group also had higher inpatient, overall physician spending, skilled nursing facility and home health spending but lower hospice spending compared to non-ACO beneficiaries ($543 vs. $783; P < 0.001). There were no differences between the two groups with respect to Part D ($3,498 vs. $3,542; P = 0.097) and outpatient spending ($4,452 vs. $4,399; P = 0.13). When we further stratified physician spending, spending on ACO patients was greater for five out of six categories, with durable medical equipment being the exception. The results were similar when we looked at spending for individual cancers. Standardized total spending among patients with solid cancers treated in an ACO was significantly higher compared to those not in an ACO. This study suggests that ACOs have not yet started targeting cancer care to lower healthcare spending or if they have, they have not yet made much headway. Given that cancer represents a substantially expensive set of conditions, it may be a primary target for future ACO interventions.