BACKGROUND: Prior to 2007, the erythropoiesis-stimulating agents (ESAs) epoetin alfa and darbepoetin alfa were indicated for use in chemotherapyinduced anemia to achieve target hemoglobin (Hb) levels of approximately 12 grams per deciliter (gm per dL), and treatment was to be withheld if Hb exceeded 13 gm per dL. In March 2007, the FDA changed the labeling of the ESAs to add boxed warnings, updated in November 2007, to include the following key points: (a) ESAs should be used only to treat anemia that occurs in patients with cancer while they are undergoing chemotherapy; (b) treatment with ESAs should be stopped when chemotherapy ends; and (c) dosing ESAs to an Hb target of 12 gm per dL or greater has resulted in more rapid cancer progression or shortened overall survival in patients with breast, head and neck, lymphoid, cervical, and non-small cell lung malignancies. In January 2008, the FDA specified that the increased risk of more rapid tumor growth or shortened survival was associated with ESAs when "administered in an attempt to achieve a Hb level of 12 gm per dL or greater, although many patients did not reach that level." A new black-box warning regarding this association was added to the labels of the ESAs in March 2008, and the FDA mandated further label changes on July 30, 2008, that ESA therapy should not be initiated in patients receiving chemotherapy at Hb levels of 10 gm per dL or higher. OBJECTIVE: To (a) assess the prevalence and predictors of ESA administrations at Hb levels above 12 gm per dL among patients with a diagnosis of solid or hematologic cancer or myelodysplastic syndrome who began their first regimen of conventional myelosuppressive chemotherapy between 2002 and 2006, and (b) describe patterns of ESA treatment subsequent to the first ESA administration at Hb above 12 gm per dL. METHODS: Using the Health Insurance Portability and Accountability Act (HIPAA)-compliant Varian Medical Oncology database of de-identified electronic medical records from 17 U.S. outpatient oncology practices, adults (aged 18 years or older) with any cancer diagnosis who began chemotherapy between January 1, 2002, and September 30, 2006, were identified. The Hb value associated with each ESA administration was defined as the closest Hb measurement within 7 days prior to the ESA administration. A first ESAHb > 12 was defined as the first time an ESA, either epoetin or darbepoetin, was given with an associated Hb greater than 12 gm per dL during the first chemotherapy regimen recorded in the database for each patient. Hb levels and ESA administrations after the first ESAHb > 12 were determined. Logistic regression models identified predictors of initial receipt of an ESAHb > 12, and of receiving further ESA treatment following the first such administration. RESULTS: Between January 1, 2002, and September 30, 2006, there were 17,731 patients on chemotherapy, the mean (SD) age was 60 (13.2) years; 58.9% were female; 24.6% had breast cancer, 22.2% had lung cancer, 15.8% had colorectal cancer, 11.8% had hematologic cancer, and 25.6% had other or multiple cancers. Of these, 8,086 (45.6%) received an ESA at any time during the regimen, and 7,606 (42.9%) received an ESA at a known Hb level (i.e., Hb measurement within 7 days prior to ESA administration). During the first recorded chemotherapy regimen, 1,844 patients (10.4% of the chemotherapy cohort, 24.2% of ESA users with a known Hb; n = 1,226 epoetin, n = 618 darbepoetin) received an ESAHb > 12. Among patients receiving ESA treatment at a known Hb level, significant predictors of receiving an ESAHb > 12 included treatment in a community-based clinic rather than a hospital-affiliated clinic (odds ratio [OR] = 2.96, 95% confidence interval [CI] = 2.40-3.65), location of practice in the eastern United States (OR for Midwest = 0.67, 95% CI = 0.57- 0.78; OR for West = 0.27, 95% CI = 0.22-0.34), hematologic cancer rather than solid tumor (OR = 1.44, 95% CI = 1.21-1.71), private health insurance (OR for public health insurance = 0.80, 95% CI = 0.70-0.93; OR for other/ unknown insurance = 0.54, 95% CI = 0.47-0.62), and year of regimen 2002- 2003 (ORs = 0.75, 0.74, and 0.71 for 2004, 2005, and 2006, respectively). Following the first ESAHb > 12, 276 (22.5%) of the patients on epoetin and 276 (44.7%) on darbepoetin received no further ESA treatment during the next 6 weeks (Pearson chi-square = 96.1, P < 0.001). CONCLUSIONS: This analysis of outpatient oncology practices between 2002 and 2006 revealed that 24% of ESA users with a known Hb level received ESAHb > 12. Dose withholding subsequently occurred in 23%- 45% of those patients. A higher proportion of patients on epoetin than darbepoetin continued ESA treatment after the first administration of ESAHb > 12.