Disparities in Care Management During Terminal Hospitalization Among Adults With Metastatic Cancer From 2010 to 2017

Key Points Question Is variation in care management during terminal hospitalization among adults with metastatic cancer associated with sociodemographic status? Findings In this cross-sectional study of 21 335 patients with metastatic cancer who died in the hospital, racial and ethnic minority patients and those with Medicare or Medicaid coverage were more likely to receive low-value, high-cost aggressive medical interventions at the end of life. Meaning This study’s findings suggest that identifying and understanding factors associated with the observed disparities will be helpful to inform communications with patients with metastatic cancer about end-of-life care.


Hospital characteristics
The HCUP obtained hospital location and teaching status from the AHA Annual Survey of Hospitals, and defined "urban" hospitals as being located in a metropolitan statistical area and "rural" hospitals as being located in a non-metropolitan statistical area. Urban hospitals were subdivided according to teaching status, with "urban teaching" hospitals defined as either having an AMA-approved residency program, being a member of the Council of Teaching Hospitals (COTH), or having a ratio of full-time equivalent interns and residents to beds of .25 or higher. Rural hospitals were not subdivided by teaching status because this category was rare. The HCUP also obtained hospital census region from the AHA Annual Survey of Hospitals, with census region defined by the U.S.
Census Bureau as either Northeast, Midwest, South, or West 2 .

Admission through the ED
The HCUP defines a data element indicating records with evidence of emergency department (ED) services, and guidelines indicate this data element as the most comprehensive indicator available in the NIS of inpatient admissions through the ED. Services captured by this data element include ED revenue codes of 450-459 on record, a positive ED charge (when revenue center codes are not available), ED CPT codes (99281-99285) reported on record, a condition code of P7 reported on record, point of origin of ED, or admission source of ED 2 . Alternative non-ED admission sources include transfers from a different acute care hospital, transfers from a different health facility including long-term care facilities, admission from court or law enforcement, admission from an outpatient facility or clinic, or physician referral 2 . The HCUP acknowledges the possibility of a patient incurring ED services with coding information that may not have been captured in the HCUP record, and therefore the possible under-representation of ED admissions based on HCUP coding. We utilized this data element as an indicator of patient admission from the ED.

Time from hospital admission to death
The HCUP codes the length of the inpatient stay in days, calculated by subtracting the admission date from the discharge date, ranging from 0 (same-day stay) to 365 days, and without subtracting leave days. The HCUP utilizes the length of stay supplied from the data source if it cannot be calculated, and codes it as missing if it cannot be calculated and is not supplied 2 .

Total charges billed to insurance
The HCUP reports total charges billed to insurance, generally not including professional fees and noncovered charges. These NIS reported total charges may include emergency department charges incurred prior to hospital admission, as Medicare requires a bundled bill for patients admitted to the hospital through the emergency department. HCUP coding rounds charges to the nearest dollar and sets zero charges as missing; HCUP coding of total charges was analyzed in this study 2 .

Systemic therapy
Patients were coded as receiving systemic therapy during their inpatient stay if one of the first 15 coded HCUP procedure CCS codes was 224 ("cancer chemotherapy"), a category which includes both systemic chemotherapy and immunotherapy procedures. The ICD-9-CM and ICD-10-CM/PCS codes corresponding to specific procedures within this category are listed in eTable 1.

Invasive mechanical ventilation
Patients were coded as receiving invasive mechanical ventilation during their inpatient stay if one of the first 15 coded ICD-9-CM and ICD-10-CM/PCS procedures was one of the codes corresponding to this intervention, listed in eTable 1. These codes were selected based on their inclusion in the HCUP procedure CCS category 216 ("respiratory intubation and mechanical ventilation"), with specific ICD codes indicating non-invasive mechanical ventilation removed.