Systemic Anticancer Therapy at the End of Life—Changes in Usage Pattern in the Immunotherapy Era

Abstract

This cohort study evaluates the rate of systemic anticancer therapy use among patients dying of cancer.

Use of systemic anticancer therapy (SACT) at end of life (EOL) is associated with increased acute care use,¹ delayed goals-of-care conversations,² late hospice enrollment, higher costs,¹ and possibly adverse quality and duration of life.³ In 2012, the American Society of Clinical Oncology and the National Quality Forum developed the quality measure Proportion Receiving Chemotherapy in the Last 14 Days of Life to promote reduction in chemotherapy and earlier integration of palliative care at EOL. Since 2012, the SACT landscape has changed owing to approvals of multiple new targeted therapies. Recent studies showed increasing checkpoint inhibitors use at EOL in patients with metastatic urothelial cancer,⁴ non–small cell lung cancer, and melanoma,⁵ despite no evidence that this practice is associated with improved outcomes. Previous studies were limited to specific cancer types and unable to associate EOL treatment rates with exact date of death.^4,5 We analyzed patterns in SACT near EOL across all cancer deaths, between 2015 and 2019, to understand changes in use of cytotoxic chemotherapy and targeted therapies.

Methods

We used the nationwide Flatiron Health electronic health record–derived database and included adult patients diagnosed with cancer beginning in 2011 who received treatment and died within 4 years of diagnosis. Flatiron Health maintains a longitudinal database containing deidentified patient-level structured and unstructured data curated via technology-enabled abstraction from approximately 280 US cancer clinics (approximately 800 sites and >2 million patients). This cohort study was approved by Flatiron Health Institutional Review Board, which waived the informed consent requirement because data were de-identified. eMethods in the Supplement provides the statistical analysis plan.

The primary outcome was SACT use at 30 days and 14 days before death. We assessed treatment subcategories, including any immunotherapy (alone or in combination with other therapy types) and chemotherapy alone. Treatment rates were estimated from multilevel logistic regression models among decedents at each practice and then summarized across all practices and cancer types, at both 30 days and 14 days before death. We examined treatment rates at EOL across the 6 most common cancer types within each treatment subcategory.

All analyses were conducted using R, version 3.6.1 (R Foundation for Statistical Computing), from February 2021 to April 2022.

Results

Over the study period, SACT use rate within 30 days of death across all cancer types combined did not change (39% in 2015 and 2019), with similar patterns observed for treatment within 14 days of death (17% in 2015 and 2019) (Figure 1). However, the type of systemic therapy received changed. We found overall decreases in chemotherapy alone (26% in 2015; 16% in 2019) and increases in immunotherapy (5% in 2015; 18% in 2019). These results were most noticeable in advanced non–small cell lung cancer and urothelial cancers, wherein we observed increases in SACT use rates at EOL that were associated with increasing checkpoint inhibitor use. Metastatic breast cancer, renal cell carcinoma, and colorectal cancer had slight decreases in overall SACT use at EOL, but there was almost no change for pancreatic cancer (Figure 2).

Figure 1. Adjusted Mean Treatment Rates Across All Cancer Types by Treatment Type and Year.

Patients were excluded if they had more than 90-day gap between index cancer diagnosis and first documented visit, more than 1 cancer diagnosis, or more than 6-month gap between last confirmed activity and date of death, or if they were treated at practices that were not in operation in 2015 or had less than 200 patient to physician ratios or fewer than 30 decedents. Multivariable models were adjusted for patient-level factors (age at diagnosis, sex, race and ethnicity, Eastern Cooperative Oncology Group performance status, insurance categories, primary cancer diagnosis, started second-line within 6 months of diagnosis, lines of therapy, and year of death) and practice-level factors (practice size, patients per physician ratio, practice type, region, proportion of Black patients, and proportion of patients with Medicaid). EOL indicates end of life.

Figure 2. Adjusted Practice-Level 30-Day End-of-Life Treatment Rates by Disease .

The same exclusion criteria were applied to the disease-specific models as those in overall cancer models (Figure 1). NSCLC indicates non–small cell lung cancer.

Discussion

We identified no difference in overall SACT use at EOL since 2015. Approval of multiple new immunotherapy agents has engendered a great replacement phenomenon, substituting immunotherapy for chemotherapy.⁶ Subsequently, although chemotherapy rates have declined in accordance with the American Society of Clinical Oncology and the National Quality Forum metric, increases in use of targeted therapies may have interfered with achieving the goal of earlier palliative care integration or reduction in acute care use.

Study limitations include the bias potential, unmeasured confounding, and misclassification in analysis. Patients may have received treatment outside the Flatiron Health network after being lost to follow-up.

One study suggests that any SACT at EOL, including immunotherapy, is associated with higher rates of downstream acute care, delayed hospice care, and higher costs.¹ This finding requires future research to examine the association of immunotherapy at EOL with downstream acute care use and quality of life in a larger, more representative sample.

Supplement.

eMethods