Abstract / Summary
While immune checkpoint inhibitors (ICIs) have reshaped cancer care, population-level patterns of infection-related involvement at death among cancer decedents remain incompletely characterized. We analyzed >25 million US cancer deaths recorded in Centers for Disease Control and Prevention Wide-Ranging Online Data for Epidemiologic Research (CDC WONDER; 1999–2023), excluding deaths with COVID-19 as the underlying cause. Infection-related involvement was defined using infection-related multiple-cause-of-death codes among decedents with cancer as the underlying cause. To approximate treatment context at the population level, malignancies were grouped by tumor-site–based likelihood of ICI exposure (a proxy rather than observed receipt), and temporal changes were evaluated using Joinpoint regression and interrupted time-series (ITS) models anchored to 2015, corresponding to the period of broad clinical adoption of ICIs. The proportion of cancer-underlying-cause deaths with infection-related multiple causes increased from 7.98% in 2009–2013 to 8.65% in 2017–2019 (+0.67 percentage points) in the prespecified period comparison. Joinpoint analysis identified a 2015 inflection, shifting from a pre-2015 decline to a sustained post-2015 increase (APC, +1.42% through 2023). Post-2015, cancers in the high–ICI-likelihood group (including melanoma, kidney, lung, and stomach cancers) showed a larger relative increase in infection-related multiple causes than low–ICI-likelihood cancers (+16.2% vs +6.7%), and ITS models supported a steeper post-2015 slope after accounting for the pandemic period. These findings indicate that infection-related conditions listed on death certificates among US cancer decedents have increased since 2015, with larger rises observed in tumor sites more likely to be treated with ICIs. Further work using patient-level treatment data is needed to determine whether these temporal patterns reflect immunotherapy exposure, shifts in case mix, or changes in certification and coding practices.