Association of proton pump inhibitor use with cancer-specific mortality in a nationally representative U.S. cohort (NHANES, 1999–2018).

10625 Background: Proton pump inhibitors (PPIs) are widely used and often continued beyond recommended durations. Evidence linking chronic PPI use to adverse cancer outcomes and mortality is mixed. Potential mechanisms include hypergastrinemia, altered gastric mucosa, and microbiome changes that may influence tumor behavior and treatment response. We assessed the association between baseline PPI use and cancer-specific mortality in a nationally representative U.S. cohort. Methods: Ten NHANES cycles (1999–2018) linked to the NHANES Linked Mortality Files (1999–2018) were pooled. Adults ≥18 years who were mortality-eligible and had examination follow-up time (PERMTH_EXM) were included. A formal written protocol was not developed; the analysis plan (study objective, exposure, endpoint, covariates, and primary models) was prespecified through discussion with the study team prior to final analyses. Baseline PPI exposure was defined from the 30-day prescription medication inventory (container-verified when available). A curated text algorithm identified PPIs using generic/brand/OTC names and fixed-dose combinations. Cancer death was defined as leading cause of death = malignant neoplasm; non-cancer deaths were censored (cause-specific hazard). Survey-weighted Cox models incorporated strata/PSU and pooled MEC weights (WTMEC2YR/10). Models used complete-case covariates per model. Results: Among 56,253 adults (1,850 cancer deaths), baseline PPI use was associated with higher cancer-specific mortality in unadjusted analysis and remained elevated after sequential adjustment (Table). Conclusions: PPI use was associated with higher cancer-specific mortality in a nationally representative U.S. cohort. These findings support careful review of PPI indication and duration and reinforce deprescribing when PPIs are not clearly indicated. Residual confounding remains possible, warranting studies using designs that better address this bias. Sequentially adjusted survey-weighted Cox models estimating the cause-specific hazard of cancer death associated with PPI use (NHANES 1999–2018). Model N Events HR (CI) Model 0: Unadjusted 56253 1850 2.63 (2.21 - 3.12) Model 1: +Age +Sex 56253 1850 1.36 (1.14 - 1.62) Model 2: +Race +PIR 51140 1695 1.40 (1.17 - 1.68) Model 3: +Diabetes +Hypertension +HLD 51140 1695 1.40 (1.18 - 1.67) Model 4: +log(cotinine)+log(metals) 9538 407 1.45 (1.10 - 1.91) PIR = Poverty Income Ratio. HLD = Hyperlipidemia. Metals = lead, cadmium, mercury. HRs represent cause-specific hazards for cancer death; non-cancer deaths were treated as censoring events.