Platelet-to-high-density lipoprotein cholesterol ratio (PHR) integrates hematologic and lipid-related information and has been associated with inflammatory and metabolic status. Its associations with Helicobacter pylori seropositivity and long-term mortality across Helicobacter pylori serostatus groups remain uncertain.

We analyzed NHANES 1999–2000 data using complex survey methods. Four progressively adjusted logistic models, categorical analyses, restricted cubic splines, and prespecified subgroup analyses assessed the association between PHR and Helicobacter pylori seropositivity. Four survey-weighted Cox models evaluated all-cause mortality according to Helicobacter pylori serostatus. Candidate prediction models were developed and internally evaluated using PSU-level nested cross-validation and a holdout test set. A contemporary hospital cohort of patients undergoing 13C-urea breath testing in 2020–2021 was analyzed as a clinical replication of the association.

Among 3190 participants, PHR was positively associated with Helicobacter pylori seropositivity after full adjustment (OR, 1.10; 95% CI, 1.05–1.15; P < 0.001). The highest versus lowest PHR quartile had an OR of 1.74 (95% CI, 1.24–2.44; P for trend < 0.001). In an independent contemporary hospital cohort of 428 participants, the positive association between PHR and Helicobacter pylori positivity was also observed after multivariable adjustment (OR, 1.117; 95% CI, 1.007–1.238; P = 0.036). Higher PHR was associated with increased all-cause mortality among Helicobacter pylori -seropositive participants (HR, 1.05; 95% CI, 1.01–1.09), but not among seronegative participants (HR, 0.94; 95% CI, 0.88–1.01). The selected weighted logistic regression model had an AUC of 0.761 and accuracy of 0.692 in the holdout test set.

Higher PHR was associated with Helicobacter pylori seropositivity in NHANES and with Helicobacter pylori positivity in a contemporary hospital cohort. Higher PHR was also associated with increased all-cause mortality among H. pylori-seropositive, but not seronegative, participants. The prediction model demonstrated internal predictive performance; however, its temporal transportability and clinical applicability in contemporary populations remain uncertain and require validation in independent contemporary cohorts.