Monocyte-Containing Inflammatory Indices Show Stronger Association with 30-Day Mortality than the Systemic Immune-Inflammation Index in Elderly Sepsis: A Single-Center Retrospective Observational Cohort Study

Background. Hematological inflammatory indices from the complete blood count have been proposed as inexpensive prognostic markers in sepsis. The systemic immune-inflammation index (SII) and neutrophil-to-lymphocyte ratio (NLR) are the most studied, but the performance of monocyte-containing alternatives (SIRI, AISI) in the elderly, in whom immunosenescence may alter the leukocyte phenotype, remains poorly characterized. Methods. In a single-center retrospective cohort of patients aged ≥65 years admitted to a tertiary ICU with Sepsis-3-defined sepsis (n = 127, 33 deaths), we compared the discrimination of six indices (NLR, PLR, MLR, SII, SIRI, AISI) for 30-day all-cause mortality using AUROC with bootstrap confidence intervals and pairwise DeLong tests. Independent associations were assessed by logistic regression adjusted for APACHE II and age; incremental value over APACHE II was explored using IDI, cNRI, calibration and decision curve analysis, with bootstrap optimism correction. Results. Thirty-day mortality was 26.0%. The monocyte-containing indices (AISI, SIRI, MLR) discriminated better than SII and NLR, and AISI was significantly superior to SII, NLR and PLR on DeLong testing, though not to SIRI, MLR or APACHE II. After adjustment for APACHE II and age, AISI, SIRI and MLR remained independently associated with mortality, whereas SII and PLR did not. Adding AISI to APACHE II improved reclassification and calibration and yielded higher net clinical benefit across clinically relevant thresholds. Conclusions. In this exploratory, single-center analysis, monocyte-containing indices, particularly AISI, were more strongly associated with 30-day mortality in elderly ICU sepsis than SII or NLR. AISI, SIRI and MLR were strongly intercorrelated and near-equivalent, and AISI did not significantly exceed APACHE II in discrimination. These hypothesis-generating findings require prospective external validation before clinical use.