Blood Inflammation Score Linked to Worse Outcomes in Liver Abscess Patients

A single blood test value calculated from routine cell counts was associated with worse short-term outcomes in patients hospitalized with pyogenic liver abscess, according to a retrospective cohort study of 670 adults treated at Linyi People’s Hospital in China between January 2021 and December 2024. The measure, called the systemic inflammation response index, or SIRI, combines three cell types from a standard complete blood count: neutrophils, monocytes, and lymphocytes. Patients whose admission SIRI was higher had greater odds of experiencing a serious clinical event within 14 days, although the study’s authors caution that the marker’s ability to actually discriminate between patients who will fare well and badly was limited.

Pyogenic liver abscess is a bacterial infection that produces a pus-filled cavity in the liver. Typical symptoms include fever, abdominal pain, and chills, and severe disease can progress to sepsis, septic shock, multiorgan dysfunction, and death. Prognostic assessment currently relies on age, comorbidities, liver and kidney function, microbiologic findings, abscess characteristics, and the response to treatment. In recent years, composite inflammatory indices derived from peripheral blood cell counts, including the neutrophil-to-lymphocyte ratio, have been evaluated as prognostic markers in infectious diseases and in liver abscess specifically, but evidence on SIRI’s role in this condition remained limited.

The study team identified 801 records from their hospital database, excluded 23 that did not meet eligibility criteria, and removed 108 verified duplicates, leaving 670 unique patients. All were adults with confirmed pyogenic liver abscess who had a complete blood count drawn within 24 hours of admission and before major therapeutic intervention, along with complete 14-day outcome data. Patients with amebic, fungal, tuberculous, or parasitic abscesses, hematologic malignancy, or recent chemotherapy, radiotherapy, or immunosuppressive therapy were excluded. SIRI was calculated from that first blood sample, with cell counts expressed per liter.

The primary outcome was a composite of adverse clinical events within 14 days of admission. Of the 670 patients, 100, or 14.9 percent, experienced at least one such event. The recorded components included four deaths, 35 intensive care unit transfers, 27 cases of septic shock, 51 escalations of antimicrobial therapy because infection remained uncontrolled, eight cases of abscess enlargement, two cases in which abscess size did not shrink by day 14, and 21 patients whose temperature remained above 38 degrees Celsius on day 14. Components could overlap, and each patient was counted once. Fourteen-day mortality was 0.60 percent, with an exact 95 percent confidence interval of 0.16 to 1.52 percent.

Median admission SIRI was markedly higher among patients with adverse outcomes: 8.95, with an interquartile range of 4.14 to 15.64, compared with 5.80, with an interquartile range of 2.93 to 11.64, among those who fared well, a difference that was statistically significant. Patients with adverse outcomes also had a much higher prevalence of sepsis at admission, 40.0 percent versus 10.0 percent, along with higher white cell counts, lower hemoglobin and platelet counts, lower albumin, and higher international normalized ratio values.

In the main statistical analysis, the researchers fitted logistic regression models across 30 multiply imputed datasets to handle missing covariate data. The base model included age, sex, admission temperature, diabetes mellitus, sepsis at admission, albumin, international normalized ratio, and alkaline phosphatase. The extended model added continuous SIRI. In the extended model, each one-unit increase in SIRI was associated with an adjusted odds ratio of 1.0187 for the composite outcome, with a 95 percent confidence interval of 1.0022 to 1.0354 and a P value of 0.0262. Sepsis at admission, a higher international normalized ratio, and lower albumin were also independently associated with adverse outcomes.

Sensitivity analyses complicated the picture. In a complete-case analysis of 606 patients, the adjusted odds ratio was 1.0163, with a confidence interval of 0.9994 to 1.0334 that included the null value of one and a P value of 0.0588. When the four patients who died were excluded, leaving 666 patients and 96 nonfatal adverse outcomes, the odds ratio was 1.0191, essentially unchanged from the primary estimate. Restricted cubic spline analyses found no clear evidence of nonlinearity, but the overall spline association test itself did not reach statistical significance, with a P value of 0.0557.

The marker’s discriminatory performance was modest. The area under the receiver operating characteristic curve for SIRI alone was 0.6072, meaning the index performed only slightly better than chance at separating adverse from favorable outcomes. The neutrophil-to-lymphocyte ratio performed better, with an AUC of 0.6782, a difference that was statistically significant in a paired comparison. All individual markers tested had AUCs below 0.70. Adding SIRI to the base clinical model raised the apparent AUC from 0.7515 to 0.7630, an increase of 0.0116, but the 95 percent bootstrap interval for that gain ranged from slightly negative to modestly positive, and every incremental-performance measure, including net reclassification improvement and integrated discrimination improvement, had an interval that included zero.

The authors interpret these findings carefully. A statistically significant regression coefficient, they note, is not equivalent to improved prediction, and the positive total reclassification improvement in their analysis was driven entirely by the nonevent component while the event component was negative. Because the composite outcome was dominated by nonfatal events of varying severity, some of which, such as antimicrobial escalation and intensive care transfer, depend partly on clinical decisions, the results should not be read as evidence that SIRI predicts mortality. With only four deaths, the exploratory mortality comparison, in which median SIRI was 14.43 among decedents versus 6.12 among survivors, could not establish or exclude an association with death.

The study also carries structural limitations typical of single-center retrospective designs: susceptibility to selection and information bias, residual confounding, and limited generalizability. Detailed abscess characteristics, the timing and adequacy of source control, and antimicrobial appropriateness could not be incorporated into the adjusted analyses, and exact time stamps linking blood sampling to each outcome event were not consistently available. Model performance was apparent rather than validated, with no calibration assessment, optimism correction, or external testing. SIRI’s appeal lies in its simplicity: it derives from an inexpensive, routinely available blood count. But the authors conclude that while higher admission SIRI was associated with greater odds of 14-day adverse outcomes in their primary model, its standalone discrimination was limited and lower than that of the neutrophil-to-lymphocyte ratio, and its clinical utility requires validation in prospective, multicenter studies before it could guide any treatment decision.

Subject of Research: Association between admission systemic inflammation response index and 14-day adverse outcomes in pyogenic liver abscess

Article Title: Admission systemic inflammation response index is associated with 14-day adverse clinical outcomes in patients with pyogenic liver abscess: a single-center retrospective cohort study

Article References: Zhou, J., Pei, G., Huang, L., Qi, L., Zhang, M., Zhao, L., Zhou, F., Wang, L., Peng, S., Liu, A., & Li, X. (2026). Admission systemic inflammation response index is associated with 14-day adverse clinical outcomes in patients with pyogenic liver abscess: a single-center retrospective cohort study. Frontiers in Cellular and Infection Microbiology, 16, Article 1961188. https://doi.org/10.3389/fcimb.2026.1961188

Image Credits: AI Generated

DOI: 10.3389/fcimb.2026.1961188

Keywords: pyogenic liver abscess, SIRI, systemic inflammation response index, inflammatory biomarker, prognosis, retrospective cohort study, neutrophil-to-lymphocyte ratio, sepsis, adverse clinical outcomes, complete blood count, Admission, systemic