Serial Changes in Routine Inflammatory, Hematologic, and Renal Laboratory Markers in Adults Hospitalized with Clinically Recorded Sepsis: A Longitudinal Observational Cohort Study

Background/Objectives: Serial laboratory measurements may capture within-episode changes not represented by a single baseline value, while routine markers from different biological domains may not evolve concordantly. Methods: We retrospectively analyzed 100 hospital admission episodes in which the source infection field explicitly recorded sepsis. A complete source-record re-audit confirmed sepsis wording in all 100 episodes and identified no severe-infection-only subgroup. Because complete SOFA components were unavailable, Sepsis-3 criteria were not retrospectively adjudicated. Procalcitonin (PCT), C-reactive protein (CRP), white blood cell count (WBC), platelet count, and serum creatinine were evaluated across three ordered recorded assessment points using random-intercept linear mixed-effects models, with Holm correction across five primary global tests and generalized estimating equations as robustness analyses. Results: PCT geometric means decreased from 0.925 to 0.306 ng/mL (Holm-adjusted p = 0.000052), CRP from 104.2 to 36.0 mg/L (p < 0.001), and serum creatinine from 1.49 to 1.10 mg/dL (p = 0.000074). Platelet count increased on the additive scale (p = 0.000237) but not after logarithmic transformation, whereas the full-cohort primary WBC model was non-significant (p = 0.0743). Conclusions: Routine laboratory markers showed heterogeneous cohort-level longitudinal patterns, supporting marker-specific rather than uniform interpretation of serial measurements.