Factors associated with maximum postoperative hemoglobin decline following knee arthroplasty: a retrospective cohort study

Postoperative hemoglobin (Hb) decline remains clinically relevant after knee arthroplasty despite enhanced recovery and patient blood management. Hb decline reflects a composite of visible blood loss, hidden blood loss, hemodilution, sampling timing, and patient-specific physiologic reserve. This study evaluated associations between routinely available preoperative inflammatory, nutritional, renal-muscle, and hemodynamic markers and maximum postoperative Hb decline after knee arthroplasty. Observed Hb decline is a concentration-based change and should not be interpreted as directly measured blood loss. This retrospective single-center cohort included patients undergoing knee arthroplasty at the Department of Orthopedics and Joint Diseases I, Wangjing Hospital, China Academy of Chinese Medical Sciences, between 1 January 2025 and 15 April 2025. After data cleaning, 289 cases were available; 224 patients had analyzable preoperative and postoperative Hb data. Exposure-specific linear models adjusted for age, sex, body mass index, procedure, operation duration, hypertension, baseline Hb, and nadir sampling day. HC3 robust confidence intervals and false-discovery-rate correction were applied. Sensitivity analyses included alternative Hb outcomes, multiple imputation, procedure-specific models, restricted cubic splines, and an exploratory Hct-derived estimate of blood loss. In 224 patients (87 total knee, 121 unicompartmental, and 16 bicompartmental arthroplasties), prealbumin and albumin were positively associated with maximum observed Hb decline: β = 2.83 g/L (95% CI, 0.99 to 4.68) and β = 3.66 g/L (95% CI, 1.75 to 5.57) per cohort IQR, respectively. Log(CRP + 1) was inversely associated with Hb decline (β = −2.41 g/L per cohort IQR; 95% CI, − 3.69 to − 1.14). All three associations remained after FDR correction across 12 primary tests. Transferrin saturation, the creatinine/cystatin-C ratio, SII, hypertension status, and blood-pressure-control categories were not associated with the primary Hb endpoint after correction. In a simultaneous model containing six laboratory markers and MAP, albumin, CRP, and MAP retained associations. These coefficient directions should not be interpreted as evidence that better nutritional status increases bleeding or that higher CRP is protective. Preoperative albumin, prealbumin, and CRP were associated with maximum observed postoperative Hb decline. The positive albumin and prealbumin coefficients may partly reflect baseline hematologic reserve and the behavior of a concentration-based change score, whereas the inverse CRP coefficient should not be interpreted as protective. Associations for transferrin saturation, the creatinine/cystatin-C ratio, and SII were not supported for the primary endpoint after multiplicity correction. These findings do not establish effects on true surgical blood loss, transfusion, causality, or individual clinical risk.