Association of Preoperative Systemic Inflammatory Indices with Early Postoperative Pain and Function After Total Knee Arthroplasty: A Retrospective Cohort Study Authors: Shubei Cui, Taoxia Wang, Weiyong Zhang, Weili Zhang, Jianlong Zhao Published: 2026 ABSTRACT AbstractBackgroundEarly postoperative pain and impaired function remain common after total knee arthroplasty (TKA). This study evaluated whether associations with early pain and 6-week function were consistent across six related preoperative complete-blood-count-derived inflammatory indices.MethodsThis retrospective single-center cohort study included 114 patients who underwent unilateral primary TKA. Neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), monocyte-to-lymphocyte ratio (MLR), systemic immune-inflammation index (SII), systemic inflammation response index (SIRI), and aggregate index of systemic inflammation (AISI) were calculated from blood samples obtained within 48 hours before surgery. Primary outcomes were numeric rating scale (NRS) pain at 24 hours and Hospital for Special Surgery (HSS) score at 6 weeks. Each index was entered separately into covariate-adjusted models; sensitivity analyses additionally adjusted for ASA grade, C-reactive protein, or erythrocyte sedimentation rate.ResultsHigher values of all six indices were associated with higher NRS-24 hour scores and lower 6-week HSS scores. NLR, SII, and SIRI were also associated with greater 72-hour opioid use. In the descriptive NLR comparison, the unadjusted between-group differences were 1.41 points for NRS-24-hour pain and 5.71 points for 6-week HSS. Sensitivity analyses for the primary outcomes were materially unchanged. Exploratory discrimination beyond the clinical base model was modest (highest AUC, 0.702 for the SII-extended model).ConclusionHigher preoperative CBC-derived inflammatory indices, particularly NLR, SII, and SIRI, were associated with greater early pain burden and lower 6-week function after TKA. They are exploratory adjunctive signals rather than validated stand-alone screening tests or substitutes for clinical judgment. The retrospective single-center design, modest sample, and short follow-up require prospective multicenter validation.Keywords: arthroplasty, replacement, knee, knee Osteoarthritis, pain, postoperative, neutrophils, recovery of Function, inflammation, biomarkers, lymphocytes Introduction Total knee arthroplasty (TKA) remains the most reliable surgical option for patients with advanced knee osteoarthritis when pain, deformity, and functional limitation are no longer adequately controlled with nonoperative care. At the group level, pain and function usually improve substantially after surgery, yet recovery is not uniform, and a meaningful proportion of patients continue to report troublesome pain, delayed functional improvement, or dissatisfaction during follow-up.1–3 This variability is clinically important because the early postoperative period influences mobilization, analgesic requirements, length of recovery, and the pace at which patients regain confidence in the operated limb. Recent work has also shown that persistent pain after TKA remains a relevant concern, with risk extending beyond surgical factors alone and involving biological and patient-related contributors that may already be present before the operation.4,5 Risk stratification before TKA therefore depends not only on demographic or radiographic information, but also on markers that may reflect the patient’s underlying pain susceptibility and recovery potential. Conventional clinical predictors such as age, body mass index, comorbidity burden, and baseline function provide useful context, but they do not fully explain the observed heterogeneity in postoperative pain and functional outcomes. Studies focused on pain sensitivity and patient-reported outcomes suggest that preoperative biological status may add clinically relevant information to these standard variables. For example, lower preoperative pain pressure thresholds have been associated with less favorable short-term satisfaction after TKA, supporting the view that preoperative pain processing may influence early recovery in ways not captured by routine clinical assessment alone.6 Inflammation provides a plausible framework for this heterogeneity. In osteoarthritis, inflammatory mediators may contribute to nociceptor sensitization and pain heterogeneity, providing a biological rationale for why a heightened preoperative inflammatory state could be associated with early postoperative pain and recovery.7,8 Surgical trauma triggers a pronounced immune response, but the magnitude and consequences of that response are likely shaped by the patient’s preoperative inflammatory state. Neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), monocyte-to-lymphocyte ratio (MLR), systemic immune-inflammation index (SII), systemic inflammation response index (SIRI), and aggregate index of systemic inflammation (AISI) can all be derived from routine blood tests at low cost and with minimal additional burden. In perioperative and broader surgical settings, elevated preoperative NLR has been associated with adverse postoperative morbidity, and perioperative changes in NLR have been linked with postsurgical pain burden and health-related outcomes.9,10 Within joint arthroplasty, observational data have also suggested that inflammatory biomarkers such as NLR, PLR, and C-reactive protein may be related to postoperative pain intensity, supporting the clinical relevance of inflammation-based indicators in this setting.11 CBC-derived indices are inexpensive and immediately available from routine preoperative blood counts, whereas CRP and ESR are established acute-phase indicators; together, these measures may capture complementary aspects of systemic inflammation.11–13 Even so, several issues remain unresolved. Much of the available TKA literature has emphasized postoperative pain trajectories, long-term prognosis, or isolated candidate predictors, whereas the comparative associations of several preoperative systemic inflammatory indices with early postoperative pain burden and short-term functional recovery have received less direct study within the same cohort.1,3 We therefore evaluated six CBC-derived indices in parallel in relation to early postoperative pain and 6-week function after primary TKA. Methods Study Design and Setting This retrospective single-center observational cohort study was conducted in the Department of Orthopedics, Handan Central Hospital. Consecutive patients who underwent elective primary unilateral TKA between January 2023 and December 2025 were identified from the institutional surgical database and electronic medical record and screened for eligibility. Data were extracted and the database was locked on 15 January 2026, after completion of the final scheduled 6-week follow-up; both primary outcomes were available for all 114 included patients. The study was designed to evaluate the prognostic associations of preoperative systemic inflammatory indices with early postoperative pain burden and short-term functional recovery after TKA. Reporting was prepared with reference to the Strengthening the Reporting of Observational Studies in Epidemiology statement.14 The protocol was approved by the institutional ethics committee of Handan Central Hospital, which waived the requirement for written informed consent owing to the retrospective design and the use of de-identified routinely collected clinical data. Participants Eligible participants were adults aged 50 years or older with advanced symptomatic knee osteoarthritis who were considered suitable for unilateral primary TKA after routine clinical and radiographic assessment. Radiographic severity was documented with the Kellgren-Lawrence grading system, and the enrolled cohort represented end-stage disease requiring arthroplasty.15 Patients were excluded if they had inflammatory arthritis, prior surgery on the index knee, active infection or recent antibiotic treatment within 2 weeks before surgery, hematologic disease, active malignant disease, severe hepatic or renal dysfunction, chronic immunosuppressive therapy or systemic glucocorticoid use, blood transfusion within 1 month before surgery, revision arthroplasty, staged bilateral procedures, or indications other than degenerative osteoarthritis. Sample Size Consideration All consecutive patients who met the eligibility criteria during the study period were included, yielding a final analytic sample of 114 patients. The adequacy of this available sample was verified with the 24-hour NRS pain score as the anchor outcome. Assuming a clinically relevant between-group difference of 1.50 points, a standard deviation of 2.50 points, a 2-sided alpha of 0.05, and 80% power, at least 90 patients were required; the available sample of 114 patients therefore provided sufficient power for the primary pain comparison. Perioperative Management All procedures were performed by the same senior arthroplasty team using a standard medial parapatellar approach under spinal anesthesia. Cemented posterior-stabilized or cruciate-retaining prostheses were selected according to intraoperative assessment. Perioperative care otherwise followed a standardized institutional pathway. Pharmacologic thromboprophylaxis consisted of enoxaparin 40 mg subcutaneously once daily starting on postoperative day 1, combined with graduated compression stockings. Postoperative rehabilitation included bedside range-of-motion training from postoperative day 1 and walker-assisted ambulation from postoperative day 2. Discharge was generally planned once patients could walk independently for 50 m and achieved active knee flexion of at least 90°. After discharge, outpatient physical therapy was recommended 2 to 3 times weekly. Rescue analgesia followed a standardized ward protocol and was documented in the medication chart; patients with postoperative NRS scores greater than 3 received intravenous parecoxib 40 mg or oral tramadol 50 mg according to clinical judgment and oral tolerance. Periarticular analgesic infiltration with ropivacaine plus epinephrine could be used at the surgeon’s discretion within this pathway. Data Collection and Inflammatory Indices Baseline demographic, clinical, laboratory, and perioperative variables were collected retrospectively from the electronic medical record and hospital information system. Complete blood count results used to derive the inflammatory indices, as well as C-reactive protein and erythrocyte sedimentation rate values used in sensitivity analyses, were obtained from routine preoperative laboratory testing performed within 48 hours before surgery. Preoperative NLR, PLR, MLR, SII, SIRI, and AISI, as previously described,16–18 were calculated from these data. NLR was calculated as neutrophil count divided by lymphocyte count, PLR as platelet count divided by lymphocyte count, MLR as monocyte count divided by lymphocyte count, SII as neutrophil count multiplied by platelet count and divided by lymphocyte count, SIRI as neutrophil count multiplied by monocyte count and divided by lymphocyte count, and AISI as neutrophil count multiplied by monocyte count multiplied by platelet count and divided by lymphocyte count. For descriptive comparisons, patients were categorized using an NLR threshold of 2.80 as a pragmatic stratification point, approximating preoperative NLR levels reported in arthroplasty cohorts.19 This threshold was not derived by outcome optimization and was not used as a clinical decision threshold. For the main regression analyses, all inflammatory indices were retained as continuous variables and standardized per 1-standard deviation increase in the analysis cohort. Outcome Assessment The primary pain outcome was the numeric rating scale (NRS) pain score at 24 hours after surgery. The NRS ranges from 0 to 10, has established clinical interpretability in postoperative pain research, and shows acceptable validity for routine perioperative pain assessment.20 Postoperative NRS scores at 12, 24, 48, and 72 hours were abstracted from routine ward assessments, with the 24-hour value prespecified as the primary pain endpoint. Cumulative opioid consumption during the first 72 postoperative hours included all systemic opioid administrations recorded in the medication chart and was converted to morphine milligram equivalents using standard equianalgesic conversions. The primary functional outcome was the Hospital for Special Surgery (HSS) knee score at 6 weeks. The HSS score is a clinician-administered instrument covering pain, function, range of motion, muscle strength, flexion deformity, and stability, and validation studies have shown good reliability and construct validity.21 The Oxford Knee Score (OKS) at 6 weeks was analyzed as a secondary functional outcome. The OKS is a 12-item joint-specific patient-reported measure with good reliability, responsiveness, and construct validity after knee arthroplasty.22 HSS and OKS were recorded at the routine 6-week postoperative follow-up assessment and extracted from the medical record. HSS below 70 was a secondary descriptive binary outcome (below-good 6-week recovery); it complements the primary continuous HSS analysis and corresponds to a score below the conventional HSS good category (70–84)23. Exploratory outcomes included postoperative NLR and SII dynamics derived from routine complete blood counts at 24 and 72 hours. Statistical Analysis Continuous variables were summarized as mean ± standard deviation or median (interquartile range), as appropriate, and categorical variables were summarized as counts and percentages. Between-group comparisons for baseline and outcome variables were performed using Welch t tests, Mann–Whitney U-tests, chi-square tests, or Fisher exact tests according to data distribution and expected cell counts. NRS at 24 hours, HSS score at 6 weeks, 72-hour morphine milligram equivalents, and OKS at 6 weeks were analyzed using multivariable linear regression, whereas below-good 6-week recovery was analyzed using multivariable logistic regression. Covariates were prespecified on clinical grounds from routinely available baseline variables. Pain models were adjusted for age, sex, body mass index, diabetes, preoperative resting NRS score, and operative time. Models for 6-week HSS and below-good 6-week recovery were adjusted for age, sex, body mass index, diabetes, and preoperative HSS score. Models for 6-week OKS were adjusted for age, sex, body mass index, diabetes, and preoperative OKS score. Because the inflammatory indices were mathematically related, each index was entered into a separate model. Exploratory discrimination analysis for below-good 6-week recovery compared the clinical base model with models expanded by NLR, SII, or SIRI using receiver operating characteristic curves and area under the curve estimates, with comparisons performed by the DeLong method. Sensitivity analyses repeated the two primary linear-regression models after additional adjustment for ASA grade, C-reactive protein, and erythrocyte sedimentation rate, individually. No formal multiplicity adjustment was applied; the two primary outcomes and covariates were prespecified. Because the six indices are mathematically related and secondary analyses involved multiple comparisons, secondary, binary, and discrimination analyses are supportive and exploratory. Analyses were performed on a complete-case basis; patients with missing data for either primary outcome or for the variables required to derive the inflammatory indices were not included in the analytic cohort, and no imputation was performed. The distributions of continuous outcomes by group were displayed using raincloud plots, which combine a half-violin kernel density estimate, a box plot, and individual data points. Statistical analyses were conducted using SPSS version 26.0 (IBM Corp) and R version 4.5.2 (R Foundation for Statistical Computing). All tests were 2-sided, and P < 0.05 was considered statistically significant. Results Patient Characteristics and Preoperative Inflammatory Status The baseline demographic, clinical, and laboratory characteristics of the study cohort are presented in Table 1. A total of 114 patients were included, of whom 49 were classified into the high-NLR group and 65 into the low-NLR group. There were no significant between-group differences in age, sex distribution, BMI, ASA grade III–IV, diabetes mellitus, hypertension, smoking history, preoperative HSS score, preoperative OKS score, or preoperative resting NRS score (all P>0.05). By contrast, all six preoperative inflammatory indices were significantly higher in the high-NLR group than in the low-NLR group. Median preoperative NLR was 3.42 in the high-NLR group and 2.28 in the low-NLR group (P<0.001). Similarly, median PLR, MLR, SII, SIRI, and AISI were all significantly elevated in the high-NLR group compared with the low-NLR group (all P<0.001). Table 1Baseline Characteristics, Preoperative Scores, and Inflammatory Indices by NLR GroupVariableHigh NLR (n=49)Low NLR (n=65)StatisticP valueAge, years66.96 ± 7.9967.94 ± 7.590.6620.510aFemale sex, n (%)40 (81.63%)47 (72.31%)1.3440.246cBMI, kg/m227.45 ± 3.6527.34 ± 3.490.1740.863aASA grade III–IV, n (%)35 (71.43%)44 (67.69%)0.1830.669cDiabetes mellitus, n (%)17 (34.69%)22 (33.85%)0.0090.925cHypertension, n (%)29 (59.18%)37 (56.92%)0.0590.809cSmoking history, n (%)8 (16.33%)6 (9.23%)1.3060.253cPreoperative HSS score54.45 ± 6.8552.94 ± 7.751.1010.273aPreoperative OKS score21.24 ± 4.7021.45 ± 4.420.2320.817aPreoperative resting NRS6.84 ± 1.016.75 ± 1.120.4150.679aPreoperative NLR3.42 (3.17–3.95)2.28 (1.92–2.47)3185.000<0.001bPreoperative PLR164.20 (129.66–205.00)108.98 (83.81–134.38)2531.500<0.001bPreoperative MLR0.35 (0.28–0.42)0.23 (0.20–0.28)2596.000<0.001bPreoperative SII789.70 (676.20–965.00)466.10 (371.30–552.80)2948.000<0.001bPreoperative SIRI1.88 (1.47–2.17)1.08 (0.83–1.25)2668.500<0.001bPreoperative AISI403.60 (302.20–522.50)217.50 (163.10–301.00)2646.500<0.001bNotes: Continuous variables are presented as mean ± SD or median (IQR); categorical variables are presented as n (%). P values marked with superscripts indicate the statistical methods used: a, Welch t test; b, Mann–Whitney U-test; c, chi-square test. ASA III–IV indicates grade III or IV. Baseline Characteristics, Preoperative Scores, and Inflammatory Indices by NLR Group Notes: Continuous variables are presented as mean ± SD or median (IQR); categorical variables are presented as n (%). P values marked with superscripts indicate the statistical methods used: a, Welch t test; b, Mann–Whitney U-test; c, chi-square test. ASA III–IV indicates grade III or IV. Early Postoperative Pain Outcomes and Opioid Consumption Early postoperative pain outcomes are summarized in Table 2. Across all postoperative time points, patients in the high-NLR group had higher pain scores than those in the low-NLR group. Mean NRS scores at 12, 24, 48, and 72 hours were 7.20±1.19, 5.69±1.37, 4.27±0.97, and 3.24±1.01, respectively, in the high-NLR group, as compared with 5.62±0.96, 4.28±0.91, 3.08±0.94, and 2.11±0.89 in the low-NLR group (all P<0.001). A similar between-group difference was observed for cumulative opioid consumption during the first 72 postoperative hours, with a mean MME of 37.77±10.13 mg in the high-NLR group and 28.85±10.74 mg in the low-NLR group (P<0.001). Table 2Early Postoperative Pain Outcomes and 72-Hour Opioid Consumption by NLR GroupVariableHigh NLR (n=49)Low NLR (n=65)StatisticP valueNRS at 12 h7.20 ± 1.195.62 ± 0.967.646<0.001aNRS at 24 h5.69 ± 1.374.28 ± 0.916.263<0.001aNRS at 48 h4.27 ± 0.973.08 ± 0.946.544<0.001aNRS at 72 h3.24 ± 1.012.11 ± 0.896.266<0.001a72 h opioid consumption, MME (mg)37.77 ± 10.1328.85 ± 10.744.535<0.001aNotes: Values are presented as mean ± SD. P values marked with superscript a were derived from Welch t tests. P values marked with superscripts indicate the statistical methods used: a, Welch t test.Abbreviation: MME, morphine milligram equivalents. Early Postoperative Pain Outcomes and 72-Hour Opioid Consumption by NLR Group Notes: Values are presented as mean ± SD. P values marked with superscript a were derived from Welch t tests. P values marked with superscripts indicate the statistical methods used: a, Welch t test. Abbreviation: MME, morphine milligram equivalents. Short-Term Functional Recovery The 6-week functional outcomes are shown in Table 3. The unadjusted between-group differences were 1.41 points for NRS-24-hour pain, 5.71 points for HSS, and 4.88 points for OKS. Patients in the high-NLR group had lower HSS and OKS values at 6 weeks than those in the low-NLR group. Mean HSS scores were 70.51±7.29 in the high-NLR group and 76.22±7.15 in the low-NLR group, whereas mean OKS scores were 31.57±6.60 and 36.45±4.79 respectively. Below-good 6-week recovery, defined as an HSS score <70 at 6 weeks, occurred in 24 of 49 patients in the high-NLR group and 11 of 65 patients in the low-NLR group (all P<0.001). Table 3Six-Week Functional Outcomes by NLR GroupVariableHigh NLR (n=49)Low NLR (n=65)StatisticP valueHSS score at 6 weeks70.51 ± 7.2976.22 ± 7.154.171<0.001aOKS score at 6 weeks31.57 ± 6.6036.45 ± 4.794.372<0.001aBelow-good 6-week recovery, n (%)24 (48.98%)11 (16.92%)13.494<0.001cNotes: Continuous variables are presented as mean ± SD and categorical variables as n (%). P values marked with superscripts indicate the statistical methods used: a, Welch t test; c, chi-square test. Six-Week Functional Outcomes by NLR Group Notes: Continuous variables are presented as mean ± SD and categorical variables as n (%). P values marked with superscripts indicate the statistical methods used: a, Welch t test; c, chi-square test. Multivariable Association Analyses The adjusted associations between each preoperative inflammatory index and the predefined clinical outcomes are presented in Table 4. For the primary pain endpoint, higher levels of all six preoperative inflammatory indices remained associated with higher NRS-24 h scores after adjustment. The adjusted β coefficients ranged from 0.426 for SIRI (95% CI, 0.187 to 0.664; P<0.001) to 0.514 for both NLR (95% CI, 0.285 to 0.743; P<0.001) and MLR (95% CI, 0.276 to 0.752; P<0.001). For the primary functional endpoint, higher preoperative NLR, PLR, MLR, SII, SIRI, and AISI were each associated with lower HSS scores at 6 weeks, with adjusted β coefficients ranging from −1.621 (95% CI, −3.086 to −0.157; P=0.032) for PLR to −2.450 (95% CI, −3.851 to −1.050; P<0.001) for NLR. Table 4Multivariable Association Analyses for Primary and Secondary OutcomesOutcomeInflammatory IndexAdjusted Effect95% CIP valueNRS-24 h†NLR0.5140.285 to 0.743<0.001PLR0.4860.251 to 0.721<0.001MLR0.5140.276 to 0.752<0.001SII0.4880.258 to 0.718<0.001SIRI0.4260.187 to 0.664<0.001AISI0.4520.215 to 0.689<0.001HSS at 6 weeks†NLR−2.450−3.851 to −1.050<0.001PLR−1.621−3.086 to −0.1570.032MLR−1.980−3.472 to −0.4880.011SII−2.229−3.630 to −0.8270.002SIRI−2.199−3.620 to −0.7780.003AISI−1.980−3.413 to −0.5460.00872 h MME†NLR3.6561.626 to 5.686<0.001PLR1.231−0.932 to 3.3940.267MLR1.298−0.912 to 3.5080.252SII2.3050.209 to 4.4000.033SIRI2.2930.167 to 4.4190.037AISI1.394−0.767 to 3.5550.209OKS at 6 weeks†NLR−2.423−3.453 to −1.393<0.001PLR−1.323−2.442 to −0.2040.022MLR−1.038−2.189 to 0.1130.080SII−2.240−3.285 to −1.196<0.001SIRI−1.568−2.667 to −0.4700.006AISI−1.669−2.766 to −0.5720.004Below-good 6-week recovery‡NLR1.7611.143 to 2.7140.010PLR1.8261.167 to 2.8570.008MLR1.5440.995 to 2.3950.052SII1.9861.276 to 3.0900.002SIRI1.5731.032 to 2.3980.035AISI1.7031.115 to 2.6010.014Notes: Each inflammatory index was entered into a separate model and standardized per 1-SD increase.† Estimated by multivariable linear regression and reported as adjusted β coefficients with 95% confidence intervals.‡ Estimated by multivariable logistic regression and reported as adjusted odds ratios with 95% confidence intervals. Multivariable Association Analyses for Primary and Secondary Outcomes Notes: Each inflammatory index was entered into a separate model and standardized per 1-SD increase.† Estimated by multivariable linear regression and reported as adjusted β coefficients with 95% confidence intervals.‡ Estimated by multivariable logistic regression and reported as adjusted odds ratios with 95% confidence intervals. For secondary outcomes, higher preoperative NLR, SII, and SIRI were each associated with greater 72-hour opioid consumption, with adjusted β coefficients of 3.656 (95% CI, 1.626 to 5.686; P<0.001), 2.305 (95% CI, 0.209 to 4.400; P=0.033), and 2.293 (95% CI, 0.167 to 4.419; P=0.037), respectively, whereas the corresponding associations for PLR, MLR, and AISI were not statistically significant. Higher preoperative NLR, PLR, SII, SIRI, and AISI were also associated with lower OKS scores at 6 weeks, while the association between MLR and 6-week OKS did not reach statistical significance (β, −1.038; 95% CI, −2.189 to 0.113; P=0.080). For below-good 6-week recovery, higher NLR, PLR, SII, SIRI, and AISI were each associated with increased adjusted odds, with odds ratios ranging from 1.573 (95% CI, 1.032 to 2.398; P=0.035) for SIRI to 1.986 (95% CI, 1.276 to 3.090; P=0.002) for SII. The association for MLR was borderline and did not meet the conventional threshold for statistical significance (OR, 1.544; 95% CI, 0.995 to 2.395; P>0.05). Sensitivity analyses for the two primary outcomes are summarized in Supplementary Table 1. Additional adjustment for ASA grade, C-reactive protein, or erythrocyte sedimentation rate did not materially change the direction, magnitude, or statistical significance of the associations between the inflammatory indices and either NRS-24 h or HSS score at 6 weeks, supporting the robustness of the primary findings (all P<0.05). Exploratory Discrimination and Graphical Results The exploratory discrimination analyses for below-good 6-week recovery are shown in Table 5 and Figure 1. The clinical base model yielded an AUC of 0.593 (95% CI, 0.479 to 0.708). The addition of NLR increased the AUC to 0.679 (95% CI, 0.576 to 0.782), corresponding to a delta AUC of 0.086 (P>0.05 vs base model). The addition of SII increased the AUC to 0.702 (95% CI, 0.604 to 0.801), corresponding to a delta AUC of 0.109 (P>0.05), whereas the addition of SIRI yielded an AUC of 0.629 (95% CI, 0.512 to 0.746), corresponding to a delta AUC of 0.036 (P>0.05). Table 5Exploratory ROC Analysis for Below-Good 6-Week RecoveryModelAUC95% CIΔAUC vs BaseP Value vs BaseClinical base model0.5930.479 to 0.708——Base model + NLR0.6790.576 to 0.7820.0860.121Base model + SII0.7020.604 to 0.8010.1090.089Base model + SIRI0.6290.512 to 0.7460.0360.516Notes: ROC analysis compared the clinical base model with marker-extended models for below-good 6-week recovery at 6 weeks. The clinical base model included age, sex, BMI, diabetes, and preoperative HSS. AUC values and 95% confidence intervals are reported to three decimal places. Comparisons versus the base model were performed using the DeLong method. Exploratory ROC Analysis for Below-Good 6-Week Recovery Notes: ROC analysis compared the clinical base model with marker-extended models for below-good 6-week recovery at 6 weeks. The clinical base model included age, sex, BMI, diabetes, and preoperative HSS. AUC values and 95% confidence intervals are reported to three decimal places. Comparisons versus the base model were performed using the DeLong method. Exploratory discrimination for below-good 6-week recovery at 6 weeks. (A) Receiver operating characteristic curves comparing the clinical base model with marker-extended models incorporating preoperative NLR, SII, or SIRI; the clinical base model included age, sex, BMI, diabetes, and preoperative HSS score. (B) Distribution of the SII-extended model’s predicted risk according to actual 6-week outcome, displayed as a raincloud plot combining a half-violin kernel density estimate, a box plot (median, interquartile range, and 1.5×IQR whiskers), and individual data points; predicted risk was higher in patients with below-good 6-week recovery. Figure 2 illustrates the postoperative pain trajectory according to preoperative NLR group. In both groups, pain scores declined over time, but the high-NLR group maintained consistently higher mean NRS values than the low-NLR group at 12, 24, 48, and 72 hours after surgery, with significant between-group differences at each observed time point (all P<0.001). Postoperative inflammatory dynamics are further shown in Figure 3. In the high-NLR group, median NLR increased from 3.42 preoperatively to 7.55 at 24 hours and then decreased to 5.47 at 72 hours; the corresponding values in the low-NLR group were 2.28, 4.57, and 3.27. Median SII showed a similar pattern, increasing from 789.70 to 1434.40 and then decreasing to 1056.90 in the high-NLR group, as compared with 466.10, 800.70, and 619.50 in the low-NLR group. Between-group differences in postoperative NLR and SII remained significant at both 24 and 72 hours (all P<0.001). Early postoperative pain distribution according to preoperative NLR group. For each group and time point, the distribution of numeric rating scale (NRS) pain scores is displayed as a raincloud plot combining a half-violin kernel density estimate, a box plot (median, interquartile range, and 1.5×IQR whiskers), and individual data points; dashed lines connect group medians across 12, 24, 48, and 72 hours after surgery. Unadjusted between-group comparisons at each time point were performed using the Welch t-test and are annotated as *** P<0.001. Patients with high preoperative NLR showed consistently higher postoperative pain scores across all observed time points. Postoperative inflammatory marker distribution according to preoperative NLR group. For NLR (neutrophil-to-lymphocyte ratio; left) and SII (systemic immune-inflammation index; right), the distribution at the preoperative time point and at 24 and 72 hours after surgery is displayed as a raincloud plot combining a half-violin kernel density estimate, a box plot (median, interquartile range, and 1.5×IQR whiskers), and individual data points; dashed lines connect group medians across time points. Unadjusted between-group comparisons at each time point were performed using the Mann–Whitney U-test and are annotated as *** P<0.001. Patients in the high preoperative NLR group maintained higher postoperative inflammatory levels across both markers. Discussion In this cohort of patients undergoing primary TKA, a higher preoperative inflammatory burden was consistently associated with a less favorable early postoperative course. Among the six indices examined, NLR, SII, and SIRI showed the most consistent relationships across pain intensity, opioid consumption, and 6-week recovery, whereas PLR, MLR, and AISI appeared to be more selective in their associations. Sensitivity analyses with additional adjustment for ASA grade, CRP, or ESR yielded materially similar estimates for the two primary endpoints, supporting the robustness of the main association findings. This pattern suggests that routine blood count-derived inflammatory indices may help identify patients whose perioperative recovery is likely to be more symptom-intensive despite similar baseline demographic and clinical profiles.9,10,12,13,24 The present findings are broadly in line with recent arthroplasty literature linking CBC-derived inflammatory ratios with postoperative morbidity, inflammatory response, and perioperative treatment effect heterogeneity.9,10,12,13,24 Our data extend that work by focusing on early pain burden and short-term functional recovery rather than medical complications alone, and by comparing several closely related inflammatory indices under the same adjustment framework. The ROC analyses also suggest that these markers should be viewed as adjunctive rather than standalone triage tools, as the marker-augmented models showed only modest improvement over the clinical base model. The ROC findings should nevertheless be interpreted as exploratory only, because the observed AUC increments were modest and did not demonstrate statistically significant improvement over the clinical base model. The observed associations are biologically plausible. Pain and disability in knee osteoarthritis are not determined solely by structural damage but are also influenced by low-grade synovial and systemic inflammation, immune-cell signaling, and pain sensitization.7,8,25 Ratios that incorporate neutrophils, lymphocytes, monocytes, and platelets may therefore capture a host state in which inflammatory signaling and surgical stress responsiveness are already amplified before the procedure. In that context, the larger postoperative rises in NLR and SII seen in the high-NLR group may reflect a more reactive perioperative inflammatory phenotype, which could be linked to greater nociceptive burden, higher analgesic requirements, and slower early functional gains. This interpretation remains inferential because cytokine profiling and sensory testing were not available in the present dataset. The functional findings are clinically relevant because the early recovery window after TKA is closely tied to pain control, mobility, and the pace of rehabilitation. Recent work has shown that patient activity and ambulation change markedly within the first 6 postoperative weeks, a period in which inadequate symptom control may plausibly hinder recovery milestones.26,27 From a practical standpoint, NLR, SII, and SIRI are inexpensive and readily available from routine preoperative testing. Although they do not replace clinical assessment, elevated values may justify closer perioperative counseling, more deliberate analgesic planning, and earlier attention to recovery targets in patients who appear otherwise similar on standard baseline evaluation. Although the between-group difference in 6-week HSS score was statistically significant, its clinical magnitude should be interpreted cautiously, particularly given the short follow-up and the absence of an established MCID framework specifically applicable to between-group HSS differences at 6 weeks. Inflammatory mediators implicated in osteoarthritis may activate NF-κB/MAPK-related signaling, and pain sensitization contributes to pain heterogeneity in knee osteoarthritis.7,8 These mechanisms provide a biological rationale for the observed associations, but CBC-derived indices are indirect systemic measures; without cytokine or synovial data, pathway activation and causality cannot be inferred. The 1.41-point NRS and 5.71-point HSS differences provide clinical context but not MCID or treatment thresholds. Likewise, NLR 2.80 was descriptive only, and the exploratory AUC gains do not support stand-alone screening. This study has several strengths. It examined a clinically homogeneous unilateral primary TKA cohort, used a relatively consistent perioperative pathway, and evaluated multiple interrelated inflammatory indices in parallel while considering pain intensity, opioid requirement, clinician-rated function, and patient-reported function. At the same time, several limitations merit consideration. The single-center, retrospective observational design limits generalizability, introduces potential for selection and information bias, and does not support causal interpretation. Because the data were obtained from routinely collected clinical records, the timing and completeness of some assessments were not standardized for research purposes, and patients with incomplete key data were excluded, which may have introduced attrition-related bias. The sample size was modest, follow-up for the main functional outcome was limited to 6 weeks, and the data structure is better suited to prognostic association analysis than to formal clinical prediction-model development. The modest sample size and absence of multiplicity adjustment increase the risk of imprecise estimates and chance findings, particularly for secondary, binary, and exploratory analyses; these results should therefore be interpreted cautiously. Residual confounding from factors such as psychological distress, sleep disturbance, frailty, radiographic inflammatory burden, and unmeasured variation in postoperative analgesic delivery cannot be excluded. The inflammatory indices were mathematically interrelated, so the present analyses do not establish the absolute superiority of any single marker. The discrimination analyses were exploratory and not externally validated. Further multicenter studies with longer follow-up, repeated biomarker sampling, and integration of symptomatic or sensitization-related measures may help determine whether these indices can be incorporated into more practical perioperative risk stratification strategies. Conclusion In patients undergoing primary TKA, higher preoperative CBC-derived inflammatory indices, particularly NLR, SII, and SIRI, were associated with greater early postoperative pain and worse 6-week function. These associations remained robust in sensitivity analyses with additional adjustment for ASA grade, CRP, or ESR. These measures may serve as exploratory adjunctive signals within a broader preoperative assessment, while larger multicenter prospective studies with longer follow-up, broader clinical phenotyping, and external validation are needed to establish their causal relevance and clinically actionable role in perioperative decision-making. Pending prospective validation, higher preoperative inflammatory-index values could prompt a fuller assessment of modifiable perioperative risks, individualized counseling and multimodal analgesia planning, and closer early rehabilitation follow-up. Funding Statement This research was funded by Hebei Medical Science Research Project (20220015). Data Sharing Statement All data generated or analyzed in this study are included in the present manuscript. Ethics Approval and Consent to Participate The study protocol was approved by the Ethics Committee of Handan Central Hospital (approval No. 2026-061). Owing to the retrospective design and the use of de-identified routinely collected clinical data, the ethics committee of Handan Central Hospital waived the requirement for written informed consent. The study was conducted in accordance with the Declaration of Helsinki. Author Contributions All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted, and agree to be accountable for all aspects of the work. Disclosure The authors declare that there are no conflicts of interest regarding this study. References Associated Data Data Availability Statement All data generated or analyzed in this study are included in the present manuscript.