Surgical wound infection (SWI) after lumbar spine surgery is a potentially catastrophic complication, with reported incidence varying by patient and procedural factors (Zhou et al., 2020). SWI is associated with prolonged hospitalization, perioperative morbidity, 90 d readmission, revision surgery, pain, disability, and financial burden (Suszczynska et al., 2025; Elsamadicy et al., 2024). Clarifying how routine inflammatory markers relate to SWI may improve interpretation of postoperative laboratory abnormalities.
Routine laboratory markers such as white blood cell (WBC) count and neutrophil-to-lymphocyte ratio (NLR) have been associated with postoperative infection in prior studies (Shen et al., 2019; Qian et al., 2023; Iwata et al., 2016). However, early WBC and NLR elevations may reflect operative stress, patient physiology, comorbidity, body mass index (BMI), extent of tissue dissection, number of operative levels, and operative duration rather than infection-specific biology (Watt et al., 2015; Kobayashi et al., 2020; Liu et al., 2018; Schoenberg et al., 2023).
In this study, SWI was defined as a pragmatic electronic-health-record-based endpoint intended to capture clinically documented wound-related events. It was not equivalent to a uniformly culture-confirmed deep infection, implant-associated infection, or standard surveillance-defined surgical site infection (SSI).
This study evaluated whether POD1 and POD3 WBC/NLR remained associated with SWI after accounting for patient and operative factors that influence postoperative inflammation.
A retrospective review was performed of prospectively collected data from adult patients who underwent elective one- to three-level lumbar fusion from 2022 to 2024 at a single academic medical center. A source registry of 462 cases was reviewed. After data cleaning and construction of the completed WBC/NLR abstraction dataset, 380 patients comprised the outcome-eligible analytic cohort. Data were obtained from the institutional electronic health record, including operative reports, laboratory records, postoperative encounters, emergency department visits, readmissions, and reoperation documentation.
Patients were included if they were 18 years of age or older, underwent elective one- to three-level lumbar fusion from 2022 to 2024, had 90 d outcome ascertainment, and had postoperative complete blood count (CBC) data available for at least one relevant inflammatory-marker analysis. Patients were excluded if follow-up was insufficient to ascertain the primary outcome, if surgery was performed for oncologic or infectious indications, or if surgery was emergent. Laboratory analyses used complete-case availability for each time point: POD1 analyses included patients with complete POD1 WBC, absolute neutrophil count (ANC), absolute lymphocyte count (ALC), and NLR values; POD3 analyses included patients with complete POD3 values; and paired trajectory analyses included only patients with complete POD1 and POD3 values.
Baseline demographic and clinical variables included age, sex, BMI, obesity class, diabetes, smoking status, race/ethnicity, insurance category, age-adjusted Charlson Comorbidity Index, and Area Deprivation Index (Petterson, 2023). Operative variables included approach, number of fused levels, operative duration, estimated blood loss, transfusion, prior laminectomy, adjacent-level laminectomy, and intraoperative complication. Postoperative variables included length of stay, ICU/step-down admission, discharge disposition, 90 d emergency department visit, readmission, and return to the operating room.
The primary laboratory values of interest were postoperative inflammatory markers measured as WBC and NLR on POD1 and POD3. Laboratory values were collected per standard hospital protocol. NLR was calculated as ANC divided by ALC from the corresponding CBC differential. When more than one laboratory value was available for a given postoperative day, the value closest to the target time point (24 h for POD1 and 72 h for POD3) was used.
The primary outcome was SWI within 90 d of surgery, defined pragmatically from the electronic health record as a documented clinical wound infection or wound-related infectious complication requiring antibiotic therapy, wound intervention, and/or reoperation. Infection status was confirmed through postoperative inpatient and outpatient encounters, emergency department visits, readmissions, wound evaluations, and operative documentation when applicable. This pragmatic endpoint was not equivalent to a uniformly culture-confirmed deep infection, implant-associated infection, or standard CDC/NHSN- or ACS-NSQIP-defined SSI. SWI cases were further categorized by treatment phenotype as return to OR, nonoperative wound complication/intervention, or nonoperative/antibiotic-only clinical treatment.
All statistics were performed using SPSS 30.0.0 (IBM Corp., Armonk, NY, USA). Continuous variables were compared using Mann–Whitney
To evaluate potential selection bias related to POD3 laboratory availability, patients with complete paired POD1–POD3 WBC/ANC/ALC/NLR values were compared with outcome-eligible patients who lacked complete paired laboratory data.
Trajectory analyses were performed among patients with complete paired POD1 and POD3 WBC/ANC/ALC/NLR values. Absolute change was calculated as POD3 minus POD1, and percentage change was calculated as absolute change divided by POD1. POD1 values, POD3 values, absolute changes, and percentage changes were compared by SWI status.
Because only 18 SWI events occurred, multivariable logistic regression models were kept parsimonious to reduce overfitting. Models focused on prespecified covariates, including BMI and operative duration, while additional covariates were evaluated descriptively. A sensitivity analysis restricted to the more definite return-to-OR/washout phenotype was considered. Because only nine such events occurred, regression modeling for this restricted phenotype was not performed because estimates would be unstable and overfit.
Of the total 462 screened source-registry cases, 380 patients were included in the outcome-eligible WBC/NLR analytic cohort. Eighteen patients developed 90 d SWI and 362 did not. Laboratory analyses used complete-case denominators specific to the marker and time point. POD1 WBC/ANC/ALC/NLR values were complete in 368 patients (18 SWI events), POD3 values were complete in 336 patients (17 SWI events), and paired POD1–POD3 values were complete in 327 patients (17 SWI events) (Table 1).
Cohort flow and laboratory analysis denominators.
Patients who developed SWI had higher BMI, more frequent morbid obesity, and longer operative duration. Other demographic, laboratory, and operative characteristics were not significantly different by SWI status (Table 2).
Baseline demographics and operative characteristics by SWI status.
Values are median (interquartile range) for continuous variables and
Among the 18 patients classified as having 90 d SWI, nine underwent return to the operating room for wound exploration, washout, irrigation and debridement, or related operative wound management. Seven additional patients had nonoperative wound-related complications or interventions documented in emergency department or readmission records, including incisional drainage, wound dehiscence, wound-vacuum/prevena management, abscess, operative-site bleeding, or wound pain/discharge. Two patients were classified as nonoperative/antibiotic-only or clinically treated SWI cases because no return to OR or procedural wound intervention was documented in the available dataset (Table 3). Because only nine return-to-OR/washout events occurred, no regression sensitivity analysis was performed for this restricted phenotype.
Clinical characterization of 90 d SWI events by treatment phenotype.
Categories were assigned using the 90 d SWI flag, return-to-OR field, readmission/ER fields, and wound-related reason text. The intermediate category is labeled nonoperative wound complication/intervention because some cases had wound-related clinical documentation without a clearly documented procedure.
Patients with complete paired POD1–POD3 WBC/ANC/ALC/NLR values (
Comparison of patients with and without complete paired POD1–POD3 laboratory values.
Values are median (interquartile range) for continuous variables and
Among 327 patients with complete paired POD1 and POD3 values, 17 developed 90 d SWI and 310 did not. WBC and NLR decreased significantly from POD1 to POD3 across the paired cohort (WBC decline
By SWI status, POD1 WBC, POD3 WBC, POD1 NLR, absolute WBC change, percentage WBC change, absolute NLR change, and percentage NLR change did not differ significantly. POD3 NLR was higher among patients who developed SWI (5.1 [4.3–8.9] vs. 3.7 [2.7–5.3],
POD1, POD3, and trajectory inflammatory markers by SWI status among patients with paired laboratory values (
Values are median (interquartile range). Delta values are calculated as POD3–POD1; negative values indicate decline from POD1 to POD3. The complete-case denominator was
POD1 WBC, POD1 NLR, POD3 WBC, and POD3 NLR were associated with SWI in univariable models. After adjustment for BMI and operative time, POD1 associations were attenuated, whereas POD3 WBC and POD3 NLR remained associated with SWI. Sensitivity models adjusting POD3 values for corresponding POD1 values attenuated these associations; delta WBC and delta NLR were not significantly associated with SWI.
Univariable, parsimoniously adjusted, and trajectory-sensitivity logistic regression models for 90 d SWI.
Adjusted models included the marker of interest, BMI, and operative time unless otherwise noted. Odds ratios are exploratory association estimates and should not be interpreted as validated diagnostic thresholds. POD1 and POD3 markers were analyzed in separate models.
In this retrospective cohort of adults undergoing elective short-segment lumbar fusion, POD3 WBC and NLR showed exploratory associations with 90 d SWI after parsimonious adjustment for BMI and operative time, whereas POD1 associations were attenuated after adjustment. These findings suggest that immediate postoperative inflammatory markers are strongly influenced by patient and operative context, while POD3 values may retain a residual exploratory association with SWI. The first 24 h after lumbar fusion are marked by leukocytosis and stress-associated differential shifts from tissue trauma, perioperative medications, hemodilution, physiologic stress, and obesity (Watt et al., 2015; Kobayashi et al., 2020; Liu et al., 2018). Isolated POD1 WBC or NLR elevation may therefore reflect surgical trauma rather than early infection.
NLR and WBC values typically begin to stabilize by postoperative day three to four as the acute inflammatory response resolves (Aono et al., 2025; Deirmengian et al., 2011). Prior work has reported associations between POD4 NLR/WBC and SWI in a heterogeneous lumbar surgery population (Shen et al., 2019). In the present cohort, POD3 elevation, particularly NLR, showed a stronger exploratory association with SWI than POD1 elevation, although interpretation still requires clinical context.
The trajectory analyses refined this interpretation. Although WBC and NLR declined significantly from POD1 to POD3, the magnitude of absolute or percentage decline did not differ by SWI status. In addition, the POD3 associations were attenuated when the corresponding POD1 value was included in sensitivity models. Thus, the most defensible trajectory-related observation was higher POD3 NLR among SWI patients, not failure of WBC/NLR to decline or independent clinical utility of the POD3 values.
These results should be interpreted cautiously. This study did not develop diagnostic thresholds, receiver operating characteristic analysis, calibration assessment, sensitivity, specificity, predictive values, or decision-curve analysis. The findings should not guide clinical protocols without prospective validation.
Postoperative spine infection definitions remain heterogeneous. CDC/NHSN and ACS-NSQIP definitions provide standardized surveillance categories but may not fully capture the clinical overlap between superficial wound problems, sterile dehiscence, deep infection, implant-associated infection, delayed presentation, and culture-negative infection after instrumented fusion (Alavi et al., 2025; Shaw et al., 2025; Warner et al., 2026). Recent postoperative spine infection frameworks emphasize integrating wound findings, microbiology, imaging, inflammatory markers, intraoperative findings, and histology when available (Alavi et al., 2025; Shaw et al., 2025). The present SWI definition was chosen to reflect real-world wound events captured in the electronic health record, but it should not be interpreted as a uniformly culture-confirmed deep infection, implant-associated infection, or standard surveillance-defined SSI. For transparency, events were further categorized as return to OR, nonoperative wound complication/intervention, or nonoperative/antibiotic-only clinical treatment.
Prior spine studies have often sought cutoffs or time points distinguishing uncomplicated recovery from infection (Shen et al., 2019), and sustained postoperative fever and C-reactive protein trajectories have also been evaluated as indicators of infectious complications (Hwang et al., 2020; Mok et al., 2008). However, adjustment for operative characteristics has been inconsistent. Iwata et al. (2016) identified lymphocyte count at POD4 and C-reactive protein (CRP) at POD7 as markers less affected by operative variables. Kraft et al. (2011) reported that BMI
This work highlights the importance of patient and operative context when interpreting early postoperative WBC and NLR. POD3 elevations showed stronger exploratory associations with SWI than immediate POD1 elevations, but those associations weakened after accounting for corresponding POD1 values and remain unvalidated.
This study has several limitations. First, it is retrospective and subject to selection bias. Patients with complete paired POD1–POD3 laboratory values had longer operative duration, greater blood loss, longer hospital stay, and more open procedures than patients without complete paired values. The paired trajectory analyses may therefore be most generalizable to patients who remain hospitalized through POD3 testing.
Second, only 18 SWI events occurred, and only 17 had complete POD1–POD3 paired values. This limited covariate adjustment and increased the risk of unstable estimates. Multivariable models were intentionally parsimonious; penalized regression, internal validation, and propensity-score methods should be considered in larger cohorts. Only nine events required return to OR or washout, precluding a stable regression sensitivity analysis restricted to this more definite phenotype.
Third, detailed microbiologic data were not available for all SWI cases. Culture status, causative organisms, polymicrobial infection, antibiotic duration, and standardized infection-depth classification could not be reported consistently. Future studies should distinguish superficial incisional infection from deep or implant-associated postoperative spine infection using standardized criteria.
Fourth, CRP, ESR, procalcitonin, and albumin were not available, limiting comparison with markers commonly used after spine surgery. WBC and NLR are nonspecific and may also reflect urinary tract infection, pulmonary complications, systemic illness, or other postoperative complications that were not fully adjudicated.
Fifth, potential confounders, including ASA class, drain use, wound class, local antibiotic use, instrumentation details, standardized revision status, and immunosuppressive therapy, were not uniformly available. Residual confounding remains possible.
Despite these limitations, the study indicates that routine laboratory values on the first postoperative day are heavily confounded by operative stress and patient factors, while POD3 values may show an exploratory association with SWI. Larger prospective, multicenter studies are needed to validate these findings and determine whether any clinically useful thresholds can be established.
Among adults undergoing elective short-segment lumbar fusion, POD3 WBC and NLR showed exploratory associations with 90 d SWI after limited adjustment, while POD1 associations were more dependent on patient and operative context. These POD3 associations were attenuated after adjustment for corresponding POD1 values, and neither absolute nor percentage POD1-to-POD3 changes were significantly associated with SWI. These results support further study of context-based interpretation of postoperative CBC markers but do not establish diagnostic thresholds, clinical utility, or a validated prediction model.
No custom code was developed for this study. Statistical analyses were performed using IBM SPSS Statistics, version 30.0.0.
The data underlying this study contain protected health information and are not publicly available. De-identified data may be made available from the corresponding author upon reasonable request and with appropriate institutional approvals.
JC and MSF conceived and designed the study. JC performed data collection and drafted the manuscript. YL contributed to statistical analysis. All authors contributed to interpretation of results, manuscript revision, and approval of the final paper.
The contact author has declared that none of the authors has any competing interests.
This retrospective cohort study was approved by the Einstein institutional review board (approval no. 2016-6896). The requirement for informed consent was waived due to the retrospective design and use of existing clinical data.
The views expressed in this manuscript are those of the authors and do not necessarily reflect the views of their institutions.Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. The authors bear the ultimate responsibility for providing appropriate place names. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.
The authors thank the clinical and research teams involved in data collection and study coordination.
This paper was edited by Fintan Moriarty and reviewed by two anonymous referees.