Acute Kidney Injury Among Patients Undergoing Orthopedic Surgery and Admitted to the ICU: A Retrospective Analysis of Nonoperative Risk Factors

Introduction

Acute kidney injury (AKI) is a common and serious complication following orthopedic procedures, occurring in 5% to 20% of patients, depending on surgical complexity and patient demographics [1,2]. Postoperative AKI is associated with prolonged hospital stays, increased healthcare costs, and a 2- to 5-fold higher risk of short- and long-term mortality [3,4]. The incidence is particularly elevated in elderly patients and those undergoing major surgeries, such as joint arthroplasty or spinal fusion, highlighting the need for early risk identification and prevention strategies [5–8].

While intraoperative factors, such as blood loss, hemodynamic instability, and nephrotoxic medications, have been well documented, emerging evidence suggests that non-surgical factors – particularly pre-existing comorbidities – play a crucial role in AKI development [9,10]. Chronic kidney disease (CKD), diabetes mellitus, hypertension, and cardiovascular disease significantly impair renal reserve, increasing susceptibility to kidney injury in the perioperative period. Additionally, modifiable factors, such as dehydration, non-steroidal anti-inflammatory drug use, and contrast exposure can further exacerbate risk [11,12]. Despite their clinical relevance, these non-surgical contributors remain understudied in orthopedic patient populations, limiting evidence-based prevention protocols.

We hypothesize that AKI following orthopedic surgery is predominantly influenced by pre-existing medical conditions, which interact with perioperative stressors to induce renal dysfunction. Specifically, patients with CKD, diabetes, or cardiovascular disease can experience exacerbated endothelial dysfunction, oxidative stress, and impaired hemodynamic regulation, making them more vulnerable to even minor intraoperative insults [13,14]. Identifying these high-risk subgroups could enable the use of targeted interventions, such as preoperative optimization, individualized fluid management, and avoidance of nephrotoxic agents, to ultimately reduce AKI incidence and improving surgical outcomes.

In this study, we aimed to (1) determine the incidence of AKI in postoperative orthopedic surgery patients admitted to the intensive care unit (ICU) and (2) identify independent non-surgical predictors using multivariable regression and diagnostic performance metrics.

Material and Methods

DATA SOURCE AND STUDY POPULATION:

This single-center retrospective cohort study was conducted using data from the Medical Information Mart for Intensive Care (MIMIC)-IV database, covering ICU admissions at Beth Israel Deaconess Medical Center between 2008 and 2019 [13]. The study was approved by the Institutional Review Board of the Beth Israel Deaconess Medical Center (2001-P001699/14) and the Massachusetts Institute of Technology (No. 0403000206). Since all patient data were anonymized, the requirement for informed consent was waived. We included adult patients (≥18 years) who underwent orthopedic surgery and were subsequently admitted to the ICU. Additionally, patients with missing essential clinical data or those diagnosed with AKI before ICU admission were excluded from the analysis.

INCLUSION AND EXCLUSION CRITERIA:

The study population consisted of postoperative orthopedic surgery patients admitted to the ICU within 24 h of their procedure. Eligible surgeries included joint replacements, fracture repairs, spinal operations, and others, and patients were required to have at least 2 serum creatinine measurements for accurate AKI assessment during their hospitalization. Exclusion criteria were systematically applied to ensure a homogeneous study cohort: patients with pre-existing advanced CKD (stages 4,5), those lacking critical baseline data (such as weight or key laboratory values), and patients who developed AKI before ICU admission. This approach allowed us to focus specifically on new-onset AKI attributable to the perioperative and critical care phases.

AKI is defined according to the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines as any of the following within 48 h [14]: an increase in serum creatinine by ≥0.3 mg/dL (≥26.5 μmol/L), an increase in creatinine to ≥1.5 times baseline (within prior 7 days), or urine output <0.5 mL/kg/h for ≥6 h.

DATA COLLECTION AND DEFINITIONS:

Clinical and demographic variables were extracted from the MIMIC-IV database using structured query language (SQL), encompassing baseline characteristics, comorbidities, vital signs, laboratory values, and illness severity scores. Demographic data collected were age, sex, height, weight, smoking status, and alcohol abuse history, while comorbidities were hypertension, diabetes, myocardial infarction, congestive heart failure, liver diseases (mild and severe), CKD (stages 1–3), chronic pulmonary disease, cerebrovascular disease, peripheral vascular disease, malignancies, metastatic solid tumors, AIDS, and rheumatic disease. The vital signs recorded within the first 24 h of ICU admission were temperature, heart rate, respiratory rate, systolic and diastolic blood pressure, mean arterial pressure, and oxygen saturation. Laboratory tests included pH, lactate, glucose, partial pressure of arterial oxygen, partial pressure of carbon dioxide, white blood cell count, platelet count, hemoglobin level, serum creatinine level, blood urea nitrogen level, prothrombin time, activated partial thromboplastin time, international normalized ratio, potassium level, and sodium level. Postoperative AKI was defined per KDIGO criteria as either an absolute serum creatinine increase of ≥0.3 mg/dL within 48 h or a ≥1.5-fold rise from baseline within 7 days. Illness severity scores were selected based on their established utility in ICU mortality prediction and potential to reflect early physiological derangements contributing to AKI. Illness severity was assessed using the following: the Simplified Acute Physiology Score II (SAPS II), Sequential Organ Failure Assessment (SOFA) score, Glasgow Coma Scale score, systemic inflammatory response syndrome criteria, Logistic Organ Dysfunction System (LODS), Charlson Comorbidity Index, Model for End-Stage Liver Disease (MELD) score, and the Oxford Acute Severity of Illness Score (OASIS). Additional clinical outcomes included sepsis occurrence, invasive mechanical ventilation, cardiopulmonary resuscitation, continuous renal replacement therapy (CRRT), and lengths of ICU and hospital stays. All data were derived from the initial 24-h ICU period, to ensure standardized comparisons.

STATISTICAL ANALYSIS:

Adult patients were admitted to the ICU between 2008 and 2019. Continuous variables were reported as medians with interquartile ranges (IQR) and compared using Mann-Whitney U tests, while categorical variables were presented as frequencies and percentages and analyzed using chi-square or Fisher exact tests, as appropriate. Multivariable logistic regression was used to identify independent risk factors for postoperative AKI, adjusting for potential confounders with the index of P<0.05 in univariate analysis. Stepwise regression (backward elimination with P<0.10 retention) further refined covariates. The variance inflation factor (<2.5 for all included variables) was used to test for multicollinearity among predictors. The predictive performance of significant risk factors was assessed using receiver operating characteristic (ROC) curve analysis, with area under the ROC curve (AUC) values, optimal cutoff points, sensitivity, and specificity reported. Sensitivity analyses were conducted by excluding patients with borderline CKD to ensure robustness. We first evaluated the extent and pattern of missingness using Little’s MCAR test and visual inspection. Complete case analysis was primarily used for variables with <5% missingness (eg, demographics). Multiple imputation via chained equations (m=20 imputations) was applied for covariates with >5% missing data (eg, laboratory values, comorbidities). There were no missing data for the primary outcome. We used standardized MIMIC-IV definitions (eg, ICD-10 codes for comorbidities, laboratory thresholds) to avoid measurement bias. Multivariate regression was used to adjust for key covariates to avoid confounding factor bias. Based on an expected 15% event rate and α=0.05, β=0.20, a sample size of 1500 was required to detect a minimum odds ratio (OR) of 1.5 for our primary exposure. All statistical tests were 2-tailed, with a P value <0.05 considered statistically significant. Data analysis was performed using R (version 4.3.0).

Results

BASELINE CHARACTERISTICS AND COMORBIDITIES:

We identified 2482 postoperative orthopedic surgery patients from the MIMIC-IV database. After excluding patients with ICU stays shorter than 24 h (n=423), pre-existing CKD stage 5 (n=148), and >20% key clinical variables (n=120), a final cohort of 1791 patients was included for analysis (Figure 1). The study included 1791 patients, with 1272 (71.1%) developing AKI postoperatively. Compared with the non-AKI group, patients with AKI were significantly older (median age 65.17 vs 58.18 years, P<0.001) and had higher body weight (84.00 vs 72.60 kg, P<0.001). Notably, patients with AKI exhibited a greater burden of comorbidities, including hypertension (41.12% vs 33.53%, P=0.003), diabetes (28.54% vs 18.50%, P<0.001), congestive heart failure (23.43% vs 10.60%, P<0.001), and CKD (18.00% vs 9.63%, P<0.001). Additionally, patients with AKI had significantly higher baseline serum creatinine (0.90 vs 0.80 mg/dL, P<0.001) and blood urea nitrogen levels (18.00 vs 14.00 mg/dL, P<0.001), suggesting pre-existing renal dysfunction. Hemodynamic and metabolic differences were also observed, with lower systolic and diastolic blood pressures (P=0.001 and P<0.001, respectively) and elevated glucose levels (135.50 vs 126.50 mg/dL, P<0.001) in the AKI group (Table 1).

DISEASE SEVERITY AND CLINICAL OUTCOMES:

Patients with AKI demonstrated significantly worse disease severity across multiple scoring systems, including higher SAPS II (35 vs 26, P<0.001), SOFA (5 vs 2, P<0.001), and LODS (4 vs 2, P<0.001) scores. The AKI group also had a higher incidence of sepsis (59.75% vs 32.56%, P<0.001) and required invasive mechanical ventilation more frequently (61.08% vs 31.41%, P<0.001). Notably, CRRT was exclusively used in patients with AKI (4.17% vs 0%, P<0.001). Furthermore, AKI was associated with prolonged hospitalization (median 11.66 vs 8.32 days, P<0.001) and extended ICU stays (3.00 vs 1.18 days, P<0.001), reinforcing the substantial clinical and economic burden of postoperative AKI (Table 2). Kaplan-Meier survival analysis demonstrated a significantly higher risk of adverse outcomes in patients who developed AKI than in the non-AKI group in 28-day mortality, there was no difference in long-term prognosis (Figure 2).

MULTIVARIABLE LOGISTIC REGRESSION ANALYSIS FOR AKI PREDICTION:

Multivariable logistic regression analysis, incorporating all variables with P<0.05 in univariate analysis, identified several independent predictors of postoperative AKI.. Weight (OR 1.03 95% CI 1.02–1.03, P<0.001), length of ICU stay (OR 1.42, 95% CI 1.29–1.56, P<0.001), OASIS score (OR 1.05, 95% CI 1.03–1.09, P<0.001), SOFA score (OR 1.13, 95% CI 1.03–1.25, P=0.012), SAPS II score (OR 1.03, 95% CI 1.01–1.06, P=0.013), severe liver disease (OR 0.23 95% CI 0.06–0.84, P=0.026), and age (OR 1.02 95% CI 1.011.03, P=0.03) emerged as significant independent predictors. Notably, higher illness severity scores (OASIS, SOFA, SAPS II) and prolonged ICU stay demonstrated particularly strong associations with AKI development, while severe liver disease showed a paradoxical protective effect. The forest plot (Figure 3) visually summarizes these associations, highlighting weight and ICU length of stay as having the widest confidence intervals among significant predictors, suggesting greater variability in their effect sizes..

PREDICTIVE PERFORMANCE OF INDEPENDENT RISK FACTORS FOR AKI:

The predictive performance of independent risk factors for postoperative AKI was assessed using ROC curve analysis (Figure 4). Among the evaluated predictors, the length of ICU stay (AUC=0.764, 95% CI: 0.741–0.787, P<0.001) and SOFA score (AUC=0.764, 95% CI: 0.741–0.787, P<0.001) demonstrated the strongest discriminatory ability, followed by SAPS II (AUC=0.709) and OASIS score (AUC=0.707). Weight (AUC=0.635) and age (AUC=0.605) showed more modest predictive value. Optimal cut-off values were identified for each parameter, with ICU stay >2.765 days (sensitivity 54.0%, specificity 83.8%) and SOFA score >3.5 (sensitivity 67.0%, specificity 72.6%) providing clinically useful thresholds for AKI risk stratification (Table 3). These findings suggest that illness severity scores and ICU resource utilization measures may be particularly valuable for early identification of high-risk orthopedic surgery patients.

Discussion

STUDY LIMITATIONS:

Despite these insights, several limitations should be acknowledged. First, the retrospective design introduces potential biases, likely resulting in an underestimation of AKI severity, due to unmeasured confounders such as intraoperative fluid management. Second, while our reliance on a single-center database can limit generalizability, the MIMIC-IV database’s standardized collection likely mitigates center-specific biases and maintains external validity for comparable ICU populations. Third, our exclusion of patients with early postoperative AKI may have led to an underestimation of incidence and introduced selection bias, particularly by omitting patients with the most acute forms of renal dysfunction. Fourth, the lack of long-term follow-up data precludes assessment of CKD progression after AKI. Finally, as the data were drawn from a single U.S. academic center, generalizability to other settings, including non-ICU orthopedic populations or international contexts, can be limited.

Conclusions

In summary, this study demonstrates the high incidence and clinical impact of AKI in orthopedic surgery patients admitted to the ICU. We identified key risk factors, including illness severity scores and prolonged ICU stay, which provide actionable targets for early intervention, such as enhanced perioperative monitoring. Notably, our analysis revealed variations in AKI risk across orthopedic injuries and sex distributions, suggesting the need for tailored risk stratification. To translate these findings into clinical practice, prospective validation of our predictive model is essential. Future studies should focus on integrating risk assessment tools into clinical decision support systems to guide individualized perioperative care. Additionally, implementation research is needed to evaluate the effectiveness of targeted prevention strategies in diverse patient populations [31].