The electronic records of 1224 consecutive patients undergoing surgery for colonic resection or stoma procedures from August 1, 2016, until December 31, 2022, at Bayhealth Medical Center (Dover, DE) were reviewed retrospectively. The database includes elective and emergent procedures, and open or minimally invasive techniques, with no exclusions for age or comorbidities. A colorectal specialist and general surgeons performed the surgery. In 2017 an early recovery after surgery program was established for elective cases. The cases were managed as previously described [13]. The study was approved by the Bayhealth Institutional Review Board.

Primary outcome

The primary outcome of interest was the relative contribution of systemic inflammation to acute kidney injury and AKI-related complications.

Definition of AKI and groups examined

AKI was defined using the KDIGO criteria, which included any creatinine increase of ≥ 0.3 mg/dl within the first 48 h after surgery, or a creatinine that was ≥ 1.5 times the baseline creatinine from day 3 to 7 [14]. Urine output was not used to determine AKI. The WBC on POD #1 and its derivatives were considered markers of inflammation. Postoperative day # 1 interleukin-6 levels correlated with the duration and invasiveness of surgery, as shown by Neff et al., so we also included procedure duration as a measure of inflammation [15].

The groups examined included the total population of 1224 patients, which included 1068 patients without concomitant infection, and 156 patients with infection on the day of surgery. This group was called the mixed population. A subgroup of patients without any concomitant infection (aseptic group) was also examined.

Any patient with a positive culture on the day of surgery was considered to have a preoperative infection; any patient with a positive culture from Day 1 through Day 30 was considered to have a postoperative infection. Any postoperative inflammation in patients with active infection was the combined result of the septic process and surgical dissection; the immediate postoperative inflammation in non-septic patients was considered aseptic inflammation.

Predictors of AKI

Previously reported causes of AKI were examined for this patient population, with an interest in causes related to inflammation. The predictors of AKI were extracted from five broad categories: demographic information, preoperative laboratories, peri-procedural laboratories, procedure-related characteristics, and medications. The following are explanations for the choice of the selected predictors.

Demographic predictors

Many predictors, such as American Society of Anesthesia (ASA) scores, body mass index (BMI), and sex were chosen because of their demonstrated importance in the literature. For demographic information, the age was categorized as elderly for sixty and greater. Diabetics were considered as those with a clinical history of diabetes.

Preoperative laboratory predictors

Some predictors were defined specifically for this study. Among the preoperative laboratories, urinary protein was considered positive for those with any level of protein identified. The average hemoglobin, including the preoperative and first three postoperative values, had a larger effect size than the preoperative value alone and was used as a predictor. Severe hyperglycemia was defined as any glucose value ≥ 180 mg/dl. Hypoglycemia was defined as glucose < 54 mg/dl. HgA1C and serum albumin were also assessed. The preoperative WBC was included to determine if there was any difference in predicting AKI in comparison with the postoperative day POD #1 WBC.

Procedure-related predictors

In procedure-related characteristics, minimally invasive surgery was laparoscopic, with some robotic or robot-assisted cases included. Ureteral stents included those with unilateral or bilateral stent placement for ureter identification. Bolus therapy with either Ringer’s lactate (RLB) or normal saline (NSB) was identified separately from continuous infusions and was considered as a binary variable. Transfusion volume was the volume of blood given per case. Any patient with net inputs and outputs (I & O’s) between 0 and 3 L on the morning after surgery was euvolemic.

Procedure related laboratoriesClinical markers of inflammation

Among the procedure-related laboratories, the WBC has two important characteristics: it is common and can be obtained at varying perioperative phases, which could exhibit different levels of inflammation. The POD #1 WBC was obtained in 82% of patients and reflected some elements of septic and aseptic inflammation, since 12.8% of patients presented with a preoperative infection.

Neutrophil-to-lymphocyte ratios (NLR) on POD # 1 were less frequently available (40.0%), but may be a more sensitive predictor for complications in patients with aseptic inflammation [16]. The NLR is less useful in patients with sepsis. The NLR at this time would not be expected to reflect a postoperative infection, since there was little time for such an infection to develop. Consequently, the POD # 1 WBC was used as a marker of inflammation for all colorectal patients, including those with sepsis, and the POD # 1 NLR was used as a marker for patients with aseptic inflammation only. It would also most likely represent peak inflammation since it was obtained within 24 h of surgery.

Procedure duration was used as an indirect marker of inflammation, as more extensive dissection, which would increase aseptic inflammation, is associated with longer operative time and surgical intensity [15, 17]. Specific markers of inflammation, such as C-reactive protein, or cytokines were not available in this retrospective study.

Medications

Medications were considered as either given or not given in the first three postoperative days. Dosages were not evaluated. Perioperative steroids were used to control postoperative nausea. Angiotensin-converting enzyme (ACE) inhibitors and angiotensin receptor blockers (ARB) are often held postoperatively for 48 h because of a link to AKI. The holding of ACE/ARBs for patients on these medications was evaluated as a predictor.

A large variety of predictors were chosen so that the relative importance of inflammation in inducing AKI could be evaluated. Only predictors that were reliably included in the electronic medical record and identified in Epic electronic health records (Verona, WI) were assessed. Not all information related to AKI could be evaluated; for instance, intraoperative fluid administration could not be determined, and the length and degree of perioperative hypotension, although important, was beyond the scope of this retrospective review.

Secondary outcomes

Perioperative complications included postoperative infections, chronic renal insufficiency (CRI), non-infectious complications, 30-day readmissions, and length of stay (LOS). Infections were calculated by documenting positive clinical cultures in the postoperative period from Day 1 to Day 30, as previously described [18]. Cultures were considered positive if pathogenic bacteria were identified, not normal flora. Mixed urogenital flora, or normal skin flora were not considered positive. A patient was considered to have a positive culture if one or more cultures were positive.

Chronic renal insufficiency was defined as an increase in creatinine of ≥ 0.3 mg/dl over the baseline creatinine from 3 to 12 months after surgery. This numerically small increase in creatinine was chosen as a cutoff because of the documented increase in complications associated with the earliest stage of AKI.

A variety of significant non-infectious in-hospital complications were also included and were extracted with software from Conduent-Midas Health Analytics Solutions (Florham Park, NJ). Respiratory and cardiopulmonary failure or arrest, acute myocardial infarction or stroke, and deep vein thrombosis or pulmonary embolus were included. Infectious and non-infectious complications were combined for an in-hospital complication rate. For combined complications, any patient was considered positive with one or more complications.

Data interpretation

Data was electronically extracted from Epic and analyzed using Excel. Data for continuous variables was expressed as mean and standard deviation; categorical variables were expressed as percentages. Two sample t-tests and tests for two proportions were used to compare groups. Mann-Whitney non-parametric tests were used for non-normally distributed data. Univariate logistic regression was used to determine significant (P < 0.05) associations between potential predictors and AKI, and multivariate regression with backward elimination was employed to determine independent predictors.

Data availability

The datasets generated or analyzed during the current study are not publicly available since they are private health information, but the data could be made available from the corresponding author upon reasonable request, consistent with IRB review and requirements.