Factors affecting perioperative serum albumin variation and short-term complications in pediatric patients undergoing major gastroenterology surgery

Background: Albumin is considered a negative acute-phase protein because its concentration decreases during injury and sepsis. The decrease in serum albumin may be important for perioperative morbidity, even in patients with normal preoperative levels in the pediatric population. Here, we intend to determine the perioperative factors associated with the reduction in serum albumin within 2 postoperative days compared with the preoperative level ( ∆ ALB) and its inuence on perioperative outcome in a pediatric general surgical cohort. Methods: This single-center retrospective review included 939 patients who underwent Roux-en-Y hepaticojejunostomy between August 2010 and August 2019. Based on the mean value of ∆ ALB (14.6%), patients were separated into two groups, a high ∆ ALB group ( ≥ 14.6%) and a low ∆ ALB group (<14.6%). Multivariable logistic regression analyses were performed to determine the independent risk factors for a reduction in serum albumin. Propensity score matching was performed to adjust for any potential selection bias for the two groups. In 366 matched patients, the inuences of operating time on perioperative outcomes, including postoperative recovery, complications measurement, and length of hospital stay between the two groups were analyzed. Results: Among the 996 patients reviewed, 939 patient records were enrolled in the nal analysis. Controlling for other factors, multivariable analysis showed that a high CRP on POD 3 or 4 (odds ratio [OR] =2.36 [95% CI, 1.51-3.86]; p =0.007), the presence of Charcot's triad (OR=1.73 [95% CI, 1.05-2.83]; p = 0.031), and a longer operating time (OR=1.18 [95% CI, 1.00 -1.53]; p=0.014) were factors that predicted a high ∆ ALB level. A high ∆ ALB level was associated with postoperative gastrointestinal functional recovery, reected by the rst defecation (p= 0.013) and rst bowel movement (p=0.019) and the of postoperative vs The mean length of postoperative stay of patients in the high ∆ ALB group was longer than that of patients in the ∆ ALB < 14.0% group, although no statistically signicant difference was observed (p=0.057). Conclusions: We showed that a change in albumin level was associated with postoperative outcome. The risk factors for ∆ ALB could be addressed in the perioperative period to permit patients to obtain a safe recovery and discharge after a major abdominal operation.


Background
As an acute-phase protein, albumin (ALB) usually negatively responds to surgical stress, injury or sepsis [1,2]. A decrease in plasma albumin concentration is considered one of the features of systemic in ammation due to the loss of albumin to the tissue spaces [3,4] and postoperative infectious complications [4]. The transcapillary leakage of albumin can be elevated by more than 300 % in systemic in ammatory pathogenesis [11,12]. Preoperative hypoalbuminemia has also been con rmed as an indicator for poor nutritional status, mortality and postoperative complications such as surgical site infections (SSIs) after spine surgery [5], GI surgery [6], and acute kidney injury [7]. Because albumin has quick kinetics after surgery, the response of albumin to surgical stress can even occur earlier than that of CRP, making it an intense focus of perioperative management [8,9].
A sharp decrease of 33% in serum ALB within 2 days has been observed after major abdominal surgery [2,10]. There are many factors that may potentially impact the reduction in ALB level (∆ALB), including the pathology being treated and the actual procedure undertaken [13]. Few studies have examined the factors that could be used to predict postoperative hypoalbuminemia in pediatric patients after major gastroenterology surgery. A clear evaluation of these factors could lead to the development of optimized perioperative care protocols.
This study aimed to investigate the risk factors for the reduction in ALB by retrospectively reviewing data on pediatric patients that had undergone major gastroenterological surgery. We further sought to clarify the association between ∆ALB and postoperative recovery and outcomes.

Population selection
The Ethics Committee of Chongqing Medical University gave expedited approval to this protocol. This retrospective study included 996 consecutive patients who underwent elective Roux-en-Y hepaticojejunostomy between August 2010 and August 2019 at the Department of General Surgery of the A liated Chongqing Children's Hospital, Chongqing Medical University. Exclusion criteria included patients with ALB infusion performed preoperatively or within postoperative day (POD 1) or with incomplete laboratory data.

Data collection and de nitions
Electronic medical records generated upon admission or referral and including clinician and nurse notes, laboratory tests, imaging exams, surgical records and histopathology results were individually reviewed by two well-trained clinical investigators who collected the relevant data. Data extraction included (1) preoperative data, including demographic data and clinical details, such as preoperative neutrophils, lymphocytes, hemoglobin, preoperative CRP and ALB, and pre-existing comorbidities; (2) intraoperative variables, including surgical procedures (surgical approach, type of resection), American Society of Anesthesiology (ASA) classi cation, operation time, duration of operation, estimated blood loss (EBL), intraoperative blood transfusion, and intraoperative hemoglobin levels; and (3) postoperative outcomes, including CRP on POD 3 or 4, ALB on POD 1, prompt postoperative biochemical pro les (hemoglobin, blood glucose, creatinine, serum electrolytes, albumin, retinol binding protein, WBC, CRP, etc.), gastrointestinal function recovery features and postoperative complications. The following laboratory data were determined preoperatively and on PODs 3 and 7: serum albumin, CRP, aspartate aminotransferase (AST), alanine aminotransferase (ALT), gamma-glutamyl transpeptidase (γ-GTP), lactate dehydrogenase (LDH), alkaline phosphatase (ALP), serum creatinine (Scr), blood urea nitrogen (BUN), hemoglobin (Hb), and white blood cell (WBC) count. Gastrointestinal symptoms were recorded for the rst 5 postoperative days, including the rst postoperative atus or defecation, gastric retention, nausea or vomiting, time to normal diet, abdominal bloating and/or cramps. All postoperative complications were recorded and ranked according to the Clavien-Dindo classi cation system [14], such as postoperative hemorrhage, anastomotic stula, anastomotic stenosis and wound infection, intraabdominal abscesses, pneumonia, bacteremia, renal failure and respiratory failure.
The relative change in serum albumin (∆ALB) was calculated as follows: (preoperative albumin levelnadir albumin level within POD 2)/albumin level before surgery × 100% [17]. We evaluated ∆ALB as normally distributed data and decided to use the mean value of the (14.6%) to dichotomize the groups. For the purpose of analysis, the patients were dichotomized into a low (<14.6%) and a high (≥14.6%) ∆ALB group based on the cutoff value (14.6%).
According to criteria reported in previous studies, all patient data were reviewed for postoperative surgical and nonsurgical outcomes, including complication rates, complication types, mortality rates, ventilator rates, ICU-stay rates, and total lengths of hospital stay (the number of days from the day of operation until the day of discharge). The primary outcome based on ∆ALB was prompt postoperative gastrointestinal function recovery. The secondary outcomes were postoperative complications and immunologic and in ammatory variables. Gastrointestinal symptoms were assessed and recorded daily for the rst 5 postoperative days, including rst bowel movement (gas and feces) after operation, abdominal bloating, abdominal cramps, diarrhea (de ned as more than three bowel movements per day), and vomiting. Wound complications consisted of wound dehiscence, erythema, swelling, and pus.

Propensity scores and matching
To minimize selection biases in the baseline characteristics between the two groups, propensity score matching was accomplished using a multivariable logistic regression model using SPSS 20.0 (IBM, Armonk, NY) or R 3.1.2 (The R Foundation for Statistical Computing). A 1:1 propensity score matching with a caliper distance of 0.2 without replacement was accomplished using nearest-neighbor analysis, which included the demographic and clinical variables. The selected variables entered into the propensity model were based on theoretical and empirical considerations according to the scienti c literature and biological plausibility, including demographic data, laboratory values, treatment protocols, surgical features, etc. We further measured the interaction among all pretest covariables. The assumption of linearity of the PS model was checked using the generalized additive model, subsequently matching 366 patients with high ∆ALB ≥ 14.6% and 366 patients with ∆ALB <14.6%. The perioperative outcomes between the two groups were compared after propensity score matching.

Statistical analysis
Statistical analyses were performed using SPSS version 19.0 (SPSS Inc., Chicago, IL, United States). Categorical data are expressed as counts with percentages and were analyzed using Fisher's exact test or Pearson's χ2 test when appropriate. Continuous data are expressed as the means ± (standard deviations) for normally distributed data and medians (interquartile ranges) for nonnormally distributed data, which were tested with Student's t-test or Mann-Whitney U test and the Wilcoxon rank-sum test, respectively. To verify independent predictors for postoperative complications, multivariate analysis was performed using multivariate logistic regression analysis after univariate analysis to identify predictors with a signi cance level of P< 0.30 (Table 3). The results of the multivariate logistic regression analysis were expressed using the P value, odds ratio (OR), and 95% con dence interval (CI). In all cases, P< 0.05 was considered signi cant.

Patient population characteristics
Among the initial 996 pediatric patients in our department who underwent hepaticojejunostomy resection, 41 did not ful ll the inclusion criteria and were initially excluded, and sixteen patients were excluded because their notes were unobtainable for data extraction. Finally, a total of 939 patient records were enrolled in the nal analysis (Table 1). CBD, Common bile duct; CRP, C-reactive protein; POD, postoperative day; SD, standard deviation; WBC, white blood cell.

Factors associated with ∆ALB
Based on the cutoff value of ∆ALB, we dichotomized the patients into two groups, the ∆ALB ≥ 14.6% and ∆ALB <14.6% groups. The baseline features of the two groups are summarized in Table 1. There were no signi cant differences in the demographic features of the patients, ASA classi cation, or preoperative CRP between the two groups (Table 1). Furthermore, the nadir of hemoglobin and operative blood loss were similar between the two groups (P > 0.05). A larger choledochal cyst size (p=0.042), worse comorbidity (Charcot's triad) (p=0.011), lower preoperative ALB (p=0.016), CRP on POD 3 or 4 (p<0.001), and a longer surgery duration (p=0.0024) were associated with a greater ∆ALB (P < 0.05) in the univariable analysis. Multivariable analysis revealed three independent risk factors. As shown in  In uences of ∆ALB on postoperative outcomes To explore the association between ∆ALB and perioperative outcome, we performed PS matching between the ∆ALB ≥ 14.6% and ∆ALB <14.6% groups to adjust for potential confounding factors and exclude the effects of correlations between factors. Under PS matching, 366 patients in the ∆ALB ≥ 14.6% group were matched to 366 patients in the ∆ALB <14.6% group. The absolute standardized mean differences reduced the values, indicating that the continuous and categorical variables were very similar and comparable between the two groups (Table 3). Overall, there were no differences in vasopressor support, diuresis usage and the number of intraoperative hypotensive events, hypokalemic episodes, metabolic acidosis (de ned by low bicarbonate) or other laboratory and hemodynamic parameters throughout the perioperative period between the ∆ALB ≥ 14.6% and ∆ALB <14.6% groups (Table 4).

Discussion
We conducted the present analysis to focus on serum albumin as an acute phase protein for a pediatric surgical cohort that underwent the same surgical procedure. This study revealed that several factors were associated with high ∆ALB, such as operative duration, disease comorbidities, and mean CBD.
Furthermore, patients with a greater relative decrease in serum albumin were associated with postoperative gastrointestinal function recovery, postoperative complications, and prolonged postoperative hospital stay.
The factors leading to hypoalbuminemia are often complicated and associated with operative case type, ALB loss, redistribution, catabolism, or theirs combination [15,16,17,18]. Numerous studies have focused on preoperative and postoperative hypoalbuminemia as risk factors for postoperative complications [5,19], while few have speci cally stressed the perioperative factors related to ∆ALB, which may be clinically signi cant for postoperative care. In the present study, several clinical factors related to the decrease in postoperative albumin were presented, including longer operating time, severe comorbidities, such as Charcot's triad, and high CRP. All these factors might be associated with an in ammatory response caused by surgical stress. During surgery, manipulation of the intestine has been proven to initiate gastrointestinal edema [20,21]. Postoperative edema might be attributed to the low colloid osmotic pressure and uid accumulation, themselves resulting from a low postoperative ∆ALB [11]. Norberg Å [2] reported that the stress response led to a reduction in postoperative albumin levels, which was consistent with the ndings in our study.
Changes in plasma volume and altered distributions between the intravascular and extravascular space also play a significant role in plasma albumin concentration. The reasons for this association may pertain to hemodilution, which might contribute to the decreased albumin level after surgery. In our previous study [22], postoperative complications were shown to be related to conventional intraoperative and postoperative uid usage, which might also affect the albumin level after surgery. Excessive uid could also promote capillary permeability and leakage of serum albumin into the extravascular space [23]. Capillary leakage is especially common in some malnourished patients with surgical trauma, followed by an increased transcapillary escape rate of ≥ 100% [24], which was not observed in the current study. Serum albumin on POD 3 has been observed to be correlated with preoperative CRP level [13,25]. In the current study, we indeed found that a high ∆ALB was associated with CRP on POD 3 or 4.
Prompt postoperative recovery serves as the main focus of all surgical specialties for postoperative rehabilitation [26]. As indicated in the current study, although uneventful recovery was present for most patients following choledochal cyst resection, unfavorable postoperative gastroenterological recovery was associated with a high ∆ALB. Previous biological investigations have demonstrated that edema has detrimental effects on intestinal function by directly affecting muscle function and force transmission [27,28]. Serum ALB is a predictor of both systemic inflammation and nutritional status, which should account for postoperative intestinal edema after a major operation or severe trauma. Furthermore, tissue oxygenation could decrease with uid accumulation, which is also unfavorable for postoperative recovery and complications. In this study, we detected an increased total number of complications, including anastomotic leakage, in patients with high ∆ALB. A possible explanation at the tissue level may be that the low ALB level reduced collagen deposition for the tissue connections and, therefore, resulted in poor structural integrity. The effect of ∆ALB on the local in ammatory response and edema recovery is also important and might also explain postoperative recovery and complications [29,30]. It remains to be determined whether ALB supplementation reduces postoperative intestinal edema and cellular swelling to and promote intestinal function recovery in patients undergoing major upper gastrointestinal surgery in the pediatric population [31,32].
There were several limitations in the current study. First, it was a retrospective observational analysis, where unmeasured differences and known selection and treatment biases may have contributed to confounding effects, which could not be entirely excluded. Second, this was a single-center study, and the general surgery procedures were performed in our hospital over a long period of time; therefore, there may have been many practice changes within both the surgical and ICU divisions, leading to different care practices between study patients, which may not re ect the outcomes from current treatment algorithms. Another point of emphasis was that although we did not nd differences in baseline characteristics after PS matching, the patients with high ∆ALB might have been more surgically challenged than those with low ∆ALB. To limit the in uence of confounding variables on the actual effects of ∆ALB, large, multicenter prospective studies must be performed to verify the conclusions of the current study.

Conclusions
In the current study, we characterized some risk factors that may predispose patients to high ∆ALB, which could negatively impact postoperative recovery after a major abdominal operation in a pediatric The dataset analyzed during the current study is available from the corresponding author on reasonable request.

* Competing interests
No potential con icts of interest relevant to this article are reported.

* Funding
This study was supported by grants from the National Natural Science Foundation of China (Nos: 30973440 and 30770950) for the design of the study, the Key Project of the Chongqing Natural Science * Authors' contributions QL, KG and XS designed the study and analyzed the data. KG, XS and CG evaluated the manuscript. CG and KG performed the statistical measurements and analyzed the data. CG analyzed the data and wrote the paper. All authors have read and approved the nal manuscript as submitted and agree to be accountable for all aspects of the work.