Effects of serum sodium and chloride levels in the outcome of critically ill pediatric patients in the post-operative period of liver transplantation

Study design and participants

We performed a retrospective analytical observational study of pediatric patients who underwent liver transplantation between January 2015 and July 2019 at the Instituto da Criança e do Adolescente do Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo (ICR-HCFMUSP), a reference center in Brazil for pediatric liver transplantation.

The inclusion criteria were: (1) age between 1 month and 18 years and (2) admission to the PICU on the immediate postoperative period after liver transplantation. All included patients were previously clinically stable and out of critical illness, being referred to the operation room from home or general infirmaries. Exclusion criteria were as follows: (1) patients with acute hepatic failure and (2) patients who underwent a second liver transplantation. These patients were excluded because of the lack of information regarding previous clinical conditions treated outside the study’s institution and the large number of complication variables related to these two conditions.

The outcomes of interest analyzed were: (1) adjusted hazard-ratios for mortality over the first 28 days of PICU and (2) development of new Kidney Disease Improving Global Outcomes (KDIGO) [17] stages 2 and 3 acute kidney injury (AKI). Both hourly urine output and worst daily serum creatinine levels were used for AKI diagnosis and stratification.

Hospital das Clínicas da Faculdade de Medicina da Universidade de Sao Paulo Ethics’ Committee – CEP (Approval number #1484/04/2019) previously approved the present study. All methods were carried out in accordance with ethical guidelines and regulations, and in accordance with the Helsinki Declaration. Due to the retrospective nature of the study, the application of written informed consent was waived by Hospital das Clínicas da Faculdade de Medicina da Universidade de Sao Paulo Ethics’ Committee - CEP.

Variables and definitions

Patient demographic and clinical data were obtained from electronic medical records and service databases. Clinical and laboratory data were collected through the PICU and evaluated at previously defined time points: admission, postoperative day 1 (D1), postoperative day 3 (D3), postoperative day 5 (D5), post-operative day 14 (D14), and discharge or death.

Baseline variables of interest were as follows: (1) age; (2) sex; (3) weight; (4) baseline diagnosis of hepatopathy; (5) type of transplantation (living or deceased donor); (6) absolute and relative size of the organ implant (defined as implant-to-body weight in %); (7) baseline creatinine values; (8) Pediatric End-Stage Disease Score (PELD) [18]; (8) intra-operatory fluid balance; and (9) Pediatric Index of Mortality 3 score (PIM-3) [19].

Baseline serum creatinine values were obtained from hospital admission laboratorial exams obtained immediately before liver transplantation. Routinely, all the patients admitted for liver transplantation in the study’s institution are submitted to a baseline laboratorial screening immediately before operation room admission for liver transplantation.

Sequentially, at the above-mentioned predefined time points, serial measures of biochemical and organic functional markers were collected: (1) serum chloride, (2) arterial pH and bicarbonate levels, (3) serum sodium, (4) serum potassium, (5) albumin levels, (6) albumin-corrected anion-gap (AG), (7) aspartate aminotransferase (AST), (8) serum creatinine (Cr), and (9) KDIGO acute kidney injury category.

For hyperchloremia, the study used [Cl] \(\ge\) 110 mEq/L. Hyponatremia was defined as [Na] < 135 mEq/L and hypernatremia was defined as [Na] > 145 mEq/L.

For data collection and management, preconceived RedCap® (Research Electronic Data Capture) data tools hosted in our institution were used.

Statistical analysis

Quantitative variables are described as means and standard deviations or medians and interquartile ratios. Qualitative and demographic variables were reported as absolute and relative frequencies. Continuous variables with normal distribution were initially evaluated using the Student’s t-test. Continuous variables without a normal distribution were analyzed using the Mann–Whitney test. Categorical variables were analyzed using chi-squared, Fischer’s exact test, or likelihood ratio tests, when applicable.

For Hazard-ratios for mortality in 28 days, the influences of baseline parameters were evaluated individually by bivariate Cox regression, including age, sex, weight, presence of hyperchloremia on PICU admission, presence of hyper or hyponatremia during PICU stay, estimated intraoperative fluid balance, baseline serum creatinine levels, PIM-3 predicted mortality, PELD/MELD scores, and type of liver donor (living or deceased). The multiple Cox regression models were adjusted for each outcome, the variables that presented a descriptive level lower than 0.20 (p < 0.20) in the bivariate analysis, other than sex, age, and presence of hyperchloremia on admission, were maintained in the complete adjusted model. Final complete model included age, sex, presence of hyperchloremia on PICU admission, PIM-3 predicted mortality and presence of hypo or hypernatremia during PICU stay.

For the evolution with KDIGO 2 or 3 (moderate/severe) AKI, associations of the qualitative variables with the outcomes were analyzed using the chi-squared or Fischer’s exact test, and for continuous variables by the Mann-Whitney test, with the influence of isolated parameters on the outcome being evaluated by bivariate logistic regressions including age, sex, weight, presence of hyperchloremia on PICU admission, presence of hyper or hyponatremia during PICU stay, estimated intraoperative fluid balance, baseline serum creatinine levels, PIM-3 predicted mortality, PELD/MELD scores, and type of liver donor (living or deceased). The models of multiple logistic regression were adjusted with the variables that presented a descriptive level on bivariate analysis of less than 0.20 (p < 0.20), other than sex, age, and presence of hyperchloremia on PICU admission. Final complete model included age, sex, presence of hyperchloremia on PICU admission, PIM-3 predicted mortality, PELD/MELD scores, type of liver donor and presence of hypo or hypernatremia during PICU stay.

Linear correlations between serum sodium (maximum and minimum) and admission chloride were analyzed through Pearson’s correlation and represented by Pearson’s correlation coefficient (r), as a measure of correlation strength.

Longitudinal data collected throughout the PICU length of stay were described by time point of evaluation and by main outcome status (survivors vs. non-survivors) and compared with generalized estimating equations (GEE) with marginal-normal distribution and identity link functions, assuming an auto-regressive correlation matrix of first order between the time points.

A type 1 error of 5% was considered for statistical significance.

Statistical analyses were performed on specific software, including SPSS (IBM SPSS Statistics®) and SAS (SAS Institute®).

留言 (0)

沒有登入
gif