Abstract
Background
Albumin, a vital component in regulating human blood oncotic pressure, plays an important role in the prediction of prognosis in pediatric patients.Previous research identified significant differences in serum albumin levels of healthy and critically ill children.
Methods
The present study aims to investigate the correlation between albumin levels measured during pediatric intensive care unit(PICU) admission and clinical outcomes. In 2022, a comprehensive analysis of the records was conducted retrospectively in four tertiary hospital PICUs in different cities in the south-east region of Turkey. We then conducted a ROC analysis to predict mortality based on hypoalbuminemia.A cut-off value was determined, and the patients were divided into two groups: hypoalbuminemic and normoalbuminemic.Based on this cut-off level, we evaluated the association between admission serum albumin levels and mortality, length of stay, laboratories, need for mechanical ventilation, and extracorporeal treatments.
Results
A total of 1329 patients admitted to the PICUs between January 1, 2022 and January 1, 2023 were included. The mortality rate amounted to 16.2%.The mean albumin level was 4.03 ± 0.69 g/dl in the surviving and 3.35 ± 0.90 g/dl in the non-surviving group(p < 0.001).The average length of stay in the PICU amounted to 8.98 ± 17.07, while the mean length of hospital stay was 11.23 ± 18.81days.Patients with hypoalbuminemia had significantly longer stays in the PICU and hospital as compared to those with normal albumin levels(p = 0.025 and p < 0.001,respectively).We found robust evidence that albumin levels are a strong predictor of mortality, with a predictive rate of 84.5%(OR:0.322, 95% CI:0.263–0.395, p < 0.001).The cut-off value was 3.785(AUC:0.731, Sensitivity:68.3%, Specificity:67%). A negative correlation was identified between albumin levels and lactate, C- reactive protein (CRP), and Blood urea nitrogen (BUN) levels across all patients. Conversely, a positive correlation was observed between neutrophil, lymphocyte, platelet, hemoglobin, sodium, and glucose levels. However, in patients with a cut-off point of 3.785 for albumin levels, a negative correlation was observed between albumin and lactate, CRP, and creatine levels. Conversely, a positive correlation was found between bicarbonate, platelet, hemoglobin, and glucose levels. Based on these findings, we conclude that hypoalbuminemia is associated with a higher rate of inotrope therapy and mechanical ventilation support. Furthermore, the rates of plasmapheresis, continuous renal replacement therapy, and extracorporeal membrane oxygenation initiation were markedly reduced in patients with elevated albumin levels (p < 0.001).
Conclusion
The serum albumin, which is an excellent and cost-effective biomarker that can be readily and economically evaluated in any center worldwide, as evidenced by the findings of our study, is a highly efficacious indicator of admission albumin levels for the prediction of mortality and outcomes in PICUs. Furthermore, it predicts the selection of appropriate treatment and mechanical supports. It is straightforward to utilise, as it does not necessitate the use of scoring systems or tests that require lengthy and intricate analyses, and it is readily available.
Keywords: Albumin, Hypoalbuminemia, Mortality, Pediatric intensive care
Background
Predicting prognosis for critically ill patients upon admission in the pediatric intensive care unit (PICU) is crucial for making prompt and safe treatment decisions. To date, several scoring systems have been investigated; however, none of these systems has been comprehensively demonstrated to be superior to others. Accordingly, there is a clear need for a reliable and precise non-invasive predictor that could be easily implemented in all centers, especially for pediatric patients [1, 2].
A promising biomarker in this respect is albumin, a vital component that regulates human blood oncotic pressure predominantly and is synthesized in the liver. Albumin also serves as a carrier for fatty acids, bilirubin, and drugs [3]. To date, extensive evidence is available showing that critically ill patients’ serum albumin levels significantly differ from those of healthy children [1, 2]. Yet, the nature of these differences remains largely unexplored. Several previous studies established that low serum albumin levels are associated with poor outcomes and high mortality rates [4–6]. Indeed, while albumin is a potent antioxidant, critically ill patients admitted to intensive care units are sensitive to increased oxidative stresses that cause injuries to human cells and tissues and activate extracellular antioxidant defense networks [7, 8]. In critical illness, capillary permeability dramatically increases, causing changes in albumin levels between intravascular and extravascular compartments [9]. Previous studies investigating the effects of albumin concentrations on mortality and severity of critical illnesses in adult patient cohorts consistently demonstrated that hypoalbuminemia is associated with increased mortality rates [10–12]. In adult studies, hypoalbuminemia was reported to be a marker of disease severity associated with increased length of stay (LOS) in intensive care, ventilator dependency, and need for respiratory and extracorporeal treatments [12, 13].
Yet, while previous research on albumin levels in adult patient cohorts is rather extensive, there is scarcity of comparable research on critically ill pediatric patients. Among the few relevant studies, a preoperative-postoperative albumin study found that low albumin levels were associated with high rates of surgical infection and increased LOS. Similarly, there is some evidence indicating that, in pediatric renal transplant patients, pretransplant low albumin levels are associated with graft failure [14, 15]. In a recent PICU-based study, Leite et al. [16] found a strong correlation between hypoalbuminemia and increased mortality, prolonged PICU stay, and the need for mechanical ventilation.
For a normal albumin level, the average plasma albumin value for the given age group needs to be known. In preterm infants, the normal range of albumin is 1.8–3.0 g/dl on the first day of life; in full-term infants, the corresponding values are 2.5–3.4 g/dl at 0–6 days and 1.9–4.9 g/dl at 0–12 months. After the first year, the normal range is 3.4–4.2 g/dl at 1–3 years and 3.5–5.6 g/dl at 3 years. At all ages, any albumin level below 3.3 g/dl is considered as hypoalbuminemia [17]. In light of the aforementioned factors, we have established a threshold albumin level that serves to indicate mortality in order to ascertain the prevalence of low albumin levels within our patient cohort, than the mortality and outcome status of the patients were examined based on the determined cut-off albumin level was aimed to investigate.
Methods
A topic that is underrepresented in the existing academic literature, the present multi-center study aims to investigate the relationship between albumin levels measured at the time of admission to the PICU and mortality, LOS, need for mechanical ventilation, extracorporeal treatment status, and outcomes in a large population of critically ill children.
We retrospectively analyzed a large pool of pediatric patients admitted in 2022 to four tertiary hospital PICUs in different cities of south-east region of Turkey. All PICUs were medical and surgical intensive care units managed by a pediatric intensive care specialist.
The following data were retrospectively collected from the patients’ records: the initial serum albumin level at PICU admission, demographic variables, the reason for hospitalization, presence of underlying disease, Pediatric Risk of Mortality Score III (PRISM III), laboratory data (complete blood count, venous or arterial blood gas analysis, sodium, glucose, urea and creatinine), mortality status, need for mechanical ventilation support, need for tracheostomy, need for inotrope and vasoactive drugs, LOS in PICU and hospital, and need for any extracorporeal support (e.g., extracorporeal membrane oxygenation (ECMO), plasma exchange (PLEX), and continuous renal replacement therapy (CRRT)).
Patients with a history of malnutrition, neonates, chronic bowel, liver or kidney disease, inborn errors of metabolism and those who had previously received albumin treatment were excluded from the sample.
In our study, which included children aged from 1 month to 18 years, binary logistic regression analysis was performed to determine the predictive ratio of albumin levels for mortality. ROC analysis was also performed to determine the sensitivity, specificity, and cut-off values of albumin levels in predicting mortality. In the ROC analysis performed for albumin levels, the area under the curve (AUC) value was detected and the sensitivity rate was calculated. Using the cut-off point for albumin levels, the patients were divided into two groups—hypoalbuminemic and normoalbuminemic—and all analyses were performed on these two groups separately.
SPSS for Windows 22.0 was used for all statistical analyses. Continuous variables were presented as means ± standard deviations. Continuous variables between groups were compared using the Mann-Whitney U test. Binary logistic regression analysis was performed with albumin levels. ROC curve analysis was also performed. Spearman correlation analysis was used to analyze correlations between albumin levels and other variables. A p-value < 0.05 was considered to be statistically significant.
Ethics approval and consent to participate
The data from all centers were collected in a single file and entered into the SPSS system for subsequent analysis. A written and signed consent form, authorising participation in the study, was obtained from all authors and their respective institutions. Subsequently, an application was submitted to the Harran University Clinical Research Ethics Committee, which is affiliated with the city where the principal investigator of the study is based. Before the study, the ethics committee approval (No. HRÜ/23.07.05) was obtained from the Harran University Clinical Research Ethics Committee in accordance with the Declaration of Helsinki. As our study involved a retrospective database analysis, parental consent was not required. Clinical trial number: Not applicable. Consent to participate declarations: Not applicable.
Results
A total of 1329 patients admitted to these four PICUs between January 1, 2022 and January 1, 2023 were included in the dataset. Mean age of the patients was 52.85 ± 61.03 months. About half of the patients (54.3%; n = 722) were male. Primary conditions of the patients at admission to the PICUs were distributed as follows: 32.3% respiratory, 15.5% neurological, 8.2% cardiac, 3.2% metabolic, 4.6% endocrine, 2.8% gastrointestinal, 1.9% hemato-oncological, 6.2% sepsis, 2.3% postoperative, 4.8% nephrological, and 18.3% miscellaneous (poisoning, scorpion, snake bite, trauma, etc.). Laboratory findings and clinical information are summarized in Table 1.
Table 1.
Laboratory findings at PICU admission and clinical information at pediatric intensive care treatment period
| Laboratory Findings (X ± SD) | ||
|---|---|---|
| Albumin (g/dl) | 3.92 ± 0.77 | |
| Lactate (mmol/l) | 3.08 ± 3.39 | |
| pH | 7.29 ± 0.17 | |
| Bicarbonate (mEq/l) | 19.17 ± 6.21 | |
| Neutrophil (*109/l) | 8.99 ± 6.51 | |
| Lymphocyte (*109/l) | 4.31 ± 3.80 | |
| Platelet (*109/l) | 353.80 ± 184.04 | |
| Hemoglobin (g/dl) | 11.12 ± 2.43 | |
| CRP (C-reactive protein) (mg/l) | 27.98 ± 54.40 | |
| Sodium (mmol/l) | 137.99 ± 6.76 | |
| Glucose (mg/dl) | 149.86 ± 116.93 | |
| Creatine (mg/dl) | 0.99 ± 0.64 | |
| BUN (Blood urea nitrogen) (mg/dl) | 33.12 ± 47.43 | |
| Clinical Information | ||
| PRISM III Scores (X ± SD)* | 17.48 ± 23.82 | |
| Inotropic treatment n (%) | Yes | 280 (21.1) |
| No | 1049 (78.9) | |
|
MV n (%) |
Yes | 367 (27.6) |
| No | 962 (72.4) | |
|
NIV n (%) |
Yes | 267 (20.1) |
| No | 1062 (79.9) | |
|
HFNC n (%) |
Yes | 148 (11.1) |
| No | 1181 (88.9) | |
|
Tracheostomy n (%) |
Yes | 49 (03.7) |
| No | 1280 (96.3) | |
|
Extracorporeal treatments (CRRT, PLEX, ECMO) |
Yes | Total: 41 (3.1) |
| CRRT: 27 (2.1) | ||
| CRRT + PLEX: 7 (0.5) | ||
| PLEX: 7 (0.5) | ||
| No | Total: 1288 (96.9) | |
*PRISM III: Pediatric Risk of Mortality Score III, MV: Mechanical Ventilation, NIV: Noninvasive Mechanical Ventilation, HFNC: High Flow Nasal Cannula Treatment, CRRT: Continuous Renal Replacement Therapy, PLEX: Plasma exchange ECMO: Extracorporeal Membrane Oxygenation
The patients included in the study, 1114 (83.8%) were discharged, while 215 (16.2%) died. Moreover, the mean PRISM III score was 10.29 ± 12.45 in patients who survived and 45.89 ± 34.52 in those who did not survive, observed. The mean PRISM III score for surviving patients was found to be lower than that for non-surviving patients (p < 0.001).
Laboratory values of the surviving patients as compared to those of the non-surviving patients revealed the following two patterns. First, albumin, pH, bicarbonate, platelet, and hemoglobin levels of the non-surviving patients were significantly lower than those of the surviving patients (p < 0.001 for all comparisons). Second, lactate, C reactive protein (CRP), creatine, and blood urea nitrogen (BUN) levels in the non-survivor group were higher than in the survivor group (p < 0.001 for all comparisons). Laboratory findings and comparison of survivor and non-survivor groups are summarized in Table 2. Furthermore, the laboratory findings of patients according to the cut-off point for albumin level 3.785 g/dl were investigated and summarized in Table 3. Furthermore, binary logistic regression analysis was performed to determine whether albumin levels were predictive of mortality. The results revealed that the predictive rate of albumin levels for mortality amounted to 84.5%. Furthermore, the odds ratio was 0.322 (95% CI: 0.263–0.395) (p < 0.001) for the predictive rate for mortality of 84.5%. Next, we performed ROC analysis to determine the sensitivity, specificity, and cut-off value of albumin levels in predicting mortality. The AUC value was found to be 0.731, the sensitivity rate was 68.3%, the specificity rate was 67%, and the cut-off value was 3.785 (see Fig. 1).
Table 2.
Laboratory findings of patients according to mortality status and comparisons
| Laboratory findings (X ± SD) | Nonsurvivor group n = 215 |
Survivor group n = 1114 |
p-value |
|---|---|---|---|
| Albumin (g/dl) | 3.35 ± 0.90 | 4.03 ± 0.69 | < 0.001 |
| Lactate (mmol/l) | 6.02 ± 5.99 | 2.49 ± 2.14 | < 0.001 |
| pH | 7.20 ± 0.23 | 7.31 ± 0.14 | < 0.001 |
| Bicarbonate (mEq/l) | 17.18 ± 7.60 | 19.56 ± 5.82 | < 0.001 |
| Neutrophil (*109/l) | 8.91 ± 6.90 | 9.01 ± 6.44 | 0.761 |
| Lymphocyte (*109/l) | 5.01 ± 4.77 | 4.17 ± 3.57 | 0.287 |
| Platelet (*109/l) | 286.06 ± 213.96 | 366.89 ± 173.53 | < 0.001 |
| Hemoglobin (g/dl) | 10.60 ± 2.98 | 11.22 ± 2.30 | < 0.001 |
| CRP (mg/l) | 51.92 ± 82.95 | 23.31 ± 45.46 | < 0.001 |
| Sodium (mmol/l) | 138.68 ± 8.79 | 137.85 ± 6.28 | 0.586 |
| Glucose (mg/dl) | 153.41 ± 133.33 | 149.16 ± 113.49 | 0.279 |
| Creatine (mg/dl) | 1.60 ± 7.39 | 0.87 ± 2.29 | < 0.001 |
| BUN (mg/dl) | 49.02 ± 61.57 | 29.77 ± 43.17 | < 0.001 |
*CRP: C-reactive protein; BUN: Blood urea nitrogen
* Values of p < 0.05 indicate statistical significance
Table 3.
Laboratory findings according to the cut-off point for albumin level 3.785 g/dl
| Laboratory findings (X ± SD) | Albumin < 3.785 (g/I) n = 501 |
Albumin > 3.785 (g/I) n = 828 |
p-value |
|---|---|---|---|
| Albumin (g/dl) | 3.15 ± 0.056 | 4.39 ± 0.42 | < 0.001 |
| Lactate (mmol/l) | 3.70 ± 4.35 | 2.73 ± 2.65 | 0.003 |
| pH | 7.29 ± 0.18 | 7.29 ± 0.16 | 0.655 |
| Bicarbonate (mEq/l) | 18.91 ± 6.41 | 19.31 ± 6.12 | 0.135 |
| Neutrophil (*109/l) | 8.53 ± 6.41 | 9.28 ± 6.57 | 0.024 |
| Lymphocyte (*109/l) | 4.15 ± 3.94 | 4.40 ± 3.71 | 0.004 |
| Platelet (*109/l) | 302.43 ± 188.29 | 384.93 ± 172.60 | < 0.001 |
| Hemoglobin (g/dl) | 10.30 ± 2.68 | 11.62 ± 2.12 | < 0.001 |
| CRP (mg/l) | 44.88 ± 70.77 | 17.79 ± 38.13 | < 0.001 |
| Sodium (mmol/l) | 137.80 ± 7.79 | 138.10 ± 6.10 | 0.079 |
| Glucose (mg/dl) | 143.97 ± 114.54 | 153.18 ± 118.20 | 0.038 |
| Creatine (mg/dl) | 1.26 ± 5.16 | 0.82 ± 2.27 | 0.139 |
| BUN (mg/dl) | 40.76 ± 54.69 | 29.19 ± 42.74 | 0.003 |
*CRP: C-reactive protein; BUN: Blood urea nitrogen; PICU: Pediatric Intensive Care Unit
* Values of p < 0.05 indicate statistical significance
Fig. 1.
The study confidently determined the sensitivity and specificity rates, as well as the cut-off value of albumin levels in predicting mortality, through ROC analysis. The albumin levels’ area under the curve (AUC) value was found to be 0.731, with a sensitivity rate of 68.30% and a specificity rate of 67.00%, using a cut-off value of 3.785
The mean length of stay in the PICU for patients with an albumin level below 3.785 g/l was found to be 10.40 ± 19.49 days, while the mean length of hospital stay was 15.66 ± 27.97 days. The group with albumin levels above 3.785 g/l exhibited an average length of stay in the PICU of 8.18 ± 15.49 days and in the hospital of 8.96 ± 10.98 days. The group with albumin levels above 3.785 g/l exhibited reduced PICU and hospital stay averages compared to the group with albumin levels below this threshold (p = 0.025, p < 0.001, respectively).
Next, correlation analysis was used to understand the relationship between albumin levels and other laboratory findings. While a negative correlation was found between albumin levels and lactate, CRP, and BUN levels in all patients, a positive correlation was observed between neutrophil, lymphocyte, platelet, hemoglobin, sodium, and glucose levels. However, if a negative correlation was found between albumin levels and lactate, CRP, and creatine levels in those with a cut-off point of 3.785 for albumin levels, a positive correlation was found between bicarbonate, platelet, hemoglobin, and glucose levels (see Table 4).
Table 4.
Correlation analysis results of changes in albumin level, albumin 3.785 cut-off point, and according to other laboratory parameters
| Laboratories | Albumin | |||
|---|---|---|---|---|
| Albumin levels | Albumin < 3.785 | Albumin > 3.785 | ||
| Lactate | Cc | -0.087 | -0.170 | 0.016 |
| p | 0.002 | < 0.001 | 0.641 | |
| pH | Cc | -0.032 | 0.064 | -0.078 |
| p | 0.249 | 0.169 | 0.025 | |
| Bicarbonate | Cc | 0.025 | 0.142 | -0.072 |
| p | 0.369 | 0.002 | 0.040 | |
| Neutrophil | Cc | 0.105 | 0.031 | 0.127 |
| p | < 0.001 | 0.482 | < 0.001 | |
| Lymphocyte | Cc | 0.069 | 0.057 | -0.009 |
| p | 0.012 | 0.206 | 0.791 | |
| Platelet | Cc | 0.255 | 0.295 | 0.053 |
| p | < 0.001 | < 0.001 | 0.125 | |
| Hemoglobin | Cc | 0.343 | 0.101 | 0.285 |
| p | < 0.001 | 0.023 | < 0.001 | |
| CRP | Cc | -0.303 | -0.227 | -0.112 |
| p | < 0.001 | < 0.001 | 0.001 | |
| Sodium | Cc | 0.106 | 0.012 | 0.166 |
| p | < 0.001 | 0.802 | < 0.001 | |
| Glucose | Cc | 0.091 | 0.214 | 0.039 |
| p | 0.001 | < 0.001 | 0.268 | |
| Creatine | Cc | 0.018 | -0.130 | 0.197 |
| p | 0.509 | 0.003 | < 0.001 | |
| BUN | Cc | -0.066 | -0.088 | 0.049 |
| p | 0.033 | 0.099 | 0.206 | |
*CRP: C-reactive protein; BUN: Blood urea nitrogen
*Cc: Correlation coefficient; p: p value
* Values of p < 0.05 indicate statistical significance
Finally, in our results, the rate of inotrope therapy and mechanical ventilation support was higher in the patients with an albumin level below the cut-off point of 3.785 than in those with an albumin level above the cut-off point of 3.785 (p < 0.001). In addition, the rates of extracorporeal treatment (CRRT, PLEX, extracorporeal membrane oxygenation (ECMO)) were significantly lower in patients with high albumin levels (p < 0.001; see Table 5).
Table 5.
Clinical information and statistical analysis of patients according to the cut-off point for albumin level 3.785 g/dl
| Clinical information n (%) |
Albumin < 3.785 (n = 501) | Albumin > 3.785 (n = 828) | p-value | |
|---|---|---|---|---|
| Inotropic treatment n (%) | Yes | 170 (33.9) | 110 (13.3) | < 0.001 |
| No | 331 (66.1) | 718 (86.7) | ||
| MV n (%) | Yes | 215 (42.9) | 152 (18.4) | < 0.001 |
| No | 962 (57.1) | 676 (81.6) | ||
| NIV n (%) | Yes | 93 (18.6) | 174 (21) | 0.319 |
| No | 408 (81.4) | 654 (79) | ||
| HFNC n (%) | Yes | 51 (10.2) | 97 (11.7) | 0.411 |
| No | 450 (89.8) | 731 (88.3) | ||
| Tracheostomy n (%) | Yes | 18 (3.6) | 32 (3.9) | 0.789 |
| No | 483 (96.4) | 796 (96.1) | ||
| Extracorporeal Treatments n (%) | No | 471 (94) | 817 (98.6) | < 0.001 |
| CRRT | 20 (4) | 7 (0.9) | ||
| PLEX | 3 (0.6) | 4 (0.5) | ||
| CRRT + PLEX | 7 (1.4) | 0 (0) | ||
| ECMO | 0 (0) | 0 (0) | ||
* MV: Mechanical Ventilation, NIV: Noninvasive Mechanical Ventilation, HFNC: High Flow Nasal Cannula Treatment, CRRT: Continuous Renal Replacement Therapy, PLEX: Plasma exchange, ECMO: Extracorporeal Membrane Oxygenation
* Values of p < 0.05 indicate statistical significance
Discussion
In this retrospective multi-center study, we investigated the prognostic role of serum albumin levels on admission for the outcomes of critically ill children in PICUs. Several other factors that could affect the patients’ outcome were also carefully evaluated. The results revealed that serum albumin concentration was associated with mortality, LOS in the PICU, discharge status, initiation of mechanical ventilation, and the need for inotropic support. Our study revealed that the albumin cut-off of 3.785 g/dl at admission to the PICUs was a sensitive and specific predictor of mortality and prognosis.
In previous research, the rate of hypoalbuminemia in critically ill pediatric patients was reported to amount to 33-57% [18]. Accordingly, there has been extensive research on albumin levels and their association with mortality, morbidity, poor outcomes, as well as LOS in the PICU and/or hospital [19–21]. For instance, in a retrospective PICU study involving 431 critically ill children aged over 3 years old, serum albumin levels were found to be significantly higher in the surviving group (3.4 g/dl [IQR: 3 to 3.8 g/dl]) than in the non-surviving group (2.9 g/dl [IQR: 2.3 to 3.4 g/dl]). Furthermore, the likelihood of survival was reported to be higher in patients with an albumin level > 3.5 g/l, and the correlation between the likelihood of survival and the albumin level was statistically significant [12]. In the present study, for the cut-off albumin level of 3.785 g/dl, the predictive rate of serum albumin levels for mortality amounted to 84.5%, while the odds ratio was 0.322 (95% CI: 0.263–0.395) (p < 0.001).
In another pediatric ICU study on the effect of hypoalbuminemia on mortality at admission and during follow-up, patients with hypoalbuminemia at admission were reported to have a high PRISM III score, prolonged PICU stay, increased need for respiratory support and mechanical ventilation, and a high rate of multiple organ failure [21]. These findings are consistent with our results. However, mortality rate did not reach statistical significance in the aforementioned study that used an albumin cut-off of 2.5 g/dl for hypoalbuminemia [21].
Similarly, a previous report on a cohort of Thai patients with hypoalbuminemia on admission found a strong correlation between low albumin levels and higher mortality, longer PICU stay, and the need for a ventilator for a prolonged period of time [22]. In that study, the cut-off value for albumin was 2.5 g/dl for those aged 0–7 months old and 3.4 g/dl for those aged > 7 months old, according to the pediatric clinical chemistry data [23]. In line with these findings, our retrospective multi-center study yielded robust evidence on the significant association between hypoalbuminemia and longer LOS in both PICU (p = 0.025) and hospital (p < 0.001), higher mortality rates (p < 0.001), and the increased use of inotropic and extracorporeal treatments (p < 0.001) in patients with multiple organ failures.
At present, many scoring systems are available to predict the prognosis of patients treated in the PICU at the time of admission and during follow-up, as well as to rapidly identify critically ill patients. In order to make these validated scoring systems more easily applicable, previous research examined albumin concentration in conjunction with several other parameters. For instance, a recent prospective descriptive study on 110 patients examined the relationships between albumin levels at admission and Pediatric Index of Mortality 2 (PIM 2), Vasoactive Inotropic Score (VIS), and Pediatric Sequential Organ Failure Assessment (pSOFA) scores in patients aged from 3 months to 16 years [17]. In the aforementioned study, hypoalbuminemia was defined as < 3.3 g/dl, and a significant relationship between patients’ low albumin levels and PIM 2, VIS, and pSOFA scores was observed. Likewise, mortality was also reported to be 4.1 times higher in hypoalbuminemic patients [17]. In our study, we examined the PRISM III scoring system, which was electronically recorded at PICU admission in all participating centers. The mean PRISM III score at admission amounted to 17.48 ± 23.82; we also observed that both low albumin levels and high PRISM III scores were statistically significant predictors of patient mortality.
Furthermore, in a study on 271 critically ill children, Leite et al. [16] found that the mean serum albumin level was 3.23 ± 0.64 g/dl in the surviving group and 2.66 ± 0.61 g/dl in the non-surviving group. In this study was investigated association between serum albumin, CRP and lactate levels with clinical severity scores at admission to pediatric ICU. In Leite et al.’s [16] research, the serum albumin level of < 3.5 g/dl was defined as hypoalbuminemia and that of ≤ 20 mg/dl was defined as normal lactate; furthermore, CRP ≤ 10 mg/dl was used as a reference normal value. Based on these criteria, the authors found that an increase of 1.0 g/dl in serum albumin on admission was associated with a 73% reduction in the risk of mortality, a 33% increase in PICU discharge, and 1.86 ventilator-free days [16]. In the present study, we found a statistically significant difference in mean serum albumin levels between surviving patients (4.03 ± 0.69 g/dl) and mortally ill patients (3.35 ± 0.90 g/dl) (p < 0.001). Moreover, along with a negative correlation between albumin levels and lactate, CRP, and BUN levels in all patients, our results also revealed a positive correlation between neutrophil, lymphocyte, platelet, hemoglobin, sodium, and glucose levels.
While most previous studies focused on the adverse effect of hypoalbuminemia, some of the available evidence indicates that high serum albumin levels are not very beneficial either. For instance, in a 5-year PICU study conducted in China, both high and low albumin levels were found to be associated with increased mortality [24]. Interestingly, however, no correlation with mortality was observed in patients with serum albumin levels ranging from 35.9 g/l (OR = 1.101; 95% CI: 0.998–1.215) to 43.2 g/l (OR = 1.139; 95% CI: 0.993–41.306). Based on these findings, the authors concluded that there is a U-shaped relationship between serum albumin levels and mortality risk [24]. The results of our study—which covers the largest population of critically ill pediatric patients reported to date— indicate that the average serum albumin level in our sample was 3.92 ± 0.77 g/dl. Surviving patients had a significantly higher average serum albumin level of 4.03 ± 0.69, while the average serum albumin level among non-survivors was significantly lower (3.35 ± 0.90 g/dl, p < 0.001).
Furthermore, in our multi-center study of a heterogeneous pediatric population, albumin, pH, bicarbonate, platelet and hemoglobin levels were lower among non-survivors than among survivors. By contrast, the lactate, CRP, creatine, and BUN levels of the patients who died were higher than those of survivors. These findings are broadly consistent with previous results showing that acute-phase proteins, which appear as a result of the development of critical diseases, either negatively or positively correlate with disease progression in human physiology [25]. Our findings also align with the results reported in another multi-center study conducted in Turkey which found an association of PRISM-4, PIM 3, CRP, CRP/albumin ratio, albumin and lactate levels with mortality risk in critically ill children. Albumin and lactate levels in that study were shown to have a stronger predictive value with regard to mortality [26].
Major limitations of the present study are retrospective design and the lack of long-term follow-up. However, by covering the largest number of pediatric patients included in the albumin-mortality research conducted to date, our study makes a significant contribution to the literature. An important characteristic of our multi-center study is that we also included tertiary care hospitals covering all surgical and medical pediatric patients, with each center having a pediatric intensive care specialist to reduce mortality.
Conclusion
As a biomarker that can be quickly, cheaply, and easily tested in any center globally, serum albumin is a valuable indicator for medical practice. In our multi-center study, we analyzed the largest sample of critically ill pediatric patients reported to date. Our results revealed that serum albumin level can serve as an effective indicator of mortality and outcomes in critically ill children. Furthermore, it predicted the situation of necessity treatment and mechanical supports. We recommended to utilise, as it does not necessitate the use of scoring systems or tests that require lengthy and intricate analyses, and it is readily available.
Acknowledgements
Thank you to all the healthcare professionals from the study hospitals who participated in treating and following up with patients during the study period.
Author contributions
Conceptualization and design: HFA, NA, DY; Methodology: NA, HFA; Data collection: HFA, EET, SY, KG, AA; Formal analysis and investigation: AK; Writing - original draft preparation: HFA, NA; Writing - review and editing: NA, DY; Funding acquisition: None; Resources: None; Supervision: HFA, EET, SY, KG, AA, AT, NA, DY.
Funding
Not applicable.
Data availability
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
Declarations
Ethics approval and consent to participate
Harran University Clinical Research Ethics Committee (HRÜ/23.07.05) approval was obtained by before the study began for the study. As our study involved a retrospective database analysis, parental consent was not required.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Bekhit OE, Yousef RM, Abdelrasol HA, Mohammed MA. Serum albumin level as a predictor of outcome in patients admitted to pediatric intensive care units. Pediatr Emerg Care. 2021;37(12):e855–860. 10.1097/PEC.0000000000002567. [DOI] [PubMed] [Google Scholar]
- 2.Straney L, Clements A, Parslow RC, et al. ANZICS paediatric study group and the paediatric intensive care audit network. Paediatric index of mortality 3: an updated model for predicting mortality in pediatric intensive care. Pediatr Crit Care Med. 2013;14:673–81. 10.1097/PCC.0b013e31829760cf. [DOI] [PubMed] [Google Scholar]
- 3.North I. Principles of critical care: fluid therapy and blood product. 2nd ed. Philadelphia: McGraw- Hill; 1999. pp. 39–40. [Google Scholar]
- 4.AL-Khafaji A, Webb AR. Should albumin be used to correct hypoalbuminemia in the critically ill? No. TATM. 2003;5(4):392–6. 10.1111/j.1778-428X.2003.tb00179.x. [Google Scholar]
- 5.Dubois MJ, Vincent JL. Use of albumin in the intensive care unit. TATM. 2002;4(3):80–4. 10.1097/00075198-200208000-00005. [DOI] [PubMed] [Google Scholar]
- 6.Nicholson JP, Wolmarans MR, Park GR. The role of albumin in critical illness. Br J Anasth. 2000;85(4):599–610. 10.1093/bja/85.4.599. [DOI] [PubMed] [Google Scholar]
- 7.Sitar ME, Aydin S, Cakatay U. Human serum albumin and its relation with oxidative stress. Clin Lab. 2013;59:945–52. [PubMed] [Google Scholar]
- 8.Abiles J, de la Cruz AP, Castano J, et al. Oxidative stress is increased in critically ill patients according to antioxidant vitamins intake, independent of severity: a cohort study. Crit Care. 2006;10:R146. 10.1186/cc5068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Fleck A, Raines G, Hawker F, et al. Increased vascular permeability: a major cause of hypoalbuminaemia in disease and injury. Lancet. 1985;1:781–4. 10.1016/s0140-6736(85)91447-3. [DOI] [PubMed] [Google Scholar]
- 10.Barchel D, Almoznino-Sarafian D, Shteinshnaider M, Tzur I, Cohen N, Gorelik O. Clinical characteristics and prognostic significance of serum albumin changes in an internal medicine ward. Eur J Intern Med. 2013;24:772–8. 10.1016/j.ejim.2013.08.004. [DOI] [PubMed] [Google Scholar]
- 11.Nagumo K, Tanaka M, Chuang VT, et al. Cys34-cysteinylated human serum albumin is a sensitive plasma marker in oxidative stress-related chronic diseases. PLoS ONE. 2014;9:e85216. 10.1371/journal.pone.0085216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Kim YS, Sol IS, Kim MJ, et al. Serum albumin as a biomarker of poor prognosis in the pediatric patients in intensive care unit. Korean J Crit Care Med. 2017;32(4):347–55. 10.4266/kjccm.2017.00437. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Rady MY, Ryan T. Perioperative predictors of extubation failure and the effect on clinical outcome after cardiac surgery. Crit Care Med. 1999;27:340–7. 10.1097/00003246-199902000-00041. [DOI] [PubMed] [Google Scholar]
- 14.Leite HP, Fisberg M, de Carvalho WB, et al. Serum albumin and clinical outcome in pediatric cardiac surgery. Nutrition. 2005;21:553–8. 10.1016/j.nut.2004.08.026. [DOI] [PubMed] [Google Scholar]
- 15.Tancredi DJ, Butani L. Pretransplant serum albumin is an independent predictor of graft failure in pediatric renal transplant recipients. J Pediatr. 2014;164:602–6. 10.1016/j.jpeds.2013.11.012. [DOI] [PubMed] [Google Scholar]
- 16.Leite HP, Rodrigues da Silva AV, de Oliveira Iglesias SB, Koch Nogueira PC. Serum albumin is an independent predictor of clinical outcomes in critically ill children. Pediatr Crit Care Med. 2016;17(2):e50–7. 10.1097/PCC.0000000000000596. [DOI] [PubMed] [Google Scholar]
- 17.Gowa MA, Tauseef U, Ahmed SH. A relation between serum albumin level and prognosis of critically ill children admitted to the pediatric ICU. J Pak Med Assoc. 2023;73:1034–42. 10.47391/JPMA.7465. [DOI] [PubMed] [Google Scholar]
- 18.Vincent JL, Dubois MJ, Navickis RJ, Wilkes MM. Hypoalbuminemia in acute illness: is there a rationale for intervention? A meta-analysis of cohort studies and controlled trials. Ann Surg. 2003;237:319–34. 10.1097/01.SLA.0000055547.93484.87. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Durward A, Mayer A, Skellett S, et al. Hypoalbuminaemia in critically ill children: incidence, prognosis, and influence on the anion gap. Arch Dis Child. 2003;88:419–22. 10.1136/adc.88.5.419. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Horowitz IN, Tai K. Hypoalbuminemia in critically ill children. Arch Pediatr Adolesc Med. 2007;161:1048–52. 10.1001/archpedi.161.11.1048. [DOI] [PubMed] [Google Scholar]
- 21.Tiwari LK, Singhi S, Jayashree M, Baranwal AK, Bansal A. Hypoalbuminemia in critically sick children. Indian J Crit Care Med. 2014;18:56. 10.4103/0972-5229.140143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kittisakmontri K, Reungrongrat S, Lao-araya M. Hypoalbuminaemia at admission predicts the poor outcomes in critically ill children. Anaesthesiol Intensive Ther. 2016;48(3):158–61. 10.5603/AIT.a2016.0028. [DOI] [PubMed] [Google Scholar]
- 23.Meites S, Buffone GJ. Pediatric clinical chemistry. 3rd ed. Washington DC: AACC; 1989. [Google Scholar]
- 24.Zhang X, Zhang L, Wei C, et al. U-shaped association between serum albumin and pediatric intensive care unit mortality in critically ill children. Front Nutr. 2022;9:931599. 10.3389/fnut.2022.931599. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Mantovani A, Garlanda C. Humoral innate immunity and acute-phase proteins. N Engl J Med. 2023;388(5):439–52. 10.1056/NEJMra2206346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Arslan G, Besci T, Özdemir G, et al. Predictive value of PRISM-4, PIM-3, CRP, albumin, CRP/albumin ratio and lactate in critically ill children. Child (Basel). 2023;10(11):1731. 10.3390/children10111731. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.