E-submission
KSPEN
KSSMN
ASSMN
JSSMNIndexed in:
Scopus, KCI, KoreaMed
Scopus, KCI, KoreaMed
Articles
- Page Path
- HOME > Ann Clin Nutr Metab > Volume 18(2); 2026 > Article
- Original Article Association of early parenteral energy provision with serum phosphorus decline and phosphorus-based refeeding syndrome in critically ill patients with liver cirrhosis: a Korean retrospective observational study
-
Jee Young Lee1
, Yeo Kon Kye2, Kwang Il Seo3, Kyung Won Seo4, Sooyong Lee5, Jesang Yu5 -
Annals of Clinical Nutrition and Metabolism 2026;18(2):162-172.
DOI: https://doi.org/10.15747/ACNM.26.0055
Published online: July 31, 2026
1Nutrition Support Team, Department of Nursing, Kosin University Gospel Hospital, Busan, Korea
2Department of Emergency Medicine, Kosin University Gospel Hospital, Busan, Korea
3Department of Internal Medicine, Kosin University College of Medicine, Busan, Korea
4Department of Surgery, Kosin University College of Medicine, Busan, Korea
5Department of Radiation Oncology, Kosin University Gospel Hospital, Kosin University College of Medicine, Busan, Korea
- Corresponding author: Jesang Yu, email: rojsyu@gmail.com
• Received: May 7, 2026 • Revised: June 16, 2026 • Accepted: June 30, 2026
© 2026 The Korean Society of Surgical Metabolism and Nutrition · The Korean Society for Parenteral and Enteral Nutrition · Asian Society of Surgical Metabolism and Nutrition
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
- 1,260 Views
- 18 Download
Abstract
-
Purpose Critically ill patients with liver cirrhosis are vulnerable to malnutrition and refeeding-related electrolyte disturbances, particularly with rapid caloric advancement. As hypophosphatemia is a key feature of refeeding syndrome (RFS) and phosphorus homeostasis may be impaired in cirrhosis, the association of early parenteral energy provision with serum phosphorus decline and RFS was evaluated in cirrhotic ICU patients.
-
Methods This retrospective study included 72 adults with liver cirrhosis admitted to a tertiary intensive care unit (ICU) between January 2021 and August 2024. Parenteral energy intake was assessed at emergency department (ED) and on ICU days 1 and 2. Serum phosphorus reduction was defined as the percentage decrease from baseline to the nadir within 5 ICU days. Phosphorus-based RFS was defined per the 2020 ASPEN consensus as a ≥10% phosphorus decrease after reinitiating or increasing energy provision, graded as mild, moderate, or severe.
-
Results Phosphorus-based RFS occurred in 53 patients (73.6%), including severe RFS in 35 (48.6%). ICU day 2 caloric intake per body weight correlated with phosphorus reduction (r=0.346, P=0.003) and was independently associated with greater decline (P=0.011) and with meeting RFS criteria (odds ratio, 1.19; 95% CI, 1.06–1.36; P=0.007), along with ED glucose load (P=0.001). Receiver operating characteristic analysis showed modest discrimination (area under the curve, 0.7061; cutoff, 10.92 kcal/kg/day).
-
Conclusion Higher caloric delivery on ICU day 2 was associated with greater phosphorus decline and RFS in cirrhotic ICU patients. These exploratory associations do not establish causality, and the cutoff requires external validation. Monitoring caloric delivery and serial electrolytes may support safer parenteral nutrition.
Graphical abstract
Introduction
Refeeding syndrome (RFS) is a serious and potentially life-threatening complication that occurs when nutrition is reintroduced after a period of malnutrition, particularly when feeding is initiated rapidly or inappropriately [1,2]. The hallmark of RFS is an abrupt decline in serum phosphorus, accompanied by shifts in potassium, magnesium, and glucose levels, which can lead to cardiac, respiratory, and neurological complications [3-5].
Malnutrition and electrolyte disturbances are common in critically ill patients with liver cirrhosis. Cirrhosis is associated with impaired metabolic homeostasis and nutrient absorption, which may contribute to nutritional imbalance during critical illness [1,2]. In patients with advanced cirrhosis, depleted hepatic glycogen stores and persistent hypermetabolism may create an accelerated starvation state, providing biological plausibility for abrupt metabolic shifts when nutrition is reintroduced [6]. In addition, cirrhosis-related sarcopenia, characterized by impaired muscle protein synthesis and altered energy metabolism, may reduce peripheral nutrient utilization capacity, whereas chronic inflammation and endotoxemia may accelerate muscle catabolism and contribute to progressive micronutrient depletion [2,4]. These cirrhosis-related metabolic derangements are presented as background rationale and biological plausibility, not as findings of the present study. Because this study did not include a noncirrhotic critically ill comparator group, it was not designed to determine whether patients with cirrhosis are more susceptible to RFS than other intensive care unit (ICU) populations.
In the ICU, total parenteral nutrition (TPN) is commonly used in patients with liver cirrhosis when enteral feeding is not feasible [7,8]. However, aggressive caloric administration through TPN, particularly early after ICU admission, has been associated with refeeding-related electrolyte disturbances in previous studies [9,10]. Although RFS is recognized as a clinical entity, empirical evidence specifically evaluating the temporal relationship between caloric intake and phosphorus-based RFS in patients with cirrhosis admitted to the ICU remains limited [4,7]. Previous investigations of refeeding complications in ICU settings have primarily examined heterogeneous critically ill cohorts, in which cirrhosis subgroups were underrepresented or evaluated only secondarily [11]. Furthermore, the optimal rate and timing of energy provision to reduce refeeding-related electrolyte disturbances in patients with cirrhosis remain unclear [4,7,8]. Recent American Society for Parenteral and Enteral Nutrition (ASPEN) and European Society for Clinical Nutrition and Metabolism (ESPEN) guidelines recommend cautious caloric advancement in patients with malnutrition or refeeding risk [4,7,8]; however, practical applications and supporting data for patients with liver cirrhosis remain sparse.
The early refeeding period is clinically important because phosphorus nadirs associated with refeeding-related metabolic shifts typically occur within the first 24–72 hours of refeeding. Evaluating early caloric load and its temporal association with phosphorus decline may therefore provide preliminary information to support safer nutritional management in patients with cirrhosis admitted to the ICU and receiving TPN.
This study aimed to evaluate the association of early parenteral caloric intake with serum phosphorus changes, phosphorus-based RFS occurrence, and RFS severity in patients with liver cirrhosis admitted to the ICU.
Methods
This retrospective observational study included 72 adult patients with liver cirrhosis who were admitted to the medical ICU of a tertiary referral hospital between January 2021 and August 2024. All patients were admitted to the ICU via the emergency department (ED) because of acute decompensation; the main reasons for ICU admission were upper gastrointestinal bleeding, such as hematemesis; altered mental status, including hepatic encephalopathy or coma; and clinically significant ascites. Eligible patients were adults with liver cirrhosis who had early parenteral energy exposure through intravenous glucose-containing fluids and/or parenteral nutrition (PN)/TPN during the early phase of ICU admission and had available serial serum phosphorus measurements and nutritional intake records. Patients were excluded if they had baseline electrolyte imbalance involving phosphorus, potassium, or magnesium; missing electrolyte data; death within 48 hours after ICU admission; or transfer to hospice.
Patients were screened retrospectively from the electronic medical records of adults with liver cirrhosis who were admitted to the medical ICU via the ED during the study period. The final analytic cohort consisted of patients who met the eligibility criteria and had available serial serum phosphorus measurements and early nutritional intake records. In total, 114 adults with decompensated liver cirrhosis who were admitted to the ICU via the ED during the study period were screened for eligibility. Of these, 42 were excluded because of initial electrolyte imbalance involving phosphorus, potassium, or magnesium (n=18), missing electrolyte data for phosphorus, potassium, or magnesium (n=21), death within 48 hours (n=1), or transfer to hospice (n=2), leaving 72 patients in the final analysis. The patient selection process is summarized in the study flow diagram (Fig. 1).
Electronic medical records were retrospectively reviewed to obtain demographic and clinical data, including age, sex, height, weight, body mass index (BMI), serum laboratory parameters, and nutrition-related variables.
For exposure classification, intravenous glucose was defined as any glucose-containing intravenous fluid, including 5% dextrose solution, 10% dextrose solution, Hartmann’s solution with dextrose, and the glucose component of TPN. PN was defined as parenteral nutrient delivery provided through intravenous formulations containing macronutrients. TPN was defined as a formal PN formulation prescribed as the main nutritional support and containing glucose, amino acids, and/or lipid. In this study, early parenteral energy exposure included intravenous glucose-containing fluids and/or PN/TPN.
Energy intake was calculated as kcal/day and kcal/kg and was assessed at three time points: the day of ED presentation, ICU day 1, and ICU day 2. Nutrition-related variables included the timing of glucose administration, timing of formal TPN initiation, total caloric intake on ICU day 1, percentage of the estimated energy requirement delivered on ICU day 1, total caloric intake on ICU day 2, percentage of the estimated energy requirement delivered on ICU day 2, and thiamine administration.
Potential nonrefeeding determinants of serum phosphorus levels, including renal replacement therapy, diuretic use, and phosphate supplementation during the first 5 ICU days, were also reviewed from the electronic medical records. However, because the timing, cumulative dose, indications, and temporal relationship of these interventions to serum phosphorus measurements were not consistently structured in the retrospective records, they could not be incorporated as fully adjusted covariates in the primary multivariable models.
Baseline nutritional assessment variables included BMI, the Nutrition Risk Screening 2002 (NRS-2002)-based initial nutritional risk assessment, and reduced dietary intake within 2 weeks before admission when available. However, comprehensive nutritional assessment tools, such as the Subjective Global Assessment (SGA) and Global Leadership Initiative on Malnutrition (GLIM) criteria, recent weight loss, and formal malnutrition diagnoses were not systematically documented in the electronic medical records and therefore could not be consistently incorporated into the primary analyses.
The primary outcome was serum phosphorus reduction, defined as the percentage decrease from the baseline phosphorus level at ED presentation to the lowest value recorded within the first 5 days of ICU admission. The secondary outcome was whether patients met the phosphorus-based RFS criterion. According to the 2020 ASPEN consensus recommendations, RFS is defined as a decrease in serum phosphorus, potassium, and/or magnesium of 10%–20%, 20%–30%, or >30% within 5 days after reinitiating or substantially increasing energy provision, corresponding to mild, moderate, or severe RFS, respectively. Severe RFS may also be supported by organ dysfunction attributable to electrolyte decreases or by thiamine deficiency. In the present retrospective study, phosphorus-based RFS was operationalized using only the serum phosphorus component of the ASPEN criteria because the study aim focused on phosphate decline and phosphorus was the only electrolyte measured serially and consistently across all patients during the predefined 5-day observation window. Therefore, phosphorus-based RFS occurrence was defined as a ≥10% decrease in serum phosphorus from the ED baseline to the nadir within 5 ICU days, and severity was classified as mild (10%–20% decrease), moderate (20%–30% decrease), or severe (>30% decrease). Potassium and magnesium values were reviewed as part of routine clinical care when available, but they were not used as independent diagnostic triggers in the primary RFS classification because serial availability was incomplete. Clinical manifestations and organ dysfunction were not used as mandatory diagnostic criteria because attribution to RFS could not be reliably determined from retrospective records in this critically ill population. Thus, the phosphorus-based RFS outcome in this study should be interpreted as an operational definition derived from the serum phosphorus component of the ASPEN framework, rather than as a full ASPEN classification using all three electrolytes and clinical criteria.
Continuous variables were presented as means±standard deviations or medians with interquartile ranges (IQRs), as appropriate, and categorical variables were presented as counts and percentages. Linear regression analysis was performed to evaluate the association between candidate predictors and serum phosphorus reduction (%). Logistic regression analysis was used to assess factors associated with meeting the phosphorus-based RFS criterion.
To reduce the risk of model overfitting, variables included in the multivariable models were selected a priori based on clinical relevance and the study hypothesis rather than through data-driven stepwise selection. The selected variables focused on early glucose exposure, timing of formal TPN initiation, and ICU day 1 and ICU day 2 caloric delivery, which were considered clinically relevant to early refeeding-related electrolyte changes. Because only 19 patients did not meet the phosphorus-based RFS criterion, the logistic regression model was considered susceptible to overfitting. Therefore, the multivariable logistic regression results were interpreted as exploratory and hypothesis-generating rather than confirmatory.
Renal replacement therapy, diuretic exposure, and phosphate supplementation were treated as clinically important determinants of serum phosphorus levels and were reviewed descriptively. However, these variables were not included as fully adjusted covariates in the primary regression models because their timing, cumulative dose, indications, and temporal relationship with the phosphorus nadir were inconsistently documented in the retrospective records. Accordingly, the observed associations should not be interpreted as causal effects of caloric delivery on serum phosphorus decline or phosphorus-based RFS status.
Thiamine administration was collected descriptively but was not included in the primary adjusted models because it was not protocol-driven, may have been influenced by clinician-perceived RFS risk, and could have introduced confounding by indication. Statistical significance was set at a two-sided P <0.05. Statistical analyses were performed using R software (version 4.5.3; R Foundation for Statistical Computing).
The study protocol was reviewed and approved by the Institutional Review Board of Kosin University Gospel Hospital (IRB No. KUGH 2026-02-010). The requirement for informed consent was waived because of the retrospective design and use of deidentified data. All procedures were conducted in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments. This manuscript follows the recommendations of the International Committee of Medical Journal Editors, the Committee on Publication Ethics, and the Guidelines on Good Publication Practice issued by the Korean Association of Medical Journal Editors.
Results
A total of 114 adults with decompensated liver cirrhosis who were admitted to the ICU via the ED were screened. After 42 patients were excluded, 72 patients were included in the final analysis (Fig. 1). The median age was 61.0 years (IQR, 54.0–71.0 years), and 49 patients (68.1%) were men. The median BMI was 23.12 kg/m² (IQR, 20.76–25.69 kg/m²), with nine patients (12.5%) having a BMI <18.5 kg/m² and two patients (2.8%) having a BMI <16 kg/m². The median Model for End-Stage Liver Disease (MELD) score was 21.0 (IQR, 14.0–24.25), and the median time from ED arrival to ICU admission was 6.98 hours (IQR, 4.41–15.68) (Table 1).
Regarding liver disease etiology, alcohol-related cirrhosis was the most common cause and was observed in 40 patients (55.6%). Hepatocellular carcinoma was present in 28 patients (38.9%), and 58 patients (80.6%) had comorbidities. The median percentage reduction in serum phosphorus was 29.68% (IQR, 8.45%–48.35%), and 53 patients (73.6%) met the phosphorus-based RFS criterion, including eight (11.1%) with mild, 10 (13.9%) with moderate, and 35 (48.6%) with severe cases based on serum phosphorus decline.
Regarding nutritional support variables, glucose was administered in the ED in 61 patients (84.7%), with a median ED glucose load of 0.40 g/kg (IQR, 0.00–0.61 g/kg) and 85.00 kcal (IQR, 0.00–106.25 kcal). TPN was initiated within 24 hours of ICU admission in 33 patients (45.8%) and after 24 hours in 36 patients (50.0%), whereas three patients (4.2%) received no formal TPN during the early ICU period. Thiamine supplementation was administered to 35 patients (48.6%). During the study period, thiamine administration was not governed by a standardized institutional protocol and was provided prophylactically or therapeutically at the discretion of the treating clinicians or nutrition support team according to perceived nutritional risk, clinical instability, alcohol-related liver disease, poor oral intake, or concern for refeeding-related electrolyte disturbance. The median total caloric intake was 835.00 kcal/day (IQR, 538.75–1,268.00 kcal/day) on ICU day 1 and 832.50 kcal/day (IQR, 734.25–1,197.75 kcal/day) on ICU day 2, whereas the median proportion of the estimated energy requirement delivered on ICU day 2 was 70.19% (IQR, 49.85%–89.94%). The median length of hospital stay was 12.0 days (IQR, 8.0–18.0 days), and in-hospital mortality occurred in 21 patients (29.2%).
A significant positive correlation was observed between ICU day 2 caloric intake per body weight and the percentage reduction in serum phosphorus (Fig. 2). In the scatter plot analysis, the correlation coefficient was r=0.346, with an explanatory power of R² = 0.120, and the association was statistically significant (P=0.003, n=72).
ICU day 2 caloric intake per body weight differed significantly according to phosphorus-based RFS status (Table 2, Fig. 3). Patients who met the phosphorus-based RFS criterion received higher ICU day 2 energy delivery than those who did not meet the criterion (15.37±6.62 kcal/kg/day vs. 10.50±5.89 kcal/kg/day; P=0.005). This finding suggests that greater caloric delivery by ICU day 2 was associated with a higher likelihood of meeting the phosphorus-based RFS criterion.
Subsequent multivariable linear regression analysis identified ED glucose load (g/kg) and ICU day 2 kcal/kg/day as independent factors associated with the percentage reduction in serum phosphorus (Table 3). A higher ED glucose load was significantly associated with a greater reduction in serum phosphorus (β, 19.71; 95% confidence interval [CI], 7.94–31.47; P=0.001), and higher ICU day 2 kcal/kg/day was also significantly associated with greater serum phosphorus reduction (β, 1.32; 95% CI, 0.31–2.34; P=0.011). In contrast, the timing of ED glucose administration, TPN initiation, and ICU day 1 kcal/kg/day were not significantly associated with serum phosphorus reduction.
Because serum phosphorus concentrations in critically ill patients can be influenced by concurrent ICU interventions, the observed serum phosphorus decline should be interpreted in the context of potential nonrefeeding determinants. Renal replacement therapy, diuretic exposure, and phosphate supplementation were reviewed when available in the medical records; however, the timing and dose of these interventions relative to the phosphorus nadir were not consistently documented. Therefore, the observed association between early caloric delivery and serum phosphorus reduction should be interpreted as exploratory rather than as evidence of a direct causal effect.
Comparison of ICU day 2 kcal/kg between patients who did and did not meet the phosphorus-based RFS criterion showed that caloric intake on ICU day 2 was significantly higher in the phosphorus-based RFS group than in the non-phosphorus-based RFS group (15.37±6.62 kcal/kg/day vs. 10.50±5.89 kcal/kg/day, respectively) (Table 2, Fig. 3). This difference was statistically significant in Welch’s t-test (t=2.994, df=35.47, P=0.005), with a 95% CI for the mean difference of 1.57–8.18.
In multivariable logistic regression analysis (Table 4), ICU day 2 kcal/kg was the only independent factor significantly associated with meeting the phosphorus-based RFS criterion. Specifically, each 1 kcal/kg/day increase in ICU day 2 kcal/kg was associated with 19% higher odds of meeting the phosphorus-based RFS criterion (odds ratio [OR], 1.19; 95% CI, 1.06–1.36; P=0.007). In contrast, timing of glucose administration within 24 hours (OR, 4.62; 95% CI, 0.65–45.47; P=0.147), ED glucose load (g/kg) (OR, 3.52; 95% CI, 0.74–27.53; P=0.170), formal TPN initiation after 24 hours compared with within 24 hours (OR, 0.32; 95% CI, 0.08–1.24; P=0.108), no formal TPN during the early ICU period compared with formal TPN initiation within 24 hours (OR, 0.34; 95% CI, 0.01–5.26; P=0.451), and ICU day 1 kcal/kg (OR, 0.92; 95% CI, 0.84–1.00; P=0.076) were not significantly associated with phosphorus-based RFS status.
To further evaluate the discriminative performance of ICU day 2 kcal/kg for the phosphorus-based RFS criterion, receiver operating characteristic (ROC) curve analysis was performed (Fig. 4). The area under the curve (AUC) was 0.7061, indicating fair but modest discriminative ability. The optimal cutoff value determined by Youden’s index was 10.92 kcal/kg/day, with a sensitivity of 77.4% and a specificity of 57.9%. Because this cutoff was derived from a single-center cohort and has not been externally validated, it should be considered an exploratory threshold rather than a clinically actionable target.
Discussion
In this retrospective cohort of critically ill patients with liver cirrhosis, greater caloric delivery on ICU day 2 was associated with a larger decrease in serum phosphorus and a higher likelihood of meeting the phosphorus-based RFS criterion. This finding is broadly consistent with the ASPEN 2020 and ESPEN 2019 and 2023 guidelines, which advocate gradual energy escalation in patients with malnutrition or refeeding risk. However, these guidelines do not provide numerical thresholds specific to patients with cirrhosis, and the underlying evidence derives largely from heterogeneous ICU or oncology cohorts with limited representation of patients with cirrhosis [4,7,12]. The 10.92 kcal/kg/day cutoff identified in this study may therefore provide a preliminary signal within this cohort, but it should be interpreted as a hypothesis-generating reference rather than as a definitive feeding threshold.
Several methodological considerations should be emphasized when interpreting these findings. First, the outcome was based on a phosphorus-based operational definition derived from the ASPEN framework, not on the full ASPEN three-electrolyte classification incorporating phosphorus, potassium, magnesium, and clinical manifestations. Therefore, the observed proportion of patients meeting the phosphorus-based RFS criterion (73.6%) likely reflects biochemical phosphorus decline rather than clinically adjudicated RFS in every case. Second, serum phosphorus concentrations in critically ill patients may be affected by factors other than refeeding physiology, including sepsis, renal replacement therapy, diuretic exposure, respiratory alkalosis, dilutional effects from fluid resuscitation, and phosphate supplementation. Third, because this study was retrospective and did not include a noncirrhotic critically ill comparator group, the findings should be interpreted as associations observed within a cirrhosis cohort, not as evidence of causality or cirrhosis-specific susceptibility to RFS.
Previous observational studies have reported associations between lower initial caloric delivery and fewer refeeding-related electrolyte disturbances, which is consistent with the direction of our findings. This concern is further supported by studies using alternative methodologies, including percentage of estimated requirements and indirect calorimetry, which similarly showed that rapid energy delivery during early critical illness was associated with metabolic complications [13-15]. In patients with cirrhosis, a recent study reported that refeeding hypophosphatemia occurred in approximately one-quarter of patients and was more common in malnourished patients receiving PN, although it was not significantly associated with Child–Pugh or MELD-defined disease severity [16]. This prior evidence provides biological plausibility for the associations observed in the present cohort.
Some studies in general ICU or surgical populations have reported that early aggressive caloric delivery may improve outcomes by reducing catabolism and promoting recovery [17,18]. A systematic review and meta-analysis reported that permissive underfeeding may confer clinical benefits, although without a definitive effect on mortality [19]. Taken together, these discrepancies support cautious, patient-specific nutritional management, particularly during the early phase of critical illness, when metabolic instability and electrolyte shifts are common.
A notable finding was the differential association between ICU day 1 and ICU day 2 kcal/kg. Although ICU day 1 kcal/kg was not significantly associated with serum phosphorus reduction or phosphorus-based RFS status, ICU day 2 kcal/kg was associated with both outcomes. This discrepancy may reflect the inherent variability in nutritional delivery on the first ICU day, when hemodynamic instability, procedures, fluid resuscitation, and fasting frequently interrupt feeding. By ICU day 2, nutritional support is generally more consistently established, making cumulative caloric delivery over the first 48 hours a more reliable indicator of metabolic refeeding burden. Thus, the rate of caloric advancement during this early window warrants close attention.
ICU day 2 kcal/kg was associated with phosphorus-based RFS status in the multivariable analysis, underscoring the tension between adequate and safe nutritional support in this population. Patients with cirrhosis admitted to the ICU frequently present with baseline malnutrition, depleted glycogen stores, protein-energy wasting, and impaired organ function, all of which create substantial nutritional demand. At the same time, rapid carbohydrate or caloric loading may precipitate electrolyte shifts that exceed metabolic adaptive capacity. In high-risk patients, careful initial caloric delivery with stepwise advancement and vigilant electrolyte monitoring, particularly serum phosphate monitoring, remains clinically prudent.
Neither the glucose infusion rate (≥2 mg/kg/min vs. <2 mg/kg/min) nor the timing of formal TPN initiation was significantly associated with outcomes in this study. Thus, cumulative caloric delivery per body weight may be more closely related to phosphorus-based biochemical changes than the rate of carbohydrate infusion or the nominal timing of formal TPN initiation. The lack of association with TPN timing may partly reflect confounding by clinical context, because the decision to initiate TPN is influenced by hemodynamic status, enteral feeding feasibility, and physician discretion. Overall, the total amount of energy delivered may be a more clinically meaningful parameter than route or timing alone.
These findings suggest several considerations for nutritional management in critically ill patients with cirrhosis. These patients often have poor chronic oral intake, malabsorption, systemic inflammation, ascites, and acute decompensation events, such as infection, gastrointestinal bleeding, or renal dysfunction, which may provide a plausible background for refeeding-related electrolyte disturbances. Refeeding-related electrolyte surveillance should ideally begin in the ED before ICU admission, particularly when glucose exposure and early parenteral energy delivery are anticipated. Early nutritional planning should consider total caloric intake, carbohydrate burden, baseline nutritional status, disease severity, organ dysfunction, and electrolyte trends. Serum phosphorus trends should also be interpreted alongside concurrent ICU interventions, including renal replacement therapy, diuretic use, and phosphate supplementation. ICU day 2 kcal/kg may serve as a practical bedside parameter for heightened surveillance during caloric advancement; however, the 10.92 kcal/kg/day cutoff should be regarded only as an exploratory, risk-informed reference rather than a rigid universal target.
This study has several limitations. First, it was a single-center retrospective study with a relatively small sample size (n=72), which may limit generalizability and raises the possibility of selection bias and model overfitting. Although variables for the multivariable models were selected a priori based on clinical relevance and the number of covariates was limited, the regression estimates should be interpreted as exploratory and require validation in larger cohorts. Second, the RFS outcome was based on a phosphorus-based operational definition derived from the ASPEN criteria rather than on a full three-electrolyte classification including phosphate, potassium, and magnesium. Potassium and magnesium were not used as independent diagnostic criteria because serial values were not consistently available, and clinical manifestations or organ dysfunction were not incorporated as mandatory diagnostic criteria because reliable attribution to RFS was not feasible using retrospective records. Therefore, the apparent proportion of patients meeting the phosphorus-based RFS criterion may have been overestimated, and outcome misclassification cannot be excluded. Third, major determinants of serum phosphorus levels were not fully captured or adjusted for in the primary models. In particular, renal replacement therapy, diuretic use, and phosphate supplementation may directly affect serum phosphorus concentrations and, consequently, both the primary outcome of serum phosphorus reduction and the classification of phosphorus-based RFS. Although these variables were reviewed from the electronic medical records when available, their timing, cumulative dose, indication, and temporal relationship to the phosphorus nadir were not consistently structured. Other factors, including sepsis severity, respiratory alkalosis, and dilutional effects from fluid resuscitation, may also have contributed to hypophosphatemia independently of refeeding physiology. Fourth, although NRS-2002-based initial nutritional screening and reduced dietary intake within 2 weeks before admission were available and summarized, comprehensive baseline nutritional assessment was limited. Detailed nutritional assessment tools, such as SGA and GLIM criteria, recent weight loss, muscle mass assessment, and formal malnutrition diagnoses were not systematically documented in the electronic medical records and therefore could not be consistently incorporated into the primary analyses. Fifth, thiamine administration was not standardized and was not included in the primary adjusted models; therefore, its potential modifying or confounding effect could not be fully evaluated. Finally, this study did not include a noncirrhotic critically ill comparator group; therefore, the findings should be interpreted as associations observed within this cirrhosis cohort and should not be considered evidence that cirrhosis itself increases RFS risk compared with other ICU populations. Overall, the findings should be interpreted as exploratory associations between early caloric delivery and phosphorus-based biochemical outcomes rather than as evidence of a causal effect of nutritional support on RFS development.
Prospective multicenter studies with larger cohorts are warranted to validate the 10.92 kcal/kg/day threshold identified in this study and to determine whether its application within structured feeding protocols improves patient outcomes. Future studies should incorporate standardized baseline nutritional assessment; serial phosphate, potassium, and magnesium monitoring; systematic documentation of phosphate replacement, renal replacement therapy, diuretic exposure, and acid-base status; and protocolized thiamine administration. Studies including both cirrhotic and noncirrhotic critically ill comparator groups are needed to determine whether the observed associations are specific to cirrhosis or reflect broader ICU-related refeeding physiology. Developing prediction models for phosphorus-based refeeding-related electrolyte disturbance and integrating nutritional surveillance into electronic medical records may further support patient-specific feeding strategies.
In critically ill patients with liver cirrhosis, higher early caloric delivery through PN, particularly on ICU day 2, was associated with greater serum phosphorus decline and a higher likelihood of meeting the phosphorus-based RFS criterion. However, given the retrospective observational design, phosphorus-based operational definition of RFS, and potential influence of concurrent ICU interventions such as renal replacement therapy, diuretic use, and phosphate supplementation, these findings should be interpreted as associations rather than causal effects. The ICU day 2 caloric threshold identified in this study should be considered exploratory and hypothesis-generating, not a definitive clinical target. Further prospective studies with standardized electrolyte monitoring and baseline nutritional assessment are needed to validate these findings.
Authors’ contribution
Conceptualization: JYL, JY. Data curation: JYL, YKK, KIS, KWS, SL. Formal analysis: JYL, JY. Methodology: JYL, JY. Writing–original draft: JYL, SL, JY. Writing–review & editing: YKK, KIS, KWS, JY. All authors read and approved the final manuscript.
Conflict of interest
The authors of this manuscript have no conflicts of interest to disclose.
Funding
None.
Data availability
Contact the corresponding author for research data availability.
Acknowledgments
None.
Supplementary materials
None.
Fig. 1.
Flow diagram of patient selection. A total of 114 adults with decompensated liver cirrhosis who were admitted to the ICU via the emergency department between January 2021 and August 2024 were screened. After 42 patients were excluded, 72 patients were included in the final analysis. ICU, intensive care unit. aInitial hyperphosphatemia or hypophosphatemia, hyperkalemia or hypokalemia, or hypermagnesemia or hypomagnesemia; bMissing initial test, or missing follow-up test within 5 days of ICU admission.
Fig. 2.
Correlation between ICU day 2 energy delivery and serum phosphorus reduction. Scatter plot showing the relationship between caloric intake per body weight on ICU day 2 (kcal/kg/day) and the percentage reduction in serum phosphorus among critically ill patients with liver cirrhosis. The solid line represents the linear regression line, and the shaded area indicates the 95% CI. ICU day 2 energy delivery was positively correlated with serum phosphorus reduction. ICU, intensive care unit; CI, confidence interval.
Fig. 3.
ICU day 2 energy delivery according to phosphorus-based RFS criterion. Box-and-whisker plot comparing ICU day 2 energy delivery between patients who did and did not meet the phosphorus-based RFS criterion. Individual dots represent observed values. The box indicates the interquartile range, the horizontal line within the box represents the median, and the whiskers indicate the data range. The diamond marker and vertical error bar represent the mean±SD. Patients meeting the phosphorus-based RFS criterion received significantly higher ICU day 2 energy delivery than those not meeting the criterion. ICU, intensive care unit; RFS, refeeding syndrome; SD, standard deviation.
Fig. 4.
ROC curve of ICU day 2 energy delivery for phosphorus-based RFS criterion. ROC curve showing the discriminative performance of ICU day 2 energy delivery for the phosphorus-based RFS criterion. The area under the ROC curve was 0.7061, indicating fair but modest discriminative ability. The optimal cutoff value was determined using Youden’s index, with a cutoff of 10.92 kcal/kg/day, sensitivity of 77.4%, and specificity of 57.9%. This cutoff should be interpreted as an exploratory threshold rather than a clinically actionable target. AUC, area under the curve; ROC, receiver operating characteristic; ICU, intensive care unit; RFS, refeeding syndrome.
Table 1.
Clinical and laboratory characteristics of the study population
| Characteristics | Overall (n=72) |
|---|---|
| Demographics | |
| Age (yr) | 61.00 (54.00–71.00) |
| Male sex | 49 (68.1) |
| Height (cm) | 167.00 (160.00–173.12) |
| Body weight (kg) | 60.50 (55.60–74.00) |
| BMI (kg/m2) | 23.12 (20.76–25.69) |
| BMI <16.0 kg/m2 | 2 (2.8) |
| BMI <18.5 kg/m2 | 9 (12.5) |
| MELD score | 21.00 (14.00–24.25) |
| Time from ED arrival to ICU admission (hr) | 6.98 (4.41–15.68) |
| Etiology and comorbidities | |
| Alcohol-related liver cirrhosis | 40 (55.6) |
| Hepatocellular carcinoma | 28 (38.9) |
| Any comorbidity | 58 (80.6) |
| Phosphorus-based RFS-related outcomesa | |
| Serum phosphorus decrease rate (%) | 29.68 (8.45–48.35) |
| Phosphorus-based RFS criterion met | 53 (73.6) |
| Mild phosphorus-based RFS | 8 (11.1) |
| Moderate phosphorus-based RFS | 10 (13.9) |
| Severe phosphorus-based RFS | 35 (48.6) |
| Phosphorus-based RFS criterion not met | 19 (26.4) |
| Nutritional assessment | |
| Initial nutritional screening | |
| Good nutritional status | 22 (30.6) |
| At risk | 41 (56.9) |
| High risk | 9 (12.5) |
| Dietary intake in the preceding 2 wk | |
| ≤30% of usual intake | 60 (83.3) |
| 30%–60% of usual intake | 4 (5.6) |
| ≥60% of usual intake | 8 (11.1) |
| Nutritional support | |
| Timing of glucose initiation | |
| Timing of glucose administration (within 24 hr) | 61 (84.7) |
| Timing of glucose administration (after 24 hr) | 11 (15.3) |
| ED-1 glucose (g/kg) | 0.40 (0.00–0.61) |
| ED-1 glucose (kcal) | 85.00 (0.00–106.25) |
| Timing of TPN initiation | |
| Timing of formal TPN initiation (within 24 hr) | 33 (45.8) |
| Timing of formal TPN initiation (after 24 hr) | 36 (50.0) |
| No formal TPN during early ICU period | 3 (4.2) |
| Nutrition delivery | |
| ICU day 1 total energy (kcal/day) | 835.00 (538.75–1,268.00) |
| ICU day 1 energy adequacy (% of requirement) | 67.81 (42.16–100.67) |
| ICU day 1 TPN amino acids (g/day) | 34.20 (19.40–46.00) |
| ICU day 2 total energy (kcal/day) | 832.50 (734.25–1,197.75) |
| ICU day 2 energy adequacy (% of requirement) | 70.19 (49.85–89.94) |
| ICU day 2 TPN amino acids (g/day) | 34.20 (31.33–47.00) |
| Thiamine supplementation | 35 (48.6) |
| Outcomes | |
| Length of hospital stay (day) | 12.00 (8.00–18.00) |
| Discharged | 42 (58.3) |
| Transferred | 7 (9.7) |
| In-hospital death | 21 (29.2) |
| Other | 2 (2.8) |
Values are presented as median (interquartile range) or number (%).
BMI, body mass index; MELD, Model for End-Stage Liver Disease; ED, emergency department; ICU, intensive care unit; RFS, refeeding syndrome; TPN, total parenteral nutrition.
aPhosphorus-based RFS was operationally defined as a ≥10% decrease in serum phosphorus from baseline to nadir within 5 ICU days. Severity was classified according to the percentage decrease in serum phosphorus: mild, 10%–20%; moderate, 20%–30%; and severe, >30%.
Table 2.
ICU day 2 energy delivery by phosphorus-based RFS criterion
| Variable | No. | Mean±SD | 95% CI | t | df | P-value |
|---|---|---|---|---|---|---|
| ICU day 2 kcal/kg | 1.57–8.18 | 2.994 | 35.47 | 0.005 | ||
| Phosphorus-based RFS yes | 53 | 15.37±6.62 | ||||
| Phosphorus-based RFS no | 19 | 10.50±5.89 |
Table 3.
Multivariable linear regression analysis for factors associated with serum phosphorus reduction
| Variable | β | SE | 95% CI | P-value |
|---|---|---|---|---|
| Timing of glucose administration (within 24 hr) | 15.64 | 9.02 | −2.37 to 33.65 | 0.088 |
| ED-1 glucose | 19.71 | 5.89 | 7.94 to 31.47 | 0.001* |
| Formal TPN initiation after 24 hr vs. within 24 hr | −4.08 | 6.05 | −16.16 to 8.01 | 0.503 |
| No formal TPN during early ICU period vs. within 24 hr | −7.39 | 14.20 | −35.76 to 20.97 | 0.604 |
| ICU 1 kcal/kg | −0.51 | 0.36 | −1.23 to 0.20 | 0.156 |
| ICU 2 kcal/kg | 1.32 | 0.51 | 0.31 to 2.34 | 0.011* |
Formal TPN initiation was entered as a three-category variable, with formal TPN initiation within 24 hours as the reference category.
β, unstandardized regression coefficient; SE, standard error; CI, confidence interval; ED, emergency department; TPN, total parenteral nutrition; ICU, intensive care unit.
*P<0.05.
Table 4.
Multivariable logistic regression analysis for factors associated with phosphorus-based refeeding syndrome status
| Variable | OR | SE | 95% CI | P-value |
|---|---|---|---|---|
| Timing of glucose administration (within 24 hr) | 4.62 | 1.06 | 0.65 to 45.47 | 0.147 |
| ED-1 glucose | 3.52 | 0.92 | 0.74 to 27.53 | 0.170 |
| Formal TPN initiation after 24 hr vs. within 24 hr | 0.32 | 0.70 | 0.08 to 1.24 | 0.108 |
| No formal TPN during early ICU period vs. within 24 hr | 0.34 | 1.43 | 0.01 to 5.26 | 0.451 |
| ICU 1 kcal/kg | 0.92 | 0.04 | 0.84 to 1.00 | 0.076 |
| ICU 2 kcal/kg | 1.19 | 0.06 | 1.06 to 1.36 | 0.007* |
- 1. European Association for the Study of the Liver. EASL clinical practice guidelines on nutrition in chronic liver disease. J Hepatol 2019;70:172-93. ArticlePubMed
- 2. Meyer F, Bannert K, Wiese M, Esau S, Sautter LF, Ehlers L, et al. Molecular mechanism contributing to malnutrition and sarcopenia in patients with liver cirrhosis. Int J Mol Sci 2020;21:5357.ArticlePubMedPMC
- 3. Reber E, Friedli N, Vasiloglou MF, Schuetz P, Stanga Z. Management of refeeding syndrome in medical inpatients. J Clin Med 2019;8:2202.ArticlePubMedPMC
- 4. da Silva JS, Seres DS, Sabino K, Adams SC, Berdahl GJ, Citty SW, et al. ASPEN consensus recommendations for refeeding syndrome. Nutr Clin Pract 2020;35:178-95. ArticlePubMedPDF
- 5. Fuentebella J, Kerner JA. Refeeding syndrome. Pediatr Clin North Am 2009;56:1201-10. ArticlePubMed
- 6. Santangeli E, Abbati C, Chen R, Di Carlo A, Leoni S, Piscaglia F, et al. Pathophysiological-based nutritional interventions in cirrhotic patients with sarcopenic obesity: a state-of-the-art narrative review. Nutrients 2024;16:427.ArticlePubMedPMC
- 7. Singer P, Blaser AR, Berger MM, Alhazzani W, Calder PC, Casaer MP, et al. ESPEN guideline on clinical nutrition in the intensive care unit. Clin Nutr 2019;38:48-79. ArticlePubMed
- 8. National Collaborating Centre for Acute Care. Nutrition support for adults: oral nutrition support, enteral tube feeding and parenteral nutrition [Internet]. National Collaborating Centre for Acute Care; 2006. [cited 2026 May 1]. Available from: https://www.nice.org.uk/guidance/cg32/evidence/full-guideline-194889853
- 9. Doig GS, Simpson F, Heighes PT, Bellomo R, Chesher D, Caterson ID, et al. Restricted versus continued standard caloric intake during the management of refeeding syndrome in critically ill adults: a randomized, parallel-group, multicentre, single-blind controlled trial. Lancet Respir Med 2015;3:943-52. ArticlePubMed
- 10. Olthof LE, Koekkoek WA, van Setten C, Kars JC, van Blokland D, van Zanten AR. Impact of caloric intake in critically ill patients with, and without, refeeding syndrome: a retrospective study. Clin Nutr 2018;37:1609-17. ArticlePubMed
- 11. Cioffi I, Ponzo V, Pellegrini M, Evangelista A, Bioletto F, Ciccone G, et al. The incidence of the refeeding syndrome: a systematic review and meta-analyses of literature. Clin Nutr 2021;40:3688-701. ArticlePubMed
- 12. Singer P, Blaser AR, Berger MM, Calder PC, Casaer M, Hiesmayr M, et al. ESPEN practical and partially revised guideline: clinical nutrition in the intensive care unit. Clin Nutr 2023;42:1671-89. ArticlePubMed
- 13. Statlender L, Raphaeli O, Shochat T, Robinson E, Hellerman Itzhaki M, Bendavid I, et al. Contributing factors to hypophosphatemia development in critically ill ventilated patients: a retrospective cohort study. Sci Rep 2024;14:19771.ArticlePubMedPMCPDF
- 14. Slingerland-Boot R, Rooijakkers E, Koekkoek K, van Blokland D, Arbous S, van Zanten A. Macronutrient intake and outcomes of ICU patients with refeeding hypophosphatemia. Clin Nutr ESPEN 2023;55:191-9. ArticlePubMed
- 15. Zusman O, Theilla M, Cohen J, Kagan I, Bendavid I, Singer P. Resting energy expenditure, calorie and protein consumption in critically ill patients: a retrospective cohort study. Crit Care 2016;20:367.ArticlePubMedPMCPDF
- 16. Can Sezgin G, Ozer NT, Sahin GG, Sahin S, Yurci MA, Guven K, et al. Frequency and related risk factors of refeeding hypophosphatemia in patients with liver cirrhosis: a pilot study. J Gastrointestin Liver Dis 2023;32:339-45. ArticlePubMedPDF
- 17. Heidegger CP, Berger MM, Graf S, Zingg W, Darmon P, Costanza MC, et al. Optimisation of energy provision with supplemental parenteral nutrition in critically ill patients: a randomized controlled clinical trial. Lancet 2013;381:385-93. ArticlePubMed
- 18. Weijs PJ, Stapel SN, de Groot SD, Driessen RH, de Jong E, Girbes AR, et al. Optimal protein and energy nutrition decreases mortality in mechanically ventilated, critically ill patients: a prospective observational cohort study. JPEN J Parenter Enteral Nutr 2012;36:60-8. ArticlePubMed
- 19. Yue HY, Peng W, Zeng J, Zhang Y, Wang Y, Jiang H. Efficacy of permissive underfeeding for critically ill patients: an updated systematic review and trial sequential meta-analysis. J Intensive Care 2024;12:4.ArticlePubMedPMCPDF
References
Figure & Data
REFERENCES
Citations
Citations to this article as recorded by 
Cite
XML DownloadAssociation of early parenteral energy provision with serum phosphorus decline and phosphorus-based refeeding syndrome in critically ill patients with liver cirrhosis: a Korean retrospective observational study
Fig. 1. Flow diagram of patient selection. A total of 114 adults with decompensated liver cirrhosis who were admitted to the ICU via the emergency department between January 2021 and August 2024 were screened. After 42 patients were excluded, 72 patients were included in the final analysis. ICU, intensive care unit. aInitial hyperphosphatemia or hypophosphatemia, hyperkalemia or hypokalemia, or hypermagnesemia or hypomagnesemia; bMissing initial test, or missing follow-up test within 5 days of ICU admission.
Fig. 2. Correlation between ICU day 2 energy delivery and serum phosphorus reduction. Scatter plot showing the relationship between caloric intake per body weight on ICU day 2 (kcal/kg/day) and the percentage reduction in serum phosphorus among critically ill patients with liver cirrhosis. The solid line represents the linear regression line, and the shaded area indicates the 95% CI. ICU day 2 energy delivery was positively correlated with serum phosphorus reduction. ICU, intensive care unit; CI, confidence interval.
Fig. 3. ICU day 2 energy delivery according to phosphorus-based RFS criterion. Box-and-whisker plot comparing ICU day 2 energy delivery between patients who did and did not meet the phosphorus-based RFS criterion. Individual dots represent observed values. The box indicates the interquartile range, the horizontal line within the box represents the median, and the whiskers indicate the data range. The diamond marker and vertical error bar represent the mean±SD. Patients meeting the phosphorus-based RFS criterion received significantly higher ICU day 2 energy delivery than those not meeting the criterion. ICU, intensive care unit; RFS, refeeding syndrome; SD, standard deviation.
Fig. 4. ROC curve of ICU day 2 energy delivery for phosphorus-based RFS criterion. ROC curve showing the discriminative performance of ICU day 2 energy delivery for the phosphorus-based RFS criterion. The area under the ROC curve was 0.7061, indicating fair but modest discriminative ability. The optimal cutoff value was determined using Youden’s index, with a cutoff of 10.92 kcal/kg/day, sensitivity of 77.4%, and specificity of 57.9%. This cutoff should be interpreted as an exploratory threshold rather than a clinically actionable target. AUC, area under the curve; ROC, receiver operating characteristic; ICU, intensive care unit; RFS, refeeding syndrome.
Graphical abstract
Fig. 1.
Fig. 2.
Fig. 3.
Fig. 4.
Graphical abstract
Association of early parenteral energy provision with serum phosphorus decline and phosphorus-based refeeding syndrome in critically ill patients with liver cirrhosis: a Korean retrospective observational study
| Characteristics | Overall (n=72) |
|---|---|
| Demographics | |
| Age (yr) | 61.00 (54.00–71.00) |
| Male sex | 49 (68.1) |
| Height (cm) | 167.00 (160.00–173.12) |
| Body weight (kg) | 60.50 (55.60–74.00) |
| BMI (kg/m2) | 23.12 (20.76–25.69) |
| BMI <16.0 kg/m2 | 2 (2.8) |
| BMI <18.5 kg/m2 | 9 (12.5) |
| MELD score | 21.00 (14.00–24.25) |
| Time from ED arrival to ICU admission (hr) | 6.98 (4.41–15.68) |
| Etiology and comorbidities | |
| Alcohol-related liver cirrhosis | 40 (55.6) |
| Hepatocellular carcinoma | 28 (38.9) |
| Any comorbidity | 58 (80.6) |
| Phosphorus-based RFS-related outcomes |
|
| Serum phosphorus decrease rate (%) | 29.68 (8.45–48.35) |
| Phosphorus-based RFS criterion met | 53 (73.6) |
| Mild phosphorus-based RFS | 8 (11.1) |
| Moderate phosphorus-based RFS | 10 (13.9) |
| Severe phosphorus-based RFS | 35 (48.6) |
| Phosphorus-based RFS criterion not met | 19 (26.4) |
| Nutritional assessment | |
| Initial nutritional screening | |
| Good nutritional status | 22 (30.6) |
| At risk | 41 (56.9) |
| High risk | 9 (12.5) |
| Dietary intake in the preceding 2 wk | |
| ≤30% of usual intake | 60 (83.3) |
| 30%–60% of usual intake | 4 (5.6) |
| ≥60% of usual intake | 8 (11.1) |
| Nutritional support | |
| Timing of glucose initiation | |
| Timing of glucose administration (within 24 hr) | 61 (84.7) |
| Timing of glucose administration (after 24 hr) | 11 (15.3) |
| ED-1 glucose (g/kg) | 0.40 (0.00–0.61) |
| ED-1 glucose (kcal) | 85.00 (0.00–106.25) |
| Timing of TPN initiation | |
| Timing of formal TPN initiation (within 24 hr) | 33 (45.8) |
| Timing of formal TPN initiation (after 24 hr) | 36 (50.0) |
| No formal TPN during early ICU period | 3 (4.2) |
| Nutrition delivery | |
| ICU day 1 total energy (kcal/day) | 835.00 (538.75–1,268.00) |
| ICU day 1 energy adequacy (% of requirement) | 67.81 (42.16–100.67) |
| ICU day 1 TPN amino acids (g/day) | 34.20 (19.40–46.00) |
| ICU day 2 total energy (kcal/day) | 832.50 (734.25–1,197.75) |
| ICU day 2 energy adequacy (% of requirement) | 70.19 (49.85–89.94) |
| ICU day 2 TPN amino acids (g/day) | 34.20 (31.33–47.00) |
| Thiamine supplementation | 35 (48.6) |
| Outcomes | |
| Length of hospital stay (day) | 12.00 (8.00–18.00) |
| Discharged | 42 (58.3) |
| Transferred | 7 (9.7) |
| In-hospital death | 21 (29.2) |
| Other | 2 (2.8) |
| Variable | No. | Mean±SD | 95% CI | t | df | P-value |
|---|---|---|---|---|---|---|
| ICU day 2 kcal/kg | 1.57–8.18 | 2.994 | 35.47 | 0.005 | ||
| Phosphorus-based RFS yes | 53 | 15.37±6.62 | ||||
| Phosphorus-based RFS no | 19 | 10.50±5.89 |
| Variable | β | SE | 95% CI | P-value |
|---|---|---|---|---|
| Timing of glucose administration (within 24 hr) | 15.64 | 9.02 | −2.37 to 33.65 | 0.088 |
| ED-1 glucose | 19.71 | 5.89 | 7.94 to 31.47 | 0.001 |
| Formal TPN initiation after 24 hr vs. within 24 hr | −4.08 | 6.05 | −16.16 to 8.01 | 0.503 |
| No formal TPN during early ICU period vs. within 24 hr | −7.39 | 14.20 | −35.76 to 20.97 | 0.604 |
| ICU 1 kcal/kg | −0.51 | 0.36 | −1.23 to 0.20 | 0.156 |
| ICU 2 kcal/kg | 1.32 | 0.51 | 0.31 to 2.34 | 0.011 |
| Variable | OR | SE | 95% CI | P-value |
|---|---|---|---|---|
| Timing of glucose administration (within 24 hr) | 4.62 | 1.06 | 0.65 to 45.47 | 0.147 |
| ED-1 glucose | 3.52 | 0.92 | 0.74 to 27.53 | 0.170 |
| Formal TPN initiation after 24 hr vs. within 24 hr | 0.32 | 0.70 | 0.08 to 1.24 | 0.108 |
| No formal TPN during early ICU period vs. within 24 hr | 0.34 | 1.43 | 0.01 to 5.26 | 0.451 |
| ICU 1 kcal/kg | 0.92 | 0.04 | 0.84 to 1.00 | 0.076 |
| ICU 2 kcal/kg | 1.19 | 0.06 | 1.06 to 1.36 | 0.007 |
Table 1. Clinical and laboratory characteristics of the study population
Values are presented as median (interquartile range) or number (%). BMI, body mass index; MELD, Model for End-Stage Liver Disease; ED, emergency department; ICU, intensive care unit; RFS, refeeding syndrome; TPN, total parenteral nutrition. Phosphorus-based RFS was operationally defined as a ≥10% decrease in serum phosphorus from baseline to nadir within 5 ICU days. Severity was classified according to the percentage decrease in serum phosphorus: mild, 10%–20%; moderate, 20%–30%; and severe, >30%.
Table 2. ICU day 2 energy delivery by phosphorus-based RFS criterion
ICU, intensive care unit; RFS, refeeding syndrome; SD, standard deviation; CI, confidence interval; df, degrees of freedom.
Table 3. Multivariable linear regression analysis for factors associated with serum phosphorus reduction
Formal TPN initiation was entered as a three-category variable, with formal TPN initiation within 24 hours as the reference category. β, unstandardized regression coefficient; SE, standard error; CI, confidence interval; ED, emergency department; TPN, total parenteral nutrition; ICU, intensive care unit. P<0.05.
Table 4. Multivariable logistic regression analysis for factors associated with phosphorus-based refeeding syndrome status
Formal TPN initiation was entered as a three-category variable, with formal TPN initiation within 24 hours as the reference category. OR, odds ratio; SE, standard error; CI, confidence interval; ED, emergency department; TPN, total parenteral nutrition; ICU, intensive care unit. P<0.05.
