Effect of A Positive Cumulative Fluid Balance On The Prognosis of Patients With Sepsis In The Xining Area

Background: The aim of this work is to analyze the effect of a positive cumulative uid balance and relative clinical indicators on the prognosis of patients with sepsis in the Xining area, China. Methods: The clinical data of 480 sepsis patients (313 males and 167 females, aged 52–77 (65) years) admitted between January 2017 and December 2019 were retrospectively analyzed. The APACHE II score, SOFA score, SIRS score and clinical laboratory test indicators of the patients were collected. Receiver operating characteristic (ROC) curves were used to analyze the sensitivity and specicity of each indicator in predicting the poor prognosis of patients with sepsis, and the maximum Youden index was used to determine threshold values. Cox regression analysis was performed to assess patient prognosis using data from patients with different uid balances. Results: The following clinical indicators were signicantly different between the 2 groups (P<0.05): APACHE II score, SOFA score, SIRS score, PCT, IL-6, BNP, CRP, PLT, BUN, CREA, Lac and total uid balance from days 1 to 5. The area under the ROC curve (AUC) for total uid balance from days 1 to 5 was 0.558, the cut-off value was 2120.5 mL, the sensitivity was 54.0%, and the specicity was 58.1%. The survival rates were different between the 2 groups (60.9% vs 48.9%, P<0.05). Total uid balance was signicantly higher in patients with septic shock and with Lac>2.0 mmol/L (P<0.05). Cox regression analysis indicated that APACHE II score, SOFA score, PLT score, Lac, and total uid balance from days 1 to 5 were independent risk factors for poor prognosis. Conclusion: uid days 5 ICU poor was poor


Background
Early and adequate uid resuscitation is the cornerstone of sepsis and septic shock treatment [1,2] .
However, recommended uid resuscitation regimens may present risks of uid overload (FO) in some patients with stable blood pressure. In a recent study, it was found that during early uid resuscitation in patients with sepsis, the rst consideration for clinicians is hemodynamic stability, that is, blood pressure stability, rather than changes in organ function. An increasing number of studies have shown that FO can affect the organ function and prognosis of patients with sepsis [3][4][5] . The altitude of the Xining area of China is 2260 m, the atmospheric pressure is 585 mmHg, the atmospheric oxygen partial pressure is 121 mmHg, and the average human arterial oxygen partial pressure is 75 mmHg. Humans in this location are in a chronic hypoxic environment; however, the human body adapts through compensatory physiological changes. For example, pulmonary arterial pressure gradually increases, and microcirculation hyperplasia occurs. Due to these physiological changes, the cumulative amount of uid resuscitation will impact the prognosis of patients with sepsis and septic shock. In this study, we retrospectively analyzed the clinical data of sepsis patients admitted to our hospital to observe whether the changes in positive uid balance in sepsis patients are related to adverse patient outcomes and to explore the possible mechanisms.

Study population
The medical records of 480 sepsis patients admitted to the intensive care unit (ICU) of Qinghai Provincial People's Hospital from January 1, 2017, to December 31, 2019, were collected. The inclusion criteria were as follows: admitted patients who met the diagnostic criteria of Sepsis 3.0 [6]; and patients older than 18 years. The exclusion criteria were patients with severe hematological diseases; patients with severe malignant tumors; patients with long-term use of hormones; patients with diabetic ketoacidosis and diabetic hyperosmolar coma; patients who were pregnant; and patients with incomplete clinical data. All patients were performed in accordance with Surviving Sepsis Campaign: International Guidelines for Management of Sepsis and Septic Shock: 2016.The informed consent was obtained from all subjects and/or their legal guardian(s).

Data collection
The following clinical data of patients at the time of admission to the ICU were collected: nationality, sex, source, diagnosis, acute physiology and chronic health evaluation II (APACHE II) score, sequential organ failure assessment (SOFA) score, systemic in ammatory response syndrome (SIRS) score, Glasgow coma scale (GCS) score, procalcitonin (PCT) concentration, interleukin-6 (IL-6) concentration, brain natriuretic peptide (BNP) concentration, C-reactive protein (CRP) concentration, white blood cell (WBC) count, platelet (PLT) count, blood urea nitrogen (BUN) concentration, creatinine (CREA) concentration, and uid from days 1 to 5 after ICU admission. The number of days of hospitalization in the ICU and the prognosis at the time of discharge from the ICU were recorded. PCT concentration was determined using a Cobas 8000 (Roche, Germany). BNP concentration was determined using a DXi 800 (Beckman Coulter, USA). Routine blood analyses were performed using an XN9000 (Sysmex Corporation, Japan). Biochemical indicators were analyzed using a Siemens ADVIA®2400 automatic biochemical analyzer.
Supporting reagents were supplied by the manufacturers.

Outcomes
Hospital mortality was the endpoint, which was de ned as the status of patient survival before ICU discharge.
Statistical analysis SPSS 22.0 was used for data processing. The 1-sample KS test (two-sided test) was used to determine whether the measurement data conformed to a normal distribution. Abnormally distributed measurement data are presented as M (Q 1 , Q 3 ), and the Mann-Whitney U test was used for comparisons between the 2 groups. Count data were analyzed using the four-grid table or row × column table X 2 test. Receiver operating characteristic (ROC) curves were used to determine the sensitivity, speci city, and maximum Youden index of each included indicator. The prognosis of patients with different uid balances was compared using Cox regression analysis. α < 0.05 indicated that a difference was statistically signi cant.
This clinical trial is registered by Chinese Clinical Trial Registry. The number is ChiCTR1900028628. This work has been approved by Ethnic Committee of Qinghai Provincial People's Hospital.

Results
Clinical data between survival group and deceased group Patients were divided into a survival group and a deceased group based on their status at the time of leaving the ICU. The clinical data of the 2 groups were compared: there was no signi cant difference between the 2 groups in nationality, sex, etiological composition, or age (P>0.05). Patient source was statistically signi cant (P <0.05), and the mortality rate for patients from wards (51.2%) was signi cantly higher than that for patients from the emergency department and from other hospitals (38.8% and 37.4%, respectively). The primary infection was not signi cantly different between groups (P>0.05). APACHE II score, SOFA score, SIRS score, PCT, IL-6, BNP, CRP, PLT, BUN, CREA, Lac, and total uid balance from days 1 to 5 were signi cantly different between the 2 groups (P<0.05) ( Table 1). This table mainly re ects the baseline data of the two groups with different nal outcomes.

Ability of the clinical indicators to predict death from sepsis
The value of the signi cant indicators in the above table (APACHE II score, SOFA score, SIRS score, PCT, IL-6, BNP, CRP, PLT, BUN, CREA, Lac, and total uid balance from days 1 to 5) in predicting the poor prognosis of patients with sepsis was analyzed using ROC curves. The sensitivity, speci city, and maximum Youden index cut-off value are shown in Table 2 and Figure 1.

Comparison of uid balance based on different clinical factors
There was no signi cant difference in the total uid balance from days 1 to 5 between sexes and among different nationalities and different sources (P>0.05). However, the total uid balance of septic shock patients and patients with Lac>2.0 mmol/L was signi cantly increased (P<0.05) ( Table 3). Note a: The Kruskal-Wallis test was used to examine whether the distributions of the 2 groups were the same and b: to examine whether the medians of the 2 groups were the same.

This table mainly shows the comparison of uid balance between 1 to 5 days in different clinical factors
Cox regression analysis Using 2120.5 ml as the cut-off point for total uid balance from days 1 to 5, the survival rates for patients with different total uid balances from days 1 to 5 was statistic difference (60.9% vs. 48.9%, P <0.05). The independent risk factors for poor prognosis included APACHE II score, SOFA score, SIRS score, PCT, IL-6, BNP, CRP, PLT, BUN, CREA, and Lac, and total uid balance from days 1 to 5 by Cox regression analysis (Table 4, Figure 2).

Discussion
Fluid resuscitation is the basis for maintaining hemodynamic stability and organ and tissue perfusion and for increasing oxygen delivery during the rescue phase of septic shock [7] . Studies have shown that early resuscitation can reduce the mortality of patients with sepsis and septic shock [8][9][10] . However, excessive uid resuscitation may be harmful to patients. Many studies have shown that a positive uid balance or FO is closely associated with the occurrence of acute kidney injury (AKI) and poor patient prognosis [11][12][13][14][15][16] . Chao et al. [17] found that a negative cumulative body uid balance from days 1 to 4 reduced the mortality of patients with severe in uenza. Judith et al. [18] found that cumulative FO during the rst 5 days of pediatric ICU (PICU) admission was an independent risk factor for poor prognosis in children with septic shock. A recent prospective study also showed that in patients who underwent cardiac or aortic surgery, continuous positive uid balance was associated with AKI and hemodialysis and that the risk of AKI in patients with a continuous positive uid balance increased by 7.1 times [19] . Our study found that the total uid balance from days 1 to 5 for patients in the deceased group was signi cantly higher than that for patients in the survival group, suggesting that total uid balance from days 1 to 5 may be associated with poor patient prognosis. Furthermore, Cox regression analysis indicated that the total uid balance from days 1 to 5 was an independent risk factor for poor patient prognosis; the risk of death increased by 0.1% for every 1 ml increase. However, a previous study [20] has shown that after adjusting for disease severity and the lactate clearance rate, cumulative uid balance was not associated with increased mortality. The reason for this result may be related to the retrospective nature of the analysis. The results of our study showed that there was no signi cant difference in the mortality rate between sexes, among different nationality groups, and among different ages (P>0.05); however, the mortality rate was signi cantly different among different sources (P<0.05), with the mortality rate for sepsis patients from wards being the highest. One explanation for this result may be that ICU physicians paid more attention to sepsis than general ward physicians or that patients in wards received more uid therapy. This is just speculation, as relevant data were not collected to address that issue. The comparison of clinical factors showed that there was no difference in uid balance between sexes, among different nationality groups, and among different sources (P >0.05), indicating that these clinical factors had no effect on uid balance. However, the uid balance for patients with shock was signi cantly higher than that for patients with stable hemodynamics at the time of ICU admission, and the mortality rate for patients with septic shock was signi cantly higher (49.85% vs. 27.7%), a nding that is consistent with the clinical situation. However, it is not clear whether the poor prognosis of patients is due to septic shock or due to a positive uid balance. More uid therapy or a more positive uid balance may be related to severe vascular leakage and third space leakage, but whether it is a direct cause of increased mortality remains unclear. Because this was a retrospective study, it can only be concluded that a more positive uid balance may be associated with poor patient prognosis. Cox regression analysis showed that APACHE II score, SOFA score, PLT, and Lac were associated with poor patient prognosis.
Studies have con rmed that APACHE II and SOFA scores are associated with poor patient prognosis [21] .
Regarding the prediction of poor patient prognosis, the ROC curve results indicated a sensitivity of 55.6% and speci city of 76.6% for APACHE II > 20 points and a sensitivity of 51.6% and speci city of 80.0% for SOFA > 8 points. The decrease in PLT may be related to the progression of sepsis and dilution caused by the increased body volume resulting from a positive uid balance. Regression analysis indicated that for every unit increase in PLT, the risk of death decreased by 0.2%. Elevated Lac represents oxygen utilization disorder in the body, which may be caused by oxygen uptake, and oxygen utilization disorder in the tissues, which may be due to the widening of the capillary space after uid treatment. Among the clinical factors, uid balance was signi cantly increased when Lac>2.0 mmol/L. Lac and uid balance may have a causal relationship. Unfortunately, this study could not determine cause and effect as it was a retrospective analysis. BUN and CREA in the deceased group were signi cantly increased at the time of ICU admission; these ndings may explain the excessive total uid load from days 1 to 5 in the deceased group.
Based on the above research results, we need more rigorous studies to re-examine uid resuscitation therapy for sepsis [22] . Degradation or damage to the glycocalyx layer of the vascular endothelial cell membrane can occur in the early stage of sepsis, and uid resuscitation therapy may further damage the glycocalyx layer, especially during rapid infusions and transfusions that result in hypervolemia [23,24] . The destruction of the glycocalyx layer of vascular endothelial cells may cause capillary leakage, local tissue edema, and oxygen utilization disorder, which may be side effects of uid resuscitation treatment, thereby affecting the prognosis of patients. The use of hypertonic uid for small volume resuscitation in patients with sepsis also requires further study.

Conclusions
This study reviewed the effect of uid balance on the prognosis of sepsis patients in Chengdong District of Xining City, Qinghai Province. We found that the cumulative uid balance from days 1 to 5 after admission to the ICU was associated with poor patient prognosis. The risk of death gradually increased when the cumulative uid balance was greater than +2120.5 ml (60.9% vs. 48.9%, P <0.05). In view of the above results, uid management, i.e., cumulative uid balance, during the treatment of patients with sepsis and septic shock in the Xining area must receive attention from clinicians. Clinicians should focus on a positive uid balance earlier (within 5 days) and optimize volume management early to achieve a negative cumulative uid balance.
This study has limitations. First, this was a retrospective study conducted in a hospital in an urban area; therefore, the results may not completely represent the clinical reality at moderate altitudes. Second, because this was a retrospective study, the causal relationship between uid balance and prognosis cannot be determined. Third, although possible confounding factors, such as sex, nationality, and origin, were compared, all confounding factors could not completely be eliminated. Fourth, the uid balance before the ICU stay was not considered. Fifth, the use of diuretics was not recorded, and their use may have affected uid management and the results. The results of this study should be further veri ed through a study with a larger sample size, and the resulting data should be paired and compared with data from low-altitude areas.  Figure 1 ROC curve for total uid balance from days 1 to 5, The AUC of total uid balance from days 1 to 5 was 0.558. The cut-off value, sensitivity and speci city of the maximum Youden index for total uid balance was 2120.5 ml, 54.0% and 58.1%.