Multiple Organ Scoring Systems for Predicting In-Hospital Mortality of Sepsis Patients in the Intensive Care Unit

doi:10.21203/rs.3.rs-4885379/v1

Objective

This study aims to comprehensively evaluate the Charlson Comorbidity Index (CCI), Logistic Organ Dysfunction System (LODS) score, Acute Physiology Score III (APS III), Simplified Acute Physiology Score II (SAPS II), and SOFA (Sequential Organ Failure Assessment) in order to provide a more scientifically rigorous and effective tool for predicting mortality risk among sepsis patients in the Intensive Care Unit (ICU).

Methods

The demographic information and sepsis-related data were extracted from the MIMIC-IV database for patients admitted to the ICU with a diagnosis of sepsis. The predictive performance of CCI, LODS, APS III, SAPS II, and SOFA scoring systems in terms of ICU mortality was evaluated by comparing receiver operating characteristic (ROC) curves. Furthermore, multivariate regression analysis was conducted to identify potential predictors which were subsequently validated using a nomogram. Finally, the dataset was divided into a training set and validation set at a ratio of 7:3 to assess the clinical utility of the prediction model through ROC curves.

Results

The study enrolled a total of 17,226 patients with a median age of 67.65 (55.79, 78.82) years; among them, males accounted for 57.69%. Within this cohort, a total of 1,115 (6.47%) individuals succumbed during their admission to the ICU. The ROC curve demonstrated that both APS III and LODS score exhibited robust predictive value for ICU mortality. Multivariate regression analysis revealed that CCI, APS III score, LODS score, respiratory rate, body temperature, and race served as potential predictors. The integration of these indicators into a nomogram demonstrated significant clinical relevance, as evidenced by an area under the curve (AUC) of 0.803 in the training set and 0.797 in the validation set.

Conclusions

Organ failure scores hold significant clinical relevance in predicting mortality among sepsis patients in the ICU. Augmenting predictive accuracy can be achieved by integrating the CCI, APS III score, LODS score, mean respiratory rate, and body temperature during the initial 24 hours following ICU admission.

Sepsis

CCI

LODS

APS III

SAPS II

SOFA

ICU

In-Hospital Mortality

The pathogenesis of sepsis involves an exaggerated host response to infection, resulting in tissue damage and organ dysfunction, thereby precipitating a systemic inflammatory response syndrome [1]. The incidence and mortality rates of sepsis are increasing, presenting a significant challenge to global public health [2]. The mortality rate of patients with sepsis in the intensive care unit (ICU) is notably high, ranging from 30–50%, rendering it one of the foremost causes of fatality among ICU patients [3–6].

In light of the significant mortality rate associated with sepsis, timely and accurate assessment of patients' condition and risk of death is crucial for improving survival rates [7]. In clinical practice, physicians commonly utilize various scoring systems to evaluate the severity and prognosis of septic patients. These scoring systems include, but are not limited to, the Charlson Comorbidity Index (CCI) [8, 9], Logistic Organ Dysfunction System (LODS) [10], Acute Physiology Score III (APS III) [11], Simplified Acute Physiology Score II (SAPS II) [12, 13], and Sequential Organ Failure Assessment (SOFA) scoring systems [14]. Scoring systems, serving as a pivotal tool in quantifying the severity of a patient's condition, assume an indispensable role in critical care medicine by employing physiological parameters, laboratory indexs, and other relevant clinical indicators to calculate a numerical value that predicts the patient's risk of mortality [15–17]. Different scoring systems place varying emphasis on different aspects [18]. For example, the CCI primarily evaluates patients' comorbidities [19], while the LODS and APS III focus more on the physiological condition of the patient [18, 20]. Conversely, SAPS II and SOFA take into consideration a multitude of factors [20, 21].

However, despite the pivotal role played by these scoring systems in predicting patient outcomes in sepsis, disparities exist in their respective levels of accuracy. Moreover, due to the intricate and dynamic nature of sepsis patients' conditions, a singular scoring system often encounters challenges when attempting to comprehensively and accurately reflect their true state [22]. Consequently, the selection and utilization of these scoring systems as well as the integration of multiple system results have emerged as prominent issues within current clinical research for more precise assessment of mortality risk in sepsis patients.

In conclusion, the objective of this study was to elucidate the prognostic value of each scoring system in assessing mortality risk among sepsis patients during their ICU stay. By conducting a comprehensive evaluation of the CCI, LODS, APS III, SAPS II and SOFA scoring systems, while considering vital signs monitoring alongside multiple scoring systems, our aim is to provide a more scientifically robust and effective clinical prediction tool for evaluating the risk of mortality in sepsis.

2.1 Study design

The data for this study were extracted from the Medical Information Mart for Intensive Care-IV (MIMIC-IV) database, encompassing comprehensive clinical information on a cohort of more than 190,000 patients and 450,000 hospitalizations admitted to Beth Israel Deaconess Medical Center (BIDMC) between 2008 and 2019 [23]. This extensive database comprises patient demographic details, laboratory test results, medication records, vital signs measurements, surgical procedures performed, disease diagnosis, drug management specifics, follow-up survival status updates, as well as other pertinent information. This study presents an analysis of a publicly available database, for which no approval from the Institutional Review Board of Jianhu County branch of Northern Jiangsu People's Hospital was deemed necessary.

Patients admitted to the ICU for the first time with a diagnosis of sepsis were included with information from MIMIC-IV. The inclusion criteria were as follows: (1) Over 18 years of age, (2) Diagnosis with sepsis. The exclusion criteria were defined as follows: patients who (1) with previous history of hospitalization, (2) duration of the ICU stay less than 24 hours, (3) with malignant tumors.

2.2 Date extraction

The diagnosis of sepsis was made in accordance with the sepsis-3 criteria [24]. The demographic information, encompassing age, gender, ethnicity, and body mass index (BMI), was extracted from the database utilizing Navicate Premium 16.0. The CCI, LODS, SOFA, APS III, SAPS II, GCS scores and laboratory parameters including creatine were evaluated 24 hours post admission to the ICU. The mean value was employed for variables with multiple measurements.

The determination of a patient's expiration in the ICU was based on comparing their time of death with their time of discharge from the ICU. If these times coincided, it was concluded that the patient deceased during their stay in the ICU; otherwise, they were considered to be alive.

2.3 Statistical analysis

Measurement data that followed a normal distribution were presented as Mean ± SD. For data with equal variances, a T test was utilized to compare two groups. In cases where the measurement data did not conform to a normal distribution, quartiles and median values were used instead, and differences between any two groups were compared using the Mann-Whitney test. Count data were expressed in terms of frequencies and proportions, with group differences assessed through chi-square testing. Statistical significance was established when p < 0.05.

The predictive efficacy of each scoring system for ICU sepsis patients' mortality was assessed using Receiver Operating Characteristic (ROC) curve analysis. Potential predictors, including vital signs, laboratory parameters, and scores, were identified through regression analysis. The variables exhibiting a significance level of P < 0.05 in the univariate regression analysis were incorporated into the multivariate regression analysis and examined using the two-way stepwise regression method. A nomogram was employed to present the final combined score and baseline characteristics of the patients. Furthermore, a training set and validation set were created by dividing the dataset in a 7:3 ratio. The performance of the final model was validated using ROC curve analysis.

3.1 Clinical characteristics of patients with sepssis

The process of data extraction and screening was carried out following the workflow illustrated in Figure 1. A total of 24,816 adult patients were diagnosed with sepsis and admitted to the hospital for their initial presentation. After excluding 2,658 patients with a history of prior hospitalization, 3,051 patients with malignancy, and 1,881 patients who stayed in the ICU for less than 24 hours, a total of 17,226 eligible patients were enrolled. Among them, 1,115 patients (6.47%) succumbed during their stay in the ICU. The enrolled patients had a median age of 67.65 (55.79, 78.82) years, with males accounting for 57.69% of the cohort. The baseline characteristics of the enrolled patients are presented in Table 1.

The death group exhibited significantly higher levels of age, CCI, LODS, SOFA, APS III, SAPS II scores, GCS scores, creatine levels, heart rate within 24 hours of admission to ICU compared to the survival group (P<0.001). However, the death group exhibited a lower BMI and mean body temperature during the initial 24-hour period in the ICU compared to the survivors. Additionally, the proportion of females (P=0.032) and patients receiving mechanical ventilation (P<0.001) in the death group were higher compared to those in the survival group.

Figure 1: The process of information extraction and the screening criteria for sepsis patients

Table 1: The baseline data of the enrolled patients

Variables	Total (n = 17226)	Survival (n = 16111)	Death (n = 1115)	Statistic	P
Age, Years, M (Q₁, Q₃)	67.65 (55.79, 78.82)	67.51 (55.69, 78.64)	70.16 (57.23, 81.64)	Z=-4.29	<0.001
Gender, n(%)				χ²=4.59	0.032
Female	7289 (42.31)	6783 (42.10)	506 (45.38)
Male	9937 (57.69)	9328 (57.90)	609 (54.62)
Race, n(%)				χ²=236.70	<0.001
Asian	438 (2.54)	411 (2.55)	27 (2.42)
Black	1306 (7.58)	1229 (7.63)	77 (6.91)
White	11450 (66.47)	10867 (67.45)	583 (52.29)
Hispanic / Latino	550 (3.19)	518 (3.22)	32 (2.87)
Other	744 (4.32)	706 (4.38)	38 (3.41)
Unknown	2738 (15.89)	2380 (14.77)	358 (32.11)
BMI, kg/m2, M (Q₁, Q₃)	28.30 (24.60, 33.10)	28.40 (24.60, 33.10)	27.90 (24.13, 32.90)	Z=2.02	0.043
SOFA, score, M (Q₁, Q₃)	3.00 (2.00, 4.00)	3.00 (2.00, 4.00)	4.00 (2.00, 6.00)	Z=-12.29	<0.001
CCI, score, M (Q₁, Q₃)	4.00 (3.00, 6.00)	4.00 (3.00, 6.00)	5.00 (3.00, 7.00)	Z=-9.43	<0.001
Creatinine, mg/dL, M(Q₁, Q₃)	1.00 (0.70, 1.50)	1.00 (0.70, 1.50)	1.30 (0.90, 2.10)	Z=-13.90	<0.001
GCS, score, M (Q₁, Q₃)	15.00 (15.00, 15.00)	15.00 (15.00, 15.00)	15.00 (15.00, 15.00)	Z=-0.81	0.415
APS III, score, M (Q₁, Q₃)	45.00 (33.00, 60.00)	44.00 (32.00, 59.00)	66.00 (50.00, 88.00)	Z=-28.63	<0.001
LODS, score, M (Q₁, Q₃)	5.00 (3.00, 7.00)	5.00 (3.00, 7.00)	8.00 (6.00, 10.00)	Z=-29.01	<0.001
SAPS II, score, M (Q₁, Q₃)	37.00 (30.00, 47.00)	37.00 (29.00, 46.00)	49.00 (39.00, 61.00)	Z=-26.20	<0.001
Heart Rate, bpm, M (Q₁, Q₃)	84.62 (75.39, 96.26)	84.42 (75.33, 95.92)	87.68 (76.44, 102.18)	Z=-5.71	<0.001
SBP, mmHg, M (Q₁, Q₃)	113.54 (105.64, 124.04)	113.79 (105.90, 124.17)	110.08 (102.32, 122.32)	Z=-7.72	<0.001
DBP, mmHg, M (Q₁, Q₃)	60.20 (54.45, 66.95)	60.23 (54.52, 66.85)	59.56 (53.03, 68.30)	Z=1.12	0.262
Respiratory rate, insp/min, M (Q₁, Q₃)	18.80 (16.65, 21.73)	18.68 (16.58, 21.54)	20.78 (18.21, 24.32)	Z=-16.17	<0.001
Temperature, ℃, M (Q₁, Q₃)	36.87 (36.59, 37.23)	36.87 (36.60, 37.23)	36.76 (36.32, 37.20)	Z=8.40	<0.001
Diabetes with complication, n(%)				χ²=0.18	0.675
NO	15636 (90.77)	14620 (90.75)	1016 (91.12)
YES	1590 (9.23)	1491 (9.25)	99 (8.88)
Mechanical ventilation, n(%)				χ²=47.05	<0.001
NO	5748 (37.51)	5454 (38.25)	294 (27.68)
YES	9574 (62.49)	8806 (61.75)	768 (72.32)

M: Median; Q₁: 1st Quartile; Q₃: 3st Quartile

Z: Mann,Whitney test; χ²: Chi,square test

Abbreviation: BMI: Body Mass Index; SOFA: Sequential Organ Failure Assessment; CCI: Charlson comorbidity index; GCS: Glasgow Coma Scale; APS III: Acute Physiology Score III; LODS: Logistic Organ Dysfunction Score; SAPS II: Simplified acute physiology score II; SBP: systolic blood pressure; DBP: diastolic blood pressure;

3.2 The effectiveness of five scoring systems in predicting ICU mortality

The ROC curve demonstrated that both APS III (AUC: 0.756, P<0.001) and LODS (AUC: 0.758, P<0.001) score exhibited robust predictive value for ICU mortality. The AUC (0.734, P<0.001) of the SAPS II score exhibited a slightly lower value compared to that of the APS III and LODS scores (Figure.2).

Figure 2: Prediction of mortality in septic ICU patients using multiple scoring systems.

SOFA: Sequential Organ Failure Assessment; CCI: Charlson comorbidity index; APS III: Acute Physiology Score III; LODS: Logistic Organ Dysfunction Score; SAPS II: Simplified acute physiology score II; AUC：Area under the receiver operating characteristic curve; CI: confdence interval.

3.3 Regressional analysis to identify potential factors

The potential predictors identified through univariate and multivariate regression analyses, with ICU mortality as the outcome, were presented in Table 2. Multivariate regression analysis revealed that CCI (OR=1.10; 95% CI:(1.06, 1.14); P<0.001), APS III score (OR=1.01; 95% CI:(1.01, 1.02); P<0.001), LODS score (OR=1.14; 95% CI:(1.09, 1.20); P<0.001), respiratory rate (OR=1.07; 95% CI:(1.05, 1.10); P<0.001), body temperature (OR=0.75; 95% CI:(0.67, 0.85); P<0.001), and unknown race (OR=2.82; 95% CI:(2.27, 3.49); P<0.001) served as potential predictors. The integration of these indicators into a nomogram demonstrated significant clinical value (Figure.3).

Table 2: Screening of covariates as potential risk factors

Variables	Univariate analysis			Multivariate analysis
Variables	β	OR (95%CI)	P	β	OR (95%CI)	P
Age, Years, M (Q₁, Q₃)	0.01	1.01 (1.01, 1.01)	<0.001
BMI, kg/m2, M (Q₁, Q₃)	-0.01	0.99 (0.98, 1.00)	0.138
SOFA, score, M (Q₁, Q₃)	0.18	1.20 (1.16, 1.24)	<0.001
CCI, score, M (Q₁, Q₃)	0.11	1.12 (1.09, 1.15)	<0.001	0.09	1.10 (1.06, 1.14)	<0.001
GCS, score, M (Q₁, Q₃)	0.01	1.01 (0.97, 1.04)	0.739
APS III, score, M (Q₁, Q₃)	0.04	1.04 (1.03, 1.04)	<0.001	0.01	1.01 (1.01, 1.02)	<0.001
LODS, score, M (Q₁, Q₃)	0.29	1.33 (1.30, 1.36)	<0.001	0.13	1.14 (1.09, 1.20)	<0.001
SAPS II, score, M (Q₁, Q₃)	0.06	1.06 (1.05, 1.06)	<0.001
Heart Rate, bpm, M (Q₁, Q₃)	0.01	1.01 (1.01, 1.02)	<0.001
SBP, mmHg, M (Q₁, Q₃)	-0.01	0.99 (0.98, 0.99)	<0.001	0.01	1.01 (1.00, 1.01)	0.102
DBP, mmHg, M (Q₁, Q₃)	-0.00	1.00 (0.99, 1.00)	0.481
Respiratory rate, insp/min, M (Q₁, Q₃)	0.11	1.12 (1.10, 1.13)	<0.001	0.07	1.07 (1.05, 1.10)	<0.001
Temperature, ℃, M (Q₁, Q₃)	-0.55	0.58 (0.52, 0.64)	<0.001	-0.28	0.75 (0.67, 0.85)	<0.001
Creatinine, mg/dL, M(Q₁, Q₃)	0.10	1.11 (1.07, 1.14)	<0.001
Gender
Male	Reference
Female	0.14	1.14 (0.99, 1.33)	0.071
Race
White	Reference			Reference
Unknown	1.06	2.90 (2.45, 3.42)	<.001	1.04	2.82 (2.27, 3.49)	<0.001
Hispanic / Latino	0.03	1.03 (0.65, 1.64)	0.891	0.21	1.24 (0.70, 2.18)	0.460
Black	0.13	1.14 (0.85, 1.53)	0.395	0.10	1.11 (0.75, 1.63)	0.613
Other	-0.11	0.90 (0.59, 1.36)	0.615	-0.12	0.89 (0.52, 1.51)	0.668
Asian	0.09	1.09 (0.65, 1.82)	0.746	0.04	1.04 (0.51, 2.14)	0.907
Diabetes with complication
NO	Reference
YES	-0.06	0.94 (0.73, 1.21)	0.634
Mechanical ventilation
YES	Reference
NO	-0.49	0.61 (0.52, 0.72)	<0.001

OR: odds ratio, CI: confdence interval.

Abbreviation: BMI: Body Mass Index; SOFA: Sequential Organ Failure Assessment; CCI: Charlson comorbidity index; GCS: Glasgow Coma Scale; APS III: Acute Physiology Score III; LODS: Logistic Organ Dysfunction Score; SAPS II: Simplified acute physiology score II; SBP: systolic blood pressure; DBP: diastolic blood pressure;

Figure 3. Nomogram for the prediction of death in the ICU

CCI: Charlson comorbidity index; APS III: Acute Physiology Score III; LODS: Logistic Organ Dysfunction Score;

3.4 Predictive performance assessed in training and validation sets.

The training set comprised 12,058 patients, while the validation set included 5,168 patients. Supplementary Table 1 presents a comparison of the characteristics between these two sets. The clinical data revealed no statistically significant disparities between the randomly allocated training set and the validation set.

The ROC curve demonstrated performance in both the training and validation datasets, with respective areas under the curve of 0.803 and 0.797. The training set demonstrated predictive sensitivities and specificities of 76.1% and 70.4%, respectively. In the validation set, corresponding sensitivities were observed to be 75.5% and 71.6% (Figure.4).

Figure 4. The performance of the prediction model in both the training and validation sets. A: Evaluation of effectiveness in the training set; B: Assessment of performance in the validation set.

The accurate prediction of mortality risk in sepsis has become imperative for optimizing patient survival, given its emergence as a predominant and lethal condition within the intensive care the ICU [25]. The objective of this study was to assess the efficacy of CCI, LODS, APS III, SAPS II, and SOFA scoring systems in predicting mortality among sepsis patients during their stay in the ICU. Retrospective analysis was conducted on clinical data from MIMIC-IV datebase to compare the predictive accuracy of these scoring systems. The findings of this investigation revealed variations in terms of accuracy among five scoring systems, indicating their prognostic capability for mortality prediction in septic patients. Notably, individual scoring systems such as APS III and LODS demonstrated significant predictive value. However, it is worth noting that a combination comprising CCI, APS III score, LODS score, average respiratory rate and body temperature within 24 hours after ICU admission along with race could further improve the accuracy of mortality risk predictions.

The APS III scoring system incorporates a range of physiological parameters, including blood pressure, body temperature, heart rate, respiratory rate, blood oxygen saturation, blood pH, serum electrolyte levels, and renal function indicators [26]. Each parameter is assigned a distinct score based on its correlation with the risk of mortality. By evaluating the APS III score of patients admitted to the ICU, physicians can promptly determine the severity of their condition and facilitate prognostication as well as evaluation of treatment response [27]. Additionally, the APS III score can serve as a valuable tool for assessing the effectiveness of various therapeutic interventions and guiding adjustments to clinical management plans [28]. The study also yielded consistent findings, demonstrating a remarkable predictive efficacy of a single APS III score in sepsis patients for ICU mortality with an AUC of 0.756.

The LODS scoring system encompasses six major organ systems, namely the respiratory, circulatory, hepatic, coagulation, nervous, and renal systems [29]. Each organ system's dysfunction is accompanied by a corresponding scoring criterion. By measuring and evaluating the patient's physiological parameters, the cumulative score can be computed to assess the extent of organ dysfunction [30]. The LODS scoring system has been shown to possess exceptional precision and reliability in predicting mortality risk among patients in the ICU, particularly those suffering from sepsis [31]. Similarly, in the present study, the prognostic efficacy of LODS on mortality in ICU septic patients was also substantial and superior to that of other scoring systems. When evaluating mortality in sepsis patients, the area under the curve (AUC) of LODS score for predicting both 28-day and 90-day prognosis outperformed that of both SOFA and qSOFA scores [32]. Moreover, the admission of sepsis patients to the ICU is associated with an increased risk of in-hospital mortality, as evidenced by a higher LODS score [6].

The CCI scoring system consists of 19 major comorbid conditions, each assigned a weight ranging from 1 to 6 points based on its impact on the risk of mortality [33]. A higher cumulative score for a patient indicates an increased probability of death within the initial year. The CCI scoring system demonstrates certain predictive value in assessing mortality risk among patients [34]; The performance of CCI, however, was not satisfactory when solely evaluating mortality in septic patients during their stay in the ICU, aligning with findings from other studies [35]. The incorporation of CCI with other scoring systems enhances its efficacy in predicting both short-term and long-term mortality [35].

However, in this study, we did not observe a significant association between SAPS II and ICU mortality in septic patients. This may be attributed to the limited inclusion of comprehensive prognostic factors, such as immune and nutritional status of patients, within the current scoring system. Furthermore, potential changes in population demographics and treatment techniques could potentially compromise the predictive accuracy of the SAPS II scoring system, highlighting the necessity for periodic calibration and updates. Additionally, numerous studies have demonstrated that the SAPS II score exhibits a superior predictive value for mortality in ICU patients with non-septic shock compared to LODS [36, 37]. The observed discrepancy in this study may be attributed to variations in the included covariates and the scoring system employed, which incorporates APS III and CCI. Zhu et al. employed six scoring systems to predict 28-day mortality in both general and specialized care units, with varying applicability across different types of ICU. For example, APS III or SAPS II can be utilized for predicting 28-day mortality in general ICUs, while the use of APS III or LODS is more effective in cardiovascular ICU for forecasting the risk of 28-day mortality [18]. The accuracy of predicting mortality risk in ICU patients with sepsis further improved by integrating multiple scoring systems with the physiological status of the patients in this study.

This study systematically compared and analyzed the value and efficacy of CCI, LODS, APS III, SAPS II, and SOFA scoring systems in predicting sepsis patient mortality within ICU settings. The findings demonstrate that although these scoring systems possess their individual strengths in prognosticating death risks among sepsis patients. Moreover, it is worth noting that a combination comprising CCI, APS III score, LODS score, average respiratory rate and body temperature within 24 hours after ICU admission along with race could further improve the accuracy of mortality risk predictions. The utilization of scoring systems not only facilitates physicians in making more precise risk assessments, but also provides guidance for clinical decision-making and resource allocation. For patients at high risk, physicians can potentially employ more aggressive treatment and intervention based on the scoring system results, thereby potentially reducing the patient's mortality risk.

This study has certain limitations. Firstly, the retrospective data from a single center may limit the generalizability of the findings. Future studies should aim to validate these results in a broader setting and with a larger sample size to enhance their reliability and applicability. Additionally, it is important to consider that variations in adaptability and sensitivity among different scoring systems may exist due to differences in their original design and scoring elements. Furthermore, this study did not extensively explore the performance of these scoring systems in various sepsis subtypes or clinical scenarios. Therefore, future research could focus on comprehensively applying these scoring systems or developing new predictive models to further optimize risk assessment and management for patients with sepsis.

Organ failure scores hold significant clinical relevance in predicting mortality among sepsis patients in the ICU. Augmenting predictive accuracy can be achieved by integrating the CCI, APS III score, LODS score, mean respiratory rate, and body temperature during the initial 24 hours following ICU admission.

Author contributions

Xuan Zhou was responsible for the initial data extraction and preprocessing; Zhenen Zhang developed the algorithms for pattern recognition and anomaly detection within the medical data; Huiming Wang led the statistical analysis and interpretation of results; Pengfei Chen facilitated the integration of findings with existing medical knowledge, formulated the outline, and authored the paper.

Declaration of conflicting interests

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding

No funding source was identified for this study.

Acknowledgements

Not Applicable

Data availability statement.

All datasets used during the present study are publicly available in the MIMIC-IV v1.0 database (https://mimic. physionet.org).

Ethical Approval

This study is an analysis of a public database. Approval from the Institutional Review Board was not required.

Supplementary Material

Supplementary Table1: The comparative analysis between the training set and validation set.

de Nooijer AH, Kotsaki A, Kranidioti E, Kox M, Pickkers P, Toonen EJM, Giamarellos-Bourboulis EJ, Netea MG. Complement activation in severely ill patients with sepsis: no relationship with inflammation and disease severity. CRIT CARE. 2023-02-16; 27 (1): 63. 10.1186/s13054-023-04344-6. PMID: 36797757.
Fleischmann-Struzek C, Mikolajetz A, Schwarzkopf D, Cohen J, Hartog CS, Pletz M et al. Challenges in assessing the burden of sepsis and understanding the inequalities of sepsis outcomes between National Health Systems: secular trends in sepsis and infection incidence and mortality in Germany. INTENS CARE MED. 2018-11-01; 44 (11): 1826–1835. 10.1007/s00134-018-5377-4. PMID: 30284637.
Thibaud P, Chow-Chine L, Gonzalez F, Bisbal M, Servan L, Sannini A et al. Septic shock and biliary sepsis: 90-day mortality and associated risk factors. HPB. 2024-02-01; 26 (2): 270–281. 10.1016/j.hpb.2023.10.012. PMID: 37940408.
Prescott HC, Harrison DA, Rowan KM, Shankar-Hari M, Wunsch H. Temporal Trends in Mortality of Critically Ill Patients with Sepsis in the United Kingdom, 1988–2019. AM J RESP CRIT CARE. 2024-03-01; 209 (5): 507–516. 10.1164/rccm.202309-1636OC. PMID: 38259190.
Kuklin V, Sovershaev M, Bjerner J, Keith P, Scott LK, Thomas OMT et al. Influence of therapeutic plasma exchange treatment on short-term mortality of critically ill adult patients with sepsis-induced organ dysfunction: a systematic review and meta-analysis. CRIT CARE. 2024-01-04; 28 (1): 12. 10.1186/s13054-023-04795-x. PMID: 38178170.
Blanco J, Muriel-Bombín A, Sagredo V, Taboada F, Gandía F, Tamayo L et al. Incidence, organ dysfunction and mortality in severe sepsis: a Spanish multicentre study. CRIT CARE. 2008-01-01; 12 (6): R158. 10.1186/cc7157. PMID: 19091069.
Liu S, He C, He W, Jiang T. Lactate-enhanced-qSOFA (LqSOFA) score is superior to the other four rapid scoring tools in predicting in-hospital mortality rate of the sepsis patients. ANN TRANSL MED. 2020-08-01; 8 (16): 1013. 10.21037/atm-20-5410. PMID: 32953813.
Derin O, Şahin M, Dumlu R, Başgönül S, Bayrak AD, Arduç Ş et al. Registry-Based Retrospective Cohort Study of Mortality among Adults Admitted to Intensive Care Units in Istanbul with Hospital Acquired Pseudomonas aeruginosa Bloodstream-Infection between 2014–2021. Antibiotics (Basel). 2024-01-17; 13 (1): 10.3390/antibiotics13010090. PMID: 38247649.
Torvik MA, Nymo SH, Nymo SH, Bjørnsen LP, Kvarenes HW, Ofstad EH. Patient characteristics in sepsis-related deaths: prevalence of advanced frailty, comorbidity, and age in a Norwegian hospital trust. INFECTION. 2023-08-01; 51 (4): 1103–1115. 10.1007/s15010-023-02013-y. PMID: 36894755.
Cui L, Bao J, Yu C, Zhang C, Huang R, Liu L et al. Development of a nomogram for predicting 90-day mortality in patients with sepsis-associated liver injury. Sci Rep. 2023-03-04; 13 (1): 3662. 10.1038/s41598-023-30235-5. PMID: 36871054.
Yang J, Peng H, Luo Y, Zhu T, Xie L. Explainable ensemble machine learning model for prediction of 28-day mortality risk in patients with sepsis-associated acute kidney injury. Front Med (Lausanne). 2023-01-01; 10 1165129. 10.3389/fmed.2023.1165129. PMID: 37275353.
Widyastuti Y, Jufan A, Widodo U, Retno Wisudarti C, Fauzi S, Ardiansyah R. F. A tertiary care center-based study of a novel ‘ICU Mortality and Prolonged Stay Risk Scoring System’ ANAESTH PAIN INTENSI. 2024-02-04; 28 (1): 100–7. 10.35975/apic.v28i1.2382
Boulet N, Boussere A, Mezzarobba M, Sofonea MT, Payen D, Lipman J et al. T. Intensive Care Unit activity in France from the national database between 2013 and 2019: More critically ill patients, shorter stay and lower mortality rate. Anaesth Crit Care Pain Med. 2023-10-01; 42 (5): 101228. 10.1016/j.accpm.2023.101228. PMID: 37031815.
Park SW, Yeo NY, Kang S, Ha T, Kim TH, Lee D et al. Early Prediction of Mortality for Septic Patients Visiting Emergency Room Based on Explainable Machine Learning: A Real-World Multicenter Study. J KOREAN MED SCI. 2024-02-05; 39 (5): e53. 10.3346/jkms.2024.39.e53. PMID: 38317451.
Serafim R, Gomes JA, Salluh J, Póvoa PA. Comparison of the Quick-SOFA and Systemic Inflammatory Response Syndrome Criteria for the Diagnosis of Sepsis and Prediction of Mortality: A Systematic Review and Meta-Analysis. CHEST. 2018-03-01; 153 (3): 646–655. 10.1016/j.chest.2017.12.015. PMID: 29289687.
Khwannimit B, Bhurayanontachai R, Vattanavanit V. Comparison of the performance of SOFA, qSOFA and SIRS for predicting mortality and organ failure among sepsis patients admitted to the intensive care unit in a middle-income country. J CRIT CARE. 2018-04-01; 44 156–160. 10.1016/j.jcrc.2017.10.023. PMID: 29127841.
Finkelsztein EJ, Jones DS, Ma KC, Pabón MA, Delgado T, Nakahira K et al. Siempos, II. Comparison of qSOFA and SIRS for predicting adverse outcomes of patients with suspicion of sepsis outside the intensive care unit. CRIT CARE. 2017-03-26; 21 (1): 73. 10.1186/s13054-017-1658-5. PMID: 28342442.
Zhu S, Chen H, Li X, Gulifeire T, Wang Y, Yang C et al. Predictive value of six critical illness scores for 28-day death risk in comprehensive and specialized intensive care unit patients based on MIMIC-IV database. Zhonghua Wei Zhong Bing Ji Jiu Yi Xue. 2022-07-01; 34 (7): 752–758. 10.3760/cma.j.cn121430-20220304-00205. PMID: 36100416.
Onur İD, Mutlu E, Sertesen E, Önder T, Duran AO, İnanç M. Evaluating the effectiveness of the Charlson Comorbidity Index in predicting immune checkpoint inhibitor-related adverse events. IMMUNOTHERAPY-UK. 2024-03-01; 16 (5): 295–303. 10.2217/imt-2023-0270. PMID: 38288692.
Pang K, Li L, Ouyang W, Liu X, Tang Y. Establishment of ICU Mortality Risk Prediction Models with Machine Learning Algorithm Using MIMIC-IV Database. Diagnostics (Basel). 2022-04-24; 12 (5): 10.3390/diagnostics12051068. PMID: 35626224.
Polok K, Fronczek J, Putowski Z, Czok M, Guidet B, Jung C et al. Validity of the total SOFA score in patients ≥ 80 years old acutely admitted to intensive care units: a post-hoc analysis of the VIP2 prospective, international cohort study. Ann Intensive Care. 2023-10-05; 13 (1): 98. 10.1186/s13613-023-01191-0. PMID: 37798561.
He Y, Xu J, Shang X, Fang X, Gao C, Sun D et al. Clinical characteristics and risk factors associated with ICU-acquired infections in sepsis: A retrospective cohort study. Front Cell Infect Microbiol. 2022-01-01; 12 962470. 10.3389/fcimb.2022.962470. PMID: 35967847.
Yao SL, Chen XW, Liu J, Chen XR, Zhou Y. Effect of mean heart rate on 30-day mortality in ischemic stroke with atrial fibrillation: Data from the MIMIC-IV database. Front Neurol. 2022-01-01; 13 1017849. 10.3389/fneur.2022.1017849. PMID: 36388223.
Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA-J AM MED ASSOC. 2016-02-23; 315 (8): 801 – 10. 10.1001/jama.2016.0287. PMID: 26903338.
Fleischmann-Struzek C, Mellhammar L, Rose N, Cassini A, Rudd KE, Schlattmann P et al. Incidence and mortality of hospital- and ICU-treated sepsis: results from an updated and expanded systematic review and meta-analysis. INTENS CARE MED. 2020-08-01; 46 (8): 1552–1562. 10.1007/s00134-020-06151-x. PMID: 32572531.
Johnson AE, Kramer AA, Clifford GD. A new severity of illness scale using a subset of Acute Physiology And Chronic Health Evaluation data elements shows comparable predictive accuracy. CRIT CARE MED. 2013-07-01; 41 (7): 1711-8. 10.1097/CCM.0b013e31828a24fe. PMID: 23660729.
Arabi Y, Goraj R, McIntyre R, Al-Abdulkareem A. The predictive value of serial acute physiology score (APS) and simplified acute physiology score II (SAPS II) in post-operative liver transplant patients CRIT CARE. 2000-01-01; 4 (Suppl 1): P240. 10.1186/cc959
Haq A, Patil S, Parcells AL, Chamberlain RS. The Simplified Acute Physiology Score III Is Superior to the Simplified Acute Physiology Score II and Acute Physiology and Chronic Health Evaluation II in Predicting Surgical and ICU Mortality in the Oldest Old. Curr Gerontol Geriatr Res. 2014-01-01; 2014 934852. 10.1155/2014/934852. PMID: 24696680.
Le Gall JR, Klar J, Lemeshow S, Saulnier F, Alberti C, Artigas A et al. The Logistic Organ Dysfunction system. A new way to assess organ dysfunction in the intensive care unit. ICU Scoring Group. JAMA-J AM MED ASSOC. 1996-09-11; 276 (10): 802 – 10. 10.1001/jama.276.10.802. PMID: 8769590.
Jang HN, Park HJ, Cho HS, Bae E, Lee TW, Chang SH et al. The logistic organ dysfunction system score predicts the prognosis of patients with alcoholic ketoacidosis. RENAL FAILURE. 2018-11-01; 40 (1): 693–699. doi: 10.1080/0886022X.2018.1491405. PMID: 30741615.
Yousefian M, Ghazi A, Amani F, Movaffagh B. Mortality rate in patients admitted to the ICU based on LODS, APACHE IV, TRIOS, SAPS II.‎J Adv Pharm Educ Res. 2022-01-01; 12 (1): 56–62. 10.51847/cghzosuz9x
Li Y, Yan C, Gan Z, Xi X, Tan Z, Li J et al. Prognostic values of SOFA score, qSOFA score, and LODS score for patients with sepsis. ANN PALLIAT MED. 2020-05-01; 9 (3): 1037–1044. 10.21037/apm-20-984. PMID: 32498525.
Tang PL, Lin HS, Hsu CJ. Predicting in-hospital mortality for dementia patients after hip fracture surgery - A comparison between the Charlson Comorbidity Index (CCI) and the Elixhauser Comorbidity Index. J ORTHOP SCI. 2021-05-01; 26 (3): 396–402. 10.1016/j.jos.2020.04.005. PMID: 32482586.
Yurkovich M, Avina-Zubieta JA, Thomas J, Gorenchtein M, Lacaille D. A systematic review identifies valid comorbidity indices derived from administrative health data. J CLIN EPIDEMIOL. 2015-01-01; 68 (1): 3–14. 10.1016/j.jclinepi.2014.09.010. PMID: 25441702.
Christensen S, Johansen MB, Christiansen CF, Jensen R, Lemeshow S. Comparison of Charlson comorbidity index with SAPS and APACHE scores for prediction of mortality following intensive care. Clin Epidemiol. 2011-01-01; 3 203 – 11. 10.2147/CLEP.S20247. PMID: 21750629.
Hu C, Hu B, Li Z, Yang X, Song H, Li J. [Comparison of four scoring systems for predicting ICU mortality in patients with sepsis]. Nan Fang Yi Ke Da Xue Xue Bao. 2020-04-30; 40 (4): 513–518. 10.12122/j.issn.1673-4254.2020.04.10. PMID: 32895135.
Zhu Y, Zhang R, Ye X, Liu H, Wei. J. SAPS III is superior to SOFA for predicting 28-day mortality in sepsis patients based on Sepsis 3.0 criteria. INT J INFECT DIS. 2022-01-01; 114 135–141. doi: 10.1016/j.ijid.2021.11.015. PMID: 34775116.

No competing interests reported.

SupplementaryMaterial.docx

Multiple Organ Scoring Systems for Predicting In-Hospital Mortality of Sepsis Patients in the Intensive Care Unit

Status:

Version 1

Abstract

Objective

Methods

Results

Conclusions

Figures

1. Introduction

2. Methods and Methods

2.1 Study design

2.2 Date extraction

2.3 Statistical analysis

3. Results

4. Discussion

5. Conclusion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1