Sepsis prognosis in the emergency department: assessment of shock indices, SOFA, qSOFA, and lactate

Abstract

Background

The aim of this study was to compare the effectiveness of the shock index (SI), age shock index (ASI), modified shock index (MSI), and reverse shock index (RSI) in predicting prognosis in patients presenting to the emergency department with sepsis and septic shock with SOFA, qSOFA, and lactate levels.

Methods

This retrospective observational study included 203 patients (139 admitted to the ward, 64 admitted to the intensive care unit) who presented to the emergency medicine clinic of a city hospital with a diagnosis of sepsis or septic shock between 1 June 2020 and 1 June 2022. Patients’ ward admission, intensive care admission, and one-month mortality outcomes were evaluated. The diagnostic performance of SI, ASI, MSI, RSI, SOFA, qSOFA, and lactate in predicting ICU admission, ward admission, and one-month mortality was assessed using age- and sex-adjusted univariate logistic regression analysis and ROC curve analysis (AUC).

Results

SOFA (AUC = 0.69, p < .001), lactate (AUC = 0.63, p = .001), and SI (AUC = 0.67, p < .001) were found to be significant in predicting ward admission. For intensive care admission, SOFA (AUC = 0.66, p = .001) and lactate (AUC = 0.63, p = .004) showed predictive value. Lactate was identified as the strongest predictor of one-month mortality (AUC = 0.74, p < .001), followed by SOFA (AUC = 0.68, p = .003) and qSOFA (AUC = 0.70, p = .033).

Conclusion

Lactate level and SOFA score were found to be the most reliable prognostic markers in sepsis patients. Among shock indices, SI showed significant value in predicting hospital admission, while other shock indices had limited predictive value.

Introduction

Sepsis is defined as a life-threatening organ dysfunction resulting from the body’s disordered response to infection and is a significant cause of morbidity and mortality worldwide [1, 2]. Early diagnosis and prompt treatment are the cornerstones of sepsis management and directly affect patient survival rates [3]. Therefore, there is a great need for simple and reliable tools that can quickly and accurately classify patient risk and predict prognosis in units where critical decisions are made, such as the emergency department (ED).

To this end, scoring systems such as SOFA (Sequential Organ Failure Assessment) and its simplified version, qSOFA (quick SOFA), have been developed. While SOFA has become a standard tool for assessing organ damage in intensive care units (ICUs), qSOFA has been recommended for rapidly identifying at-risk patients in the ED due to its ease of application at the bedside [4, 5]. Serum lactate level, on the other hand, is a biomarker that has been used for many years to determine prognosis in sepsis and septic shock as an indicator of tissue hypoperfusion [6].

In addition to these established markers, the shock index (SI) and its derivatives have also gained attention as indicators of hemodynamic instability. First defined in 1967 by Allgower and Burri, the SI (heart rate/systolic blood pressure) has been widely used in conditions such as trauma and hypovolemic shock due to its ease of calculation and lack of cost [7, 8]. Over time, various variations have been developed, such as the Age-Adjusted Shock Index (ASI), which accounts for the effect of age on the hemodynamic response, the Modified Shock Index (MSI), which uses mean arterial pressure, and the Reverse Shock Index (RSI), which reverses the formula [9, 10]. Although studies have been conducted on the prognostic value of these indices in sepsis patients, the results are variable, and the comparative effectiveness of these indices with standard markers such as SOFA, qSOFA, and lactate remains a subject of research [11, 12].

The originality of this study lies in evaluating multiple clinical outcomes (ward admission, ICU admission, and one-month mortality) within the same cohort using various shock indices compared with established markers (SOFA, qSOFA, and serum lactate). This comparative assessment across multiple outcomes has not been extensively reported in previous literature.

We hypothesized that shock indices (SI, ASI, MSI, RSI) would demonstrate comparable or superior prognostic performance to established markers (SOFA, qSOFA, and lactate) in predicting clinical outcomes in sepsis patients.

The aim of this study is to compare the prognostic value of various shock indices with traditional scoring systems such as SOFA, qSOFA, and lactate levels in patients presenting to the emergency department with sepsis and septic shock.

Materials and methods

This retrospective, observational study was conducted using data from patients diagnosed with sepsis or septic shock who were admitted to and followed up at the Emergency Medicine Clinic of Basaksehir Çam and Sakura City Hospital between 1 June 2020 and 1 June 2022.

Ethics committee approval was obtained for the study.

Patients over 18 years of age who had been diagnosed with sepsis or septic shock and whose data were complete were included in the study. Diagnosis of sepsis according to Sepsis-3 criteria: suspected or documented infection plus acute increase in SOFA score ≥ 2 points.

Septic shock defined as: sepsis with requirement for vasopressors to maintain mean arterial pressure (MAP) ≥ 65 mmHg AND serum lactate > 2 mmol/L despite adequate fluid resuscitation [1]. Patients under 18 years of age and those with incomplete data were excluded from the study.

Demographic data (age, gender), vital signs (heart rate, systolic and diastolic blood pressure, respiratory rate, body temperature), Glasgow Coma Scale (GCS) and laboratory parameters (lactate, white blood cell count, bilirubin, creatinine, PaO₂, FiO₂, platelets) were collected retrospectively via the hospital information management system (HIMS). The following indices were calculated:

• Shock Index (SI) = Heart rate / Systolic blood pressure.

• Age-Adjusted Shock Index (ASI) = SI × Age.

• Modified Shock Index (MSI) = Heart rate / Mean arterial pressure.

• Reverse Shock Index (RSI) = Systolic blood pressure / Heart rate.

• SOFA and qSOFA scores were also calculated using the relevant parameters.

Statistical analysis

All analyses were performed using Jamovi v2.6.44.0 (The Jamovi Project, 2023) software. Continuous variables were summarized as mean ± standard deviation or median (Q1–Q3) according to their distribution characteristics; categorical variables were summarized as number (n) and percentage (%). The normality of continuous variables was assessed using the Shapiro–Wilk test and graphical methods (histogram, Q–Q plot).

For intergroup comparisons, Student’s t-test was used when assumptions were met, otherwise the Mann–Whitney U test; for categorical variables, Pearson’s chi-square test (with Yates’ correction when necessary) was used. All p-values were two-tailed, and the level of statistical significance was set at α = 0.05.

In the logistic regression analyses, each variable was evaluated in a separate model adjusted for age and sex. The discriminatory performance of each model was assessed using receiver operating characteristic (ROC) curve analysis, and the area under the curve (AUC) was calculated. ROC analyses were performed separately for each of the three outcomes (ward admission, intensive care unit admission, and 1-month mortality) to evaluate the predictive performance of each variable. In addition, for each outcome, the AUC values of predictors that were found to be statistically significant in the ROC analysis were compared using the DeLong test to determine whether there were significant differences in discriminatory performance between them. As no correction for multiple testing was applied, some statistically significant findings should be interpreted with caution because of the potential risk of type I error.

Results

Clinical variables were compared between the intensive care and non-intensive care groups using the Mann–Whitney U test. Findings indicated that Heart rate (HR) values were significantly higher in the intensive care group (U = 3574.0, p =.025, r = .20). Furthermore, the Glasgow Coma Scale (GCS) was found to be significantly lower in the intensive care group (U = 3834.5, p <.001, r = − .14). No significant differences were observed between the two groups for other variables (p >.05) (Table 1).

Table 1.

Comparison of clinical variables between patients admitted to and not admitted to the intensive care unit

Variable	No Admission (n = 139)	Admitted (n = 64)
	Med. (Q1-Q3) / Mean ± SD / n (%)	Med. (Q1-Q3) / Mean ± SD / n (%)	Statistics	p	Effect
Age (years)	68 (57–78)	69.50 (58.00–77.25)	4417.5a	0.938	0.01
HR (beats/min)	96 (82.5–108.5)	103.5 (87.75–139.25)	3574.0a	0.025	0.20
SBP (mmHg)	121 (105–136)	122 (101.75–139.25)	4347.5a	0.797	0.02
DBP (mmHg)	71 (62–79)	73.5 (60.5–85.25)	4145.5a	0.437	0.07
GCS	15 (15–15)	15 (15–15)	3834.5a	< 0.001	–0.14
Body temperature (°C)	36.0 (36.0–36.4)	36.0 (36.0–36.5)	4241.5a	0.623	0.04
Respiratory rate (per minute)	20 (20–22)	20 (20–22)	4327.0a	0.744	–0.03
WBC (10³/µL)	11.8 (8.5–15.3)	11.3 (7.0–15.4)	4104.0a	0.477	–0.06
Bilirubin (mg/dL)	0.42 (0.27–1.00)	0.71 (0.29–1.00)	3962.5a	0.272	0.10
Creatinine (mg/dL)	0.98 (0.72–1.44)	1.18 (0.82–1.62)	3746.0a	0.071	0.16
PaO₂ (mmHg)	90 (82–95)	90 (85–94)	4299.0a	0.702	–0.03
FiO₂	0.21 (0.21–0.22)	0.21 (0.21–0.23)	4179.5a	0.438	0.06
PLT	224.826 ± 56.190	208,434 ± 60,807	1.88b	0.061	0.28
Gender
Male	80 (57.6)	37 (57.8)	0.00c	1.000	0.00
Female	59 (42.4)	27 (42.2)

Med: Median, Q1-Q3: Quartile values; HR: Heart rate, SBP: Systolic blood pressure, DBP: Diastolic blood pressure, GCS: Glasgow Coma Scale, WBC: White blood cell count, ^a: Mann Whitney U, ^b: Student t-test, ^c: χ² (Continuity correction)

Effect sizes; rank biserial correlation coefficient for Mann Whitney U, Cohen d coefficient for Student t test, phi-coefficient for 2 × 2 table

In univariate logistic regression analyses adjusted for age and gender, qSOFA (B = − 0.54, OR = 0.58, 95% CI = 0.35–0.97, p =.037) was found to be significantly associated with a reduced likelihood of admission to the service. Similarly, SOFA (B = − 0.35, OR = 0.71, 95% CI = 0.59–0.85, p <.001) and lactate level (B = − 0.34, OR = 0.72, 95% CI = 0.59–0.87, p =.001) significantly reduced the likelihood of admission to the service.

In contrast, high RSI values increased the risk of hospitalization (B = 1.32, OR = 3.75, 95% CI = 1.30–10.81, p =.014). MSI (B = − 1.45, OR = 0.23, 95% CI = 0.08–0.66, p =.006) and SI (B = − 2.27, OR = 0.10, 95% CI = 0.02–0.40, p <.001) significantly reduced the likelihood of hospitalization with their low values. ASI was also statistically significant but its effect was relatively small (B = − 0.03, OR = 0.97, 95% CI = 0.95–0.99, p =.004).

The AUC values indicating the discriminatory power of the models ranged from 0.58 to 0.69, with the highest AUC value observed for SOFA (AUC = 0.69). The Nagelkerke R² values ranged from 4% to 12.

According to the DeLong test conducted to compare the models, a significant difference was only detected between three models. There was a significant difference between SOFA and qSOFA (AUC difference = 0.11; p =.032), SI and ASI (AUC difference = 0.03; p = .028), and SI and RSI (AUC difference = 0.02; p =.015) models in terms of their ability to predict hospitalization (Table 2), (Fig. 1).

Table 2.

Adjusted univariate logistic regression models for predicting hospitalization by age and gender

Variables	Cut-Off	B	Z	p	OR	%95 CI	AUC	R 2 N
qSOFA	≤ 1	-0.54	-2.09	0.037	0.58	0.35–0.97	0.58	0.04
SOFA	≤ 1	-0.35	-3.77	< 0.001	0.71	0.59–0.85	0.69	0.12
Lactate	≤ 4	-0.34	-3.27	0.001	0.72	0.59–0.87	0.63	0.09
RSI	≥ 1.10	1.32	2.45	0.014	3.75	1.30-10.81	0.65	0.06
MSI	≤ 1.10	-1.45	-2.75	0.006	0.23	0.08–0.66	0.64	0.06
ASI	≤ 74.9	-0.03	-2.85	0.004	0.97	0.95–0.99	0.64	0.07
SI	≤ 1.10	-2.27	-3.30	< 0.001	0.10	0.02–0.40	0.67	0.09

Each row represents a separate model; age (continuous) and gender (binary) were included as covariates in all models, and the indicator specified in the row (qSOFA, SOFA, Lactate, Shock Index (SI), Age-Adjusted Shock Index (ASI), Modified Shock Index (MSI), Reverse Shock Index (RSI)), was added individually. Event = admission to the service (n = 162) AUC: Area under the curve, CI: Confidence interval, R²_N: Nagelkerke R²

Fig. 1.

ROC curves for predicting hospitalization

Each row represents a separate model; age (continuous) and gender (binary) were included as covariates in all models, and the indicator specified in the row (qSOFA, SOFA, Lactate, RSI, MSI, ASI, SI) was added individually. Event hospitalization (n = 162) AUC: Area under the curve, CI: Confidence interval, R²: Nagelkerke R²

Following univariate logistic regression analyses adjusted for age and gender, the SOFA score was found to be significant in predicting admission to the ward (B = 0.28, OR = 1.32, 95% CI = 1.12–1.56, p = .001), and the model’s discriminatory power was moderate (AUC = 0.66, R²_N = 0.08). Similarly, lactate level also significantly increased the likelihood of admission to the ward (B = 0.29, OR = 1.34, 95% CI = 1.10–1.62, p =.004; AUC = 0.63, R²_N = 0.07).

Other scores (qSOFA, RSI, MSI, ASI, and SI) were not statistically significant in predicting admission to the intensive care unit, and the AUC values of these models ranged from 0.52 to 0.58, with low explanatory power (R²_N ≤ 0.02).

In the DeLong analysis conducted to compare SOFA and lactate models, no statistically significant difference was found (AUC difference = 0.03; p = .715) (Table 3), (Fig. 2).

Table 3.

Adjusted single-variable logistic regression models for predicting admission to intensive care by age and gender

Variables	Cut-Off	B	Z	p	OR	%95 CI	AUC	R 2 N
qSOFA	≥ 2	0.19	0.83	0.200	0.44	0.12–1.55	0.52	0.01
SOFA	≥ 2	0.28	3.27	0.001	1.32	1.12–1.56	0.66	0.08
Lactate	≥ 1.90	0.29	2.91	0.004	1.34	1.10–1.62	0.63	0.07
RSI	≤ 1.3	-0.36	-0.91	0.361	0.70	0.32–1.51	0.55	0.01
MSI	≥ 1.5	0.72	1.56	0.118	2.06	0.83–5.10	0.56	0.02
ASI	≥ 79.9	0.01	1.36	0.173	1.01	0.99–1.03	0.55	0.01
SI	≥ 1.2	1.09	1.85	0.064	2.97	0.94–9.39	0.58	0.02

Each row represents a separate model; age (continuous) and gender (binary) were included as covariates in all models, and the indicator specified in the row (qSOFA, SOFA, Lactate, Shock Index (SI), Age-Adjusted Shock Index (ASI), Modified Shock Index (MSI), Reverse Shock Index (RSI)), was added individually. Event = intensive care admission (n = 64), AUC: Area under the curve, CI: Confidence interval, R²N: Nagelkerke R²

Fig. 2.

ROC curves in predicting admission to intensive care

In univariate logistic regression analyses adjusted for age and gender, qSOFA was found to be significantly associated with mortality (B = 0.82, OR = 2.27, 95% CI = 0.07–4.80, p =.033; AUC = 0.70, R²_N = 0.08). Similarly, a high SOFA score increased the risk of mortality (B = 0.33, OR = 1.40, 95% CI = 1.12–1.74, p =.003; AUC = 0.68, R²_N = 0.13). Lactate level was also identified as a strong predictor and showed a significant increase in mortality risk (B = 0.44, OR = 1.56, 95% CI = 1.21–2.00, p <.001; AUC = 0.74, R²_N = 0.16).

The other scores (RSI, MSI, ASI, and SI) were not statistically significant, with AUC values for these models ranging from 0.64 to 0.66 and R²_N values remaining at 5–6%.

When the AUC values of the variables that were significantly predictive in the logistic regression analysis were compared using the DeLong test, no statistically significant difference was found between the models in predicting a 1-month mortality (p >.05) (Table 4), (Fig. 3).

Table 4.

Adjusted single-variable logistic regression models for predicting one-month mortality according to age and gender

Variables	Cut-Off	B	Z	p	OR	%95 CI	AUC	R 2 N
qSOFA	≥ 1	0.82	2.14	0.033	2.27	1.07–4.80	0.70	0.08
SOFA	≥ 2	0.33	2.95	0.003	1.40	1.12–1.74	0.68	0.13
Lactate	≥ 1.80	0.44	3.47	< 0.001	1.56	1.21-2.00	0.74	0.16
RSI	≤ 1.9	-0.94	-1.19	0.234	0.39	0.08–1.84	0.64	0.05
MSI	≥ 1.4	1.21	1.63	0.103	3.35	0.78–14.30	0.66	0.06
ASI	≥ 42.9	0.02	1.17	0.241	1.02	0.99–1.05	0.64	0.05
SI	≥ 1.10	1.32	1.35	0.178	3.75	0.55–25.66	0.65	0.05

Each row represents a separate model; age (continuous) and gender (binary) were included as covariates in all models, and the indicator specified in the row (qSOFA, SOFA, Lactate, Shock Index (SI), Age-Adjusted Shock Index (ASI), Modified Shock Index (MSI), Reverse Shock Index (RSI)) was added individually. Event = one-month mortality (n = 15), AUC: Area under the curve, CI: Confidence interval, R²N: Nagelkerke R²

Fig. 3.

ROC curves for predicting 1-month mortality

Discussion

This study compared the predictive performance of various shock indices and standard prognostic markers (SOFA, qSOFA, lactate) in predicting different clinical outcomes in patients presenting to the emergency department with sepsis and septic shock. Our findings indicate that SOFA, qSOFA, and lactate are more consistent and reliable than the shock indices examined in predicting more serious outcomes such as intensive care admission and one-month mortality.

Lactate level is considered one of the most important indicators of tissue hypoperfusion [13]. In our study, lactate had the highest AUC value (0.74) for mortality prediction, which is consistent with recent studies. A study published by Liu and colleagues in 2020 reported that the lactate-enriched qSOFA (LqSOFA) score outperformed the traditional qSOFA [14]. Similarly, a recent study published by Althunayyan and colleagues in 2025 confirmed that lactate is a strong predictor of mortality and ICU length of stay in sepsis patients [15].

These findings provide supportive evidence for the association between lactate levels and sepsis prognosis. However, it should be noted that this study was based on single time-point measurements at emergency department presentation. Recent literature emphasizes the prognostic value of dynamic or serial lactate measurements, which were not assessed in the current study.

The SOFA score has found widespread use in sepsis patients as a comprehensive scoring system that assesses multiple organ dysfunction [16, 17]. In our study, the reported AUC values (0.66–0.68) represent moderate discrimination, and this should be considered when interpreting SOFA’s prognostic utility. The clinical applicability of these findings may be limited by the single center, retrospective design and the use of single time-point measurements qSOFA is a simplified screening tool developed in conjunction with sepsis-3 definitions [18]. In our study, the AUC value of qSOFA for mortality was found to be 0.70, which is similar to the results reported by Askim and colleagues in their 2017 study [4]. However, Zhou and colleagues reported in their 2020 study of community-acquired pneumonia sepsis patients that the prognostic value of qSOFA was limited [19]. These differences suggest that qSOFA may exhibit variable performance depending on the patient population and clinical context. In the current study, the AUC of 0.70 represents moderate, not high, prognostic accuracy.

There are conflicting results in the literature regarding the prognostic value of shock indices in sepsis patients. A study published by Jouffroy et al. in 2024 reported that the modified shock index may be useful in predicting mortality in sepsis patients [20]. However, a study conducted by Lee et al. in 2024 on septic shock patients showed that shock indices were associated with in-hospital mortality in patients with normal left ventricular ejection fraction, but that this relationship was complex [21].

The absence of statistically significant associations between shock indices and mortality in this cohort may be due to the fact that these indices are based on only two vital parameters (heart rate and blood pressure) and therefore cannot fully reflect the multifactorial pathophysiology of sepsis. It is important to note that this finding represents the specific results of this single center study and does not necessarily indicate a universal failure of shock indices across all clinical contexts. Similarly, a study conducted by Yeşiltaş and colleagues in 2023 on COVID-19 patients reported that shock indices have limited prognostic value [22].

In recent years, there has been increasing interest in the age-adjusted shock index (ASI). In a study published by Matsuda and colleagues in 2025, it was demonstrated that the 24-hour dynamic changes in ASI could be valuable in predicting 30-day mortality in patients with septic shock [23]. However, this study evaluated dynamic or longitudinal measurements, whereas the current study relied on single time-point values at emergency department presentation. This methodological distinction should be clearly acknowledged when comparing findings across studies.

In a study conducted by Vakhshoori and colleagues in 2024 on patients with acute ST-elevation myocardial infarction, it was reported that the ASI had a significant association with in-hospital mortality [24]. These findings emphasise the importance of considering the effect of age on the hemodynamic response.

Our findings provide supportive evidence that lactate levels and SOFA scores may be useful markers in the risk stratification of sepsis patients in the emergency department setting. The rapid measurability and prognostic associations of lactate are important for early intervention decisions in the emergency department. The SOFA score provides a more detailed assessment of organ dysfunction. However, these recommendations should be considered in the context of the study’s methodological limitations, particularly the retrospective single center design and limited adjustment variables.

Study limitations

There are some important limitations to consider when interpreting this study.

First, the retrospective and single center design of the study imposes a significant constraint on the generalizability of the findings.

Second, the fact that prognostic markers are based solely on measurements taken at a single point in time at the time of emergency department presentation is another important methodological limitation.

This approach prevents the evaluation of time-dependent (dynamic) changes, such as serial lactate levels or shock index trends, which could better reflect changes in the patient’s clinical condition. Therefore, an instantaneous snapshot of the prognosis is presented, but its course over time is ignored.

In light of these limitations, multicenter and prospectively designed advanced studies are needed to obtain more robust results.

Conclusion

Lactate levels and SOFA scores have shown the most consistent relationships with clinical outcomes in patients with sepsis and septic shock. SI showed a significant relationship with ward admission, while other shock indices showed limited predictive value in this sample.

Emergency department physicians may consider lactate levels and SOFA scores as supportive tools for risk stratification in patients with sepsis.

Author contributions

GA and SPB designed and planned the study. HAO supervised the conduct of the study and the data collection process. GA and AEB managed the data, including quality control. GA, MH, and ASY analysed the data. GA and AK conducted the literature review. GA and SD drafted the manuscript, and all authors contributed significantly to the revision of the manuscript. GA is responsible for the final version of the manuscript.

Funding

No funding support was received for this study.

Data availability

All data generated or analysed during this study are included in this published article.

Declarations

Ethics approval and consent to participate

Approved by the Ethics Committee of Istanbul Basaksehir Çam and Sakura City Hospital with the number KAEK/2023.07.291, Istanbul, Turkey. Due to the retrospective nature of the study, informed consent has been waived and this has been approved by the ethics committee. All procedures were conducted in accordance with the principles of the Declaration of Helsinki.

Patient consent

Not applicable.

Competing interests

The authors declare no competing interests.