Comparison of SOFA Score, SIRS, qSOFA, and qSOFA + L Criteria in the Diagnosis and Prognosis of Sepsis

Objective: Sepsis has been defined as a life-threatening organ dysfunction that develops as a result of impaired host response to infection. This study aimed to investigate sequential organ failure assessment (SOFA) score, systemic inflammatory response syndrome (SIRS), quick SOFA (qSOFA), and qSOFA + lactate criteria (qSOFA+L) in the diagnosis and prognosis of sepsis. Materials and Methods: A retrospective study was performed that included all patients diagnosed with sepsis between January 1, 2013 and December 31, 2017 in Izmir Tepecik Training and Research Hospital Infectious Diseases and Clinical Microbiology Clinic. Results: A total of 976 patients diagnosed with sepsis (mean age 72.5±13.7 years, 52.7% women) over five years were included in this study. Of all patients admitted to the emergency department and diagnosed with sepsis, 37.4% (n=365) were hospitalized and 52.3% (n=191) of these patients died. Emergency department mortality was 12.5% (n=122). The mortality rate was higher in patients with qSOFA and qSOFA+L criteria ≥2 in the emergency department. There was no statistically significant difference in terms of SIRS, qSOFA, or qSOFA+L criteria among patients who died in the hospital. The SOFA score (area under receiver operator characteristic curve, AUC=0.89) was highly discriminative in predicting sepsis. When the SOFA score was>11, its sensitivity and negative predictive values were both 100%. The SOFA score (AUC=0.75 and 0.72, respectively) was also highly discriminative in predicting emergency and in-hospital mortality. When the SOFA score was>11, the sensitivity and specificity of predicting emergency department mortality were 63.5% and 78.8%, respectively. The sensitivity was 65.8% and the specificity was 75.5% when describing in-hospital mortality for SOFA scores>9. Conclusion: The SOFA score was highly sensitive and predictive in the diagnosis of sepsis. The SOFA score had a high discriminative ability to predict emergency and in-hospital mortality.


Introduction
Sepsis is one of the oldest and most difficult syndromes in the history of medicine. Despite the emergence of modern antibiotics, germ theory cannot fully explain the pathogenesis of sepsis,and why so many patients dieafter the causative pathogen has beeneradicated. Therefore, researchers posit that the host response was effective in the pathogenesis of sepsis, not the microbes, and subsequently, a fittingdefinition of sepsis was developed by international consensus [1].
TheThird International Consensus Definitions for Sepsis and Septic Shock in 2016 defined sepsis as life-threatening organ dysfunction that develops as a result ofan impaired host response to infection. Thesequential organ failure assessment (SOFA)is a scoring system usedfor clinical evaluation purposes; a score of 2 or moreis associated with an in-hospital mortality riskover 10%. Septic shock is a subset of sepsis that progresses with deep circulatory, cellular, and metabolic abnormalities with higher mortality risk. Patients with septic shock need vasopressors to achievean average arterial pressure of 65 mm Hg or more, and in the absence of hypovolemia, serum lactate levels exceed 2 mmol/L; the in-hospital mortality rateinthese patients isover40% [2].
The quick SOFA (qSOFA) can be used toinforma sepsis-induced prognosis in adult patients with suspicious infections in nonhospital, emergency, or general hospital conditions. A positive qSOFA score involves having a respiratory rate of 22 or more per minute, altered mental state, or systolic blood pressure of, or below, 100 mm Hg [2]. However, thesystemic inflammatory response syndrome (SIRS) criteria are superior to qSOFAforthe clinical diagnosis of sepsis, and qSOFA is superior to SIRS forpredictinginhospital mortality [3].
In thisstudy, we investigated the predictive capacityof the SOFA score, SIRS, qSOFA, and qSOFA + lactate criteria (qSOFA+L) criteria in the diagnosis and prognosis of sepsis.

Definitions
Infection: An inflammatory response to the invasion of microorganisms into sterile host tissue.
Bacteremia: The presence of live bacteria in the blood, diagnosed byblood culture.

SIRS:
The first attempt to standardize sepsis terminology was made in 1991 at the American Chest Diseases Association/Society of Critical Care Medicine (SCCM) Consensus Conference. Sepsis is conceptually defined as a systemic inflammatory response to the presence of infection [4][5][6]. SIRS is diagnosed by the presence ofat least two of the following four SIRS criteria: 1. body temperature > 38°C or < 36°C, 2. heart rate > 90 beats/min, 3. respiratory rate > 20 breaths/min or PaCO 2 < 32 mm Hg, 4. leukocyte count over12,000mm -3 or under4,000mm -3 or immature band/neutrophil ratio exceeds 10%.
Sepsis-1 (1991): Clinical manifestation of an infection with SIRS is defined as sepsis. were abandoned in this consensus [2].
Septic shock(a subset of sepsis with higher mortality): Patients with septic shock need vasopressors to achieve average arterial pressure of 65 mm Hg or more, and in the absence of hypovolemia, serum lactate levelsexceed2 mmol/L. In-hospital mortality rates amongthese patients are over40% [2].

SOFA:
The SOFA score identifies organ failure in six systems and assigns0-4 points for each system. It was created in 1996 by consensus [7] ( Table 1).
qSOFAcriteria: qSOFA criteriadetermine sepsis-induced prognosis in adult patients with suspicious infections in nonhospital, emergency, or general hospital conditions. A positive qSOFAscorerequires at least two of the following criteria: respiratory rate 22 breaths/minor more, altered mental state, or systolic blood pressure under 100 mm Hg [2] (Table 1). All the study participantswere diagnosed with sepsis by aninfectious disease and clinical microbiology specialist who came to the emergency department.

Statistical Analysis
In the statistical analyses, categorical data were summarized asfrequencies and percentages, and continuous data were summarized asmean ± standard deviation or median values (minimum-maximum) depending on the distribution of the data. Forcontinuous variablesthat were normally distributed, independent sample t-tests were used in the comparison of the two groups, whileanalysis of variance was used for comparisons among more than two groups. For variables that werenot normally distributed, the nonparametric Mann-WhitneyU test was used fortwo-group comparisonsand the Kruskal-Wallis test was used to comparemore than two groups. In addition, Spearman and Pearson correlation coefficients were used to control for the relationship between continuous variables. Normality controls for the continuous measurements were assessedwith the Shapiro-Wilk test. Receiver operator characteristic (ROC) analyses of SIRS, SOFA, qSOFA, and qSOFA+L criteria were performed. In addition, the ROC curves of the parameters were compared. For the descriptive statistics, the area under ROC curve (AUC) value, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (LR + ), negative likelihood ratio (LR -), and 95% confidence interval values are given. P<0.05 was considered statistically significant.

Results
This study included a total of 976 patientswho wereadmitted to the emergency department during the five-year study periodand diagnosed with sepsis by aninfectious diseaseand clinical microbiology specialist. Of those, 52.7% (n=514) were female and 47.3% (n=462) were male. Their mean age was 72.5±13.7 years. The • The important feature of our study was that SOFA score, SIRS, qSOFA, and qSOFA+L criteria were compared in both ICU patients and emergency department patients in the diagnosis of sepsis and prognosis.
• Despite the limitations described above, we found that the diagnostic and prognostic value of the SOFA score was more valuable than that of other scores.
• We have also found that the qSOFA criterion may be suitable for emergency department.

Main Points
Eurasian J Med 2021; 53(1): 40-7 mean age of the female patients was 72.6±14 years, and the mean age of the male patients was 72.3±13.3 years. The most common comorbiditywas chronic kidney disease (CKD) at18.9% (n=184). The demographic characteristics of the patients are shown in Table 2.
Overall, 37.4% (n=365) of all patients admitted to the emergency departmentand diagnosed with sepsis were hospitalized and 52.3% (n=191) of these patients died. Emergency department mortality was 12.5% (n=122). The median length of hospital stay was 137.5 h.
The most common source of infection was the respiratory system (24.5%, n=239), followed by the urinary system (23.8%, n=232). The source of infection could not be determinedin 31.1% (n=323) of the patients. Gram-negative bacillus (22.7% and 51%, respectively) was observed in the cultures of emergency cysts (n=396) and urine (n=414). The clinical and laboratory features of the patients are shown in Tables 2 and 3.
There was a statistically significant difference between patients who died in the emergency department and those who survived in terms of qSOFA and qSOFA+L criteria (P<0.05). The mortality rate was higher in patients with qSOFA and qSOFA+L criteria ≥2 (Table 4). There was no statistically significant difference between the patients who died in the hospital and those who survived in terms of SIRS, qSOFA, and qSOFA+L criteria (P>0.05) ( Table 4).
There was a statistically significant difference in the SOFA scores of patients who died in the emergency department and those who survived (P<0.001). Patients who died had a higher SOFA score than patients who lived (Table 5).
According to the results of the ROC analysis in terms of emergency department mortality, the ability to predict SOFA score, qSOFA, and qSOFA+L criteria was found to be statistically significant (P<0.0001, 0.009, and 0.006, respectively). The ability of theSIRS-criteria-based test to predict emergency department mortality was not statistically significant (P=0.303). The SOFA score (AUC=0.75) had a high distinctive ability topredict emergency department mortality (P<0.0001). When the SOFA score was >11, its sensitivity was 63.5%, specificity was 78.8%, PPV was 46.6%, NPV was 88.2%, LR + was 3, and LRwas 0.46. When the qSOFAscorewas >1, sensitivity was 81.7% and specificity was 28.6% (Table 6; Figure 2).

Discussion
Several studies have already compared the diagnostic criteria for sepsis in emergency departments and intensive care units (ICUs). Those that primarily consider mortality were generally performed in ICU patients [8,9]. Our study is novel because it compared the predictive capacity of SIRS, qSOFA, qSOFA +L, and SOFA score in both the diagnosis and prognosis of patients admitted to the emergency department with sepsis.
In our study, patients who died in the emergency department and in the hospital had higher SOFA scores than those who survived. Mortality rates were higher in the emergency department for patients with qSOFA and qSOFA+L criteria ≥2. There was no statistically significant difference between the patients who died and those who survived in the hospital in terms of SIRS, qSOFA, and qSOFA+L criteria. The SOFA (AUC=0.89) had a high distinctive ability to predict sepsis. When the SOFA score was >11, sensitivity and NPV were both 100%. The SOFA score (AUC=0.75 and 0.72, respectively) was highly distinctive in predicting emergency and in-hospital mortality. When the SOFA score was >11, the sensitivity was 63.5% and the specificity was 78.8% in determining the emergency department mortality. When the SOFA score was>9, the sensitivity was 65.8% and the specificity was 75.5% in determining the in-hospital mortality. When the qSOFA criteria score was >1, the sensitivity was 81.7% and the specificity was 28.6% in determining the emergency department mortality. In terms of the ability to predict emergency department mortality, the SIRS-criteria-based testwas not statistically significant. SIRS, qSOFA, and qSOFA+L criteria were not significant, while the SOFA score was highly distinctive in predicting in-hospital mortality.
Over 20 years ago, sepsis was described as a combination of an infection and SIRS. However, research has since shown that sepsis is not only a proinflammatory condition butalso a convergence of early anti-inflammatory responses. SIRS has been criticized for a long time, as it covers even mild conditions (e.g., influenza) without any organ dysfunction. In 2015, Churpek and colleagues showed that almost half of adult patients met at least once, often two SIRS criteria during their hospital stay [10]. Kaukonen et al. [11] showed that approximately 12% of adult ICU patients with infections and at least one organ disorder are negative according to theSIRS-criteria-based test, and their mortality rates are high. These results indicate that the SIRS-criteria-based testisnot suitable for screening at-riskpatients and does not accurately reflect the risk of mortality/organ dysfunction.
Risk factors for sepsis include advanced age, comorbidity (e.g., diabetes, kidney failure, respiratory failure, etc.), infection source, infection location (e.g., nosocomial), and the patient' s ward (e.g., ICU, emergency department, etc.). Severe sepsis is especially common among elderlypatients,and more than half of the patients with sepsis are over 65 years old. Furthermore, the majority of sepsis cases involve at least one chronic disease. Severe sepsis is more likely to develop in patients with Chronic obstructive pulmonary disease (COPD), malignancy, CKD, and diabetes [12,13]. A total of 976 patients diagnosed with sepsis during a five-year period were included in our study. The mean age was 72.5 years, and 52.7% (n=514) of the patients were women. The most common comorbidity was CKD, which affected 18.9% (n=184) of all patients.
Sepsis is the greatest financial burden for hospitals andthe leading cause of death in noncoronary ICU cases, contributing to 30-50%   of all in-hospital deaths [9]. In accordance with the literature, in our study, the mortality rate in patients admitted to the emergency department with sepsis was 12.5%, and the mortality rate of the sepsis patients hospitalized was 52.3%. In addition, in our study, the median value of the hospital stay was 137.5 h.
There are multiple causative pathogens in the etiology of sepsis. However, the factor responsible could not be isolated in 30-50% of the cases, depending on the study. With the prevalent use of antibiotics, gram-negative bacteria have increasingly become a factor insepsis.Recently, an increase has been observed in gram-positive bacteria, especially in cases of staphylococcal sepsis [14]. In our study, the most frequently detected causative agent wasgram-negative bacilli, followed by gram-positive cocci. In addition, the most common source of infection was the respiratory system, followed by the urinary system. In our study, the source of infection could not be determined in 31.1% of patients, which is in line with the literature.
A 2016 international prospective cohort study compared the old and new sepsis criteria and found that the accuracy of the qSOFA criteria was higher than both the SIRS-criteria-and the severe sepsis-criteria-based tests in predicting in-hospital mortality [15]. In a retrospective largecohort study in patients admitted to ICUs in Australia and New Zealand in 2017 with primary diagnoses associated with infection, SOFA scores over 2 had higher prognostic accuracy for in-hospital mortality than SIRS or qSOFA criteria [16]. Both of these studies support the Sepsis-3 recommendations to use the qSOFA criteria in non-ICU patients and the full SOFA score in ICU patients to effectively identify high-risk individuals among potentially infected patients. In our study, the SOFA had a high distinctive ability in the diagnosis of sepsis.   Abandoning the SIRS-criteria-based test and focusing on the qSOFA criteria, the SOFA isa major concern owing to the possibility of delaying early diagnosis and treatment [17,18]. Despite its weaknesses, the SIRS-criteria-based test is helpful in the early detection of an infection and in preventing progression to organ dysfunction. Quality improvement studies worldwide have relied on the SIRS-sepsis structure for years [19][20][21]. The use of the SIRS-criteria-based test is credited with contributing to the decrease in sepsis-related mortality seen over the past 20 years [22][23][24].
Another potential weakness of the new sepsis definitions is that they arebased on the SOFA score. The SOFA was created primarily for research purposes in ICUs in 1996. As scoring systems such as SOFA are not easy to memorize, their clinical use is generally limited. Many components of the SOFA (e.g.,PaO 2 /FiO 2 ratioand vasopressor requirement) are specific to the ICU or not routinely administered initially in septic patients (e.g., the Glasgow Coma Scale). However, the initial treatment of many septic patients begins in the emergency department or hospital services [25,26].
The Sepsis-3 consensus definitions should not be interpreted as replacing the SIRS criteria with qSOFA. In previous definitions, the SIRScriteria-based test has been a prerequisite for the diagnosis of sepsis, but this is not the case with the qSOFA. The qSOFAscore is presented only as an additional clinical criterion in identifying suspected infection patients at risk of a pooroutcome andinforming earlyintervention choices [9]. In a recent cohort study of suspect-ed emergency department patients, qSOFAhad a high specificity for predicting organ dysfunction (96.1%) but with low sensitivity (29.7%), while SIRS was less specific (61.1%) but more sensitive (72.3%) [27].
In a prospective large study ofICU patients in 16 countries, SOFA scores greater than 15 were associated with a 90% mortality rate [28].
In a prospective study conducted in another ICU, the first increase occurred in the SOFA score for 96 h; regardless of baseline score, the mortality rate was at least 50% as the score increased, and 27-35% when remained unchanged, and less than 27% when decreased [29]. In addition, the SOFA has been validated and administered in various ICU patient groups, including medical, surgical, and heart and burn patients [30]. Although, in our study, the SIRScriteria-based test was not statistically significant in terms of its ability to predict emergency department mortality, the qSOFA criteria and the SOFA score were significant. SIRS, qSOFA, and qSOFA+L criteria were not significant, while the SOFA score was highly distinctive in predicting in-hospital mortality.
This study has several limitations. The first limitation of our study is its retrospective nature. The second limitation, the current status of the patients who came to the emergency department and transferred, is not known. In addition, since many seriously infected patients are sent to our hospital, our mortality rate is high. The third limitation of this study is a single center, so the results may not be representative. Therefore, multicenter prospective studies are needed.
In conclusion, as a result, in our study, the SOFA score had a high sensitivity and negative prediction in the diagnosis of sepsis. In the emergency department and in-hospital mortality estimations, the SOFA was again highly capable. However, the qSOFA score hadhigher sensitivity in the emergency department mortality estimate than the SOFA score, but less specific.
In terms of the ability to predict emergency department mortality, the SIRS-criteria-based test was not significant. SIRS, qSOFA, and qSOFA+L criteria were not significant in predicting in-hospital mortality.