Clinical outcomes and prognostic factors of patients with sepsis caused by intra-abdominal infection in the intensive care unit: a post-hoc analysis of a prospective cohort study in Korea

This post-hoc analysis of a prospective cohort study was performed by the Korean Sepsis Alliance (KSA) encompassing 16 tertiary or university-affiliated hospitals in Korea. The Steering Committee developed the research data collection methods to manage the sepsis data platform, periodically reviewed the progress of each study, and supervised the overall research progress in association with the Korea Disease Control and Prevention Agency (KDCA). The data used in this study were screened from all consecutive patients who visited the participating hospitals for six months (September 1, 2019, to February 29, 2020).

This prospective cohort study analyzed data from the Korean sepsis registry, and this study was already deliberated as a service project by the KDCA. The study was approved by the institutional review boards (IRB) of all participating hospitals, including the IRB of Asan Medical Center (approval number 2018-0675). Data were collected and analyzed in an ethical manner while protecting the patients' right to privacy. The requirement for informed consent was waived owing to the non-interventional, observational characteristics by the IRB of all participating hospitals, including the IRB of Asan Medical Center (approval number 2018-0675).

Our study enrolled a total of 2126 patients with sepsis who were admitted to the participating hospitals for 6 months. Of these patients, 901 were treated in ICUs. Finally, the study included 219 patients who were admitted to the ICU due to sepsis caused by IAI (Fig. 1). Sepsis patients aged ≥ 19 years were included and followed up until death or discharge. As defined by the Clinical Criteria of the Third International Consensus Definition (Sepsis-3), sepsis was defined as a life-threatening organ dysfunction resulting from an uncontrolled host response to infection [11]. Organ dysfunction was included in the definition of sepsis, and the presence or absence of organ dysfunction was determined using a Sequential Organ Failure Assessment (SOFA) score. In this study, sepsis was diagnosed if the patient met the following two conditions: (1) suspicion or confirmation of infection and (2) increase in SOFA score by two points or more when an event occurred. The base SOFA score of each organ was assumed to be zero for patients with no known pre-existing organ dysfunction, and patients with a SOFA score of two or higher at the time of the event were enrolled in the study. Peritonitis caused by perforation of a hollow viscus, intra-abdominal abscess, biliary tract infection, pancreatic infection, enteritis, and colitis were defined as having IAI. The Eastern Cooperative Oncology Group (ECOG) scale is a measurement index created to check performance status, determine appropriate treatment policy, and predict prognosis in cancer patients. Using the ECOG scale, it is possible to objectively evaluate the patient's level of functioning in terms of their ability to care for themselves, perform activities of daily living, and their physical ability. It is divided into six scales: (0) fully active and no performance restrictions; (1) strenuous physical activity restricted, fully ambulatory and able to carry out light work; (2) capable of all self-care but unable to carry out any work activities; up and about > 50% of waking hours; (3) capable of only limited self-care and confined to bed or chair > 50% of waking hours; (4) completely disabled and cannot carry out any self-care; and (5) death. Continuous renal replacement therapy (CRRT) refers to patients who underwent CRRT for the treatment of acute kidney injury, and excludes patients who underwent conventional hemodialysis as renal replacement therapy. In community-acquired sepsis, time zero was defined as the time of an emergency room triage visit. In-hospital-acquired sepsis, time zero was defined as the first time the rapid response team recognized the sepsis. Source control was defined as non-surgical treatments such as percutaneous drainage, and surgical treatment such as debridement or laparotomy. Patients who were treated with antibiotics and received intensive care, without any such procedures, were considered not to have received source control.

All trained research coordinators in each participating hospital completed their entry into the shared data platform, and the coordinating hospital evaluated the quality of the data for completeness and logical errors. The data collected retrospectively in each hospital were as follows: (1) patient characteristics, including age, sex, body mass index (BMI), Charlson Comorbidity Index (CCI), physiological status, medical history, SOFA score, simplified acute physiology score III (SAPS 3), and laboratory data at time zero; (2) clinical results, including the duration from time zero to antibiotic administration, source control implementation, and duration from time zero to source control implementation; (3) infection and microbiological data, including the type of isolated bacteria and fungi, occurrence of bacteremia and multidrug resistance (MDR), and location of infection; and (4) organ dysfunction data, including the type of organ dysfunction results and analysis of the number of organ dysfunctions. Additional data compared with those in previous studies were collected. However, these were not used in our study.

Statistical analysis was conducted by the staff of the coordinating center and by a professor at the Department of Clinical Epidemiology and Biostatistics, Asan Medical Center, who did not participate in data collection.

Data are presented as numbers (percentages) or as mean ± standard deviation. The characteristics and clinical data of the survivor and non-survivor groups were compared using the chi-square test or Fisher’s exact test (for categorical variables) and Student’s t-test (for continuous variables). Risk factors associated with 28-day mortality were analyzed by univariate and then multivariate logistic regression, and the degree of association with 28-day mortality was presented as an independent factor by means of odds ratios (ORs) with their corresponding 95% confidence intervals. Infection profile data were analyzed using the number (percentage) or mean ± standard deviation. The microbiological data were described by total numbers and proportions, which showed the distribution of bacteria and fungi in patients with sepsis due to IAI. Multivariate logistic regression analysis was used to estimate the association between organ dysfunction and 28-day mortality, with unadjusted and adjusted evaluations. We adjusted 28-day mortality for age, sex, CCI, and lactic acid levels. We used OR to present the impact of organ dysfunction on 28-day mortality and analyzed the impact of the number of organ dysfunctions on 28-day mortality using ORs in the logistic regression analysis. Also, the multiple logistic regression by stepwise selection was used to explain the 28-day mortality using independent variables. The logistic regression performance is denoted by c-statistics and Hosmer–Lemeshow tests. Differences were considered statistically significant at P < 0.05. All analyses were conducted using SPSS software (version 25.0; IBM Corp., Armonk, NY, USA).

During the 6-month study period, total 2126 patients were admitted owing to sepsis at the participating hospitals; 219 (10.3%) were admitted to the ICU owing to sepsis caused by IAI according to collected data review. The 28-day mortality rate of all enrolled patients with sepsis was 28.1% (n = 598/2126). Among these patients, the mortality rate was 28.8% (n = 63/219) for sepsis due to IAI in the ICU.

Patient characteristics and clinical data at the time of sepsis diagnosis are summarized in Table 1. The mean age of the patients was 69.4 ± 13.1 years, and 53.9% were men. The mean BMI was 23.0 ± 4.2 kg/m²; no significant difference was noted in BMI between the survivor and non-survivor groups. The mean CCI in the non-survivor group was higher than that of the survivor group, with significant differences (5.9 ± 3.0 vs. 5.1 ± 2.3, p = 0.011). The initial Eastern Cooperative Oncology Group scores ranged from 0 to 4. However, there was no significant difference in the scores between the two groups (p = 0.075). The incidence of septic shock was 47%, and was significantly higher in the non-survivor group than in the survivor group (58.7% vs. 42.3%, p = 0.028). The initial lactate level was significantly higher in the non-survivor group than in the survivor group (7.2 ± 5.0 vs. 4.5 ± 3.1, p < 0.001). In addition, the initial SOFA score was significantly higher in the non-survivor group than in the survivor group (12.2 ± 4.6 vs. 8.4 ± 3.1, p < 0.001). Moreover, patients who underwent CRRT had a higher mortality rate than those who did not undergo CRRT (68.3% vs. 16%, p < 0.001). By contrast, no significant difference was noted in the number of patients who were prescribed steroids to treat sepsis between the two groups (26.9% vs. 34.9%, p = 0.239). In the initial laboratory tests at time zero, total bilirubin, C-reactive protein, procalcitonin, and brain natriuretic peptide levels were higher in the non-survivor group.

In addition, there was no significant difference between the two groups in the duration from time zero to the administration of antibiotics and source control. However, the number of patients who underwent source control through non-surgical treatment such as percutaneous drainage, or surgical treatment was significantly higher in the survivor group than in the non-survivor group (46.8% vs. 27%, p = 0.007). The mean number of organ dysfunction was significantly higher in the non-survivor group than in the survivor group (2.3 ± 1.4 vs. 1.4 ± 1.2, p = 0.019).

Table 2 shows the distribution of the isolated microbiological pathogens expressed by per species percentage of bacteria and fungi from patients with sepsis due to IAI. Of the 219 patients, 157 (71.7%) were identified with a causative pathogen. Among the patients with isolated pathogens, gram-negative bacteria were found in 81.5%, gram-positive bacteria in 32.5%, and fungi in 4.5% of patients. Among the isolated causative pathogens in our study, the most common pathogen was Escherichia coli (48.4%), followed by Klebsiella pneumoniae (22.9%), Enterococcus faecium (8.9%), and Enterococcus faecalis (5.7%). Additionally, Acinetobacter baumannii (4.5%), Enterobacter cloacae (3.8%), Klebsiella oxytoca (3.2%), and Pseudomonas aeruginosa (3.2%) were other gram-negative bacteria, and Staphylococcus aureus (1.8%) and Corynebacterium striatum (0.5%) were other gram-positive bacteria. Moreover, Candida albicans (3.2%), Candida glabrata (2.5%), and Candida krusei (0.6%) were the identified fungi.

Bacteremia occurred in 49.3% of all patients, and did not differ significantly between the survivor and non-survivor groups (48.7% vs. 50.8%, p = 0.781). Among the identified pathogens, there were no statistically significant differences in the types of bacteria and fungus between the survivor and the non-survivor groups. Mixed growth was defined as the detection of more than one type of gram-positive, gram-negative, and fungal pathogens. Of the patients with identified causative pathogens, 19.1% had a mixed growth on culture. The proportion of patients with mixed growth was higher in the non-survivor group than in the survivor group, although the difference was not statistically significant (27.5% vs. 16.2%, p = 0.304). MDR was defined as the antimicrobial resistance of a microorganism to at least one antibiotic in three or more antimicrobial categories [12]. The overall prevalence of MDR pathogens did not differ significantly between the survivor and non-survivor groups (47.9% vs. 40.0%, p = 0.134). Among the MDR pathogens, the most common were Enterobacteriaceae (54.2%), followed by Enterococcus spp. (12.5%), Acinetobacter spp. (4.2%), Staphylococcus aureus (2.8%), and Pseudomonas aeruginosa (2.8%). There was no difference in the mortality rate according to the location of infection (Table 3).

In the organ dysfunction analysis, the most common type of organ dysfunction in patients with sepsis due to IAI was respiratory dysfunction (36.5%), followed by renal (36.1%), coagulation (34.2%), cardiovascular (25.6%), central nervous system (CNS) (19.6%), and liver (18.3%) dysfunctions. Among the dysfunctional organs, the mortality rate associated with each organ dysfunction was the highest in patients with CNS dysfunction, followed by coagulation, renal, respiratory, cardiovascular, and liver dysfunctions (Table 4).

Multivariate logistic regression analysis was used to estimate the association between organ dysfunction and 28-day mortality. Among the organ dysfunctions, the 28-day mortality rate was more significantly affected with coagulation (OR = 2.99 [1.63–5.48], p < 0.001), CNS (OR = 2.38 [1.19–4.75], p = 0.014), renal (OR = 1.99 [1.09–3.61], p = 0.024), and respiratory (OR = 1.93 [1.06–3.5], p = 0.031) dysfunctions. Furthermore, after adjusting for age, sex, CCI, and lactic acid level, only coagulation dysfunction appeared to affect 28-day mortality significantly (OR = 2.78 [1.47–5.23], p = 0.001) (Table 5).

On studying the association between the number of organ dysfunctions and the 28-day mortality rate, mortality was found to be 24.7% (n = 20/81, p = 0.307) in patients with one organ dysfunction and 18.4% (n = 9/49, p = 0.068) in patients with two organ dysfunctions. The 28-day mortality was 55.6% in patients with three organ dysfunctions (OR = 9.06 [2.64–31.15], p = 0.001) and 50% in patients with four or more organ dysfunctions (OR = 7.25 (1.85–28.36], p = 0.004). In patients with three or more organ dysfunctions, 28-day mortality was more than twice that of single organ dysfunction (Table 6). In time, the mortality rate in patients with more than three organ dysfunctions distinctly increased.

CCI, septic shock, initial lactate level, initial SOFA score, SAPS 3 score, CRRT, source control, and the number of dysfunctional organs were associated with 28-day mortality in the univariate analyses. In the multivariate logistic regression analysis, after adjusting for each risk and confounding factor, only the SAPS 3 score (p < 0.001) and CRRT (p < 0.001) were independently associated with higher 28-day mortality (Table 7).