2.1 Study Design, Period, and Area

A prospective observational study was conducted from January 1, 2023, to September 30, 2023, among septic shock patients admitted to the adult medical ICU of Yekatit 12 Hospital Medical College (Y12HMC), a teaching hospital in Addis Ababa, Ethiopia. It is one of the largest hospitals in the country, with over 500 beds, treating approximately 310,000 patients each year, and has 21 departments. The hospital has a 10-bed adult medical ICU that serves about 350–400 patients per year. Septic shock patients who were admitted to the adult medical ICU were managed by a team of nurses, residents, anesthesiologists, emergency medicine and critical care physicians, and internists.

2.2 Source and Study Population

The source population was all septic shock patients who were admitted to the adult medical ICU in public hospitals in Addis Ababa. The study population was all septic shock patients aged ≥ 15 years who were admitted to the adult medical ICU at Y12HMC.

2.3 Eligibility Criteria

Eligible patients were those aged 15 years or older admitted to the adult medical ICU at Y12HMC with septic shock, defined as a subset of sepsis with a suspected or confirmed focus of infection, requiring a vasopressor to maintain a MAP ≥ 65 mmHg. Patients with septic and cardiogenic shock were excluded to reduce the potential for misclassification between the two conditions and to ensure that our analysis was focused specifically on septic shock cases. Additionally, patients who had incomplete documentation of essential clinical or laboratory data and those who were unable to provide consent were excluded from the study.

2.4 Study Variables

The dependent variable was the incidence of in-hospital mortality. Independent variables were demographic data and medical history (age, sex, prior hospitalization, and comorbidities), clinical characteristics (duration of illness, vital signs, GCS, and source of infection), microbiological evidence, treatment-related factors (timing of antimicrobial initiation, type of antimicrobial agent, type of vasopressor, and steroid use), length of hospital stay, and systemic complications.

2.5 Operational Definitions

SIRS was defined by the presence of at least two of the following four clinical characteristics: axillary body temperature < 36°C or > 38°C; heart rate (HR) > 90 beats/min; hyperventilation, evidenced by a RR > 20 breaths/minute or partial arterial pressure of carbon dioxide < 32 mmHg; and white blood cell (WBC) count > 12,000 cells/μL or < 4000 cells/μL or with > 10% immature forms [1].

The SOFA score was a 24-point measure of organ dysfunction that used six organ systems (cardiovascular, pulmonary, renal, hepatic, neurologic, and hematologic), where 0–4 points were assigned per organ system [2].

Septic shock was defined as a subset of sepsis with a suspected or confirmed focus of infection, in which a vasopressor was required to maintain a patient's MAP ≥ 65 mmHg [2].

Low GCS was defined by a GCS score less than or equal to 14, indicating mild to severe impairment of consciousness or neurological function [2].

Low SpO2 was defined by peripheral oxygen saturation levels below 90%, indicating hypoxemia [19].

Acute lung injury (ALI) was defined by acute respiratory insufficiency with tachypnea, peripheral oxygen saturation < 90%, and diffuse pulmonary infiltrates on chest X-ray [20].

The timing of antimicrobial initiation was defined by the time between septic shock diagnosis documented by a physician and the first dose of antimicrobial administration.

Outcomes included systemic complications, in-hospital mortality, and the length of hospital stay.

2.6 Sample Size Determination and Sampling Technique

The sample size was calculated as 374 using the single population proportion formula: $$n\,=\,[{(Za/2)}^{2}(p \mbox{-} q)]/{d}^{2}$$, using a 42% in-hospital mortality rate (p = 0.42) derived from a study conducted at the University of Gondar Hospital in Ethiopia [18], a 95% confidence level (Zα/2 = 1.96), and a 5% margin of error (d = 0.05). After adjusting for an estimated population of 225 patients over 9 months, the sample size was reduced to 140 using the correction formula $$n/(1+n/N)$$. Including a 15% nonresponse rate, the final sample size was 161. A convenient sampling technique was used for data collection and all patients with septic shock who fulfilled the inclusion criteria were included.

2.7 Data Collection Procedure and Data Quality Control

Two internal medicine residents, who were trained on how to interview patients and extract data from electronic health records, filled out a prestructured questionnaire, which was prepared by reviewing different literature. Data regarding age, medical history, and duration of illness were collected on the day of admission to the adult medical ICU by interviewing patients and/or their legal guardians, and these were cross-checked for consistency with the patients' electronic health records. Additional clinical characteristics, source of infection, laboratory findings, management-related factors, and outcomes were extracted from the EMR system on the day of admission to the adult medical ICU and during the subsequent in-hospital stay. The data collection process was supervised and monitored by the principal investigators. The completeness and consistency of the collected data were ensured on a daily basis.

2.8 Statistical Analysis

Epi Info, version 7, and SPSS, version 25 (developed by IBM's team based in Chicago, Illinois, USA) were utilized for data entry and statistical analysis, respectively. Data cleaning and consistency checks were performed to ensure the accuracy and reliability of the data set. The characteristics and outcomes were summarized using descriptive statistics. Categorical variables were reported as frequencies and percentages, while continuous variables were presented as means with standard deviations for normally distributed data, and as medians with interquartile ranges for non-normally distributed data. Crude hazard ratios (CHR) and adjusted hazard ratios (AHR) with a 95% confidence interval (CI) were calculated using Cox proportional hazards regression models to identify the predictors of in-hospital mortality associated with septic shock. Variables with a p value < 0.25 in the crude Cox regression analysis were included in the multivariable Cox regression analysis, and a p value < 0.05 in the final model was considered statistically significant. The model fit was assessed using the omnibus test of model coefficients. The −2 Log Likelihood value (754.5) indicates a good fit, and the significant χ² test (p < 0.01) suggests that at least one coefficient in the model is non-zero. The incidence of in-hospital mortality was illustrated using the hazard function curve.

2.9 Ethical Approval and Consent to Participate

Ethical approval was obtained from the Institutional Review Board of Y12HMC (protocol number: 314/22). For adult patients (≥ 18 years), written informed consent was taken from themselves and/or their caregivers. For those patients under the age of 18 years, written informed consent was obtained from their parents or legal guardians.

3.1 Demographic Data and Medical History

A total of 156 patients were assessed for eligibility, of which 12 were excluded for not meeting the inclusion criteria. Finally, 144 patients were included for analysis (Figure 1).

The mean age (standard deviation) was 46 ± 19 years with a range of 15–95 years. The majority of patients (55.6%) were male. Approximately one-third of the patients had a history of hospitalization 3 months before the current study. The most common comorbidity was human immunodeficiency virus (HIV) (18.8%), followed by diabetes mellitus (DM) (17.4%) and chronic lung disease (14.6%). Furthermore, the majority (68%) had one or more comorbidities (Table 1).

3.2 Clinical and Laboratory Profiles

The most frequently identified SIRS variables were RR > 20 breaths per minute and HR > 90 beats per minute, observed in 97.9% and 97.2% of the patients, respectively. Among the SOFA variables, SBP ≤ 100 mmHg was present in all patients, while a SpO2 < 90% and a GCS score < 15 were recorded in 93.1% and 56.9% of the patients, respectively. Additionally, the majority of the patients (63.9%) experienced electrolyte imbalances (Table 2).

3.3 Sources of Infection

The respiratory system was the most common source of infection, identified in 70.2% of septic shock patients, followed by the gastrointestinal system in 34.7% and the central nervous system (CNS) in 10.4% (Table 3). Of note, more than a quarter of the patients (27.8%) had more than one source of infection.

3.4 Microbiologic Evidence

Blood cultures were performed on a total of 83 patients (57.6%). Samples for blood cultures were obtained before the administration of antimicrobials in 78 patients (94%). The available antimicrobials for culture and sensitivity testing were amoxicillin-clavulanic acid, ceftriaxone, ceftazidime, cefepime, ciprofloxacin, clindamycin, cotrimoxazole, erythromycin, gentamycin, meropenem, and vancomycin.

Positive blood cultures were identified in only seven patients (8.4%), with four showing gram-negative isolates and three showing gram-positive isolates. Two species of Klebsiella, one species of Escherichia coli, and one species of Streptococcus pneumoniae were isolated, all sensitive to the tested antimicrobials. A species of Acinetobacter baumannii was also isolated, exhibiting resistance to all tested antimicrobials. Additionally, two species of coagulase-negative staphylococci were identified, although drug susceptibility results were indeterminate. Sputum GeneXpert testing was conducted on 14 patients, with three testing positive for Mycobacterium tuberculosis. Importantly, none of the positive cases demonstrated resistance to rifampicin.

3.5 Treatment Modalities

All patients received crystalloid fluid, antimicrobials, and vasopressors. In 95% of the cases, antimicrobial therapy was initiated within the first 3 h of diagnosis. The three commonly utilized antimicrobials were vancomycin (97.9%), ceftazidime (51.4%), and cefepime (29.9%). Only four patients had their antimicrobial regimens revised based on bacterial isolates that showed sensitivity to specific tested antibiotics, while the vast majority (97.2%) continued to receive empiric antibiotics. The most widely used vasopressor was adrenaline (73.6%), followed by noradrenaline (27.8%). Add-on vasopressor was required for 12.5% of the patients, while hydrocortisone was needed for 64.6% of the patients due to refractory septic shock, as shown in (Table 4). All of these therapies were administered via peripheral intravenous access.

3.6 Outcomes of Septic Shock Patients

Nearly two-thirds of the patients suffered from multiple organ dysfunctions due to sepsis. The most prevalent complications were acute kidney injury (AKI) in 51.4% and acute lung injury (ALI) in 50.7% of the cases (Table 5). In addition, five patients developed venous thromboembolism during their hospital stay.

The in-hospital mortality rate of septic shock patients was 66.7%. The median length of hospital stay was 6 days (interquartile range: 2–11 days). The minimum and maximum lengths of hospital stay ranged from 1 to 49 days. Nearly a quarter of the cases died in the first 10 days of admission. Half of the total cases died within the first 3 weeks of admission, with a median survival time of 22 days (Figure 2).

3.7 Predictors of Septic Shock In-Hospital Mortality

The bivariable Cox regression analysis identified that a low GCS score (CHR: 1.84, 95% CI, 1.19–2.86, p = 0.01) and ALI (CHR: 1.52, 95% CI, 1.22–2.32, p = 0.04) were significantly linked to an increased risk of in-hospital mortality in septic shock patients. Additionally, low SpO2 showed a borderline significant association with mortality (CHR: 5.24, 95% CI, 4.71–7.73, p = 0.05) (Table 6).

The multivariable Cox regression analysis further demonstrated that both low GCS (AHR: 2.23, 95% CI: 1.35–3.69, p = 0.01) and low SpO2 (AHR: 8.74, 95% CI: 1.18–14.84, p = 0.03) at presentation were independent predictors of in-hospital mortality from septic shock (Table 6). Patients with low GCS and SpO2 had approximately 2 and 8 times higher risks of death, respectively, compared to those with normal values. Other variables such as age, sex, comorbidities, clinical findings (except low GCS and SpO2), laboratory results, infection source, treatment approaches (e.g., antimicrobials, vasopressors, steroids), timing of antibiotics, and systemic complications (except ALI) did not qualify for inclusion in the multivariable analysis.