A total of 62 of the 192 patients underwent a CT examination to search for a septic focus within 96 h of ED presentation. The CT group and no-CT group of patients did not differ regarding sex (proportion of women, 38.7% (n = 24/62) vs 39.2% (n = 51/130), p 0.945), qSOFA score (median of 1 (IQR 1–2) vs. 2 (IQR 1–2), p = 0.512), and comorbidities (median Charlson index of 3 (1–4) vs. 2 (2–4), p = 0.303). However, the median age of patients in the CT group was lower than in the no-CT group (median of 64.5 years (IQR, 51.8–73.0) vs. 72.5 years (IQR 61.8–79.3), p = 0.002). Furthermore, septic shock occurred more often in the CT group in comparison to the no-CT group (17.7% (n = 11/62) vs. 6.2 (n = 8/130), p = 0.012). The most common final septic focus was pulmonary (51.0%; n = 98/192), followed by abdominal foci in 10.9% (n = 21/192), and other foci (Table 2).

In the CT group, the majority of the patients 43.5% (n = 27/62) had a pulmonary focus. Of these patients, 56% (n = 15/17) presented with atypical pneumonia, 33% (n = 9/27) with typical pneumonia, and for 11% (n = 3/27) patients it was not specified. Among the most common abdominal sources of sepsis were retroperitoneal abscesses (n = 6/17), colitis/gastroenteritis (n = 5/17), and pancreatitis (n = 4/17). Less common were cholecystitis, diverticulitis, spontaneous bacterial peritonitis, etc. In four patients, abdominal emergencies, specifically perforations, were reported. No postsurgical patients were identified in the no-CT group.

In the CT group (n = 62/192), 96.8% of the patients received a microbiological test to identify the responsible agent. In 20%, no agent was identified. A detailed description of the etiological agents can be found in the supplementary Table 2.

The median time-to-CT after sepsis diagnosis (ttCTsd) was 3.83 h (IQR 1.50–29.97 h), whereas the median time-to-CT after emergency department arrival (ttCTeda) was 7.44 h (IQR 3.68–30.77 h). With regard to organ regions covered by the CT scan, approximately half of the patients (n = 32) received a CT scan of more than one organ region. Specifically, in 33.9% of the patients (n = 21), more than two organ regions were scanned, in 12.9% (n = 8), more than 3, and in 4.8% (n = 3), four organ regions were scanned. The most commonly examined region was the chest (95.2%; n = 49/62), followed by the abdomen/pelvis (62.9%; n = 39/62), and the head (19.4%; n = 12/62). The extremities were the organ region least commonly examined at 4.8% (n = 3/62) (Fig. 2). The percentage of positive findings on CT, i.e., graded as possible or certain foci, per organ region was 57.1% (n = 28/49) for the chest and 53.9% (n = 21/39) for abdomen/pelvis. An infectious focus was least commonly found in the head (8.3%, n = 1/11). In 58.1% (n = 36/62) of focus-CTs, secondary findings were documented. Most commonly, other findings (55.6%; n = 20/36), tumor (11.1%; n = 4/36), and trauma (8.3%; n = 3/36) were noted. Only one patient had a fracture as an incidental finding. In 25.0% (n = 9/36), several secondary findings were documented. In 41.9% (n = 26/62), no secondary findings were noted.

Diagnostic certainty and body regions examined by focus-CT in the CT group of patients (n = 62/192). Diagnostic certainty indicates whether there was no focus detected, a possible or a definite infectious focus based on the radiological report

Comparison of infectious foci between the CT group versus the no-CT group revealed that performing a focus-CT was significantly associated with the final septic focus being pulmonary or intraabdominal, with p < 0.0001 (Table 2). Patients with no-focus-CT showed a significant association with a urogenital focus or unclear death with p < 0.0001.

Among patients who received a CT scan, the most common infectious foci that led to sepsis were pulmonary (43.3%; n = 27), abdominal (25.8%; n = 16), and urogenital (6.4%; n = 4). The consistency of the final sepsis focus with the focus detected by CT was 77.8% (n = 21/27) when the focus was pulmonary and 87.5% (n = 14/16) when the focus was abdominal. In patients with urogenital sepsis, the focus was confirmed in 50% (n = 2/4) of the cases by CT (Fig. 3; Table 2). For focus detection, a sensitivity of 81.1% (95% confidence interval (CI), 68.0–90.6%) was calculated, with a specificity of 55.6% (95% CI 21.2–86.3%).

Infectious focus consistency between focus-CT and final septic focus of the CT group of patients (n = 62/192)

Surgical source control, e.g., debridement of infected tissues, was performed in 10.8% (n = 21/192) of the patients within 96 h after ED admission. Patients with CT underwent surgical source control more often than patients with no-focus-CT (20.9% vs 6.1%; p = 0.02). Thirteen percent (n = 25/192) of all patients died within 28 days after study enrollment. The length of hospital stay was found to be longer in the focus-CT group (median 15 d, IQR 10 d–24 d vs median 9 d, IQR 7 d–14 d 06 h; p = 0.001). Mortality did not differ significantly between the focus-CT (12.9%, n = 8) and no-focus-CT group (13.1, n = 17; p = 0.973) (Table 3) or between patients with a short ttCTsd versus long ttCDsd (16.1%, n = 5/31 vs 9.7, n = 3/31; p = 0.449) (Table 4). Median survival time in patients with a short ttCTsd (17 days; IQR 4 d 12 h–23 d 00 h) and long ttCTsd (8 days; IQR 1 d 00 h–11 d 00 h) did not differ with p = 0.651. There were no differences between short and long ttCTsd regarding length of hospital stay (median 16 days, IQR 9 d 12 h–23 d 18 h vs median 13 days, IQR 10 d 00 h–24 d 00 h; p = 0.863) or duration of intensive care (median 3 d 12 h, IQR 2 d 6 h–7 d 18 h vs median 5 d, IQR 2 d–11 d; p = 0.800). Linear regression analysis of relevant covariates revealed that higher PCT and lower GCS were associated with lower ttCTsd, with (beta =  − 839.80, p = 0.551 and beta = 12,328.93, p = 0.261), respectively. Higher age and higher qSOFA score were associated with higher ttCTsd (beta = 2334.99, p = 0.218 and beta = 57,919.86, p = 0.240), respectively (supplementary Table 2).