Radiological detection rate and clinical impact of thoracoabdominal computed tomography in suspected infection in the emergency department: a retrospective cohort study Authors: Abdullah Enes Ataş, Mustafa Emir, Fatih Cemal Tekin, Abdurrahman Koç Published: 2026 ABSTRACT AbstractBackgroundIdentifying the anatomical source of infection in emergency department (ED) patients with suspected sepsis is challenging. This study evaluates the radiological detection rate of thoracoabdominal computed tomography (TACT) and its association with downstream clinical management.MethodsIn this retrospective cohort study, 267 adult ED patients who underwent TACT for suspected severe infection were evaluated. The primary outcome was the radiological detection of an infectious focus, confirmed by a composite reference standard (CT findings, microbiology, clinical follow-up, and treatment response). Demographic data, quick Sequential Organ Failure Assessment (qSOFA) scores, and biomarkers were analyzed alongside clinical outcomes, including hospital admission and the need for interventional procedures.ResultsThe radiological detection rate of TACT for a definitive infectious focus was 41.2% (n = 110), while non-infectious acute pathologies mimicking sepsis were identified in 18.0% (n = 48). Thoracic infections were the most common (54.5%). Multivariate logistic regression identified a qSOFA score ≥ 2 (OR = 3.14; p = 0.009) and elevated C-reactive protein (OR = 1.009; p = 0.034) as independent predictors of a positive CT scan. Patients with a radiologically confirmed infectious focus underwent source-control interventions more frequently than those with normal CT findings (14.8% vs. 2.9%; OR = 5.72, p = 0.014). The detection of any pathological finding on CT was strongly associated with hospital admission.ConclusionTACT is a valuable diagnostic tool in suspected infections. High qSOFA scores and CRP levels correlate with a higher probability of detecting an infectious focus on CT. Beyond primary diagnosis, positive TACT findings are strongly associated with hospital admission and facilitate targeted source-control interventions.Keywords: Computed tomography, Sepsis, Diagnostic imaging, Emergency medicine, Infection Introduction Among the conditions that pose diagnostic challenges in emergency medicine, fever or suspected infection rank at the top. Although identifying the source of infection is critical for targeted treatment, it can be clinically challenging, particularly in conditions such as sepsis-associated encephalopathy where the patient’s mental status is impaired [1]. While standard definitions of sepsis (SIRS criteria, including current definitions incorporating markers of organ dysfunction) are available [2], identifying the source of the infection remains a separate challenge. It is known that the pathophysiology of sepsis involves a heterogeneous structure comprising inflammation, coagulation, and endothelial interactions [3]. Timely and accurate identification of the site of infection is critical for initiating appropriate antimicrobial therapy, planning potential source control procedures (e.g., abscess drainage, surgical intervention), improving patient outcomes, and, particularly in sepsis, reducing mortality. Delays in diagnosis can adversely affect the prognosis [4]. The initial evaluation typically includes a detailed history, physical examination, basic laboratory tests (complete blood count, C-reactive protein, erythrocyte sedimentation rate, blood and other body fluid cultures, liver function tests), and a chest X-ray [4]. If these initial steps do not identify the source of infection, advanced imaging methods are employed. Modalities used include ultrasound (particularly focused abdominal sonography in trauma or sepsis—FAST), computed tomography (CT), magnetic resonance imaging (MRI), and nuclear medicine studies [5]. The diagnostic accuracy of initial imaging methods, such as chest X-rays or focused ultrasound, may be limited due to operator dependency, patient factors such as obesity, or limited anatomical coverage [6]. Thoracoabdominal CT (TACT) rapidly scans a broad anatomical area encompassing the thorax, abdomen, and pelvis, providing detailed information about regions where hidden infections frequently occur [1]. TACT can detect specific signs of infection and complications such as abscesses, phlegmons, fluid collections, inflammatory changes, lymphadenopathy, and organomegaly [7]. Contrast-enhanced CT scanning can improve diagnostic accuracy, particularly in distinguishing abdominal foci and fluid collections [8]. TACT is a widely used imaging method and is frequently employed in emergency situations to investigate sepsis of unknown origin or fever of unknown origin. Early studies have emphasized the superiority of abdominal CT over abdominal ultrasound in the evaluation of fever of unknown origin [7]. In this study, we aimed to demonstrate the radiological detection rate of TACT in identifying the source of infection in a high-risk ED cohort and to evaluate its association with subsequent clinical management. Materials and methods Study design and patient population This single-center, retrospective cohort study was conducted at the Emergency Department of Necmettin Erbakan University Hospital between 1.1.2024 and 1.1.2025. The study protocol was approved by the Institutional Ethics Committee (Approval Number: 2026/6282, Date: 09.01.2026). Due to the retrospective nature of the study, the requirement for written informed consent was waived. All procedures were performed in accordance with the ethical standards of the institutional research committee and the 1964 Helsinki Declaration and its later amendments. The target population consisted of adult patients (aged ≥ 18 years) who presented to the ED with clinical suspicion of severe systemic infection or sepsis (e.g., fever, altered mental status, hemodynamic instability). To represent a cohort requiring advanced imaging, inclusion was restricted to patients who underwent TACT because routine initial evaluations (physical exam, X-ray, point-of-care ultrasound) failed to localize the source, or because the patient experienced sudden clinical deterioration. Patients with trauma, pregnant patients, those referred with an already established diagnosis, and patients with missing data were excluded. Consequently, a total of 267 eligible patients were included (Fig. 1). Patient selection flowchart Clinical and laboratory data collection Baseline demographic variables, including age and gender, alongside comprehensive clinical and laboratory data, were extracted from the electronic medical records. To quantify the baseline comorbidity burden of the study population, the Charlson Comorbidity Index (CCI) was calculated for each patient based on their pre-existing chronic conditions documented upon admission. Clinical severity and the risk of sepsis on admission were assessed using the quick Sequential Organ Failure Assessment (qSOFA) score. The qSOFA score (ranging from 0 to 3) was calculated by assigning one point for each of the following clinical criteria met at the time of ED triage: respiratory rate ≥ 22 breaths per minute, altered mentation (Glasgow Coma Scale score < 15), and systolic blood pressure ≤ 100 mmHg. Concurrent laboratory parameters drawn upon admission prior to the CT scan were recorded. These included white blood cell (WBC) count, neutrophil-to-lymphocyte ratio (NLR), C-reactive protein (CRP), procalcitonin, and serum creatinine levels. Computed tomography imaging protocol and image analysis All imaging examinations were performed using a 256-slice dual-source CT scanner (Siemens Somatom Drive, Erlangen, Germany). The decision to administer intravenous (IV) iodinated contrast material was not standardized; it was made at the discretion of the attending physician based on the patient’s renal function (e.g., baseline creatinine) and clinical suspicion of vascular or ischemic events. Consequently, the cohort included both non-contrast CT (n = 191) and IV contrast-enhanced CT (n = 76) scans. CT images were evaluated by two radiologists with 4 and 10 years of experience, blinded to the final clinical discharge diagnoses. Due to the retrospective nature of the study, a formal interobserver agreement was not calculated. Instead, when searching for the infectious source, any discrepancies in interpretation between the two radiologists were resolved by reaching a consensus to determine the final finding. To establish the final presence of an infectious focus, a composite reference standard was strictly applied. This standard integrated consensus CT findings with microbiological data (blood, urine, and fluid cultures), formal consultation notes, and the documented clinical response to targeted antimicrobial therapy during hospitalization. Based on this composite standard, CT outcomes were classified into three categories: (1) Focus of infection detected, (2) Normal CT, and (3) Non-infectious pathological finding (e.g., occult malignancies, bowel ischemia). Clinical outcomes and management To assess the impact of CT findings on clinical decision-making, subsequent patient management data were collected. The primary clinical outcomes included the final patient disposition (hospital admission vs. outpatient follow-up), length of hospital stay (LOS) in days, necessity for antibiotic modification/adjustment following the CT report, and the requirement for source-control interventional procedures (defined as percutaneous drainage, surgical intervention, or urgent surgical consultation) during the hospital stay. Statistical analysis Statistical analyses were performed using SPSS v24.0 (IBM Corp., Armonk, NY, USA) and Jamovi v2.6 (Sydney, Australia). The normality of continuous variables was assessed using analytical methods (Shapiro-Wilk test) and visual inspection of histograms. Since the continuous data were not normally distributed, they were presented as medians with interquartile ranges (IQR). Categorical variables were expressed as frequencies and percentages. For the comparison of continuous variables across the three CT finding groups, the Kruskal-Wallis H test was utilized. When significant differences were observed, post-hoc pairwise comparisons were conducted using the Mann-Whitney U test. The comparison of categorical variables between groups was performed using the Pearson Chi-square (χ²) test or Fisher’s exact test, as appropriate. The diagnostic performance of clinical and biochemical markers in distinguishing patients with a definitive focus of infection from those with normal CT findings was evaluated using Receiver Operating Characteristic (ROC) curve analysis. The Area Under the Curve (AUC) was calculated for each parameter. The optimal cut-off values were determined by maximizing the Youden index (Sensitivity + Specificity − 1), from which the corresponding sensitivity and specificity metrics were derived. To identify independent predictors of detecting an infectious focus on CT, as well as predictors of hospital admission, multivariate logistic regression analyses were conducted. Variables demonstrating potential clinical relevance or a p-value < 0.10 in univariate analyses were included in the multivariable models. Odds ratios (OR) with 95% confidence intervals (CI) (including bootstrap 95% CIs where applicable) were reported. A two-sided p-value of < 0.05 was considered statistically significant for all analyses. Results Patient demographics and baseline clinical characteristics A total of 267 patients were included in the study [median age: 71 years, interquartile range (IQR): 61–80; age range: 19–104 years]. The study cohort comprised 156 (58.4%) males and 111 (41.6%) females. Comorbidity burden was evaluated using the Charlson Comorbidity Index (CCI); the median CCI was 5 (IQR: 4–8; mean ± SD: 5.6 ± 2.8), indicating a high comorbidity burden within the study population. Upon evaluation of the quick Sequential Organ Failure Assessment (qSOFA) score, which predicts the risk of sepsis on admission, 36.7% of the patients (n = 98) had a qSOFA score of 0, 24.7% (n = 66) had a score of 1, 32.2% (n = 86) had a score of 2, and 6.4% (n = 17) had a score of 3. The overall rate of patients with a qSOFA score of ≥ 2 was 38.6% (n = 103). Baseline laboratory values on admission were as follows: median white blood cell (WBC) count was 12.6 × 10³/µL (IQR: 9.2–17.4), median neutrophil-to-lymphocyte ratio (NLR) was 9.6 (IQR: 5.5–17.4), median C-reactive protein (CRP) was 187.4 mg/L (IQR: 116.0–263.0), median procalcitonin was 1.07 ng/mL (IQR: 0.31–6.58), and median creatinine was 1.58 mg/dL (IQR: 1.08–2.60). Regarding the imaging protocols, 191 patients (71.5%) underwent non-contrast thoracoabdominal computed tomography (CT), while 76 patients (28.5%) underwent intravenous (IV) contrast-enhanced thoracoabdominal CT. Baseline demographic and clinical characteristics are summarized in Table 1. Table 1Baseline demographic and clinical characteristics of the patientsVariableAll Patients (n = 267)Focus of Infection Present (n = 110)Normal CT (n = 109)Age, median [IQR]71 [61–80]70 [61–78]72 [61–81]Gender: Male, n (%)156 (58.4%)68 (61.8%)58 (53.2%)CCI, median [IQR]5 [4–8]5 [4–7]5 [4–7]qSOFA, median [IQR]1 [0–2]2 [1–2]*1 [0–1]qSOFA ≥ 2, n (%)103 (38.6%)77 (70.0%)*14 (12.8%)WBC, median [IQR]12.6 [9.2–17.4]13.2 [9.5–18.2]12.4 [9.6–16.0]NLR, median [IQR]9.6 [5.5–17.4]9.7 [5.7–18.6]8.2 [5.1–15.4]CRP, median [IQR]187.4 [116.0–263.0]229.9 [136.8–306.9]*159.3 [71.8–220.0]Procalcitonin, median [IQR]1.07 [0.31–6.58]1.90 [0.50–13.55]*0.76 [0.24–2.48]Creatinine, median [IQR]1.58 [1.08–2.60]1.54 [0.95–2.50]1.64 [1.19–2.59]IV Contrast-enhanced CT, n (%)76 (28.5%)36 (32.7%)17 (15.6%)*Note: * p < 0.05 (Comparison between the ‘Focus of Infection Present’ and ‘Normal CT’ groups). IQR: interquartile range; CCI: Charlson Comorbidity Index; WBC: white blood cell (×10³/µL); NLR: neutrophil-to-lymphocyte ratio; CRP: C-reactive protein (mg/L); CT: computed tomography; IV: intravenous Baseline demographic and clinical characteristics of the patients Note: * p < 0.05 (Comparison between the ‘Focus of Infection Present’ and ‘Normal CT’ groups). IQR: interquartile range; CCI: Charlson Comorbidity Index; WBC: white blood cell (×10³/µL); NLR: neutrophil-to-lymphocyte ratio; CRP: C-reactive protein (mg/L); CT: computed tomography; IV: intravenous Radiological detection rate and CT findings Three categories were defined for the evaluation of CT examinations: (i) detection of an infectious focus, (ii) normal CT appearance, and (iii) non-infectious pathological findings. Based on the composite reference standard, an infectious focus was identified in 110 (41.2%) patients, CT was considered normal in 109 (40.8%), and non-infectious pathological findings were detected in 48 (18.0%). Thus, the radiological detection rate of TACT for an infectious focus was 41.2%. The impact of the imaging protocol on CT findings was statistically significant (χ²=18.77; p < 0.001). While the rate of detecting an infectious focus was 47.4% in patients who underwent IV contrast-enhanced CT, this rate decreased to 38.7% in non-contrast CT; however, the rate of non-infectious pathological findings was also significantly higher with contrast-enhanced CT (30.3%), which was considered the primary factor leading to the statistical difference when compared to the rate in non-contrast CT (13.1%). When analyzing the organ/system distribution of the 110 patients with a detected focus of infection, thoracic infections (pneumonia, empyema, abscess, etc.) were by far the most predominant (n = 60; 54.5%) (Fig. 2). This was followed by urinary tract infections (n = 24; 21.8%), gastrointestinal tract infections (n = 15; 13.6%), and other foci (skin/soft tissue, musculoskeletal system; n = 9; 8.2%). Eleven patients (10.0%) had multiple foci of infection (e.g., thorax + urinary tract: n = 4). Hepatobiliary (n = 2), pancreatic (n = 1), and gynecological (n = 1) foci were observed less frequently. CT finding categories and system distributions are detailed in Table 2. (a) An axial CT image, optimized for the interrogation of the pulmonary parenchyma, delineates the manifestation of multifocal bronchopneumonic infiltrative processes. (b) A corresponding axial cross-section exhibits expansive regions of conspicuous parenchymal consolidation consistent with pneumonia Table 2CT finding categories and organ/system distribution in cases with a focus of infectionCT Finding Category / Infection System n % CT Findings Focus of infection detected11041.2Normal CT10940.8Non-infectious pathological finding4818.0 System Distribution in Cases with a Focus of Infection ( n = 110) Thorax (pneumonia, empyema, etc.)6054.5Urinary system2421.8Gastrointestinal system1513.6Other (skin-soft tissue, musculoskeletal)98.2Thorax + Urinary (multiple foci)43.6Hepatobiliary system21.8Gastrointestinal + Thorax (multiple foci)21.8Pancreas / Gynecological / Other multiple32.7 CT finding categories and organ/system distribution in cases with a focus of infection Clinical and laboratory differences between CT finding groups In the intergroup comparison, no statistically significant differences were observed in terms of age (Kruskal-Wallis H = 1.49; p = 0.476) and WBC count (H = 2.30; p = 0.316). However, the CCI value was significantly higher in the non-infectious pathology group compared to the other groups [median: 7 (IQR: 5–10) vs. infection: 5 (4–7) and normal CT: 5 (4–7); H = 18.16; p < 0.001], suggesting that non-infectious pathologies are more frequently detected on CT in patients with a severe comorbidity burden. The qSOFA score emerged as the strongest differentiating variable among the groups (H = 54.92; p < 0.001). The median qSOFA score was 2 (IQR: 1–2) in the infection focus group, while it was 1 (IQR: 0–1) in the normal CT group and 1 (IQR: 0–1) in the non-infectious pathology group. The rate of patients with qSOFA ≥ 2 was 70.0% (n = 77) in the infection group compared to only 12.8% (n = 14) in the normal CT group, demonstrating a highly significant difference (Mann-Whitney U = 9,189.5; p < 0.001). CRP levels were significantly higher in the infection group [median: 229.9 mg/L (IQR: 136.8–306.9)] compared to the normal CT group [median: 159.3 mg/L (IQR: 71.8–220.0)] (U = 8,214; p < 0.001). The overall Kruskal-Wallis test also indicated a significant difference between the groups (H = 22.53; p < 0.001). Similarly, procalcitonin levels were significantly elevated in the infection group [median: 1.90 ng/mL (IQR: 0.50–13.55)] compared to the normal CT group [median: 0.76 ng/mL (IQR: 0.24–2.48)] (H = 7.50; p = 0.024; pairwise comparison: U = 6,652; p = 0.006). NLR and creatinine levels did not show statistically significant differences among the groups (p > 0.05). Diagnostic performance of clinical and biochemical markers in identifying the focus of infection The diagnostic performance of clinical and laboratory parameters in differentiating between the infection focus group and the normal CT group was evaluated using Receiver Operating Characteristic (ROC) curve analysis. The qSOFA score exhibited the highest diagnostic discrimination capacity with an AUC of 0.748. When applying a clinically relevant threshold of ≥ 2, qSOFA demonstrated robust association with positive CT findings. For CRP, the AUC was 0.643, with an optimal cut-off of > 210.2 mg/L yielding a sensitivity of 57.2% and a specificity of 68.7%. Procalcitonin demonstrated a moderate diagnostic value with an AUC of 0.588. Conversely, the contributions of WBC and NLR to diagnostic discrimination were found to be insufficient (AUC = 0.544 and 0.515, respectively; p > 0.05). The ROC analysis results for all parameters are presented in Table 3; Fig. 3. Table 3Diagnostic performance of clinical and biochemical markers (ROC analysis)MarkerAUC (95% CI) p Optimal Cut-offSensitivitySpecificityWBC (x 10³)0.5440.21≥ 16.734.5%78.7%CRP (mg/L)0.643< 0.001> 210.257.2%68.7%Procalcitonin (ng/mL)0.5880.01> 2.4947.6%70.6%NLR0.5150.67> 6.9367.2%40.0%qSOFA score0.748< 0.001≥ 1.570.0%83.7%Note: ROC analysis represents the comparison between the “Focus of infection detected” group and the “Normal CT” group. The optimal cut-off was determined according to the Youden index. CI: confidence interval; NLR: neutrophil-to-lymphocyte ratio; WBC: white blood cell; CRP: C-reactive protein Diagnostic performance of clinical and biochemical markers (ROC analysis) Note: ROC analysis represents the comparison between the “Focus of infection detected” group and the “Normal CT” group. The optimal cut-off was determined according to the Youden index. CI: confidence interval; NLR: neutrophil-to-lymphocyte ratio; WBC: white blood cell; CRP: C-reactive protein Receiver operating characteristic (ROC) curve analysis of clinical and biochemical markers for the detection of an infectious focus by CT Independent predictors of infection focus detection: logistic regression analysis A multivariate logistic regression analysis was performed (n = 209) to identify the independent variables differentiating the infection group from the normal CT group. Age, gender, CCI, qSOFA, WBC, NLR, CRP, procalcitonin, creatinine, and CT contrast protocol were included in the analysis. In the multivariate analysis, the qSOFA score was found to be the strongest independent predictor of detecting an infectious focus [OR = 3.14; 95% bootstrap CI: 2.34–5.26; p = 0.009]. The CRP level also met the threshold for statistical significance as an independent predictor (OR = 1.009; p = 0.034). Although the use of IV contrast-enhanced CT nearly doubled the likelihood of detecting an infectious focus (OR = 1.98), it did not reach statistical significance in the multivariate analysis due to wide confidence intervals (p = 0.285). Age, gender, WBC, NLR, procalcitonin, and creatinine did not achieve significance as independent predictors in the multivariate analysis (p > 0.05 for all). Hospitalization and clinical management Of the patients, 213 (80.1%) were hospitalized, and 53 (19.9%) were followed up in the outpatient clinic. The hospitalization rate varied significantly among the CT finding groups (χ²=13.69; df = 2; p = 0.001): 91.7% (n = 44/48) in the non-infectious pathology group, 85.5% (n = 94/110) in the infection group, and 69.4% (n = 75/108) in the normal CT group were hospitalized. The median length of stay (LOS) was 10 days [IQR: 6–15] for all groups, with no statistically significant difference observed between the groups (Kruskal-Wallis H = 2.06; p = 0.357). In the logistic regression analysis, CCI (OR = 1.13; p < 0.05) and qSOFA score (OR = 1.44; p < 0.05) were identified as independent predictors of hospitalization. Notably, while the detection of non-infectious pathological findings (OR = 2.61) and the detection of an infectious focus (OR = 1.64) on CT increased the probability of admission compared to a normal CT finding, the OR value in the non-infectious pathology group was distinctly higher than that of the infection group (Fig. 4). (a) An axial CT section manifests the presence of prominent hilar lymphadenopathy (red arrow). (b) Axial CT image demonstrates the accumulation of pathological fluid, indicative of bilateral pleural effusions (red arrows). (c) An axial projection optimized for pulmonary parenchyma reveals a cavitating malignant mass lesion, localized to the superior lobe of the right lung (red arrow) Impact of CT findings on the treatment process and interventional approaches In the 195 patients evaluated for clinical management (those who were hospitalized and had available treatment data), antibiotic adjustment was the most frequently performed therapeutic intervention. Need for interventional procedures (percutaneous drainage, surgical consultation, etc.) among hospitalized patients differed significantly between the groups (χ²=6.74; df = 2; p = 0.034). Interventional procedures were required in 14.8% (n = 13/88) of the infection group and 15.4% (n = 6/39) of the non-infectious pathology group, whereas this rate was only 2.9% (n = 2/68) in the normal CT group (Fig. 5). A pairwise comparison using Fisher’s exact test revealed that the infection group required interventional procedures five times more frequently than the normal CT group (OR = 5.72; 95% CI: 1.22–26.82; p = 0.014). Similarly, the rate of interventional procedures in the non-infectious pathology group was significantly higher than in the normal CT group (OR = 6.00; 95% CI: 1.04–34.67; p = 0.049). Clinical management data according to CT finding groups are summarized in Table 4. (a) Axial contrast-enhanced CT image demonstrating an abscess formation in the cholecystectomy bed (red arrow). (b) Axial non-contrast CT image revealing pneumobilia and intrahepatic biliary dilatation consistent with cholangitis (yellow arrows), accompanied by multiple cholangitic abscess foci within the left lobe of the liver (red arrow). Both patients were subsequently treated with percutaneous drainage Table 4Comparison of hospitalization and clinical management according to CT finding groupsClinical Management ParameterFocus of Infection (n = 110)Normal CT (n = 109)Non-infectious Finding (n = 48)Hospitalization, n (%)94 (85.5%)75 (69.4%)44 (91.7%)Length of Stay, median days [IQR]10 [5–15]9 [3–14]10 [4–13]Antibiotic Modification/Adjustment, n (%)*75 (85.2%)66 (97.1%)33 (84.6%)Interventional Procedure, n (%)*13 (14.8%)†2 (2.9%)6 (15.4%)‡Note: * Rate among hospitalized patients. † Infection group vs. Normal CT: Fisher’s OR = 5.72 (p = 0.014). ‡ Non-infectious finding group vs. Normal CT: Fisher’s OR = 6.00 (p = 0.049) Comparison of hospitalization and clinical management according to CT finding groups Note: * Rate among hospitalized patients. † Infection group vs. Normal CT: Fisher’s OR = 5.72 (p = 0.014). ‡ Non-infectious finding group vs. Normal CT: Fisher’s OR = 6.00 (p = 0.049) Antibiotic modification was highly prevalent across all groups. The 97.1% modification rate in the normal CT group suggests that in the absence of a localized, drainable focus, clinicians may frequently resort to empirical escalation or broad-spectrum adjustments, though the precise clinical rationale cannot be definitively established from retrospective data. Discussion In this study, we evaluated the radiological detection rate of TACT and its association with the clinical management of ED patients presenting with suspected severe infection. Our findings indicate an overall radiological detection rate for identifying a definitive infectious focus, which aligns with intensive care literature highlighting the challenge of localizing an anatomical target in suspected sepsis [8, 9]. A notable finding was the high prevalence of non-infectious pathological findings that clinically mimicked sepsis. Comprehensive imaging often reveals unexpected pathologies that explain SIRS-like presentations [10, 11]. In our cohort, identifying these indeterminate pathologies proved to be highly relevant for patient disposition. The impact of IV contrast administration requires careful interpretation. While contrast-enhanced CT detected a larger proportion of non-infectious pathologies, it did not emerge as a statistically significant independent predictor for identifying an infectious focus in our multivariate analysis. Given the lack of a standardized protocol, this observation is undoubtedly confounded by selection bias, as physicians likely reserved contrast for patients with specific vascular concerns or robust renal function [12]. However, beyond selection bias, contrast-enhancement inherently improves the delineation of complex intra-abdominal fluid collections and soft-tissue hyperemia. Consequently, the significantly higher detection rate in the contrast group likely reflects a combination of targeted patient selection and the true diagnostic superiority of contrast for specific intra-abdominal etiologies. Nevertheless, recent studies suggest that unenhanced CT may be sufficient for initial localization of macroscopic infectious processes (e.g., pneumonia, large abscesses) in patients at risk for contrast-induced nephropathy [13, 14]. Thoracic infections predominated in our cohort (54.5%), reflecting an elderly demographic burdened by comorbidities. This echoes established epidemiological data identifying the respiratory tract as the leading source of severe sepsis [15]. Interestingly, the qSOFA score emerged as a strong independent predictor of a positive CT scan. While originally designed to predict mortality and severity [9, 16], our data suggest a hypothesis-generating observation: patients with a higher systemic severity burden (qSOFA ≥ 2) have a significantly higher probability of harboring a radiologically visible infectious focus. We do not propose qSOFA as an exact localization tool, but rather as a clinical severity marker that correlates strongly with imaging positivity. Clinicians might use lower qSOFA scores to exercise caution and minimize unnecessary radiation in low-risk scenarios. Furthermore, while our regression analysis identified CRP as an independent predictor, its clinical utility in isolation is limited. The effect size was marginal (OR = 1.009), and its discriminative ability in the ROC analysis was weak (AUC = 0.643). Indeed, while qSOFA demonstrated an acceptable discriminative capacity (AUC = 0.748) for predicting a positive scan, CRP was weak, and procalcitonin provided poor utility (AUC = 0.588) for anatomical localization. Elevated CRP reflects systemic inflammation but lacks the anatomic specificity required to serve as a standalone diagnostic triage tool. Similarly, traditional systemic markers such as WBC count and NLR were unable to independently predict the presence of a localized focus. While these markers reflect overall physiological stress [17, 18], their lack of anatomical specificity further underscores the utility of advanced imaging when the source is unclear. Regarding clinical management, TACT findings were strongly associated with downstream actions. Patients with a radiologically confirmed site of infection were substantially more likely to undergo surgical or percutaneous interventions. While our observational design precludes causal claims (i.e., CT alone dictating admission independent of clinical severity), the detection of structural abnormalities or drainable fluid collections serves as an essential guide for achieving effective source control [19]. From a practical standpoint, integrating qSOFA scores and baseline biomarkers into the initial ED evaluation algorithm can help clinicians identify which patients are most likely to yield actionable findings on TACT, thereby optimizing resource utilization and expediting targeted interventions. Our study has several limitations. First, the single-center, retrospective design carries an inherent selection bias, as we specifically included a high-risk ED cohort that underwent CT; thus, our findings cannot be broadly generalized. Second, the unstandardized administration of IV contrast acts as a significant confounder. Third, while we utilized a composite reference standard, this standard inherently incorporated the CT findings themselves. This introduces an incorporation bias that can significantly overestimate the diagnostic performance; therefore, our reported detection rate of 41.2% should be interpreted as a context-dependent finding rather than the intrinsic test performance of TACT. Fourth, the lack of a formal interobserver agreement (kappa) analysis limits the reproducibility and external validity of our radiological assessments. Without an estimate of diagnostic variability, it is important to recognize that our use of consensus reading likely resolved ambiguous cases favorably, which may further inflate the reported accuracy compared to real-world, single-reader emergency settings. Future prospective studies should incorporate independent, blinded re-evaluations to formally assess inter-radiologist reliability. Finally, our study establishes an association between CT findings and clinical management, rather than a direct causative pathway, as sicker patients are simultaneously more likely to receive CT scans and require hospital admission. Conclusion Thoracoabdominal CT yields a substantial radiological detection rate in ED patients with suspected infection, particularly among those with high qSOFA scores and elevated CRP levels. By identifying infectious sources and unmasking critical non-infectious mimics, TACT findings are strongly associated with hospital admission decisions and facilitate targeted source-control interventions. Acknowledgements None. Abbreviations Area Under the Curve Charlson Comorbidity Index Confidence Interval C-reactive protein Computed Tomography Emergency Department Interquartile Range Intravenous Length of Stay Neutrophil-to-lymphocyte ratio Odds Ratio quick Sequential Organ Failure Assessment Receiver Operating Characteristic Thoracoabdominal Computed Tomography White blood cell Author contributions A.E.A.: Conceptualization, methodology, formal analysis, writing-manuscript; M.E.: Methodology, investigation, data curation, formal analysis; F.C.T.: Formal analysis, writing-review and editing, supervision; A.K.: Formal analysis, writing-review and editing, supervision. Funding This research received no funding. Data availability The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate The study protocol was approved by the Necmettin Erbakan University Ethics Committee (Approval Number: 2026/6282, Date: 09.01.2026) and conducted in accordance with the principles of the Declaration of Helsinki, which waived the requirement for informed consent because of the retrospective nature of the study. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.