*2.2. Blood Cultures*

Blood samples were obtained either via sterile venipuncture or from a central venous catheter (CVC) after thorough disinfection, according to the institutional standard. At least two pairs of BCs (aerobic and anaerobic, volume 8–10 mL each) were obtained and inoculated (BACTEC, Becton Dickinson, Heidelberg, Germany). The BC bottles were sent to the institutional laboratory and analyzed as previously described [21]. Samples were incubated for up to seven days and the institutional average time to positivity for this method was 13 h.

#### *2.3. Next-Generation Sequencing*

Blood samples were drawn under the same conditions as mentioned above, collected into stabilizing blood tubes (Cell-Free DNA BCT CE, Streck, La Vista, NE, USA), and shipped at ambient temperature by a medical logistics service provider to a specialized laboratory (Noscendo GmbH, Reutlingen, Germany). Blood samples were separated to plasma by centrifugation at 1600× *g* for 10 min at 4 ◦C and the plasma supernatant was transferred to a fresh reaction tube. Then, a second centrifugation step at 16,000× *g* for 10 min at 4 ◦C was performed, supernatants were again transferred, and plasma aliquots were further stored. Nucleic acid isolation, quality controls, and library preparation were carried out as previously described [22]. Adequate positive and negative controls accompanied all laboratory and sequencing procedures. All data generated were analyzed using Noscendo's DISQVER® platform, which comprises a curated microbial genome reference database of over 16,000 microbial species, including bacteria, DNA viruses, fungi, and parasites, while potential contaminations and commensals were discriminated from infective agents based on statistical calculations [23]. The analysis time for this method is specified as less than 24 h after the sample is received by the laboratory. Reports were accessible to the treating clinician via an online portal after email notification.

#### *2.4. Virology*

Additional diagnostic tests for viruses from blood samples were only conducted if viral DNA was detected by NGS and considered as potentially clinically relevant, or if a viral infection was clinically suspected by the attending physician. The routine virus detection panel for blood samples was performed by real-time PCR, as per institutional protocol, and included herpes simplex virus type 1 and 2 (HSV-1/2), Epstein–Barr virus (EBV), and cytomegalovirus (CMV).

## *2.5. Data Collection*

Data were collected retrospectively through a standardized case report form from electronic and paper medical records. Data included patient demographics, length of ICU and hospital stay, major comorbidities, and discharge data. Clinical data obtained on admission and on the day of sample collection included relevant laboratory data, such as C-reactive protein (CRP), procalcitonin (PCT), and leucocytes, as well as the Sequential Organ Failure Assessment (SOFA) score, therapeutic measures, such as mechanical ventilation, renal replacement therapy, and vasopressor support, antimicrobial treatment, and infectious source control measures. Results from other routine microbiological tests (tracheal secretions, drainage samples, urine, orsamples from surgical sites) were performed within three days, either, before or after blood sample collection for NGS diagnostics were included in the evaluation. Changes in antimicrobial therapy and infectious source control procedures within seven days of NGS sampling were reviewed. Additionally, in patients with viruses detected by NGS, medical records were screened for virological tests and antiviral therapies.

### *2.6. Data Review*

A panel composed of at least two intensivists and one microbiology specialist examined medical files, including clinical parameters, previous course of the disease, consultations with infectious disease specialists, antimicrobial treatments, and results from pathogen diagnostics. Results of the NGS analysis were assessed for clinical relevance and categorized as to whether the results (1) provided any additional or unique findings, (2) led to further diagnostic measures, or (3) affected antimicrobial therapy. The therapeutic impact was further distinguished between (I) initiation of additional antimicrobial treatment, (II) confirmation of therapy already initiated, or (III) discontinuation of ongoing antimicrobial treatment. Identification of typical BC contaminants, such as CNS, was assessed for clinical relevance. Contamination was assumed if the suspected isolates were considered as such, either according to clinical documentation or were present in only one BC and no further action was taken.

#### *2.7. Statistics*

Statistical analysis was performed using IBM SPSS statistics software version 28.0 (IBM Corp., Armonk, NY, USA). Continuous data are presented as the median and interquartile range, categorial data are presented as counts and percentages. Quantitative variables were compared using the Mann-Whitney test, qualitative variables were compared using the Chi-square test or Fisher's exact test, as appropriate. A *p*-value < 0.05 was considered statistically significant.

### **3. Results**

#### *3.1. Study Population*

A total of 25 cases with simultaneous BC and NGS sampling were identified for analysis. Demographic data and clinical variables are presented in Table 1. The median (IQR) age was 70 years (56.5–76.5) and patients were predominantly male (16/25, 64%). Patients with positive NGS results were older (75 vs. 59.5 years; *p* = 0.03) and a history of cardiovascular disease was more common in these patients (77% vs. 33%; *p* = 0.03). Other demographical or clinical variables were similar in cases with positive or negative NGS results. The median (IQR) length of ICU stay was 20 days (11–33.5), and in-hospital mortality was 64% (16/25). At the time of sampling, the median (IQR) SOFA score was 8 (6.5–10.5) with 52% of patients (13/25) requiring invasive mechanical ventilation and 88% of patients (22/25) depending on vasopressor support. Antimicrobial therapy was administered in 56% of cases (13/25) and did not significantly affect NGS or BC positivity rates (NGS *p* = 0.32, BC *p* = 0.67). All samples were collected ≥ 48 h after hospital admission.

**Table 1.** Demographic data and clinical variables at the time of sampling. Demographic and clinical data showed no relevant differences regarding NGS positivity. Data are presented as the median and interquartile range or as counts and percentages, as appropriate. CRP: C-reactive protein; ICU: intensive care unit; IL-6: interleukin 6; NGS: next-generation sequencing; PCT: Procalcitonin; SOFA: sequential organ failure assessment.


#### *3.2. NGS and BC Results*

Results from 25 NGS tests and 61 sets of BC (minimum of two sets per case) from 25 COVID-19 ICU patients with suspected BSIs were assessed. At least one isolate was detected by NGS or BC in 64% of cases (16/25) and the combination of NGS/BC found 31 microorganisms, including 22 bacteria, 1 fungus, and 8 viruses. An overview of all detected isolates is presented in Table 2.

**Table 2.** Microorganisms detected by NGS, BC, and PCR. NGS detected 23 isolates (14 bacteria, 1 fungus, and 8 viruses) and provided positive results in 52% (13/25) of cases, whereas BC identified eight bacteria in 28% (7/25) of cases. The most frequently detected bacteria were Enterococcus species in NGS and coagulase-negative staphylococci in BC. Following the identification of viruses by NGS, PCR confirmed four out of five viruses and detected two further isolates of HSV-1. † Coagulasenegative staphylococci. BC: blood culture; HSV-1: herpes simplex virus type 1; NGS: next-generation sequencing; PCR: polymerase chain reaction.


Accounting for all pathogens, NGS showed a statistically non-significant higher positivity rate than BC (NGS: 52%, 13/25 vs. BC: 28%, 7/25; *p* = 0.083). NGS identified 23 isolates in total (14 bacteria, 1 fungus, and 8 viruses), whereas BC only detected eight bacterial species (*p* = 0.20).

Contamination of positive samples was less frequent in NGS (15%, 2/13) than in BC (57%, 4/7; *p* = 0.12). Contaminants found in NGS were *Xanthomonas campestris* (*n* = 1) and *Corynebacterium imitans* (*n* = 1), whereas contamination in BC was only caused by CNS (*n* = 5). These isolates were excluded from the analysis.

The sensitivity for potentially relevant pathogens for NGS and BC combined was 48% (12/25). A comparison of NGS and BC regarding potentially clinically relevant results revealed that NGS provided significantly more positive results than BC (*p* = 0.01). NGS returned positive results in 48% of cases (12/25) and identified 12 bacteria, 1 fungus, and 8 viruses. Three bacteria considered relevant were detected by BC in 12% of cases (3/25).

#### *3.3. Direct Comparison of NGS and BC*

Excluding viruses from NGS results for a more direct comparison, positive results were detected in 36% of cases (9/25) by NGS and in 12% of cases (3/12) by BC (*p* = 0.05). Both methods returned positive results in three cases and agreed on two bacterial species (*Escherichia coli* and *Enterococcus faecium*). In the remaining specimen, NGS and BC provided inconsistent findings (NGS: *Staphylococcus aureus*, *E. faecium*, *Serratia marcescens*; BC: *Staphylococcus epidermidis*). In nine cases with positive NGS results (seven bacteria, one fungus, and seven viruses), BC analysis remained negative. In 13 cases, no clinically relevant pathogen was found by either method.

#### *3.4. Diagnostic Benefit of NGS*

Compared to BC alone, additional NGS diagnostics provided further information in 44% of cases (11/25). NGS identified ten bacteria and one fungus that remained undetected in simultaneously collected BC samples.

Even when including results from other routine microbiological tests, such as surgical swabs, tracheal secretions, and urine, NGS identified four bacteria and one fungus, which were not found in conventional diagnostics (*E. faecium*, *n* = 2; *Bacteroides fragilis*, *n* = 1; *Helicobacter pylori*, *n* = 1; *Candida parapsilosis*, *n* = 1). NGS confirmed a systemic infection in four cases by detecting six bacteria in the bloodstream, which were otherwise only identified in samples taken directly from the septic focus (abdomen, *n* = 1, *E. faecium*, *Enterococcus. raffinosus*; lung *n* = 3, *E. coli*, *S. marcescens*, *S. aureus*, *Klebsiella pneumoniae*).

Since routine screening for viraemia was not routinely performed, NGS revealed eight viral isolates in seven specimens (EBV: *n* = 4; HSV-1: *n* = 2; CMV: *n* = 2), which would otherwise have remained undetected.

#### *3.5. Additional Viral Diagnostic*

In four out of seven cases with positive viral NGS, additional PCR confirmed four out of five viruses. EBV and HSV-1 were each confirmed in two out of two cases by PCR. Additionally, HSV-1 was detected by PCR in two further samples, which were missed by NGS. In one patient, positive HSV-1 NGS results led to the suspicion of herpes simplex encephalitis, which was consequently excluded by PCR from cerebrospinal fluid after lumbar puncture.
