This retrospective, single-center, observational study was conducted at a tertiary care center at Tohoku University Hospital, Sendai, Miyagi, Japan. We investigated electronic clinical charts and hospital records to gather research variables from patients with HNC who underwent blood culture (BC) at the Department of Otolaryngology, Head and Neck Surgery, Tohoku University Hospital, between June 2009 and May 2023. All patients with HNC diagnosed with BSI were eligible for inclusion in this study. Exclusion criteria included possible contaminants, such as coagulase-negative staphylococci, Propionibacterium spp., and Corynebacterium spp.

The clinical characteristics were retrospectively examined using electronic clinical charts, hospital records, and microbiological data. Detailed information regarding the collected anamnestic and clinical data is provided in the Supplementary method S1. Microbial data (associated infection sites, organisms, and susceptibility) were extracted from the Infectious Diseases Department database. The focus of infection was identified by infectious disease specialists based on detailed physical examinations and confirmation of information in medical records. The primary outcome of this study was the 90-day mortality rate. Secondary outcomes included the risk factors for PB and mortality.

This study was conducted in accordance with the Declaration of Helsinki and approved by the Human Ethical and Clinical Trial Committee of Tohoku University Hospital (2018–1-736). The requirement for patient consent was waived due to the retrospective nature of the study.

The definitions of BC collection, BSI, PB, follow-up BC, PB duration, PB clearance, contamination, neutropenia, intravascular devices, source control, and adequacy of antimicrobial therapy were adopted from our previous report (13). Bacteremia was classified as nosocomially acquired, healthcare-related, or community-acquired by applying previously described criteria (17). The quick Sequential Organ Failure Assessment (qSOFA) score was calculated based on previously established definitions (18). Comorbidities were identified using the Charlson Comorbidity Scoring System (19). The 90-day mortality rate was defined as death within 90 days of the first BC collection.

The histological type and tumor grading were assessed in accordance with the International Classification of Diseases for Oncology, third edition (20), third edition first revision (21), and third edition second revision (22). Tumor staging was performed based on the Union for International Cancer Control TNM classification, including the sixth (2002) (23), seventh (2010) (24), and eighth (2017) editions (25). Immunohistochemical analysis of p16 expression in patients with oropharyngeal cancer served as an indicator of human papillomavirus status. Oral and laryngopharyngeal mucositis and dermatitis were classified according to the Common Terminology Criteria for Adverse Events Version 4.0 (26).

Details pertaining to the methodology for identifying microorganisms, measuring antimicrobial susceptibility, and performing statistical analyses are provided in the Supplementary methods S2, S4.

Figures 1A,B illustrate the temporal changes in hospitalization morbidity rates per 1,000 patient days and mortality rates for the PB and TB groups. Hospitalization morbidity rates per 1,000 patient days and mortality rates exhibited similar trends over time, with both groups peaking in 2011. In more recent years, an increased trend was observed in hospitalization morbidity and mortality rates for TB cases. Catheter-related bloodstream infection (CRBSI) accounted for a significant proportion of the TB and PB groups, serving as a focal point of infection.

The clinical characteristics of PB and TB in patients with HNC are shown in Table 1. During the study period, there were 97 cases of positive BCs in patients with HNC; of these, 14 (14%) exhibited PB. Significantly fewer cases of inappropriate antimicrobial use were observed in the PB group (0 cases; 0%) than in the TB group (23 cases; 28%; OR, 0; CI not applicable). In contrast, the number of cases with insufficient source control measures was significantly higher in the PB group than in the TB group (odds ratio [OR], 6.2; 95% confidence interval [CI] 1.4–27.1, p = 0.023).

In terms of infection sites, CRBSI was the most common infection in the PB and TB groups (8 cases [57%] and 36 cases [43%], respectively). In the PB group, pyogenic spondylitis (4 cases; 29%) and thrombophlebitis (2 cases; 14%) were the next most prevalent, whereas respiratory (10 cases; 12%) and urinary tract infections (8 cases; 10%) were more frequent in the TB group. The proportion of patients with pyogenic spondylitis was significantly higher in the PB group than in the TB group (OR, 16.2; 95% CI 2.6–100, p = 0.004). Additionally, the incidence of thrombophlebitis tended to be higher in the PB group than in the TB group, although it was not statistically significant (OR, 4.4; 95% CI 0.7–29.4).

Regarding the pharmacological treatment regimen, the proportion of biotherapy, including cetuximab + fluorouracil + cisplatin, cetuximab-RT, and nivolumab, showed a tendency to be higher in the PB group than in the TB group, although the difference was not statistically significant (OR, 4.4 [95% CI 0.7–29.4], OR, 6.3 [95% CI 0.4–107.2], and OR, 6.3 [95% CI 0.4–107.2], respectively). When comparing the 90-day mortality rates between the PB and TB groups, the PB group tended to exhibit a higher mortality rate (OR, 2.6; 95% CI 0.6–11.1), although no statistically significant difference was observed. Additionally, the group in which PB clearance was not achieved tended to have a higher mortality rate than the group in which PB clearance was achieved (OR, 9; 95% CI 0.5–155.2).

Regarding the causative microorganisms of PB and TB in patients with HNC, an overview of the bacterial species based on Gram staining is shown in Figure 2. Detailed bacterial genera and species names are presented in Supplementary Tables S2, S3, highlighting the variation in bacterial species between PB and TB cases, as well as between instances of mortality and survival. GPC tended to occur in a higher proportion of patients in the PB group than in the TB group (67% vs. 55%; OR, 1.6 [95% CI 0.5–5]). Conversely, there was a significantly lower proportion of GNR in the PB group than in the TB group (7% vs. 34%; OR, 0.1 [95% CI 0–1.1], p = 0.037). In the PB group, the most frequently identified causative pathogen was methicillin-resistant S. aureus (MRSA; 33%), followed by methicillin-susceptible S. aureus (MSSA) and Candida parapsilosis (13%). In contrast, the most frequently identified causative pathogen in the TB group was Staphylococcus epidermidis (20%), followed by MRSA (13%), MSSA (9%), and Pseudomonas aeruginosa (7%).

Table 2 presents the multivariate logistic regression analysis results. The factor independently associated with PB in patients with HNC was pyogenic spondylitis (adjusted odds ratio [aOR], 22.1; 95% CI 2–239.2], p = 0.011). Factors independently associated with mortality in patients with HNC were serum albumin levels (aOR, 0.1; 95% CI 0–0.8, p = 0.023) and clinical stages I–II (aOR, 12.9; 95% CI 1.8–93.4, p = 0.011).

The hospitalization morbidity rate per 1,000 patient days and hospitalization mortality rate both peaked in 2011. Furthermore, the hospitalization morbidity and mortality rates in the TB group has exhibited an increasing trend since 2022. Since 2011, the medical team managing HNC has actively implemented infection control measures, specifically by increasing opportunities for BC collection and ensuring the collection of two sets of BCs (Supplementary Figure S1). Additionally, until mid-2011, the primary chemotherapeutic regime was docetaxel + cisplatin + fluorouracil, resulting in a high incidence of febrile neutropenia due to hematologic toxicity. These factors may have contributed significantly to the notable increase in hospitalization morbidity rate per 1,000 patient days and hospitalization mortality rate in 2011. More recently, there has been a renewed increase in the hospitalization morbidity and mortality rates in the TB group in our study. It is imperative to monitor infection control measures and treatment management within hospital wards to prevent further increases in the mortality rate of patients with BSIs and HNC.

Recent technological advancements in the healthcare industry have improved surgical techniques, including those used for treating HNC. In the field of chemotherapy, molecular targeted therapy and immunotherapy have emerged alongside conventional treatment regimens (27, 28). Despite the presumed improvement in hygiene management during patient care, the number of CRBSI cases in our department remains high. In the Department of Otolaryngology and Head and Neck Surgery, laryngoscopy is frequently performed during hospitalization to monitor the treatment progress and airway management. Additionally, procedures such as suction with sputum dispersion and postoperative wound care are often performed within the examination units of the ward. In patients with HNC, central venous catheters or ports are frequently inserted for chemotherapy and nutritional supplementation. In these patients, there is a potential risk of contracting CRBSI, especially during techniques involving aerosol dispersion, where droplets may adhere to the catheter surroundings or be transmitted through contact with healthcare personnel. Although our department generally follows the CRBSI guidelines (29), infection prevention strategies focused solely on managing the catheter insertion site have limitations. Thus, it is desirable to reduce the environmental bacterial load alongside catheter management. Our hospital’s infection control team regularly inspects various medical and surgical units, providing infection management guidance. Particularly for patients with antimicrobial-resistant bacteria, the infection control team recommend isolated treatment units, frequent cleaning and disinfection of equipment and chairs, and environmental disinfection using ultraviolet irradiation devices (30). In addition to managing the catheter-insertion site, implementing these environmental disinfection measures to lower the environmental bacterial load is crucial to decrease the incidence of infections, including CRBSI, in patients with HNC. Furthermore, fostering collaboration between the infection control team and medical staff to increase infection management awareness, informed by current issues, proves to be an effective approach.

Malnutrition or enhanced vascular permeability can lead to low serum albumin levels, which are linked to heightened infection and BSI risks (16). Patients with esophageal cancer with low pretreatment serum albumin levels reportedly have a higher incidence of complications, such as BSIs, respiratory failure, arrhythmias, and heart failure (40). Furthermore, patients with low pretreatment serum albumin levels had higher postoperative mortality rates, and pretreatment serum albumin levels have been identified as predictors of survival (40). In patients with HNC, low albumin levels are also a BSI risk factor, escalating both early and overall mortality rates (16). Our multivariate analysis identified low albumin levels as a mortality risk factor. Thus, in patients with HNC, serum albumin levels are crucial prognostic factors and valuable for prognostic assessment alongside other clinical metrics.

Patients in clinical stages I and II of HNC exhibited a higher mortality risk compared to those in clinical stages III and IV. However, since the number of cases in clinical stages I and II in this study was small (n = 18 cases), the results may have been influenced by sample size bias (Supplementary Table S4).

In conclusion, to the best of our knowledge, this is the first study to investigate clinical features of PB in patients with HNC. Although this study was a single-center retrospective investigation with limitations in generalizability, it was conducted over a long period of 15 years, making the results significant. The key findings are summarized as follows: (1) Peaks in hospitalization morbidity rates per 1,000 patient days and mortality rates in the PB and TB groups were observed in 2011. An increasing trend has recently been observed in hospitalization morbidity and mortality rates for the TB group. (2) CRBSI was the main infectious focus in both groups, with a consistently high contribution to BSI. (3) Patients with HNC with pyogenic spondylitis had an elevated PB risk, particularly without source control. (4) MRSA-related BSI in patients with HNC led more often to PB and higher mortality than did MSSA-related BSI. (5) Pyogenic spondylitis increased PB risk, while hypoalbuminemia increased mortality in patients with HNC. (6) PB had a higher mortality rate than did TB, thereby underscoring the need for PB clearance to improve prognosis.