Study subjects were recruited through retrospective medical chart review. Patients who primarily visited Asan Medical Center (a tertiary medical center) via the outpatient clinic or emergency room and were subsequently diagnosed with deep neck infection from January 2007 to December 2015 were reviewed. Patients with deep neck infection were defined as those who had undergone at least one surgical drainage/debridement of the infection source in addition to being hospitalized for more than 7 days. Patients under the age of 18 and those with deep cervical infection due to underlying malignancy, tuberculosis, trauma, and foreign material insertion to the head and neck area were excluded. Additionally, patients who had undergone contrastenhanced computed tomography (ceCT) or surgical debridement in other medical centers, as well as those referred from other medical centers, were excluded. This study was approved by the Institutional Review Board of Asan Medical Center (IRB No. 2017-0031).

All patients underwent ceCT of the head and neck area, as well as the chest area in cases where infectious spread along the mediastinum was evident. The diagnosis of deep neck infection in each patient was confirmed by an official reading from a boardcertified radiologist. Furthermore, two sub-categories of deep neck infections were established. Accordingly, patients whose ceCT showed an evident abscess pocket(s) in the deep cervical space were defined as the “deep cervical abscess” group, whereas those suffering from severe necrosis with soft-tissue infection spreading along the deep cervical fascia and cervical muscles were defined as the “necrotizing fasciitis” group. Abscess pockets exclusively located at the peritonsillar space, oropharyngeal/laryngeal mucosa, and salivary glands were not considered as true “deep neck infections” and therefore not included in the study.

Each patient had undergone surgical debridement followed by hospital admission. Decisions regarding patient treatment were made by a board-certified head and neck surgeon. Upon initial patient encounter, prompt resuscitation of vital signs, securing of the upper airway, and peripheral blood laboratory/microbiology tests had been performed. Intravenous (IV) antibiotics were immediately administered empirically based on established guidelines. Upon surgical drainage of deep cervical abscesses, a paired swab and tissue culture had been obtained for the identification of underlying microorganisms. Thereafter, the regimen and dosage of antibiotic agents were modified based on the results of the bacterial culture study and consultations with an infectious disease specialist. Cases where in the infection had spread to the mediastinum through the prevertebral space were consulted with a cardiothoracic surgeon, and surgical intervention and drainage of pus were performed as needed. Some patients with severe bacteremia and sepsis despite adequate surgical and medical treatment needed aggressive intervention, including critical patient care in intensive care unit admission. In cases with persisting hypoalbuminemia, single-to-multiple IV administrations of 100 mL of 20% albumin solution were considered. However, this could only be performed upon the informed consent from the patient or their guardians. Patients were discharged when they showed no evidence of active infection/inflammation and were tolerant to reasonable amounts of oral feeding.

Demographic information (age and sex) was reviewed. The severity of the underlying comorbidity in each patient was evaluated using the Charlson comorbidity index (CCI) in addition to reviewing the patients’ medical history of hypertension, diabetes, chronic liver disease, chronic pulmonary disease, end-stage/chronic renal disease, and coronary artery disease. The ceCT images were used to determine the extent of involvement (unilateral/bilateral), number of involved cervical levels, and presence of simultaneous mediastinitis. Peripheral blood laboratory results, including a complete blood count, chemical battery, and CRP, were available for all patients. Follow-up for the complete blood count, chemical battery, and CRP was done daily or every other day for the first and second week of treatment. For the remaining hospitalization period, blood tests were performed twice a week. Initial serum albumin and BMI were used to assess the nutritional state of each patient. Moreover, 1-week follow-up CRP and the duration for an over 50% decrease in the initial CRP were used to determine responses to treatment. Clinical outcomes were evaluated using the total number of surgical drainage, total length of hospitalization, and overall survival.

Values for or proportion of patients with each clinical factor according to the initial and 2-week nutritional status were compared using Student t-test, chi-square test, and Fisher exact test. Odds ratios (ORs) for the development of each clinical manifestation and worse clinical outcomes with poorer nutritional status, older age, and poorer underlying conditions were calculated using univariate logistic regression analysis. Multivariate analysis of the ORs for each clinical result was performed after adjusting for age and CCI. Survival curves were plotted using the Kaplan-Meier method, while the log-rank test was used to compare overall survival and elapsed time for a 50% reduction in serum CRP according to initial serum albumin levels. All of the tests were two-sided, and a P-value <0.05 was considered statistically significant. Statistical analyses were performed using IBM SPSS ver. 21.0 (IBM Corp., Armonk, NY, USA).

A total of 135 patients satisfied the inclusion criteria for the current study (Table 1), among which 50 (31.6%) and 85 (53.8%) accounted for males and females, respectively, with a median age of 62 years. Accordingly, 19 patients (14.1%) had hypertension, 46 (34.1%) diabetes, four (3.0%) chronic hepatitis, five (3.7%) end-stage renal disease (ESRD), two (1.5%) chronic pulmonary disease, and 19 (14.1%) coronary artery disease, resulting in 73 patients (54.1%) with a CCI ≥1. The median serum albumin level upon initial presentation was 3.1 g/dL with 62 patients (45.9%) having initial hypoalbuminemia (serum albumin <3.0 g/dL). Moreover, the median BMI value was 23.0 kg/m² with 17 patients (12.6%) being underweight (BMI <18.5 kg/m²). A total of 30 patients (19.0%) showed bilateral disease involvement, while 85 (63.0%) had an infection that extended to more than one cervical level. The ceCT revealed a definite abscess pocket in 126 patients (93.3%) and necrotizing fasciitis in nine (6.7%). Simultaneous mediastinitis was observed in 19 patients (14.1%). The mean number of surgical drainage was 1.7 (standard deviation, 1.0) with a minimum of 1 and a maximum of 7, while the median duration of hospitalization was 14 days. Despite aggressive resuscitation and adequate control of the infection source, 11 patients (8.1%) died from sepsis followed by multiple organ failure.

Values for and proportion of patients with each clinical finding were compared according to initial serum albumin levels and BMI (Table 2). The proportion of patients with older age, simultaneous mediastinitis, necrotizing fasciitis, disease extent of more than one cervical level, and higher initial CRP was significantly greater in the hypoalbuminemia group than in the normoalbuminemia group (all P<0.05). The hypoalbuminemia group also showed poorer treatment responses and clinical outcomes, which resulted in longer duration of hospitalization, higher 1-week follow-up CRP, and longer time interval for a >50% decrease in the initial CRP level (all P<0.05). However, the number of surgical drainage and overall mortality rate did not differ between the two groups. On the other hand, no significant differences were found between the underweight and normalto-overweight BMI groups. ORs showed that individuals with initial serum albumin levels below 3.0 g/dL had a significant increase in the risk of developing simultaneous mediastinitis (2.96), necrotizing fasciitis (10.67), disease extent of more than one cervical level (2.18), and initial serum CRP over 20 mg/dL (4.20) (all P<0.05) (Table 3). In terms of clinical outcomes, the hypoalbuminemia group showed a significant increase in the risk for hospitalization of more than 14 days (4.90), 1-week followup CRP over 5 mg/dL (3.43), and longer time interval for an over 50% decrease in the initial CRP (3.10) (all P<0.05). Multivariate analysis showed significantly increased OR for more severe initial presentation and worse clinical outcomes as well (Table 4). To determine possible confounding clinical variables, univariate analysis was conducted for patients over 70 and those having a CCI >1 (Supplementary Table 1). In the elderly group (age >70 years), ORs suggested a significant increase in risk for necrotizing fasciitis (4.30), initial CRP >20 mg/dL (2.10), >14 days of hospitalization (3.11), and >1 week for a >50% decrease in the initial CRP (2.26). On the contrary, no significant increase/decrease in risk was observed for patients with CCI >1. Patients diagnosed with hepatic or renal failure showed a significant increase in the risk for necrotizing fasciitis (2.13), mediastinitis (2.16), and CRP >20 mg/dL (4.06) but not for clinical outcomes. Subsequently, a log-rank test was conducted for overall survival and time interval for a >50% decrease in initial CRP according to initial serum albumin levels, both of which showed worse outcomes and slower responses to treatment in the hypoalbuminemia group (both P<0.05) (Fig. 1).

A total of 73 patients hospitalized for longer than 14 days were assessed for their 2-week follow-up serum albumin levels and were subsequently categorized based on such findings. Hypoalbuminemia was observed in 41 patients (56.2%), whereas 32 (43.8%) had serum albumin levels greater than or equal to 3 g/dL (Supplementary Table 2). Despite the significantly higher proportion of patients receiving IV albumin replenishment in the normoalbuminemia group than in the hypoalbuminemia group (P<0.05), no significant differences in the demographic or clinical outcome parameters had been observed (Supplementary Table 2). In addition, IV albumin replenishment resulted in a 3.44-fold increase in the OR for 2-week normoalbuminemia and a 2.44-fold increase in the OR for less than 30 days of hospitalization (both P<0.05). However, ORs for overall survival and number of surgical drainage were not significant (Supplementary Table 3).