In the present study, we investigated the characteristics of OCNF in comparison with OCSC and BNF. Multivariate analyses showed that NLR ≥15.3 and LRINEC score ≥6 points were significantly related to OCNF in OCSI. On the other hand, univariate analyses of NF showed that CRP, WBC and clinical stages were significant factors. Furthermore, the OCNF group had much lower mortality than the BNF group.
In the patients with OCSI, univariate analyses showed that age, compromised host, CK, CRP, WBC, Hb, Cr, Alb, NLR ≥15.3, LRINEC score ≥6 points were significantly associated with OCNF. Simonart et al. reported that CK is important for the early diagnosis of BNF [23]. Several reports have shown that infection-driven inflammation correlates with markers of malnutrition and inflammation, such as CRP, WBC, Hb, Cr, and Alb in blood tests [24, 25]. When inflammation becomes severe, CRP, WBC, and Cr tend to increase, while Hb and Alb tend to decrease [24, 25]. In the present study, OCNF showed such a tendency significantly when compared with OCSC. Several reports have shown that LRINEC score is useful for predicting BNF [26-30]. Most of the reports set LRINEC score cutoff value at 6 points, which is similar to the present study. However, only 2 reports have investigated the usefulness of LRINEC score for predicting OCNF [31, 32]. Although both reports both set LRINEC score cutoff value at 6 points and the same definition of OCNF and OCSC, opinions on the usefulness of the score were exact opposite [31, 32]. This may be because both reports were quite different in the proportion of OCNF in OCSI patients (17 vs. 70 and 16 vs. 595) [31, 32]. In this study, the proportion was 14 vs. 217 and the definitions of OCNF and OCSC were also similar to previous studies [31, 32]. No reports have investigated the relationship between NLR and NF. Several reports have shown that NLR can predict the outcomes of patients with cancers [33-37]. We found that NLR is also useful for predicting OCNF in OCSI.
In the patients with NF, the OCNF group had significantly higher inflammatory markers than the BNF group, but significantly lower clinical stages when diagnosed clinically. Furthermore, the OCNF group (7.7%) had much lower mortality than the BNF group (35.3%) These findings showed that when compared with BNF, OCNF can be detected at a lower clinical stage and therefore, the surgeons may be better prepared to save lives. This may be because, compared to BNF, OCNF is easily noticed at an earlier stage (in the state of OCSC or early stage of OCNF), by patients themselves or their family members, as a complication in the oro-cervical region, such as trismus, dysphagia, and skin flare. NF makes superficial-fascia with poor blood flow the main base of bacterial infection; however, several reports showed that because of the abundant blood supply in oro-cervical regions, OCNF is rare in comparison to BNF [13, 38, 39]. In addition, some reports have shown that the superficial cervical fascia is thinner than those of other parts including limbs and trunk [40, 41]. The thinness of the superficial cervical fascia may be associated with the patient's increased awareness of symptoms and, consequently, the rareness of OCNF. In fact, OCNF is reported in 2.6-5.0% among all cases of NF [1, 42]. Next, we considered the difference between OCNF and BNF from the aspect of NF type. Type 1 NF which is 70-80% of NF, occurs in immunocompromised individuals, such as patients with diabetes mellitus or chronic renal failure, infection of the oro-cervical region, abdominal wall, and surgical wounds is common, and gas production can be sometimes seen [16-18]. Type 2 NF can be caused by trauma to the limbs (especially lower limbs) even in young healthy people, with a very high mortality rate of 40% when it occurs [16-18]. In the present study, the causative bacteria were identified in half of the OCNF group (7 out of 14 cases), in which multiple bacterial species including anaerobic bacteria, such as Prevotella spp. and Peptostreptococcus spp. Most of the OCNFs were probably type 1 NF because the cases in which causative bacteria were not detected were also accompanied by a strong anaerobic odor and anaerobic bacteria are actually difficult to culture. In the BNF group, the causative strain was identified in 14 out of 17 cases. Of these, 9 were Streptococcus pyogenes, 5 were multiple strains including Bacteroides fragilis and Escherichia coli. In other words, more than half of the BNFs were type 2 NFs. In fact, the OCNF group (57.1%) had a higher number of compromised host patients than the BNF group (29.4%). In addition, the OCNF group (35.7%) had much more gas production than the BNF group (17.6%). This may be because OCSI is a dental infection that spreads through "spaces" of the oro-cervical regions, unlike BNF. When the infection reaches the fascia (i.e., OCNF), the infection has already spread to the spaces, which may also have been accompanied by gas production due to some anaerobes. This infection route of OCNF, which is different from that of BNF, may cause severe local symptoms early and may have led to different results regardless of the same NF.
However, this study has some limitations. First, there is a possibility of unknown confounding factors because this was a retrospective study. Next, a few BNF patients were excluded because they transferred to other hospitals for various reasons during this study. If they were added to the BNF group, other results may have been obtained (e.g., mortality of BNF group may have been significantly higher than that of the OCNF group).