Bacteriology of peritonsillar abscess: the changing trend and predisposing factors

Introduction Peritonsillar abscess is the most common deep neck infection. The infectious microorganism may be different according to clinical factors. Objective To identify the major causative pathogen of peritonsillar abscess and investigate the relationship between the causative pathogen, host clinical factors, and hospitalization duration. Methods This retrospective study included 415 hospitalized patients diagnosed with peritonsillar abscess who were admitted to a tertiary medical center from June 1990 to June 2013. We collected data by chart review and analyzed variables such as demographic characteristics, underlying systemic disease, smoking, alcoholism, betel nut chewing, bacteriology, and hospitalization duration. Results A total of 168 patients had positive results for pathogen isolation. Streptococcus viridans (28.57%) and Klebsiella pneumoniae (23.21%) were the most common microorganisms identified through pus culturing. The isolation rate of anaerobes increased to 49.35% in the recent 6 years (p = 0.048). Common anaerobes were Prevotella and Fusobacterium spp. The identification of K. pneumoniae increased among elderly patients (age > 65 years) with an odds ratio (OR) of 2.76 (p = 0.03), and decreased in the hot season (mean temperature > 26 °C) (OR = 0.49, p = 0.04). No specific microorganism was associated with prolonged hospital stay. Conclusion The most common pathogen identified through pus culturing was S. viridans, followed by K. pneumoniae. The identification of anaerobes was shown to increase in recent years. The antibiotics initially selected should be effective against both aerobes and anaerobes. Bacterial identification may be associated with host clinical factors and environmental factors.


Introduction
Peritonsillar abscess (PTA), or quinsy, is the most common deep neck infection. 1 The abscess may spread into the parapharyngeal space of other deep neck spaces, to the adjacent structure, and to the bloodstream. It rarely occurs but PTA is potentially life threatening. Early diagnosis of PTA is extremely crucial, and appropriate antibiotics and surgical intervention to remove the abscess are required. 2 Antibiotics result in a substantial reduction in the progression of this disease. The empirical antibiotic used should be effective against the possible causative pathogen of PTA.
Our objectives were to investigate the microbiology of PTA and to identify its relationship with clinical variables including the underlying systemic disease of patients; habits such as smoking, alcoholism, and betel nut chewing; and hospitalization duration.

Study design and sample population
This retrospective study included 415 patients with PTA who were admitted to a tertiary medical center located in Southern Taiwan from June 1990 to June 2013. Inclusion criteria were hospitalized patients who were clinically diagnosed with PTA (ICD-9 code 475) by positive pus aspiration or computed tomography (CT) imaging. We reviewed the chart of each patient to collect the following data: admission date, age, sex, height, weight, host clinical factors (diabetes mellitus [DM], hypertension, smoking habit, alcoholism, and betel nut chewing), pus culture result, antibiotic treatment, surgery, and hospitalization duration. The study was approved by the institutional review board.
We classified the bacteria into different categories according to the characteristics of Gram staining and anaerobic properties. We defined prolonged hospitalization as hospitalization duration of more than 6 days. Obesity was defined as a body mass index of more than 27, and elderly patients were defined as those aged older than 65 years. We defined the hot season as the months from May to October when the average temperature in Southern Taiwan was above 26 • C according to the record of the Central Weather Bureau of R.O.C.

Statistical analysis
All data were analyzed using the SPSS statistical software (IBM Corp., Armonk, NY, USA), except for the Cochran---Armitage test, which was performed using the SAS program (SAS Institute, Cary, NC, USA). The association with each independent variable was statistically analyzed among the different groups. Categorical variables were compared using the Pearson's Chi-square test or the Fisher's exact test, as appropriate. Odds ratios (ORs) and their 95% confidence intervals (CIs) were calculated. Trends of isolated pathogens    were analyzed using the Cochran---Armitage test. A p-value less than 0.05 was considered statistically significant.

Ethic statement
This study has been approved by the Institutional Review Board; the approval protocol number is VGHKS14-CT7-01.

Demographraphic characteristics
This study included 415 patients. The results of pus cultures from either surgery or needle aspiration were available for 266 patients. Adjustments for sample submitted to tonsil surgery or PTA drainage was performed, as shown in Table 1.
There is no patient with history of AIDS or HIV infection in this study.  Table 3 and  Table 2.

Bacteriology
Host clinical factors were associated with several isolated pathogens. Betel nut chewing was associated with the isolation of gram-positive cocci (GPC) (OR = 2.67, p = 0.04). The association of bacterial isolation with smoking habit and alcoholism was not statistically significant. Elderly patients (age > 65 years) had higher K. pneumoniae isolation (OR = 2.76, p = 0.03). Obesity (BMI > 27) was associated with a higher isolation of Peptostreptococcus (OR = 4.19, p = 0.04), as shown in Table 4.
In addition, in the hot season, we found that the risk of isolating gram-positive bacilli (GPB) increased (OR = 3.22, p = 0.02), but that of K. pneumoniae isolation decreased (OR = 0.49, p = 0.04), as shown in Table 4. There was no specific microorganism associated with prolonged hospital stay.
Searching from the PubMed database, there were 30 studies involved in bacteriology of PTA during 1980---2016. The timeframes, the geographical locations and the predominant bacterial species identified in these studies were listed in Table 5.
Several broad-spectrum antibiotics such as penicillin or cefazolin combined with gentamycin (GM) and metronidazole, clindamycin plus GM, or augmentin along were used in our series. All these antibiotics were effective without any significant difference.  ---, indicates ''not disclosed''.

Discussion
In our study, the most common pathogen identified through pus culturing in patients with PTA was S. viridans, followed by K. pneumoniae; commonly isolated anaerobes in our study were Prevotella and Fusobacterium spp. We reviewed the bacteriology data from previous studies, as shown in Table 5. Most of the studies 3--- 16 have reported group A Streptococcus as the most common aerobic pathogen in PTA; some studies 12,17,18 have reported that common aerobic pathogens were S. viridans, followed by group A ␤-hemolytic streptococci. The prevalence of K. pneumoniae has been rarely reported in previous studies. In previous studies, Fusobacterium nucleatum, 3,8,11,12,15,19,20 Prevotella, 3,12,19---21 Bacteroides, 7,8,19 Peptostreptococcus, 8,9,20 and anaerobic streptococcus 12 were the most common anaerobic pathogens. The divergence of bacterial culture may be owing to different geographical location. With difference between diets and lifestyle, the bacterial flora within each people may be also different. K. pneumoniae and Streptococcus spp. are common oral flora normally found in the mouth and are odontogenic pathogens of deep neck infection. 22---24 The S. viridans group is the etiological agent of dental caries, pericoronitis, or, if introduced into the bloodstream, endocarditis. In Taiwan, K. pneumoniae has been linked to lung infection in aspiration patients or a liver abscess 25 in immunocompromised patients or those with diabetes. 26 Patients with old age 27 or diabetes mellitus 28 are considered to be immunocompromised and have more chance to get infection. DM and elder are also linked with more complications and higher mortality rate in deep neck infection. 29,30 Thus PTA patients with above characteristics often have longer hospital stay. 30 We reported the microbiology of PTA in such immunocompromised patients. Patients with DM had no increased risk of isolating K. pneumoniae as the causative pathogen of PTA. By contrast, elderly patients with PTA in the current series had a higher risk of K. pneumoniae isolation. Such a phenomenon might result from a real change in pathogens; the alteration of antibiotics used, or improved culture methods for anaerobic pathogens. In our series, no major alteration of antibiotics used or improvement of the culture methods was observed. Physicians should prescribe empirical antibiotics to cover anaerobes.
PTA is often a polymicrobial infection. Polymicrobial growth was observed in the pus cultures of 57.39% of patients. The rationale of using empirical antibiotics was to cover GPCs, GNBs, and respiratory anaerobes. If necessary, suitable antibiotics should be chosen on the basis of culture results. However, the management of most uncomplicated patients may not be affected by the culture result. 31 Repanos et al. 32 suggested that using broad-spectrum antibiotics such as cephalosporin or penicillin combined with metronidazole was effective. In our study, no significant difference was found among several combinations of broadspectrum antibiotics.
Smoking habit has been commonly observed in patients with PTA in several studies 17,18,33,34 ; these studies have reported smoking as a risk factor for PTA. Marom et al. 17 reported a significantly higher incidence for S. viridans, other gram-positive cocci isolates, and anaerobes. In our study, no statistical significance was observed in the causative pathogen between smokers and nonsmokers with PTA, similar to the findings of the study by Klug. 34 Betel nut chewing is a popular habit in Southeast Asia. To the best of our knowledge, no study has found an association between the bacteriology of PTA and betel nut chewing. In our series, this habit was associated with a higher risk of GPC as a pathogen. In the study by Ling et al., 35 it was associated with a likelihood of subgingival infection by Actinobacillus actinomycetemcomitans and Porphyromonas gingivalis.
In our study, elderly patients (older than 65 year-old) had a high risk of K. pneumoniae isolation. The study by Marom 17 reported a significantly higher isolation rate for infection by GPC (mixed Streptococcus species) and gram-negative rods in older patients (40 year-old or older) than in younger patients.
The hot season increased the risk of GPB infection and reduced the risk of K. pneumoniae infection in patients with PTA in our current study. Our institute is located in a tropical region that has approximately six months (May to October) of hot weather, with a mean temperature of 27 • C. By contrast, Klug et al. 15 from another institute located in a temperate zone reported a higher incidence of F. nucleatum infection during summer than during winter. It also reported Group A streptococcus was significantly more frequently identified from in the winter and spring. The study in French 36 reported PTA caused by S. pyogenes or anaerobes were more prevalent in the winter and spring than summer. Such fluctuation in the microbiology of PTA might be weather related.
In our series, no specific microorganism was associated with the poor prognosis of PTA. This finding is considerably similar to the reports by Marom 17 and Mazur. 18 Our study has several limitations. Because we retrospectively collected data by chart review, data from the medical record might be lost during the early years. As we used several small populations of isolated pathogens, a larger sample size is necessary to determine the relationship between the isolated pathogen and the predisposing factors.

Conclusions
The most common causative pathogen of PTA was S. viridans, followed by K. pneumoniae. The isolation of anaerobes significantly increased in recent years. The common ones were Prevotella and Fusobacterium spp. Empirical antibiotics targeting both aerobes and anaerobes should be appropriate as treatment. Bacterial isolation may be associated with host clinical factors, environmental factors, and hospitalization duration.

Conflicts of interest
The authors declare no conflicts of interest.