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. 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. 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.

Methods

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.

Results

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 1

Demographraphic characteristics of patients with peritonsillar abscess.

Age	Overall, (n = 266)	Diabetes mellitus	Smoking	Alcoholism	Betel-nut chewing
<18 y/o	22 (8.27)	0	6 (27.27)	5 (22.73)	2 (9.09)
18–64 y/o	215 (80.83)	19 (8.84)	111 (51.63)	76 (35.35)	37 (17.21)
≥65 y/o	29 (10.90)	4 (13.79)	5 (17.24)	7 (24.14)	3 (10.34)
Total	266 (100)	23 (8.65)	122 (45.86)	88 (33.08)	42 (15.79)

Data are presented as n (%).

y/o indicates “year old”.

Bacteriology

Within these patients with pus obtained, 230 (230–266, 86.47%) showed bacterial growth in their pus culture. The pus culture of the remaining 36 patients showed no bacterial growth. Of the 230 patients, 132 (132–230, 57.39%) had polymicrobial pus, including 62 cases merely reported as “normal flora” or “mixed flora” (62–230, 26.96%). Pus cultures of 168 patients (168–266, 63.15%) showed positive results for pathogen isolation. More than a single pathogen was isolated in 64 patients (64–168, 38.10%). Aerobic bacteria were isolated from 85.7% (144/168) of positive cultures, anaerobic or facultative aerobic bacteria from 44.0% (74–168), and mixed aerobic and anaerobic bacteria from 29.8% (50–168).

The most common pathogen identified through pus culturing was Streptococcus viridans (48–168, 28.57%), followed by Klebsiella pneumoniae (39–168, 23.21%) and the beta-hemolytic Streptococcus group (17–168, 10.12%), as shown in Table 2. We divided patients by the 4 periods of 1990–1995, 1996–2001, 2002–2007, and 2008–2013; the isolation rate of the anaerobes was 25%, 23.81%, 45.45%, and 49.35%, respectively. The isolation rate of anaerobic pathogens increased significantly between 1990 and 2013 (Cochran–Armitage test, p = 0.048), as shown in Table 3 and Fig. 1. The isolation rates of gram-positive bacteria and gram-negative bacteria in these 4 periods were 100% and 25%, 57.14% and 47.62%, 62.12% and 48.48%, and 51.95% and 49.35%, respectively. Most of the anaerobic pathogens were Prevotella spp. (24–168, 14.29%) and Fusobacterium spp. (16–168, 9.52%), as shown in Table 2.

Table 2

Bacteriology of 168 patients with peritonsillar abscess with definite isolation of pus culture.

Causative pathogen	Overall (n = 168)	DM (n = 17)	HTN (n = 22)	Smoking (n = 78)	Alcoholism (n = 59)	Betel-Nut chewing (n = 27)	Obesity (n = 30)
Aerobic GNB	51 (30.36)	10 (58.82)	8 (36.36)	22 (28.21)	15 (25.42)	8 (29.63)	8 (26.67)
Klebsiella pneumoniae	39 (23.21)	7 (41.18)	7 (31.82)	19 (24.36)	13 (22.03)	7 (25.93)	7 (23.33)
Aerobic GPB	22 (13.10)	10 (58.82)	5 (22.73)	12 (15.38)	11 (18.64)	4 (14.81)	6 (20.00)
Aerobic GNC	4 (2.38)	1 (5.88)	1 (4.55)	1 (1.28)	1 (1.69)	1 (3.70)	2 (6.67)
Aerobic GPC	99 (58.93)	10 (58.82)	12 (54.55)	46 (58.97)	32 (61.02)	21 (77.78)	19 (63.33)
Staphylococcus spp.	9 (5.36)	0 (0.00)	1 (4.55)	5 (4.6)	5 (8.47)	1 (3.70)	2 (6.67)
Beta-hemolytic streptococcus group	18 (10.71)	1 (5.88)	1 (4.55)	15 (19.23)	10 (16.95)	4 (14.81)	4 (13.33)
Streptococcus melliri group	24 (14.29)	4 (23.53)	4 (18.18)	10 (12.82)	8 (13.56)	9 (33.33)	4 (13.33)
Streptococcus viridans group	48 (28.57)	5 (29.41)	7 (31.82)	16 (20.51)	13 (22.03)	7 (25.93)	7 (23.33)
Anaerobic cocci	34 (20.23)	5 (29.41)	5 (22.73)	17 (21.79)	13 (22.03)	7 (25.93)	11 (36.67)
Peptostreptococcus spp.	9 (5.36)	1 (5.88)	1 (4.55)	7 (8.97)	6 (10.17)	3 (11.11)	4 (13.33)
Anaerobic GPB	14 (8.33)	1 (5.88)	5 (22.73)	5 (6.41)	4 (6.78)	2 (7.41)	3 (10.00)
Anaerobic GNB	45 (26.79)	1 (5.88)	6 (27.27)	24 (30.77)	18 (30.51)	7 (25.93)	6 (20.00)
Fusobacterium spp.	17 (10.12)	0 (0.00)	1 (4.55)	10 (12.82)	6 (10.17)	3 (11.11)	2 (6.67)
Prevotella spp.	24 (14.29)	1 (5.88)	4 (18.18)	11 (14.10)	10 (16.95)	4 (18.18)	4 (13.33)

Data are presented as n (%).

Aerobic isolates included aerobic and facultative anaerobic isolates.

DM, diabetes mellitus; GNB, gram-negative bacilli; GNC, gram-negative cocci; GPB, gram-positive bacilli; GPC, gram-positive cocci; HTN, hypertension.

Table 3

Isolation rate of different types of bacteria during each 6 year interval, 1990–2013.

Years, number of patient types (%) of bacteria	1990–1995	1996–2001	2002–2007	2008–2013	Total (1990–2013)	Test for trends (p-value)
Gram positive (% of total patient)	4 (100)	12 (57.14)	41 (62.12)	40 (51.95)	97 (57.74)	0.120
Gram negative (% of total patient)	1 (25)	10 (47.62)	32 (48.48)	38 (49.35)	81 (48.21)	0.569
Anaerobes (% of total patient)	1 (25)	5 (23.81)	30 (45.45)	38 (49.35)	74 (44.05)	0.048a
Patient	4	21	66	77	168

Denotes for p-value less than 0.05.

Figure 1

Isolation rate of different types of bacteria during each 6 year interval.

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.

Table 4

Association between the predisposing factors and the pathogen.a

Predisposing factors	Causative pathogen	OR	95% CI	p-value
Elderb	KP	2.76	1.10–6.93	0.03e
Obesityc	Peptostreptococcus	4.19	0.98–17.88	0.04e
Hot seasond	GPB	3.22	1.13–9.19	0.02e
	KP	0.49	0.23–1.01	0.04e
Betel-Nut chewing	GPC	2.67	1.02–7.02	0.04e

No statistically difference was observed among bacterial isolates and smoking, alcoholism, and DM.

Elderly indicates patient's age was more than 65 years old.

Obesity indicates patient's body mass index was more than 27.

Hot season indicates the admission date was between May and October, during which time the average temperature in southern Taiwan was more than 27 °C.

Denotes for p-value less than 0.05.

CI, confidence interval; DM, diabetes mellitus; GPB, gram-positive bacilli; GPC, gram-positive cocci; HTN, hypertension; KP, Klebsiella pneumoniae; OR, odds ratio.

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.

Table 5

Studies involved in bacteriology of PTA during 1980–2016.

Investigator	Country	Year	Positive culture	Predominant aerobes	Predominant anaerobes
Brook et al. (1981)19	U.S.	–	16	Gamma-hemolytic streptococci	Bacteroides sp.
				Alpha-hemolytic streptococci	Anaerobic GPC
Jokipii et al. (1988)8	Finland	–	42	Group A streptococcus	Peptostreptococcus sp.
				Streptococcus viridans group	Bacteroides sp.
Brook et al. (1991)37	U.S.	1978–1985	34	Staphylococcus aureus	Bacteroides sp.
				Streptococcus pyogenes	Peptostreptococcus sp.
Snow et al. (1991)38	UK	–	55	Beta hemolytic streptococci	–
				Staphylococcus aureus
Jousimies-Somer et al. (1993)20	Finland	–	122	Streptococcus pyogenes	Fusobacterium necrophorum
				Streptococcus milleri group	Prevotella melaninogenica
Mitchelmore et al. (1995)9	UK	1982–1992	45	Group A streptococcus	Peptostreptococcus sp.
					Prevotella
Muir et al. (1995)39	New Zealand	1990–1992	39	Group A streptococcus	–
Prior et al. (1995)40Cherukuri (2002)	UK	–	45	–	–
	USA	1990–1999	82	Streptococcus sp.	–
				Haemophilus sp.
Matsuda et al. (2002)10	Japan	1988–1999	386	Alpha-hemolytic streptococci	Anaerobic gram-negative rods
				Neisseria sp.	Porphyromnas sp.
Hanna et al. (2006)41	Northern Ireland	2001–2002	37	Group A streptococci	Bacteroides sp.
Sakae et al. (2006)42	Brazil	2001	26	Streptococcus viridans	Peptostreptococcus sp.
				Streptococcus pyogenes	Prevotella sp.
Zagolski et al. (2007)	Poland	–	12	Streptococcus sp.	Bacteroides sp.
Megalamani et al. (2008)	India	2003–2006	39	Beta hemolytic streptococcus	–
				Pseudomonas
Sunnergren et al. (2008)7	Sweden	2000–2006	67	Group A streptococcus	Bacteroides sp.
Klug et al. (2009)11	Denmark	2001–2006	405	Group A streptococcus	Fusobacterium sp.
				Groups C or G streptococci
Gavriel et al. (2009)6	Israel	1996–2002	137	Streptococcus pyogenes	Prevotella sp.
				Streptococcus intermedius	Peptostreptococcus sp.
Segal et al. (2009)4	Israel	2004–2007	64	Group A streptococcus	–
				Group C streptococcus
Repanos et al. (2009)32	UK	1998–2005	107	Streptococcal sp.	–
Rusan et al. (2009)	Denmark	2001–2006	623	Group A streptococcus	Fusobacterium sp.
Acharya et al. (2010)5	Nepal	2007–2008	18	Streptococcus pyogenes	–
				Staphylococcus aureus
Marom et al. (2010)17	Canada	1998–2007	180	Streptococcus viridans	–
				Group A streptococcus
Hidaka et al. (2011)43	Japan	2002–2007	65	Streptococcus milleri group	Prevotella sp.
				Other Streptococcus sp.	Peptostreptococcus sp.
Klug et al. (2011)12	Denmark	2005–2009	36	Streptococcus viridans	Prevotella sp.
				Neisseria sp.	Fusobacterium sp.
Love et al. (2011)13	New Zealand	2006–2008	147	Group A streptococcus Other beta-hemolytic streptococci	Fusobacterium sp.
Albertz et al. (2012)16	Chile	2000–2012	112	Streptococcus pyogenes	Bacteroides sp.
				Other streptococci	Peptostreptococcus sp.
					Fusobacterium sp.
Takenaka et al. (2012)3	Japan	2005–2009	50	Streptococcus pyogenes	Anaerobic
					Streptococcus
					Fusobacterium sp.
Sowerby et al. (2013)14	Canada	2009–2010	42	Group A streptococcus	–
				Streptococcus anginosus
Gavriel et al. (2015)	Israel	1996–2003	132	Streptococcus pyogenes	Prevotella sp.
					Peptostreptococcus sp.
Mazur et al. (2015)18	Poland	2003–2013	45	Streptococcus viridans group	Fusobacterium sp.
				Streptococcus pyogenes	Prevotella sp.
Plum et al. (2015)44	USA	2002–2012	69	Streptococcus milleri in adults	–
				β-hemolytic streptococcus in children
Lepelletier et al. (2016)36	French	2009–2012	412	Group A streptococci	Fusobacterium spp.
Tachibana et al. (2016)45	Japan	2008–213	100	Streptococcus viridans	Fusobacterium sp.
Vaikjarv et al. (2016)46	Estonia	2011–2012	22	Streptococcus sp.	Streptococcus spp.
Present study (2017)	Taiwan	1990–2013	168	Streptococcus Viridans	Prevotella sp.
				Klebsiella Pneumoniae	Fusobacterium sp.

–, indicates “not disclosed”.

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.

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 studies3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16 have reported group A Streptococcus as the most common aerobic pathogen in PTA; some studies12, 17, 18 have reported that common aerobic pathogens were S. viridans, followed by group A β-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, 20, 21 Bacteroides,7, 8, 19 Peptostreptococcus,8, 9, 20 and anaerobic streptococcus 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, 23, 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 in immunocompromised patients or those with diabetes.

Patients with old age or diabetes mellitus 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. 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.

A trend toward a higher isolation rate for anaerobes was observed during 2002–2013 (p = 0.048). Gavriel reported a significant increase in anaerobic growth during 1996–1999 and then a slow nonsignificant decline until 2002. Takenaka reported no change in the percentage of cases with anaerobic growth between 2 periods (2005–2007 and 2008–2009). 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. Repanos et al. 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 broad-spectrum antibiotics.

Smoking habit has been commonly observed in patients with PTA in several studies17, 18, 33, 34; these studies have reported smoking as a risk factor for PTA. Marom et al. 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.

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., 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 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. 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 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 and Mazur.

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.