Flynn 2006

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

J Oral Maxillofac Surg

64:1104-1113, 2006

Severe Odontogenic Infections, Part 2:


Prospective Outcomes Study
Thomas R. Flynn, DMD,* Rabie M. Shanti,† and
Catherine Hayes, DMSc, DMD‡

Purpose: The purpose of this study was to identify significant predictors of 4 outcomes in patients with
severe odontogenic infections: abscess formation, penicillin therapeutic failure (PTF), length of hospital
stay (LOS), and need for reoperation.
Patients and Methods: We used a prospective case series study design and enrolled 37 consecutive
patients admitted for severe odontogenic infection between March 1996 and June 1999. Treatment
consisted of intravenous penicillin (PCN) or clindamycin in PCN-allergic patients, surgical incision and
drainage, and extraction(s) as soon as possible. Study variables were categorized as demographic,
preadmission, time-related, preoperative, anatomic, treatment, microbiologic, and complications. The
primary outcome variables were abscess formation, PTF, LOS, and reoperation. Multivariate linear and
logistic regression techniques were used to measure associations between study variables and the
outcome variables.
Results: The sample consisted of 37 subjects (23 male, 14 female) with a mean age of 34.9 ⫾ 15.8 years.
Multivariate analyses, controlling for confounding variables, indicated that culture of Peptostreptococci
was a negative predictor of abscess formation. LOS was predicted by the number of infected spaces and
duration of operation. There was no significant predictor of PTF or reoperation on multivariate analysis,
although PCN-resistant organisms were isolated in all cases of PTF.
Conclusion: Increased LOS in severe odontogenic infections is predicted by the anatomic extent and
severity of the infection and the occurrence of complications such as PTF and the need for reoperation.
PTF is significantly associated with later identification of PCN-resistant organisms. The role of Peptostrep-
tococci in abscess formation warrants further investigation.
© 2006 American Association of Oral and Maxillofacial Surgeons
J Oral Maxillofac Surg 64:1104-1113, 2006

Dodson et al,1 Biederman and Dodson,2 Peters et al,3 The predictor variables identified in these studies
and Huang et al4 identified predictor variables associ- included admission temperature, admission white
ated with increased hospital stay and other unfavor- blood cell count (WBC), systemic diseases, and in
able outcomes, such as the need for surgical drainage. pediatric patients over the age of 5, lower face infec-
tions and odontogenic etiology.
The specific aim of this study was to identify pre-
*Assistant Professor of Oral and Maxillofacial Surgery, Harvard dictors of unfavorable outcomes in patients with se-
School of Dental Medicine, Boston, MA; and Associate Visiting vere odontogenic infections (OI): 1) abscess forma-
Surgeon, Massachusetts General Hospital, Boston, MA. tion; 2) penicillin therapeutic failure (PTF); 3) length
†Howard Hughes Medical Institute-National Institutes of Health Re- of hospital stay (LOS); and 4) complications. We hy-
search Scholar, National Institutes of Health, Bethesda, MD; and Pre- pothesize that we will be able to identify 1 or more
doctoral Candidate, Harvard School of Dental Medicine, Boston, MA. variables associated with these unfavorable out-
‡Associate Professor of Oral Health Policy and Epidemiology, comes.
Harvard School of Dental Medicine, Boston, MA.
Supported in part by the Montefiore Medical Center Department
of Dentistry and the Massachusetts General Hospital Department of
Patients and Methods
Oral and Maxillofacial Surgery Education and Research Fund.
Address correspondence and reprint requests to Dr Flynn: Har- STUDY DESIGN/SAMPLE
vard School of Dental Medicine, 188 Longwood Ave, Boston, MA In this study we used a prospective case series
02115; e-mail: [email protected] design in which consecutive patients with OIs severe
© 2006 American Association of Oral and Maxillofacial Surgeons enough to justify hospitalization were treated with
0278-2391/06/6407-0016$32.00/0 intravenous (IV) penicillin (PCN) or clindamycin (in
doi:10.1016/j.joms.2006.03.031 PCN-allergic patients) and surgical incision and drain-

1104
FLYNN, SHANTI, AND HAYES 1105

age (I&D) of all anatomic deep fascial spaces affected Table 1. SEVERITY SCORES FOR SEVERE
by cellulitis or abscess as soon as possible during their ODONTOGENIC INFECTIONS
hospital stay.
The subjects enrolled in this study presented for Severity Score Anatomic Space
care between March 1996 and June 1999 at 1 of 4 Severity score ⫽ 1 Vestibular
large urban hospitals served by the Montefiore Medi- (low risk to Subperiosteal
cal Center Department of Dentistry, including Monte- airway or vital Space of the body of the mandible
fiore Medical Center, Jack Weiler Hospital at the Al- structures) Infraorbital
Buccal
bert Einstein College of Medicine, North Central Severity score ⫽ 2 Submandibular
Bronx Hospital, and Jacobi Medical Center in New (moderate risk Submental
York, NY. To be eligible for enrollment, subjects had to airway or Sublingual
to be admitted to the hospital for evaluation and vital structures) Pterygomandibular
management of a severe maxillofacial infection of Submasseteric
Superficial temporal
odontogenic origin, determined by the attending oral Deep temporal (or infratemporal)
and maxillofacial surgeon. The patient had to consent Severity score ⫽ 3 Lateral pharyngeal
to enrollment using forms and procedures developed (high risk to Retropharyngeal
for this institutional review board-approved study. airway or vital Pretracheal
The criteria for hospital admission were: OI causing structures) Danger space (space 4)
Mediastinum
swelling in 1 or more of the deep fascial spaces of the Intracranial infection
head and neck, impending threat to the airway or vital
structures, fever greater than 101°F, need for general NOTE. The severity score for a given subject is the sum of the
severity scores for all of the spaces involved by cellulitis or abscess,
anesthesia, or the need for inpatient control of a based on clinical and radiographic examination.
concomitant systemic disease. Potential subjects were Flynn, Shanti, and Hayes. Severe Odontogenic Infections. J Oral
excluded from this study according to the following Maxillofac Surg 2006.
criteria: pregnancy, nonodontogenic cause (eg, trau-
ma-related or upper respiratory infection), and refusal
of consent. Previously published nomenclature and The demographic variables recorded were age, gen-
descriptions of the deep fascial spaces were used for der, and race. Preadmission variables were: smoking,
the purposes of this study.5,6 drug allergies, preadmission antibiotic therapy, and
the presence of immunocompromising diseases. The
TREATMENT METHODS time-related variables were: the number of preopera-
All patients were subjected to the treatment proto- tive days of pain, preoperative days of swelling, LOS,
col described in our previous report.7 IV PCN 2 mil- operating room time, time between admission and
lion units was administered every 4 hours. PCN-aller- surgery, and season of the year. To maximize the
gic patients received IV clindamycin 900 mg every 8 accuracy of anamnestic data, such as the number of
hours. Broad-spectrum antibiotic therapy with genta- days of preoperative swelling, questions were care-
micin, metronidazole, and clindamycin was used in fully scripted on the data recording form, limited to
patients with necrotizing fasciitis. All patients were the current episode of infection, and verified by the
taken to the operating room for I&D of all anatomic attending surgeon.
spaces affected by cellulitis or abscess as soon as Preoperative clinical variables included causative
possible after admission. Culture and sensitivity test- teeth, number of teeth involved, dental diagnosis
ing was performed at surgery for all patients. (such as caries, periodontal disease, or pericoronitis),
PTF was defined as the need to discontinue PCN dyspnea, dysphagia, trismus (maximum interincisal
because of: 1) toxic or allergic reactions; 2) develop- opening ⱕ20 mm), WBC, and admission core temper-
ment of necrotizing fasciitis; or 3) failure of clinical ature.
improvement 48 hours or more after I&D plus post- The anatomic variables were: deep fascial spaces
operative computed tomography imaging demon- involved by cellulitis or abscess, number of spaces
strating adequate drainage of all affected anatomic affected, and severity score (SS). A severity rating of 1,
spaces. 2, or 3 was given to each anatomic space according to
its low, moderate, or high severity, respectively, as
STUDY VARIABLES detailed in Table 1.
Study data were collected for each patient using a The recorded treatment variables were the ana-
standardized data form. The variables were catego- tomic spaces that were drained and the presence or
rized as demographic, preadmission, time-related, absence of pus at surgical drainage. If pus was
preoperative, anatomic, treatment, microbiologic, present, the stage of infection was recorded as ab-
and complications. scess; if not, then the stage of infection was recorded
1106 SEVERE ODONTOGENIC INFECTIONS

as cellulitis. The product at I&D was categorized as BIVARIATE ANALYSES


pus if it could be described as a creamy yellow to grey The potential predictor variables associated with
fluid, or flecks or curds of pus suspended in a bloody the 4 outcomes are listed in Table 7. The only poten-
fluid drained from any of the spaces that were ex- tial predictor variable significantly associated with
plored at surgery. Cellulitis was recorded when the abscess formation was the identification of Pep-
product at I&D consisted only of serosanguineous tostreptococci in culture (Fisher’s exact, P ⫽ .04).
fluid. Treatment variables also included the number of Other potential predictor variables were: age, culture
spaces drained, type and number of drains used, num- of Prevotella or Porphyromonas species, autumn oc-
ber of teeth extracted, and antibiotic(s) used. currence, infection of non-third molar upper poste-
The microbiologic variables recorded were: genus rior teeth, and infection of non-third molar lower
and species identification, oxygen requirements (an- posterior teeth. Interestingly, there was no statisti-
aerobic or aerobic, including facultative), PCN and cally significant association between the stage of in-
clindamycin sensitivity, and number of species iso- fection and the number of days of preoperative swell-
lated per case. Only the initial culture and sensitivity ing (t test, P ⫽ .3). Further, there was no significant
results, sampled at the outset of treatment, were used association between stage of infection and SS, smok-
for statistical analyses. ing, trismus, dysphagia, dyspnea, WBC, initial core
Complications were recorded as present or absent
temperature, preadmission antibiotics, and immune
and included: PTF, facial nerve deficit, need for reop-
system compromise. There was no statistically signif-
eration, emergency airway management, spread of
icant association between stage of infection and any
infection into the chest or the brain, trigeminal nerve
of the demographic, preadmission, dental, anatomic,
deficit, and death.
treatment, or any of the other microbiologic variables.
For purposes of statistical analysis, certain variables
In the 6 patients with PTF (for whom sensitivity
were grouped together. Teeth were categorized as
data were available), 1 or more PCN-resistant strains
maxillary or mandibular anterior, maxillary or man-
were later identified by culture and sensitivity testing
dibular non-third molar posterior, and maxillary or
(Fisher’s exact, P ⫽ .001). In the 10 patients on PCN
mandibular third molar. Species of bacteria were cat-
that were later found to harbor 1 or more PCN-
egorized into aerobic (including facultative) and an-
resistant strains, PTF occurred in 6 cases (60%).
aerobic, as well as into groups of similar species: S.
milleri group, S. viridans group (excluding S. mil- The only potential predictor variables that were
leri), beta-hemolytic streptococci, other aerobic spe- statistically significantly associated with PTF were the
cies, Prevotella and Porphyromonas, Fusobacte- presence of PCN-resistant organisms (Fisher’s exact, P
rium, Peptostreptococcus, and other anaerobic ⫽ .001) and LOS (t test, P ⬍ .001). Other potential
species. predictor variables were age, gender, white ethnic-
ity, and dyspnea on admission. There was no statis-
tically significant relationship between PTF and any
DATA MANAGEMENT AND ANALYSES
of the other variables, including especially pread-
Data were recorded prospectively on standardized mission antibiotics and the other microbiologic
collection forms. A database was constructed using variables.
Microsoft Excel (Microsoft, Redmond, WA) and im- Several potential predictor variables for LOS were
ported into SPSS 12.0 (SPSS, Inc, Chicago, IL) for identified in this study. The WBC on admission was
statistical analysis. Bivariate analyses (Fisher’s exact, significantly associated with an increased LOS (linear
chi-square, t test, linear regression, and Spearman’s
regression, P ⬍ .004). Several anatomic variables
correlation, as appropriate) were used to evaluate the
were also associated with an increased LOS. These
relationship among selected study variables. Values of
included SS (linear regression, P ⫽ .001), number of
P ⱕ .05 were considered statistically significant. All
infected spaces (linear regression, P ⫽ .001), number
relevant demographic variables and variables that had
of infected teeth (linear regression, P ⫽ .02), and
P ⱕ .15 in the bivariate analyses were considered for
infection of non-third molar lower posterior teeth (t
inclusion in the multivariate models. Values of P ⱕ .05
test, P ⫽ .05). The amount of time spent in the
were considered statistically significant.
operating room also was associated with an increased
LOS (linear regression, P ⫽ .007). The other potential
predictor variables were: space 3 infection (t test, P ⫽
Results .053), immune system compromise (t test, P ⫽ .11),
The descriptive statistical results in this study have and infection of upper third molars (t test, P ⫽ .11).
been reported in detail separately.7 A brief summary Subjects who developed PTF had a mean LOS of 7.9
of those results can be found in Tables 2 through 5. ⫾ 3.2 days, while subjects without PTF had a mean
Selected data for each case are reported in Table 6. LOS of 4.2 ⫾ 1.8 days (t test, P ⬍ .001). Subjects who
FLYNN, SHANTI, AND HAYES 1107

Table 2. CATEGORICAL STUDY VARIABLES Table 2. CATEGORICAL STUDY VARIABLES (CONT’D)

No. of % of No. of % of
Cases Cases Cases Cases

Gender Infratemporal 1 3
Male 23 62 Maxillary sinus 1 3
Female 14 38 Parotid 1 3
Ethnicity Groups of spaces affected
African-American/black 20 54 Masticator 29 78
White 8 22 Perimandibular (submandibular,
Hispanic 8 22 sublingual, and submental) 22 60
Asian 1 3 Space 3 (lateral pharyngeal,
Penicillin allergy 3 8 retropharyngeal, and pretracheal) 16 43
Immune system compromise Stage of infection
Diabetes 2 5 Cellulitis (no pus at I&D) 9 24
HIV seropositivity 1 3 Abscess (pus at I&D) 28 76
Smoking Antibiotics used
Not recorded 4 11 Penicillin 33 89
Yes 15 41 Clindamycin 3 8
No 18 49 Gentamicin ⫹ metronidazole ⫹
Preadmission antibiotics clindamycin 1 3
No antibiotic 17 46 Airway management techniques
Penicillin 10 27 Fiberoptic intubation 18 49
Clindamycin 5 14 Direct laryngoscopic intubation 16 43
Penicillin ⫹ cephalexin 2 5 No intubation 3 8
Erythromycin 1 3 Tracheotomy (at reoperation) 1 3
Penicillin ⫹ clindamycin 1 3 Penicillin therapeutic failure (N ⫽ 33)* 7 21
Penicillin ⫹ metronidazole 1 3 Facial nerve deficit 3 8
Dental etiology Need for reoperation 3 8
Caries 24 65 Death 0 0
Periodontitis 8 22
Pericoronitis 8 22 *Thirty-three of 37 subjects received penicillin. Three subjects
received clindamycin because of penicillin allergy and 1 received
Needle track infection (after dental clindamycin, gentamicin, and metronidazole because of necrotizing
procedures) 2 5 fasciitis.
Postoperative infection (third molar
exodontia) 1 3 Flynn, Shanti, and Hayes. Severe Odontogenic Infections. J Oral
Maxillofac Surg 2006.
Teeth involved
Lower third molars 25 68
Other lower posteriors 18 49
Upper third molars 3 8 required reoperation had a mean LOS of 10.0 ⫾ 4.0
Other upper posteriors 3 8 days, while patients who did not require reoperation
Upper anteriors 0 0 had a mean LOS of 4.62 ⫾ 2.5 days (t test, P ⫽ .001).
Lower anteriors 0 0 There was no statistically significant association be-
Dyspnea 5 14
Dysphagia 29 78
Trismus (MIO ⱕ 20mm) 27 73
Season of occurrence Table 3. CONTINUOUS STUDY VARIABLES
Summer 15 41
Spring 8 22 Mean Years
Autumn 7 19 ⫾ SD Range
Winter 7 19 Age 34.9 ⫾ 15.8 14–76
Spaces affected White blood cell count on
Pterygomandibular 22 60 admission (⫻103) 14.9 ⫾ 4.2 5.9–26.0
Submandibular 20 54 Admission core temperature (°F) 101.3 ⫾ 1.3 98.0–104.4
Lateral pharyngeal 16 43 Number of teeth involved 1.5 ⫾ 0.9 1–5
Buccal 15 41 Days of preoperative swelling 8.2 ⫾ 14.6 1–71
Space of body of mandible 13 35 Time between admission and
Submasseteric 9 24 OR (hours) 5.1 ⫾ 7.5 0.2–23.3
Deep temporal (including infratemporal) 6 16 Duration of surgery (hours) 2.1 ⫾ 0.7 0.9–3.8
Sublingual 6 16 Number of spaces affected 3.3 ⫾ 1.5 1–8
Submental 4 11 Severity score 6.0 ⫾ 3.1 1–16
Superficial temporal 3 8 Number of spaces drained 3.1 ⫾ 1.8 1–8
Infraorbital 2 5 Length of hospital stay (days) 5.1 ⫾ 3.0 1–14
Retropharyngeal 2 5
Flynn, Shanti, and Hayes. Severe Odontogenic Infections. J Oral
Maxillofac Surg 2006.
1108 SEVERE ODONTOGENIC INFECTIONS

Table 4. MICROBIOLOGIC DATA (N ⴝ 24)


of pus at I&D (abscess formation) were the identifi-
cation of Peptostreptococci in culture (Fisher’s exact,
% of Strains P ⫽ .04) and autumn occurrence (Fisher’s exact, P ⫽
No. of % of Penicillin .052). Other potential predictor variables were: age,
Strains Cases Resistant*
culture of Prevotella or Porphyromonas species, in-
No growth (2 cases) None 8 fection of non-third molar upper posterior teeth, and
Aerobes infection of non-third molar lower posterior teeth.
S. milleri group species 12 50 0 Interestingly, there was no statistically significant as-
Other S. viridans species
sociation between the stage of infection and the num-
(excluding S. milleri) 2 8 0
Other streptococci 3 13 33 ber of days of preoperative swelling (t test, P ⫽ .3).
Other aerobic/facultative Further, there was no significant association between
species 12 50 58 stage of infection and SS, smoking, trismus, dyspha-
Anaerobes gia, dyspnea, WBC, initial core temperature, pread-
Prevotella and
mission antibiotics, and immune system compromise.
Porphyromonas species 23 63 35
F. nucleatum 5 21 25 There was no statistically significant association be-
Peptostreptococcus species 12 50 0 tween stage of infection and any of the demographic,
Other anaerobic species 21 88 0 preadmission, dental, anatomic, treatment, or any of
*Not all strains were tested for antibiotic sensitivity. the other microbiologic variables.
Flynn, Shanti, and Hayes. Severe Odontogenic Infections. J Oral MULTIVARIATE ANALYSES
Maxillofac Surg 2006.
The 4 clinically significant outcome variables se-
lected for this study were: abscess formation, PTF,
tween an increased LOS and any of the demographic LOS, and the need for reoperation. We attempted to
or other preadmission variables, including smoking, construct multivariate models for each of these 4
the use of preadmission antibiotics, and season of the outcome variables. Table 7 lists the potential predic-
year or an increased admission core temperature. tor variables that had a significance level of P ⱕ .15 in
There was no clinically significant association be- the bivariate analyses described above.
tween an increased LOS and involvement of any of To create a multivariate model for abscess forma-
the specific deep fascial spaces. However, patients tion (the presence or absence of pus at I&D), we
who had a space 3 infection (involvement of any or all selected a set of variables that were biologically im-
of these spaces: lateral pharyngeal, retropharyngeal, portant (age, gender, and race) or were nearly or
and pretracheal) had a mean LOS of 6.1 ⫾ 4.0 days, actually statistically significantly associated with ab-
while patients who did not have space 3 infection had scess formation at P ⱕ .15. Using logistic regression
a mean LOS of 4.2 ⫾ 1.4 days (t test, P ⫽ .053). This analysis, only a positive culture of Peptostreptococci
association was nearly statistically significant. Interest- remained significantly associated with the presence
ingly, there was a lack of significant association be- or absence of pus at I&D, controlling for other vari-
tween the LOS and the presence or absence of pus ables. In this model, the odds ratio (OR) and associ-
and any of the microbiologic variables. ated 95% confidence interval (CI) for abscess forma-
The only significant predictor of the need for re- tion, with a positive culture of Peptostreptococci was
operation was the WBC on admission. The mean ⫾ 0.86 (CI, 0.008 to 0.878; P ⫽ .04). Subjects yielding
SD admission WBC in the 34 patients who did not
require reoperation was 14.3 ⫾ 3.7, while the mean
SD admission WBC in the 3 patients who did require Table 5. SUMMARY MICROBIOLOGIC DATA (N ⴝ
reoperation was 20.7 ⫾ 4.7 (t test, P ⫽ .009). The 24)
only other potential predictor variable was spring
occurrence (Fisher’s exact, P ⫽ .11). There was a No. of % of
Culture and Sensitivity Results Cases Cases
clinically significant lack of association between need
for reoperation and SS, initial temperature, dyspnea, No growth 2 8
dysphagia, smoking, trismus, immune system com- Oxygen requirements
promise, PTF, and the presence or absence of pus at Aerobes only 2 8
Anaerobes only 4 17
I&D. Further, there was no association between the Mixed aerobes and anaerobes 16 67
need for reoperation and any of the demographic, Antibiotic resistance
dental, anatomic, treatment, or microbiologic vari- Penicillin 13 54
ables. Clindamycin 4 17
The only potential predictor variables that were Flynn, Shanti, and Hayes. Severe Odontogenic Infections. J Oral
significantly associated with the presence or absence Maxillofac Surg 2006.
FLYNN, SHANTI, AND HAYES 1109

Table 6. SELECTED DATA FOR ALL CASES

No. of No. of OR Pus


Case Gender/Age WBC Severity Infected Infected Time at PCN-Resistant LOS
No. Race (Yrs) (⫻103) Score Spaces Teeth (min) I&D PTF Strains (Days) Reoperation

1 B F/22 12.6 7 4 2 99 Y N 3 N
2 W M/22 11.6 4 2 1 145 N N 4 N
3 B M/23 13.0 3 2 1 82 Y N 3 N
4 A F/45 13.7 4 2 1 94 Y N 3 N
5 W M/26 13.8 3 2 2 116 Y N 3 N
6 H F/43 15.6 3 3 3 102 Y N 6 Y
7 B M/21 15.4 7 4 1 70 Y N 3 N
8 B M/33 14.2 5 3 2 75 Y N 4 N
9 W M/76 15.3 5 4 1 55 Y C 4 N
10 B M/54 13.0 3 3 2 60 Y N 4 N
11 B F/27 17.0 13 6 1 200 Y Y 9 N
12 H M/46 11.7 2 2 3 113 Y N 4 N
13 B F/21 26.0 8 4 2 154 Y N 7 N
14 B F/45 6.7 5 2 1 110 Y Y Y 5 N
15 W F/60 14.7 6 3 3 90 Y Y Y 6 N
16 B M/14 12.1 2 1 1 115 Y C Y 3 N
17 B M/41 15.1 8 4 2 175 Y N N 6 N
18 B F/23 21.6 4 5 1 100 N Y Y 14 Y
19 W F/22 13.6 6 3 3 120 N Y Y 7 N
20 B M/42 16.1 9 4 1 225 N Y Y 9 N
21 W M/31 12.4 7 2 1 115 Y C N 2 N
22 H M/45 16.9 16 8 5 182 Y NF Y 14 N
23 H F/29 10.4 6 6 1 180 N N N 7 N
24 B F/19 16.2 5 2 1 135 Y N N 4 N
25 H M/25 18.2 12 6 1 190 Y N N 6 N
26 B M/24 22.1 5 4 1 110 Y N N 3 N
27 B M/39 24.9 9 5 1 157 Y N N 10 Y
28 W M/28 13.8 5 2 1 185 Y N Y 4 N
29 B F/31 16.8 6 4 2 130 Y N Y 3 N
30 B M/23 12.7 8 4 1 110 N N N 3 N
31 B F/75 5.9 1 1 1 80 Y N N 3 N
32 H M/36 11.3 5 3 1 160 N N N 3 N
33 B M/29 14.8 5 2 1 90 N N N 4 N
34 B F/21 15.3 6 3 1 120 N N Y 4 N
35 H M/28 19.8 9 3 1 175 Y N N 4 N
36 W M/71 12.6 3 2 1 106 Y Y Y 5 N
37 H M/30 13.0 5 2 1 112 Y N Y 1 N
Abbreviations: A, Asian; B, African-American/black; H, Hispanic; W, white; WBC, white blood cell count on admission; OR, operating room;
I&D, incision and drainage; PTF, penicillin therapeutic failure; PCN, penicillin; LOS, length of hospital stay; C, subject received clindamycin;
NF, subject received multiple antibiotics for necrotizing fasciitis.
Flynn, Shanti, and Hayes. Severe Odontogenic Infections. J Oral Maxillofac Surg 2006.

Peptostreptococci in culture had only 86% of the risk allergy or necrotizing fasciitis were excluded from the
of abscess formation seen in subjects that are culture- model because they had no possibility of developing
negative for Peptostreptococci. When all 37 cases (in- PTF. Using logistic regression analysis, no significant
cluding those without complete culture and sensitiv- predictor of PTF was found when all potentially con-
ity data) were included in this model, the results were founding variables were controlled for.
similar (OR ⫽ 0.10; CI, 0.01 to 1.005; P ⫽ .05). To create a multivariate model for LOS, we selected
To create a multivariate model for PTF, we selected a set of variables that were biologically important
a set of variables that were biologically important (age, gender, and race) or were nearly or actually
(age, gender, and race) or were nearly or actually statistically significantly associated with LOS (P ⱕ
statistically significantly associated with PTF (P ⱕ .15). The potentially significant predictor variables
.15). The potentially significant predictor variables were: WBC, immune system compromise, SS, number
were: age, gender, white ethnicity, presence of PCN- of infected spaces, time in the operating room, space
resistant organisms, LOS, and dyspnea on admission. 3 infection, number of involved teeth, infection of
The 4 subjects that did not receive PCN because of upper third molars, infection of non-third molar lower
1110 SEVERE ODONTOGENIC INFECTIONS

Table 7. POTENTIAL PREDICTOR VARIABLES


dictor variables were infections occurring during the
spring, LOS, and WBC. In the multivariate logistic
P Value regression model, the only significant predictor that
remained was LOS, after adjusting for age, gender,
Stage of infection
Culture of Peptostreptococci .04 race, WBC, and infections occurring during the spring
Autumn occurrence .052 (OR ⫽ 1.55; CI, 1.1 to 2.3; P ⫽ .021).
Culture of Prevotella or Porphyromonas .09
Age .11
Infection of non-third molar lower Discussion
posterior teeth .12
Infection of non-third molar upper The specific aim of this study was to identify pre-
posterior teeth .14 dictors of abscess formation, PTF, LOS, and compli-
PTF cations in patients with severe OI.
Culture of PCN-resistant bacteria .001 A limitation of this study may be the inability to
LOS .001
detect statistically and clinically significant differ-
Dyspnea .05
White ethnicity .09 ences because of the small sample size (type II statis-
Gender .11 tical error). Our sample size was limited by the exclu-
Age .13 sion of relatively minor infections, which were
LOS treated on an outpatient basis, by the clinical need to
SS .001
avoid PCN in 4 cases, by the availability of standard-
Need for reoperation .001
PTF .001 ized culture and sensitivity methods for only the last
Number of infected spaces .001 24 cases, and by the inherent inaccuracies of anam-
WBC .004 nestic data. Statistical analyses that involved culture
Time in operating room .007 and sensitivity data included only the last 24 cases.
Number of infected teeth .022
Another limitation of this study is the possible mis-
Infection of non-third molar lower
posterior teeth .04 classification of the anatomic location of the infection
Space 3 infection .053 in some cases, which may have been improved by the
Infection of upper third molars .11 routine use of preoperative computed tomography in
Need for reoperation all cases.
LOS .001
Although certain results and conclusions of this
WBC .009
Spring occurrence .11 study may seem clinically obvious, the value of this
prospective study is to provide empiric data to sup-
Abbreviations: PTF, penicillin therapeutic failure; PCN, penicil-
lin; LOS, length of stay; SS, severity score; WBC, white blood cell
port some clinical impressions, question others, and
count. to provide a scientific basis for further research.
Flynn, Shanti, and Hayes. Severe Odontogenic Infections. J Oral A positive culture of Peptostreptococci was, after
Maxillofac Surg 2006. controlling for other potentially significant variables,
a significant negative predictor of abscess formation.
This finding suggests that future research to identify
posterior teeth, PTF, and reoperation. Using linear virulence factors unique to Peptostreptococci might
regression analysis, PTF (Beta ⫽ 0.503; CI, 1.7 to 3.6; be helpful to understand the etiology of cellulitis in
P ⬍ .0001), reoperation (Beta ⫽ 0.423; CI, 3.1 to 5.9; these cases. Other oral pathogens that were isolated
P ⫽ .0001), number of infected spaces (Beta ⫽ 0.251; in this study, including the Streptococcus milleri
CI, 0.14 to 0.895; P ⫽ .008), operative room time group and Prevotella and Porphyromonas species,
(Beta ⫽ 0.275; CI, 0.006 to 0.03; P ⫽ .004), and have been associated with abscess formation in other
infection of non-third molar lower posterior teeth studies,8-10 but this study did not support those find-
(Beta ⫽ 0.170; CI, 0.04 to 1.7; P ⫽ .041) were iden- ings.
tified as significant predictor variables of LOS, when It is interesting to note that abscess formation was
age, gender, race, WBC, space 3 infection, number of not associated with any of the time variables in this
infected teeth, and infection of upper third molars study, most notably the number of days of preopera-
were controlled for. The adjusted R-squared of the tive swelling. This is in contrast to the commonly
model was 0.84, which means that 84% of the vari- reported observation of a time-related progression
ability of LOS could be explained by the model. from cellulitis to abscess formation during the course
To create a multivariate model for reoperation, we of OI.10-12 Abscess formation was likewise not corre-
selected a set of variables that were biologically im- lated with any of the variables that measure the se-
portant (age, gender, and race) or were nearly or verity of infection, as might be expected. We failed to
actually statistically significantly associated with re- find any preoperatively identifiable clinical factor that
operation (P ⱕ .15). The potentially significant pre- predicts the presence of an abscess at operation. In
FLYNN, SHANTI, AND HAYES 1111

another study, by using contrast-enhanced computed ported no increased incidence of severe OI in HIV-
tomography combined with physical examination, seropositive subjects, but did find an increased inten-
Miller et al13 detected abscess formation with 95% sity of care required in these cases. In a retrospective
sensitivity, 80% specificity, and 89% accuracy. series of 185 cases, Huang et al4 found a statistically
There were no preadmission, timing, anatomic, or significant correlation of medically compromising dis-
preoperative clinical variables that predicted PTF by eases, including diabetes, renal insufficiency, hepatic
multivariate analysis. It is also interesting to note that cirrhosis, myeloproliferative disorders, and chemo-
none of the species or groups of related species iden- therapy, with advancing age, LOS, complications, the
tified in our cultures was significantly associated with need for tracheotomy, and death. Chen et al,20 in a
PTF, in spite of recent reports of increasing PCN retrospective analysis of 214 cases, found significant
resistance among oral pathogens.14-16 However, all associations of immunocompromising diseases, such
patients who experienced PTF were later found to as diabetes, renal failure, and malignancy, with com-
harbor PCN-resistant organisms (Fisher’s exact, P ⫽ plications of infection such as death, shock, medias-
.001). tinitis, and necrotizing fasciitis. The lack of statisti-
The identification of PCN-resistant pathogens is of- cally significant association between LOS and immune
ten delayed for as much as 2 weeks when conven- system compromise found in the current study may
tional culture and sensitivity methods are used for be because of the small sample size, because the
slow-growing oral bacteria. Therapeutic decisions relationship was nearly significant (t test, P ⫽ .11).
must often be made in an earlier time frame. This Grodinsky and Holyoke21 found that infection in-
points out the need for more rapid methods of bac- volving space 3, which includes any or all of the
terial identification and antibiotic sensitivity testing, lateral pharyngeal, retropharyngeal, and pretracheal
such as molecular techniques based on the polymer- spaces, nearly significantly predicted LOS on bivariate
ase chain reaction.17 analysis (t test, P ⫽ .053.) Because the SS assigns a
In other studies of orofacial infections, several clin- greater weight to space 3 infections, its value as a
ically useful predictors of the LOS have been identi- means of assessing and quantitating the severity of
fied, including admission WBC and temperature, cases is supported by this finding. SS was significantly
lower face infection, and medical or immune system associated with LOS on bivariate analysis (linear re-
compromise.1-4 In this study, only complications (PTF gression, P ⫽ .007), but this relationship was not
and re-operation) and variables related to the ana- significant when other potential confounding vari-
tomic extent or location of the infection (number of ables were controlled for on multivariate analysis.
infected spaces, time spent in the operating room, PTF was also correlated significantly with an in-
and infection of non-third molar lower posterior creased LOS. These findings are readily explained by
teeth) predicted LOS. Thus, these variables appear to this study’s protocol to allow at least 48 hours of IV
be valid measures of the severity of infection, which PCN therapy before changing antibiotics.
in turn appears to be directly related to LOS. Reoperation was significantly associated with an
Admission temperature did not predict LOS in this increased LOS. The additional time required for recu-
study, as in Peters et al.3 These results do not support peration from the second surgery necessarily pro-
the findings of Dodson et al,1 who reported that longed LOS. Interestingly, measures of the severity of
admission temperature and admission WBC can pre- infection, such as WBC, SS, space 3 infection, and
dict LOS; but our results are consistent with the asso- admission temperature did not predict reoperation.
ciation of deep infection and operating room use with Stage of infection (abscess or cellulitis) did not pre-
an increased LOS in those 2 studies. This inconsis- dict the need for reoperation, nor did preadmission
tency is almost certainly explained by the fact that antibiotics or any microbiologic variable, as might be
Dodson et al were studying children who commonly expected. Recently, Ylijoki et al22 found that a rising
develop high fevers, which complicates their manage- C-reactive protein on the second day after I&D was
ment and contributes to LOS. Wall et al18 found that predictive of the need for reoperation. C-reactive pro-
WBC ⬎ 15.4 ⫻ 109/L and serum sodium ⬍ 135 tein levels were not measured in the current study.
mmol/L were associated with necrotizing fasciitis in The use of oral antibiotics before admission pre-
general surgical infections. Although serum sodium dicted none of the significant outcomes in this study.
was not evaluated in this study, future research may Apparently, preadmission antibiotics did not hasten
fruitfully investigate its role in predicting LOS. abscess formation or shorten hospital stay. In contrast
Immune system compromise was not significantly to other studies, preadmission antibiotics were not
associated with LOS in the current study. Only 3 (8%) associated with culture identification of PCN-resistant
of the subjects in the present study were immuno- organisms.23-25 This may be because of the relatively
compromised (2 insulin-dependent diabetics and 1 small number of cases (24) for which culture and
HIV-seropositive individual). Carey and Dodson19 re- sensitivity data were available.
1112 SEVERE ODONTOGENIC INFECTIONS

Biederman and Dodson2 found lower face infection third molar lower posterior teeth also predicted the
to be a predictor of the need for use of the operating LOS. No clinically useful variables predicted the need
room. In their pediatric study, lower face infection for reoperation.
also correlated with odontogenic cause; upper face Directions for future research include development
infection correlated with sinus and other respiratory of biochemical and clinical predictors of outcomes in
tract infections. This may be explained by the deep severe OI, and molecular methods for bacterial iden-
portal of entry (bone) and the increased likelihood of tification and antibiotic sensitivity testing.26
abscess formation in OI. All cases in the current study
were odontogenic. Acknowledgment
Chen et al20 performed a retrospective multivariate The authors thank Drs Richard Kraut, Norman Trieger, Arthur
analysis of mortality in severe head and neck infec- Adamo, Mauricio Wiltz, and all of the Oral and Maxillofacial Surgery
residents at the Montefiore Medical Center for their assistance in
tions of various etiologies. They found that systemic the care of patients and gathering of data.
disease, neck swelling, and an increased time be-
tween the onset of symptoms and treatment were
significant predictors of complications, which in- References
cluded septic shock, mediastinitis, pleural empyema, 1. Dodson TB, Barton JA, Kaban LB: Predictors of outcome in
upper airway obstruction, disseminated intravascular children hospitalized with maxillofacial infections: A linear
coagulation, acute respiratory distress syndrome, nec- logistic model. J Oral Maxillofac Surg 49:838, 1991
2. Biederman GR, Dodson TB: Epidemiologic review of facial
rotizing fasciitis, pericarditis, and 1 death. No deaths infections in hospitalized pediatric patients. J Oral Maxillofac
occurred in the present study. Surg 52:1042, 1994
It is impossible to compare the complication rate in 3. Peters ES, Fong B, Wormuth DW, et al: Risk factors affecting
hospital length of stay in patients with odontogenic infections.
this study with other studies because of differences in J Oral Maxillofac Surg 54:1386, 1996
study design, patient population, cause of infection, 4. Huang TT, Liu TC, Chen PR, et al: Deep neck infection: Analysis
and the lack of a common method of calibrating of 185 cases. Head Neck 26:854, 2004
5. Flynn TR: Anatomy and surgery of deep fascial space infec-
severity. Such calibration may be possible in the fu- tions, in Kelly JJ: Oral and Maxillofacial Surgery Knowledge
ture by using C-reactive protein, WBC, SS, or a com- Update 1994. Rosemont, IL, American Association of Oral and
bination thereof. Maxillofacial Surgeons, 1994, pp 79-107
6. Flynn TR: Anatomy of oral and maxillofacial infections, in
In this study, we did not find empiric evidence to Topazian RG, Goldberg MH, Hupp JR (eds): Oral and Maxillo-
support the clinical observation that cellulitis tends to facial Infections. Ed 4. Philadelphia, PA, Saunders, 2002, pp
occur in the first 1 to 5 days of swelling, and that 188-213
7. Flynn TR, Liu TC, Shanti RM, Levy M, et al: Severe Odontogenic
abscess formation tends to follow at 3 to 7 days. The Infections, Part One: Prospective Report. J Oral Maxillofac Surg
number of days of preoperative swelling did not sig- 64:1093, 2006
nificantly correlate with the presence or absence of 8. MacDonald JB, Socransky SS, Gibbons RJ: Aspects of the patho-
genesis of mixed anaerobic infections of mucous membranes.
pus found at I&D. In the cellulitis group, 9 subjects J Dent Res 42:529, 1963
had a mean preoperative period of swelling of 3.8 ⫾ 9. Heimdahl A, Von Konow L, Satoh T, et al: Clinical appearance
6.9 days versus 9.7 ⫾ 16.2 days in the 28 subjects in of orofacial infections of odontogenic origin in relation to
microbiological findings. J Clin Microbiol 22:299, 1985
the abscess group (t test, P ⬎ .3). Further, there was 10. Lewis MAO, MacFarlane TW, McGowan DA: Quantitative bac-
no correlation between the stage of infection and any teriology of acute dento-alveolar abscesses. J Med Microbiol
of the microbiologic variables, except a positive cul- 21:101, 1986
11. Fisher LE, Russell RRB: The isolation and characterization of
ture for Peptostreptococci, which was significantly milleri group streptococci from dental periapical abscesses. J
negatively associated with abscess formation. The fact Dent Res 72:1191, 1993
that stage of infection did not correlate with expected 12. Flynn TR: The swollen face. Emerg Med Clin North Am 15:481,
2000
variables, such as the number of days of preoperative 13. Miller WD, Furst IM, Sandor GKB, et al: A prospective blinded
swelling and the presence of S. milleri group organ- comparison of clinical examination and computed tomography
isms in culture may be because of the difficulty of in deep neck infections. Laryngoscope 109:1873, 1999
14. Sixou J-L, Magaud C, Jolivet-Gougeon A, et al: Microbiology of
selecting variables that measure these parameters ac- mandibular third molar pericoronitis: Incidence of betalacta-
curately or to small sample size. mase-producing bacteria. Oral Surg Oral Med Oral Pathol Oral
Predictors of significant clinical outcomes of severe Radiol Endod 95:655, 2003
15. Sobottka I, Cachovan G, Sturenbur E, et al: In vitro activity of
OI were identified using multivariate analysis. Abscess moxifloxacin against bacteria isolated from odontogenic infec-
formation was negatively predicted by the presence tions. Antimicrob Agents Chemother 46:4019, 2002
of Peptostreptococci. No statistically significant pre- 16. Flynn TR, Halpern LR: Antibiotic selection in head and neck
infections. Oral Maxillofac Surg Clin N Am 15:17, 2003
dictor of PTF was found. PTF and the need for re- 17. Kim Y, Flynn TR, Donoff RB, et al: The gene: The polymerase
operation predicted an increased LOS. Indicators of chain reaction and its clinical application. J Oral Maxillofac
the anatomic extent and location of infection, includ- Surg 60:808, 2002
18. Wall DB, Klein SR, Black S, et al: A simple model to help
ing the number of deep fascial spaces infected, time distinguish necrotizing fasciitis from non-necrotizing soft tissue
spent in the operating room, and infection of non- infection. J Am Coll Surg 191:227, 2000
FLYNN, SHANTI, AND HAYES 1113

19. Carey JW, Dodson TB: Hospital course of HIV-positive patients pallens genotypes and in vitro susceptibilities to selected
with odontogenic infections. Oral Surg Oral Med Oral Pathol antimicrobial agents. Antimicrob Agents Chemother 43:2383,
Oral Radiol Endod 91:23, 2001 1999
20. Chen MK, Wen YS, Chang CC, et al: Predisposing factors of 24. Heimdahl A, von Konow L, Nord CE: Beta-lactamase producing
life-threatening deep neck infection: Logistic regression analy- Bacteroides species in the oral cavity in relation to penicillin
sis of 214 cases. J Otol 27:141, 1998 therapy. J Antimicrob Chemother 8:225, 1981
21. Grodinsky M, Holyoke EA: The fasciae and fascial spaces of the 25. Kuriyama T, Nakagawa K, Karasawa T, et al: Past administra-
head, neck, and adjacent regions. Am J Anat 63:367, 1938 tion of beta-lactam antibiotics and increase in the emergence of
22. Ylijoki S, Suuronen R, Jousimies-Somer H, et al: Differences ␤-lactamase-producing bacteria in patients with orofacial odon-
between patients with or without the need for intensive care togenic infections. Oral Surg Oral Med Oral Pathol Oral Radiol
due to severe odontogenic infections. J Oral Maxillofac Surg Endod 89:186, 2000
59:867, 2001 26. Flynn TR, Stokes LN, Lee AM, et al: Molecular microbiology
23. Matto J, Asikainen S, Vaisanen M-L, et al: Beta-lactamase produc- of orofacial infections. J Oral Maxillofac Surg 60:72, 2002
tion in Prevotella intermedia, Prevotella nigrescens, and Prevotella (suppl 1)

You might also like