Download Factors Associated With the Mediastinal Spread of Cervical

Factors Associated With the Mediastinal Spread of
Cervical Necrotizing Fasciitis
Franck Petitpas, MD, Jean-Philippe Blancal, MD, Joaquim Mateo, MD, Iyed Farhat, MD,
Walid Naija, MD, Raphael Porcher, PhD, Catherine Beigelman, MD,
Mourad Boudiaf, MD, Didier Payen, PhD, MD, Philippe Herman, PhD, MD, and
Alexandre Mebazaa, PhD, MD
GENERAL THORACIC
Departments of Anesthesiology and Critical Care, ENT, and Radiology, Lariboisière University Hospital, Paris; Department of
Biostatistics and Clinical Epidemiology, Saint-Louis University Hospital, Paris; Department of Radiology, La Pitié Salpétrière
University Hospital, Paris; University Paris 7, Pres Sorbonne Paris Cité, Paris; and INSERM U942, Paris, France
Background. We conducted a study to determine factors associated with the occurrence of mediastinitis in
patients hospitalized for cervical necrotizing fasciitis
(CNF).
Methods. We reviewed the medical records of 130
consecutive patients in an intensive care unit (ICU) who
were hospitalized with a diagnosis of CNF. Two radiologists reviewed cervical and thoracic computed tomography (CT) scans to determine the source and extension of
the infection in each patient.
Results. Among the cohort of 130 patients with CNF, 37
(28%) had mediastinitis at the time of their admission
(which in 13 cases was superior, or above the aortic arch,
and in 24 cases inferior). Cervical necrotizing fasciitis
complicated by mediastinitis resulted in a longer stay in
the ICU than did CNF without mediastinitis, of a mean of
29 days (range, 18 to 39 days) versus 14 days (range, 9 to
19) days, respectively (p < 0.0001). Multivariate analysis
revealed that the presence of mediastinitis was associated with oral intake of glucocorticoids before admission
(odds ratio [OR], 2.17; range, 0.99 to 4.76), a pharyngeal
focus of CNF (OR, 2.17; range, 1.04 to 4.53), or gas seen on
an initial CT scan (OR, 4.49; range, 2.15 to 9.38). Both a
pharyngeal focus of fasciitis and the presence of gas were
strong independent predictors of inferior mediastinitis
(OR, 15.1; range, 4.9 to 46.4; p < 0.0001).
Conclusions. The present study is the first to describe
three independent factors associated with extension of
cervical fasciitis to the thoracic cavity, including glucocorticoid intake before admission, and confirms previous
reports of a high incidence of mediastinitis in patients
with CNF.
(Ann Thorac Surg 2012;93:234 –9)
© 2012 by The Society of Thoracic Surgeons
N
Material and Methods
ecrotizing fasciitis is an acute soft-tissue infection
that usually involves the extremities and perianal
areas, and more rarely the head and neck, including the
periorbital region [1]. Typically, necrotizing fasciitis is the
result of a local infection spreading throughout the skin
and resulting in dermal, subcutaneous, and deep fascial
acute pyogenic infection. Necrotizing fasciitis can also
rapidly expand into adjacent tissues and organs in a
fulminant course of infection. Factors causing the distant
spread of necrotizing fasciitis are unknown.
Cervical necrotizing fasciitis (CNF) is a life-threatening
infection originating chiefly from dental or oropharyngeal infection. It may spread to the chest, leading to a
“descending” mediastinitis, which is the most virulent
complication of CNF. We conducted a study to assess
factors that facilitate the spread of CNF into the mediastinum, for which we retrospectively analyzed data for 130
consecutive patients with CNF.
Accepted for publication Sept 2, 2011.
Address correspondence to Dr Mebazaa, Department of Anesthesiology
and Critical Care, Lariboisière Hospital, University Paris 7, UMR 942
INSERM, 2 rue Ambroise Paré, 75010 Paris, France; e-mail: alexandre.
[email protected].
© 2012 by The Society of Thoracic Surgeons
Published by Elsevier Inc
Lariboisière Hospital is the reference center for medical
and surgical management of CNF in Paris and the surrounding region. The 130 patients whose data we reviewed were patients admitted to the Hospital for CNF
between 1999 and 2006. Approval for the retrospective
collection of data and the data analysis for the study was
given by the Institutional Review Board of Paris North
Hospitals.
The diagnosis of CNF was confirmed with contrastenhanced computed tomography (CT) of the cervical
neck and chest upon the admission of each patient to the
hospital. After confirmation of the diagnosis, antibiotic
therapy was administered in the following regimen before surgery was done: ceftriaxone (4 g/d loading dose for
2 days, followed by maintenance doses of 2 g/d) plus
ornidazole (3 g/d loading dose for 2 days, followed by
maintenance doses of 1.5 g/d), with both drugs continued
for 14 days. In patients with CNF-related mediastinitis,
the antibiotic regimen was extended for up to another 7
days. A bilateral cervicotomy was then done for the
collection of samples for microbiologic examination, together with the early debridement of all necrotic tissue
and drainage of all fascial planes. Debridement of the
0003-4975/$36.00
doi:10.1016/j.athoracsur.2011.09.012
cervical region was then repeated, chiefly in the ICU, at
least twice daily until healthy fascia was encountered.
Surgical management was guided by the findings on
chest CT scanning. If the extension of CNF was limited to
the area above the aortic arch, in which case it was
designated as “superior” mediastinitis, or type I mediastinitis as defined by by Endo and colleagues [2], drainage
of mediastinal infection was performed through the cervicotomy described above. When signs of infection were
found below the aortic arch, in what was termed “inferior” mediastinitis, or type II mediastinitis as defined by
Endo and colleagues, immediate sternotomy was done
(including the removal of a necrotic thymus in the case of
anteroinferior mediastinal extension) or thoracotomy
was done (in the case of posteroinferior mediastinal
extension), typically at the time of cervicotomy or, in a
very few cases, within 12 hours of cervicotomy. Because
the mediastinum is in contact with the “open” cervical
area, no vacuum-assisted losure (VAC) system can be
used in the latter procedure. However, drains were put in
place in the retrosternal or the posterior region of the
mediastinum or both, to allow continuous irrigation.
Data Collection
To assess the factors that lead to distal expansion of CNF
in the chest, the study patients’ data were reviewed.
Severity of illness was assessed with the Simplified Acute
Physiology Score II (SAPS II) and the Sequential Organ
Failure Assessment (SOFA) score [3, 4]. To determine the
source of CNF and the level of mediastinal involvement,
the initial CT scan was retrospectively reviewed by two
radiologists blinded to the surgical findings and the outcome of the patient. Although no recording was made of
the interval between the suspicion of infection and the
beginning of antibiotic treatment given by the physician treating the patient before the patient’s admission, the interval between antibiotic treatment and
admission to the hospital was available. A diagnosis of
CNF was based on the finding of diffuse thickening of
subcutaneous fat, cervical fascia, and cervical muscles.
The presence of fluid or gas collections or both was
recorded. The results of this were compared with the
findings at surgery.
Statistical Analysis
Results are expressed as mean ⫾ standard deviation
(SD), median and first and third quartiles (Q1 to Q3), or
counts and percentages, as appropriate. Study outcomes
were mediastinitis and inferior mediastinitis, the data for
each of which were analyzed separately.
Marginal association between single variables and
each outcome was assessed with Wilcoxon’s rank-sum
test for quantitative variables and with the ␹2 or Fisher’s
exact test when appropriate for qualitative variables.
Multiple logistic regression analysis was used to identify
a set of variables independently associated with each
outcome. Variables associated with the occurrence of
mediastinitis at a level of p ⬍ 0.15 were considered in the
multiple models. A backward stepwise procedure was
used for the selection of variables, with a cut-off value of
PETITPAS ET AL
FACTORS ASSOCIATED WITH MEDIASTINITIS IN CNF
235
p ⫽ 0.05. First-order interactions between selected variables were then tested. For inferior mediastinitis, Because of the moderate number of instances of inferior
mediastinitis, data reduction through heuristic shrinkage
was done before the use of multivariate analysis, without
considering the association of variables with the outcome
[5]. The first-order interaction between variables marginally associated with the outcome was then tested before
the backward stepwise procedure. The Cessie and van
Houwelingen goodness-of-fit test [6] was used to verify
the validity of the logistic regression model.
Internal validation of the models was performed by
bootstrapping [7]. The procedure was based on a random
generation of 200 samples from the original data, which
were drawn with replacement. The whole-variableselection procedure was applied to these samples, and
the performance measures of the derived model were
calculated. The discriminative ability of the models was
evaluated with the c index (which is identical to the area
under the receiver operating characteristics [ROC] curve)
[8] and through calibration with the calibration slope [9].
The performances of the models estimated on the basis of
each bootstrap sample were then evaluated in the original sample. The differences between the performance of
the selected model in the bootstrap sample and the
original sample were taken as a measure of the overoptimism of the model. The performance of the final
model was then corrected by this over-optimism (ie, the
area under the ROC curve was reduced accordingly), and
estimated regression coefficients were multiplied by the
calibration slope. All tests were two-sided. Analyses were
done with R statistical software (R Foundation for Statistical Computing, Vienna, Austria).
Results
During the 8-year period in which the study data were
collected, our hospital admitted 130 patients for CNF.
Most were referred from emergency rooms, general
practitioners, or ear, nose, and throat practices in Paris
and its suburbs. The patients had been given oral antibiotics (64%), nonsteroidal anti-inflammatory drugs
(NSAIDs) (40%), or glucocorticoids (26%) for an average
of 4 days before their admission to the hospital (Table 1).
Most of the patients were self-ventilating and were not
taking catecholamines at the time of their admission.
Mediastinitis was always discovered by CT scanning
rather than being suspected on the basis of clinical or
biologic signs. Patients with high illness severity scores
(SAPS II score of 35 ⫾ 13 and median SOFA of 3 [1–5])
were transferred postoperatively to the intensive care
unit (ICU).
Imaging with CT showed signs of CNF in all of the
patients studied, and allowed the assessment of pharyngeal and dental infections as the two most frequent
causes of CNF (42% and 37%, respectively). Notably, gas
was present in the CT scans of 31% of the patients. Signs
found in CT scanning were confirmed by surgical findings.
GENERAL THORACIC
Ann Thorac Surg
2012;93:234 –9
236
PETITPAS ET AL
FACTORS ASSOCIATED WITH MEDIASTINITIS IN CNF
Ann Thorac Surg
2012;93:234 –9
Table 1. Patient History, Sociodemographic, and Clinical Data at Admission
CNF With Mediastinitis
Patient Data
GENERAL THORACIC
Number of patients
Epidemiologic findings
Age (years)
Male
Smoking
Alcoholism
Diabetes
Medications before admission
Antibiotics
Glucocorticoids
NSAIDs
Severity score at admission
SAPS II
SOFA [median (Q1–Q3)]
CNF source
Pharyngeal
Dental
Other
Presence of gas
Bacteria, n (%)
Number of bacteria
isolated/patient
Streptococcus milleri
Anaerobic bacilli
Coagulase-negative
Staphylococcus
Streptococcus pyogenes
Streptococcus oralis
Enterobacter
Staphylococcus aureus
Fungi
Others
Outcome [median (Q1–Q3)]
Duration of mechanical
ventilation (d)
Length of stay in ICU (d)
Mortality at 28 d [n (%)]
CNF With
Superior
Mediastinitis
CNF With
Inferior
Mediastinitis
Presence
(n ⫽ 37) vs
Absence of
Mediastinitis
(n ⫽ 93)
Inferior (n ⫽ 24)
vs Absence of
Mediastinitis/
Superior
Mediastinitis
(n ⫽ 106)
Total
CNF
Without
Mediastinitis
130
93 (72%)
13 (10%)
24 (18%)
48 ⫾ 18
75 (58%)
58 (45%)
25 (19%)
8 (6%)
47 ⫾ 17
56 (60%)
45 (48%)
21 (23%)
6 (6%)
45 ⫾ 15
6 (46%)
4 (31%)
0 (0%)
1 (8%)
58 ⫾ 19
13 (54%)
9 (38%)
4 (17%)
1 (4%)
0.065
0.43
0.0086
0.82
83 (65%)
35 (27%)
52 (40%)
59 (64%)
19 (20%)
37 (40%)
7 (58%)
5 (38%)
6 (46%)
17 (71%)
11 (46%)
9 (38%)
0.84
0.015
0.84
0.64
0.039
0.82
35 ⫾ 13
3 (1–6)
34 ⫾ 14
2 (1–5)
31 ⫾ 10
4 (2–7)
40 ⫾ 11
4 (2–7)
0.49
0.046
0.0047
0.07
55 (42%)
48 (37%)
27 (21%)
39 (31%)
33 (35%)
40 (43%)
20 (21%)
18 (20%)
5 (38%)
2 (15%)
6 (46%)
7 (54%)
17 (71%)
6 (25%)
1 (4%)
14 (58%)
0.018
0.027
0.0026
0.24
⬍0.0001
0.0025
2.7 ⫾ 1.5
2.6 ⫾ 1.6
2.2 ⫾ 1.5
3.3 ⫾ 1.8
0.3
0.043
70 (54%)
80 (62%)
16 (12%)
49 (53%)
56 (60%)
12 (13%)
6 (46%)
6 (46%)
2 (15%)
15 (62%)
18 (75%)
2 (8%)
0.85
0.69
1.0
0.5
0.17
0.73
12 (9%)
4 (3%)
6 (5%)
4 (3%)
7 (5%)
25 (19%)
9 (10%)
3 (3%)
4 (4%)
3 (3%)
5 (5%)
16 (17%)
3 (23%)
0 (0%)
1 (8%)
0 (0%)
1 (8%)
3 (23%)
0 (0%)
1 (4%)
1 (4%)
1 (4%)
1 (4%)
6 (25%)
1.0
1.0
1.0
1.0
1.0
0.46
0.12
0.57
1.0
0.57
1.0
0.57
12 (8–18)
10 (7–16)
16 (10–24)
20 (14–27)
⬍0.0001
⬍0.0001
16 (10–25)
11 (8 %)
14 (9–19)
5 (5 %)
21 (12–33)
2 (15 %)
30 (22–46)
4 (17 %)
⬍0.0001
0.075
⬍0.0001
0.12
CNF ⫽ cervical necrotizing fascitis;
ICU ⫽ intensive care unit;
NSAIDs ⫽ nonsteroidal antiinflammatory medications;
Q1 ⫽ first quartile;
Q3 ⫽ third quartile;
SAPS II ⫽ Simplified Acute Physiology Score II;
SOFA ⫽ Sequential Organ Failure Assessment Score.
As previously described, the most frequent bacteria found
in necrotic tissues were Streptococcus milleri and anaerobic
bacilli (in 54% and 62% of samples, respectively) [10].
Expansion of CNF toward the mediastinum was present in 37 patients (28%). All cases of mediastinitis were
detected on a first contrast-enhanced CT scan, done at
the time of admission to our center, and later confirmed
by surgery. These 37 patients had a higher incidence of
oral glucocorticoid ingestion before admission than did
patients without mediastinitis (43% versus 20%, respectively; p ⫽ 0.015) (Table 1). The source of CNF associated
with mediastinitis was predominantly pharyngeal (59%),
and included gas-producing bacteria in more than 50% of
the cases of such disease.
On admission, 13 of the 37 cases of CNF-related mediastinitiswere superior and 24 were inferior (above and
below the aortic arch, respectively) according to the
definitions of Endo and colleagues [2]. No patient had
newly discovered mediastinitis in the 90 days after admission for CNF, if the mediastinitis was present at
admission. Furthermore, no “superior” mediastinitis extended to inferior mediastinum during ICU stay.
PETITPAS ET AL
FACTORS ASSOCIATED WITH MEDIASTINITIS IN CNF
Patients with mediastinitis had a longer duration of
mechanical ventilation and ICU stay than those without
it (both, p ⬍ 0.0001) (Table 1). Any mediastinal involvement also resulted in a higher, albeit not statistically
significant, mortality (16% vs 5%, respectively, p ⫽ 0.075).
Patients with inferior mediastinitis were sicker on admission (SAPS II, 40 ⫾ 11) and required a longer ICU stay
(mean, 30 days; range, 22 to 46 days) than did those with
superior mediastinitis (Table 1).
Multivariate analysis (Table 2) showed that extension
of CNF to the chest was independently associated with
oral glucocorticoid use before admission (OR, 2.17; range,
0.99 to 4.76), a pharyngeal source of CNF (OR, 2.17; range,
1.04 to 4.53), or the presence of gas in both the source of
infection and its extension on the initial CT scan (OR,
4.49; range, 2.15 to 9.38). The combination of a pharyngeal
source of CNF and the presence of gas (test for interaction: p ⫽ 0.023) was a strong predictor of the extension of
CNF to below the aortic arch, also called inferior mediastinitis (OR, 15.1; range, 4.9 to 46.4; p ⬍ 0.0001) (Table 2).
Comment
Our study showed that the production of gas was associated with a greater incidence of expansion of CNF into
the chest. Indeed, the combination of a pharyngeal infection with the production of gas increased the risk of
extension into the inferior mediastinum by 15-fold.
This study involved the largest population of patients
with CNF who were treated with surgery and had follow-up in the ICU. Expansion of CNF toward the mediastinum was found in 28% of the patients. This was in
accordance with prior reports in the literature, which had
had several limitations in terms of etiology and disease
course.
Cervical necrotizing fasciitis results from the extension
of infection from a pharyngeal or odontogenic source.
The infection spreads along the fascia, often in an ipsilateral fashion, but not hematogenously. The infection
Table 2. Factors Associated With Mediastinitis in Patients
Admitted for Cervical Nectrotizing Fasciitis
Multivariate Analysis
Patient Data
OR
All Mediastinitis
Variables
Production of gas by
infecting organism
Pharyngeal origin
Oral glucocorticoid intake
Inferior Mediastinitis
Parameters
Pharyngeal source and
gas production
SAPS II
CI ⫽ confidence interval;
Acute Physiology Score II.
95% CI
p Value
4.49
2.15–9.4
⬍0.0001
2.17
2.17
1.04–4.58
0.99–4.76
0.039
0.054
4.9–46.4
⬍0.0001
15.1
1.5
OR ⫽ odds ratio;
1.1–
2.2
0.016
SAPS II ⫽ Simplified
237
may further expand toward the mediastinum into pretracheal, perivascular, and prevertebral spaces, bound anteriorly by the alar fascia and posteriorly by the prevertebral fascia, and known collectively as the “danger
space” because of its patency from the base of the skull to
the diaphragm [11].
Both dental and pharyngeal infections can extend into
the mediastinum[12, 13]. In our study, pharyngeal infections were more liable to cause mediastinitis, a finding
not demonstrated in previous studies based on smaller
case series[14 –17]. Reynolds and associates emphasized
that pharyngeal infections can spread through retropharyngeal and prevertebral spaces from the base of the
skull to the diaphragm [11], as noted above. This lack of
inherent protection brings about the rapid onset of a
life-threatening mediastinitis.
The mechanisms by which CNF expands into the
mediastinum remain unclear. The expansion may be due
to mechanical or chemical properties or both of the gas
produced by the bacteria causing the fasciitis, or may
directly indicate virulence of the causative bacteria. If it is
produced in large quantities, such gas may dissect the
tissues, especially along the fascial planes. The anaerobic
bacteria found in our patients, such as Prevotella spp.,
Peptostreptococcus spp., also produce non-olatile fatty acids (including succinic and phenylacetic acids) that may
decrease extracellular pH in the surrounding tissues
leading to cellular alterations that include leukocyte
inactivation [18].
The present study shows that oral glucocorticoid intake, but not NSAIDs, put patients at an increased risk of
expansion of CNF into the mediastinum (p ⫽ 0.015). The
mechanism responsible for such expansion of CNF in
patients taking glucocorticoids is unknown. Glucocorticoid use may have diminished initial clinical signs of this
expansion and delayed the surgical debridement of affected tissues, or may have favored the spread of infection to the mediastinum, at least partly via these drugs’
immunosuppressive properties. Hence, until new data
appear, and for the sake of safety, the use of glucocorticoids should not be advocated for the early resolution of
pain in acute pharyngitis, despite the lack of any previour
report of a complication such as the mediastinal spread
of CNF.
The present study has important clinical implications
in the management of CNF. It revealed gas on CT
imaging in 31% of our CNF patients, and the absence of
gas in the remaining 69% confirming that the absence of
gas does not rule out CNF, in contrast with the findings in
another recent study [19]. We also confirm that there is a
continuum between CNF and descending mediastinitis:
cervical infection may spread first to the superior and
eventually to the lower mediastinum. We strongly recommend CT scanning at the time of admission of patients with CNF to assess the source of infection, production of gas, and any expansion of infection [20]. Early and
aggressive treatment based on the findings made with
CT may stop further downward expansion of the infection [21]. This strategy, along with experience (more than
16 cases of CNF per year seen in our center versus fewer
GENERAL THORACIC
Ann Thorac Surg
2012;93:234 –9
238
PETITPAS ET AL
FACTORS ASSOCIATED WITH MEDIASTINITIS IN CNF
GENERAL THORACIC
than 5 cases per year in other studies [22]), may reduce
the mortality rate and length of stay in the ICU for
patients with CNF below those in prior studies [22–24].
Despite monocentricity, our findings lead us to conclude that: (1) mediastinitis was always present in patients admitted to our hospital for CNF, (2) no further
expansion of CNF was observed when patients received
appropriate and aggressive treatment beginning at the
time of admission, and (3) two subgroups of mediastinitis
can be defined, consisting of superior and inferior mediastinitis. Superior mediastinitis was limited to the area
above the aortic arch and was treated through cervicotomy[25]. Inferior mediastinitis, below the aortic arch,
required a combined cervical and thoracic surgical approach [24] with a much longer ICU stay. Three factors
independently predicted extension of CNF to the mediastinum: (1) a pharyngeal origin of CNF, (2) the presence
of gas in the source site of infection, and (3) oral glucocorticoid intake before admission to the hospital.
In summary, our study confirms a high incidence of
mediastinitis in a considerable number of patients with
CNF. As noted above, three independent factors appeared
to indicate the expansion of cervical infection toward the
mediastinum: (1) a pharyngeal source of infection, (2) the
presence of gas in the source site of infection, and (3) oral
intake of glucocorticoids before hospital admission. Our
study thus challenges the use of glucocorticoids in local
infections such as dental or pharyngeal infections, and
emphasizes the need for additional studies of mediastinitis
in CNF. Should our results be confirmed by other studies,
the use of glucocorticoids in acute pharyngitis or dental
infection should be re-examined.
The authors thank Professor Luc Dubreuil (Lille) for his advice,
and Professors Bernard Regnier (Université Paris 7) and Mervyn
Singer (London) for editing this paper. The authors also wish to
thank the assistants in the Ear, Nose, and Throat Department of
the Lariboisiere University Hospital who contributed to the
surgical management of these patients, as well as all health
providers in the Ear, Nose, and Throat Department, Intensive
Care Unit, and Recovery Unit of the Lariboisière Hospital for
their hard work in treating the critically ill patients in this study.
The authors also thank their colleagues for their dedication,
which made this work possible: Elisabeth Sauvaget, Franck
Schaulievge, Alain Bel, Anne-Claire Lukascewicz, Sophie Dagois, Corinne Dematons, Peter Karpati, and Romain Kania. In
memoriam of Christine Berton.
This study was partly supported by grant EA 322, DP from the
Ministère de l’Enseignement et de la Recherche of France.
References
1. Swartz MN. Clinical practice. Cellulitis. N Engl J Med
2004;350:904 –12.
2. Endo S, Murayama F, Hasegawa T, et al. Guideline of
surgical management based on diffusion of descending
necrotizing mediastinitis. Jpn J Thorac Cardiovasc Surg
1999;47:14 –9.
3. Le Gall JR, Lemeshow S, Saulnier F. A new Simplified Acute
Physiology Score (SAPS II) based on a European/North
American multicenter study. JAMA 1993;270:2957– 63.
Ann Thorac Surg
2012;93:234 –9
4. Vincent JL, Moreno R, Takala J, et al. The SOFA (Sepsisrelated Organ Failure Assessment) score to describe organ
dysfunction/failure. On behalf of the Working Group on
Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intens Care Med 1996;22:707–10.
5. Harrell FE, Jr., Lee KL, Mark DB. Multivariable prognostic
models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors.
Stat Med 1996;15:361– 87.
6. Le Cessie S, van Houwelingen JC. A goodness of fit test for
binary regression models, based on smoothing methods.
Biometrics 1991;47:1267– 82.
7. Steyerberg EW, Eijkemans MJ, Harrell FE, Jr., Habbema JD.
Prognostic modeling with logistic regression analysis: in
search of a sensible strategy in small data sets. Med Decis
Making 2001;21:45–56.
8. Harrell FE, Jr., Califf RM, Pryor DB, Lee KL, Rosati RA.
Evaluating the yield of medical tests. JAMA 1982;247:2543– 6.
9. Miller ME, Hui SL, Tierney WM. Validation techniques for
logistic regression models. Stat Med 1991;10:1213–26.
10. Fihman V, Raskine L, Petitpas F, et al. Cervical necrotizing
fasciitis: 8-years’ experience of microbiology. Eur J Clin
Microbiol Infect Dis 2008;27:691–5.
11. Reynolds SC, Chow AW. Life-threatening infections of the
peripharyngeal and deep fascial spaces of the head and
neck. Infect Dis Clin North Am 2007;21:557–76, viii.
12. Righini CA, Motto E, Ferretti G, Boubagra K, Soriano E, Reyt
E. [Diffuse cervical cellulites and descending necrotizing
mediastinitis]. Ann Otolaryngol Chir Cervicofac 2007;124:
292–300.
13. Thomas SJ, Atkinson C, Hughes C, Revington P, Ness AR. Is
there an epidemic of admissions for surgical treatment of
dental abscesses in the UK? BMJ 2008;336:1219 –20.
14. Freeman RK, Vallieres E, Verrier ED, Karmy-Jones R, Wood
DE. Descending necrotizing mediastinitis: an analysis of the
effects of serial surgical debridement on patient mortality.
J Thorac Cardiovasc Surg 2000;119:260 –7.
15. Levine TM, Wurster CF, Krespi YP. Mediastinitis occurring
as a complication of odontogenic infections. Laryngoscope
1986;96:747–50.
16. Panda NK, Simhadri S, Sridhara SR. Cervicofacial necrotizing fasciitis: can we expect a favourable outcome? J Laryngol
Otol 2004;118:771–7.
17. Sancho LM, Minamoto H, Fernandez A, Sennes LU, Jatene
FB. Descending necrotizing mediastinitis: a retrospective
surgical experience. Eur J Cardiothorac Surg 1999;16:200 –5.
18. Hanly EJ, Aurora AA, Shih SP, et al. Peritoneal acidosis
mediates immunoprotection in laparoscopic surgery. Surgery 2007;142:357– 64.
19. Lee JW, Immerman SB, Morris LG. Techniques for early
diagnosis and management of cervicofacial necrotising fasciitis. J Laryngol Otol;124:759 – 64.
20. Pinto A, Scaglione M, Scuderi MG, Tortora G, Daniele S,
Romano L. Infections of the neck leading to descending
necrotizing mediastinitis: role of multi-detector row computed tomography. Eur J Radiol 2008;65:389 –94.
21. Boyer A, Vargas F, Coste F, et al. Influence of surgical
treatment timing on mortality from necrotizing soft tissue
infections requiring intensive care management. Intens Care
Med 2009;35:847–53.
22. Mathieu D, Neviere R, Teillon C, Chagnon JL, Lebleu N,
Wattel F. Cervical necrotizing fasciitis: clinical manifestations and management. Clin Infect Dis 1995;21:51– 6.
23. Mohammedi I, Ceruse P, Duperret S, Vedrinne J, Bouletreau
P. Cervical necrotizing fasciitis: 10 years’ experience at a
single institution. Intensive Care Med 1999;25:829 –34.
24. Roccia F, Pecorari GC, Oliaro A, et al. Ten years of descending necrotizing mediastinitis: management of 23 cases. J Oral
Maxillofac Surg 2007;65:1716 –24.
25. Chen KC, Chen JS, Kuo SW, et al. Descending necrotizing
mediastinitis: a 10-year surgical experience in a single institution. J Thorac Cardiovasc Surg 2008;136:191– 8.