Accuracy of computerized tomography in deep neck infections in the pediatric population

Accuracy of Computerized Tomography in Deep Neck Infections in the Pediatric Population Cetin Vural, MD,* Anil Gungor, MD,† and Susan Comerci, MD‡ Objectives: Computerized tomography (CT) is used widely to diagnose deep neck infections (DNIs), and, generally, the decision of surgical intervention is based on findings of a CT study. This study examines the accuracy of CT in differentiating abscess versus cellulitis in DNIs (lateral pharyngeal and retropharyngeal). Study Design: This is a retrospective chart review study with re-evaluation of the CT scans by a blinded observer. Methods: A retrospective review of medical records of 80 patients with DNIs who were evaluated with a CT study was performed. CT scans of these patients were reviewed by a radiologist who was blinded to the clinical and surgical findings and to the original CT study report. To diagnose the infection and differentiate abscess from cellulitis, our radiologist scored the CT scans regarding the following variables: low-density core, rim enhancement, soft-tissue swelling, obliterated fat planes, and mass effect. Radiologic diagnosis was compared with operative findings (whether pus found at surgery or not) in all cases treated surgically. Accuracy, sensitivity, specificity, and positive and negative predictive values of CT study were calculated. Results: Thirty-nine (49%) patients were treated medically with intravenous (IV) antibiotics alone, and 41 (%51) patients were treated both surgically and medically. The overall accuracy of CT in DNI was 63%. The sensitivity, specificity, and positive and negative predictive values were 68%, 56%, 71%, and 53%, respectively. Conclusion: CT study has important limitations in differentiating abscess versus cellulitis in DNIs. Clinical findings as well as CT diagnosis should guide the decision of surgery. (Am J Otolaryngol 2003;24:143-148. © 2003 Elsevier Inc. All rights reserved.)

The incidence of deep neck infections (DNIs) has decreased during the last decades, but they still continue to cause serious morbidity and mortality. Life-threatening airway compromise, septicemia, internal jugular vein thrombosis, carotid artery erosion, mediastinitis, pericarditis, and epidural abscesses are possible serious complications.1 The treatments of choice for deep neck abscesses are needle aspiration or incision and drainage in addition to intravenous (IV) antibiotics. When the cellulitis stage is accurately identified, DNIs can be treated with IV antibiotics alone.2 Radiologic studies including plain radiographs, ultrasound, computed tomography

From the *Ear Nose and Throat Department, Sisli Children’s Hospital, Istanbul, Turkey; and Departments of †Pediatric Oncology, and ‡Radiology, Children’s Hospital of Pittsburgh, Pittsburgh, PA. Address correspondence to: Anil Gungor, MD, Children’s Hospital of Pittsburgh 3705 Fifth Avenue, Pittsburgh, PA 15213. E-mail: [email protected]. © 2003 Elsevier Inc. All rights reserved. 0196-0709/03/2403-0003$30.00/0 10.1016/S0196-0709(03)00008-5

(CT), and magnetic resonance imaging are used to evaluate these patients. CT has become the most accurate and widely used imaging technique in the evaluation of children with DNI.2 However, our clinical experience3 shows that with CT scans, false-positive as well as false-negative studies are not infrequent. The purpose of this study is to determine the accuracy of CT in identifying an abscess in DNIs. BACKGROUND AND SIGNIFICANCE The retropharyngeal space is a potential space extending from the skull base to the superior mediastinum at about the level of the second thoracic vertebra. It lies between the prevertebral fascia (deep layer of the deep cervical fascia) posteriorly, the buccopharyngeal fascia (which covers the pharyngeal constrictors) anterosuperiorly, and the fascia covering the esophagus anteroinferiorly. The adherence of this fascia to the prevertebral fascia in the midlineis clinically important. Through this adherence, the retropharyngeal

American Journal of Otolaryngology, Vol 24, No 3 (May-June), 2003: pp 143-148

143

144

fascia is divided into a left and a right side. A retropharyngeal infection therefore will have laterality, whereas a prevertebral space infection will be midline. The prevertebral space extends between the skull base and the diaphragm, posterior to the retropharyngeal space.4,5 The lateral pharyngeal (parapharyngeal, pharyngomaxillary) space is also a potential space extending from the base of the skull to the hyoid cartilage and bone. The fascia of the pharyngeal constrictor muscles are medial, and the mandible and the pterygoid muscles are lateral. The styloid process divides this inverted pyramid-shaped space into anterior and posterior compartments. The internal carotid artery; internal jugular vein; cranial nerves IX, X, XI; and the cervical symphatetic trunk are in the posterior compartment. The anterior compartment contains regional muscles. The lateral pharyngeal space communicates with all other spaces formed by the deep layer of the deep cervical fascia. Children with infections in the lateral pharyngeal and prevertebral spaces often present with irritability, fever, dysphagia, muffled speech or cry, noisy breathing, stiff neck, and cervical lymphadenopathy. Stridor and drooling accompanies symptoms in advanced stages of infection. Lateral pharyngeal space infections may present with cranial nerve involvement (posterior compartment), Horner’s syndrome (cervical sympathetic chain in the posterior compartment), or trismus (anterior compartment). Regardless of the starting point, an infection in one of these spaces can spread to involve the other. Etiology is usually limited to respiratory and dental infections and blunt and penetrating trauma involving layers of the deep cervical fascia.2 Both anaerobes and aerobes are implicated (Streptococcus pyogenes, Staphylococcus aureus, Moraxella catarrhalis, Bacteroides species) Many are polymicrobial, and ␤-lactamase production is common. After the initial diagnosis is made, IV antibiotic treatment is indicated. Recommended antimicrobial agents are those effective against the bacteria commonly isolated in these infections. ␤-Lactamase stable antibiotics or those that inhibit ␤-lactamase are preferred. If anaerobes are thought to be involved, clindamycin would be an appropriate alterna-

VURAL, GUNGOR, AND COMERCI

tive. However, this drug is not recommended as monotherapy because of the involvement of gram negatives, such as Haemophilus influenzae, sulbactam ampicillin, and cefuroxime, which have excellent in vitro activity against the usual pathogens. If an abscess is suspected, management should include planning for surgical drainage. At this point, the differential diagnosis between abscess and cellulitis becomes an important issue in management. Even with the help of sophisticated imaging techniques, differentiating the two still can be a diagnostic challenge. In the past, the noninvasive diagnosis of DNIs mostly involved standard radiography. Diagnosing abscess is easy when air or air-fluid levels are present in soft-tissue neck radiograph films. Otherwise plain films are of limited help for the surgical decision-making process.1,5,6 A CT scan provides the most accurate information on the extent and exact anatomical location of the infection. In the last 2 decades, CT has replaced plain films in the diagnosis of DNIs. Holt and colleagues7 reported that CT scans had 100% accuracy in the diagnosis of deep neck abscesses. Some early studies reported similar reliable results of CT scans in infections of head and neck region.8-13 Although CT scan is the most accurate tool in diagnosing and delineating the localization of the infection in the head and neck, the accuracy of CT in distinguishing between abscess and cellulitis for DNIs has been questioned.1-3,5,6,14,15 In most reports, sample sizes were small. To date, the largest studies are reported by Elden et al2 and Ungkanont et al3 (110 and 117 patients, respectively). We are reporting a series of 80 subjects and contribute a large series to the literature. This study may also serve as a continuation to the study by Ungkanont et al, also from our institution.3 MATERIAL AND METHODS A retrospective review of medical records of children admitted to the Children’s Hospital of Pittsburgh from November 1992 to March 2001 with the diagnosis of DNI was performed (Ungkanont et al reporting up to 1992).3 Eighty patients were eligible and were included in the study. All subjects had a diagnosis of DNI (lateral pharyngeal and retropharyngeal infections and concomitant peritonsillar infection in some subjects) and had a CT scan of the head and neck with contrast during

DEEP NECK INFECTIONS

145

TABLE 1. Presenting Signs and Symptoms of DNI in 80 Children Treated at Children’s Hospital of Pittsburgh From November 1992 to March 2001

TABLE 2. Anatomic Sites of DNI in 120 Children Treated at Children’s Hospital of Pittsburgh From November 1992 to March 2001

Signs and Symptoms

Number of Children (%)

Anatomic Site

Cellulitis

Abscess

Total (%)*

Fever Limitation of neck motion Sore throat Neck Pain Neck swelling Decreased appetite Drooling Odynophagia Trismus Otalgia Irritability Difficulty breathing

46 (58) 35 (44) 30 (38) 27 (34) 15 (19) 14 (18) 10 (13) 7 (9) 6 (8) 5 (6) 3 (4) 3 (4)

Retropharyngeal Lateral pharyngeal Peritonsillar Total

137 37 6 80

16 10 0 26

53 (50) 47 (44) 6 (6) 106 (100)*

their admission. Patients with isolated peritonsillar infections (without lateral pharyngeal and retropharyngeal infections) were excluded. Gender, age, anatomic site of infection, presenting signs and symptoms, bacteriology of infection, and duration of hospital stay were recorded. CT scans were reviewed by the author (SC) and scored for the presence of the following variables: low-density core, rim enhancement, soft-tissue swelling, obliterated fat planes, and mass effect. The reviewer was blinded to the clinical and surgical findings and the original CT report. As a measure of internal consistency, the agreement between the initial CT report and the evaluation by our blinded observer was calculated for 65 patients, whose initial CT reports could be located.

RESULTS Eighty patients were included in this study. Forty-five (56%) were men and 35 (44%) were women. The mean age was 4.9 years (range 3 months-14 years). Forty-six children (58%) had fever on admission. Table 1 shows presenting signs and symptoms of the 80 children. The distribution of DNIs classified by site and by the presence of abscess and/or cellulitis is shown in Table 2. Thirty-nine (49%) patients were treated medically with IV antibiotics, whereas 41 (51%) were treated with surgical intervention (needle aspiration and/or incision and drainage within 24 hours) and IV antibiotics. Sulbactam ampicillin and clindamycin were the most commonly used IV antibiotics. Sulbactam ampicillin was used in 66%, and clindamycin was used in 31% of the patients alone or in combination with other antibiotics to cover ␤-lactamase producing strains and anaerobes. Antimicrobial treat-

*Twenty-six children had infections in 2 sites. †Patients who were treated medically only were classified as cellulitis in this table.

ment continued with oral antibiotics (mostly clavulanic acid amoxicillin or clindamycin) after a minimum of 5 days of IV treatment. All patients improved with their respective treatments. The average length of hospital stay was 4.2 days for medically treated patients and 6.7 days for patients treated surgically. Culture results of 27 specimens are shown in Table 3. The most frequently cultured pathogenic organisms were Streptococcus pyogenes (22%) and Prevotella species (15%). Inhabitants of the oral flora including ␣-hemolytic streptococcus and Neisseria species were identified in 44% and 11% of the cultures, respectively. Radiology The CT criteria used to diagnose and differentiate abscess from cellulitis are shown in Table 4.2,3,7 Table 5 shows the agreement between radiologic diagnosis and operative

TABLE 3. Organisms Isolated in 27 Cultures Obtained from 80 Children Treated at Children’s Hospital of Pittsburgh from November 1992 to March 2001

Organism

No. of cases (%)

␤-Hemolytic streptococcus (S pyogenes) Prevotella species Staphylococcus coagulase negative Haemophilus parainfluenzae Staphylococcus aureus Bacteroides species Pseudomonas aeruginosa Candida albicans ␣-Hemolytic streptococcus* Neisseria species* No growth

6 (22) 4 (15) 3 (11) 2 (7) 1 (4) 1 (4) 1 (4) 1 (4) 12 (44) 3 (11) 5 (19)

*Member of oral, oropharyngeal flora.

146

VURAL, GUNGOR, AND COMERCI

TABLE 4. CT Criteria Used in This Study to Diagnose Abscess and Cellulitis Criteria

Abscess

Cellulitis

Rim enhancement (complete) Low-density core Soft-tissue swelling Obliterated fat planes Mass effect

x x x x x

x x x x

findings in 41 DNI patients who underwent surgery. Radiologic diagnosis was in agreement with the operative findings in 26 (63%) patients (accuracy). There were 7 (17%) falsepositive studies (CT diagnosed abscess but no pus identified at surgery). Pus was identified during surgery in 8 (20%) patients with DNI, despite negative CT diagnosis (false negative). Sensitivity of CT was 68% and specificity was 56%. Positive and negative predictive values were 71% and 53%, respectively. Four out of 39 (10%) medically treated cases were improved without surgical intervention, despite radiologic diagnosis of abscess. Original reports of 3 of these 4 patients could be reached, and all were in agreement with the diagnosis of our reviewer. CT diagnoses of remaining 35 medically treated cases were cellulitis. Sixty-three of 65 (97%) original CT reports were in agreement with the diagnosis of our reviewer. Two (3%) of the original reports (both were reported to be abscess) were different from our reviewer’s diagnosis of cellulitis. Operative findings were in agreement (pus was found) with the original CT reports in these 2 patients. DISCUSSION In this study, 39 (49%) patients were treated with IV antibiotics alone, and 41 (51%) were treated with surgical drainage in addition to antibiotics. The percentage of medically treated cases is higher than it is reported in other studies (15%-25%).1,3,16 However, there are similar high success rates with medical treatment only in the medical literature.8,17 Culture results of the specimens taken in the operation room were in agreement with results of previous studies.2,3,16,18,19 Streptococcus pyogenes (22%) and Prevotella species, which are anaerobic organisms (15%),

were the most common microorganisms. Haemophilus parainfluenzae (7%) and Pseudomonas aeruginosa (4%) were the cultured gram-negative pathogens. Some of these organisms are potential ␤-lactamase producers, and initial empirical antimicrobial therapy in our patient group mostly included ␤-lactamase–stable antibiotics effective against ␤-lactamase–producing strains and anaerobes as recommended by authors.3,18,19 In this retrospective study, the sensitivity, specificity, and negative and positive predictive values of CT in DNI are lower than the values reported in previous studies. Specificity rates are lower than sensitivity rates (56% v 68%), and this is consistent with the results of others.3,6 The overall accuracy rate is 63%, indicating that CT diagnoses were not confirmed by the operation findings in 37% of the patients. Ungkanont et al3 reported sensitivity, specificity, and positive and negative predictive values of CT in distinguishing abscess from cellulitis in 16 children with head and neck space infections to be 91%, 60%, 83%, and 75%, respectively. Boucher et al6 reported the statistics of their study performed with 15 patients with retropharyngeal infection. The sensitivity was 100%, specificity was 45%, and positive and negative predictive values were 40% and 100%, respectively, in their study. High false-positive results of CT in neck infections were reported before.2,6 However, even higher false-negative results than false-positive results in our study (20% v 17%) were surprising. This showed that the physician should suspect an abscess with or without radiologic confirmation if clinical findings indicate an abscess or there is no improvement with antibiotic therapy. TABLE 5. Agreement Between CT Study and Incision and Drainage Findings in 41 Children With DNI Treated at Children’s Hospital of Pittsburgh From November 1992 to March 2001 Pus found at surgery (Incision and Drainage) CT scan

Yes

No

Positive (abscess) Negative (cellulitis)

17 8

7 9

NOTE. Sensitivity: 17/25 ⫽ 68%, specificity: 9/16 ⫽ 56%, positive predictive value: 17/24 ⫽ 71%, negative predictive value: 9/17 ⫽ 53%, and accuracy: 26/41 ⫽ 63%.

DEEP NECK INFECTIONS

Holt et al7 defined the CT findings of deep neck abscess as single cystic or multiloculated appearance, low-density CT number, air and/or fluid at the center of the abscess, contrast enhancement of the abscess wall, and anatomic boundaries that fit fascial spaces. We used 5 criteria (mentioned in Table 4) to diagnose DNIs and to differentiate abscess from cellulitis. The most important criterion to distinguish these 2 pathologies was complete rim enhancement. Our radiologist diagnosed abscess when she saw complete rim enhancement surrounding a low-density core. Elden et al2 stated that partially or thinly enhancing rims presented in both early abscess and cellulitis, and more rarely was absent in some abscesses. This can explain the false-negative and false-positive results. Low attenuation was shown not to be a reliable differentiating criterion, and there was no statistically significant difference in CT numbers between abscess and adenitis in the retropharyngeal region.14 Elden et al2 reported higher accuracy rates of CT in lateral neck infections (80%) versus DNI (72%), without any comment on this subject. Visceral vascular space lateral to the carotid is larger than deep spaces of the neck. Infection progresses more easily toward a large abscess formation in this space. Thus, the radiologist can diagnose an abscess when he/she sees a large low-density core surrounded with an enhanced rim. However, to diagnose an abscess without the help of a radiologic study is easier in the lateral neck than it is in the deep neck. Changes in the skin color, softening, or fluctuation of the mass give clues for abscess. NA can be used more easily in lateral infections for diagnosis and treatment in older children and can reduce the need for radiologic examination. In an ideal study designed to determine the accuracy of a diagnostic tool, each patient with the diagnosis of abscess or cellulitis would have to be subjected to open surgical intervention just after the diagnostic procedure. In such a study, the true accuracy of the diagnostic procedure could be calculated as comparing preoperative diagnosis and operative findings, which would be possible for all subjects and all diagnoses. Such a study design would be highly unethical and unacceptable. For these reasons, no such study will

147

ever be conducted. For calculations of accuracy of the CT studies, our study design precludes the inclusion of data from children who received only medical treatment. Some may argue that when surgical validation of CT diagnosis cannot be obtained and the subject improves with medical treatment, it would be safe to assume that all these subjects actually had cellulitis. Therefore, specificity and positive and negative predictive values would be affected. We are, however, most interested in data in which validation could be obtained through surgery, which was decided on based on the diagnostic test itself. If we accepted the medically treated subjects to be cellulites, the specificity, negative predictive value, and accuracy would have been higher in our study (80%, 85%, and 76%, respectively). In the scope of literature data and data provided from this study, in the management of a child with suspected DNI, radiologic assessment can be delayed for 48 hours, if there is no suspicion of complication and airway compromise. This might increase the accuracy of radiologic examination and decrease the rate of false-positive results. After 48 hours of IV antibiotic treatment, if no clinical improvement occurs or there is suspicion of complication or airway compromise, decision of surgery should be made even with radiologic diagnosis of cellulitis. A negative operative result is more preferable than a catastrophic complication, which is caused by a further delay in the decision of surgery. CONCLUSION Although CT study gives exact information for the diagnosis and the localization of the DNI, it has some limitations in differentiating abscess from cellulitis because of the falsenegative and false-positive results. Therefore, the decision of surgery must be based on not only the radiologic diagnosis but clinical findings as well. Especially our high false-negative results indicated that if the infection does not respond to medical treatment or there is a suspicion of complication, surgery should be considered with or without radiological confirmation of the clinical diagnosis.

148

REFERENCES 1. Lazor JB, Cunningham MJ, Eavey RD, et al: Comparison of computed tomography and surgical findings in deep neck infections. Otolaryngol Head Neck Surg 111: 746-750, 1994 2. Elden LM, Grundfast KM, Vezina G: Accuracy and usefulness of radiographic assessment of cervical neck infections in children. J Otolaryngol 30:82-89, 2001 3. Ungkanont K, Yellon RF, Weissman JL, et al: Head and neck space infections in infants and children. Otolaryngol Head Neck Surg 112:375-382, 1995 4. Yellon RF: Infections of the fascial spaces of the head and neck and cervical adenopathy in children, in de Souza C, Stankiewicz J, Pellitteri PK (eds): Textbook of Pediatric Otorhinolaryngology—Head and Neck Surgery San Diego, CA, Singular Publishing Group, Inc, 1999, pp 111-125 5. Bluestone CDRetropharyngeal and lateral pharyngeal space infections: in Kaplan SLCurrent Therapy in Pediatric Infectious Disease, (ed 3). St. Louis, MO, MosbyYear Book, Inc, 1993, pp 33-37 6. Boucher C, Dorion D, Fisch C: Retropharyngeal abscesses: A clinical and radiologic correlation. J Otolaryngol 28:134-157, 1999 7. Holt RG, McManus K, Newman RK, et al: Computed tomography in the diagnosis of deep-neck infections. Arch Otolaryngol 108:693-696, 1982 8. Gianoli GJ, Espinola TE, Guarisco JL, et al: Retropharyngeal space infection: changing trends. Otolaryngol Head Neck Surg 105:92-100, 1991 9. Endicott JN, Nelson RJ, Saraceno CA: Diagnosis and

VURAL, GUNGOR, AND COMERCI

management decision in infection of the deep fascial spaces of the head and neck utilizing computerized tomography. Laryngoscope 92:630-633, 1982 10. Barratt GE, Koopmann CF Jr, Coulthard SW: Retropharyngeal abscess—A ten-year experience. Laryngoscope 94:455-463, 1984 11. Nyberg DA, Jeffrey RB, Brant-Zawadzki M, et al: Computed tomography of cervical infections. J Comput Assist Tomogr 9:288-296, 1985 12. Welsh LW, Welsh JJ, Gregor FA: Radiographic analysis of deep cervical abscesses. Ann Otol Rhinol Laryngol 101:854-860, 1992 13. Brodsky L, Belles W, Brody A, et al: Needle aspiration of neck abscesses in children. Clin Pediatr 31:7176, 1992 14. Glasier CM, Stark JE, Jacobs RF, et al: CT and ultrasound imaging of retropharyngeal abscesses in children. Am J Neuroradiol 13:1191-1195, 1992 15. Matt BH, Lusk RP: Delineation of a deep neck abscess with magnetic resonance imaging. Ann Otol Rhinol Laryngol 96:615-617, 1987 16. Thompson JW, Cohen SR, Reddix P: Retropharyngeal abscess in children: A retrospective and historical analysis. Laryngoscope 98:589-592, 1988 17. Broughton RA: Nonsurgical management of deep neck infections in children. Pediatr Infect Dis J 11:14-18, 1992 18. Brook I: Microbiology of abscesses of the head and neck in children. Ann Otol Rhinol Laryngol 96:429-433, 1987 19. Asmar BI: Bacteriology of retropharyngeal abscess in children. Pediatr Infect Dis J 9:595-597, 1990