Deep cervical abscesses in children: efficacy of the cefotaxime–rifampicin combination

The objective is to determine whether a medical treatment, the combination of cefotaxime and rifampicin, is effective in avoiding surgery for managing deep cervical abscesses in children and to determine prognostic factors in the efficacy of this medical treatment. This is a retrospective analysis of all patients under 18 presenting with para- or retro-pharyngeal abscess over the period 2010–2020 in two hospitals' pediatric otorhinolaryngology departments. One hundred six records were included. Multivariate analyses were performed to study the relationship between the prescription of the Cefotaxime-rifampicin protocol at the onset of the management and the use of surgery and to evaluate the prognostic factors of its efficacy. The 53 patients who received the cefotaxime-rifampicin protocol as first-line treatment (vs. 53 patients receiving a different protocol) required surgery less frequently: 7.5% versus 32.1%, validated by a Kaplan–Meier survival curve and a Cox model analysis adjusted for age and abscess size (Hazard Ratio = 0.21). This good outcome of the cefotaxime-rifampicin protocol was not demonstrated when it was instituted as a second-line treatment after the failure of a different protocol. An abscess larger than 32 mm at hospitalization was significantly associated with more frequent use of surgery in multivariate analysis adjusted for age and sex (Hazard Ratio = 8.5). Conclusions: The cefotaxime-rifampicin protocol appears to be an effective first-line treatment in managing non-complicated deep cervical abscesses in children. What is Known: • Nowadays, medical treatment is preferred for managing deep neck abscesses in children. There has yet to be a consensus on the antibiotic therapy to be proposed. • Staphylococcus aureus and streptococci are the most frequent causative organisms. What is New: • The cefotaxime-rifampicin protocol introduced at first intention is effective, with only 7.5% of patients requiring drainage surgery. • The only risk factor for failure of the medical treatment is the initial size of the abscess. What is Known: • Nowadays, medical treatment is preferred for managing deep neck abscesses in children. There has yet to be a consensus on the antibiotic therapy to be proposed. • Staphylococcus aureus and streptococci are the most frequent causative organisms. What is New: • The cefotaxime-rifampicin protocol introduced at first intention is effective, with only 7.5% of patients requiring drainage surgery. • The only risk factor for failure of the medical treatment is the initial size of the abscess.


Introduction
Parapharyngeal abscesses (PPA) and retropharyngeal abscesses (RPA), also called deep cervical abscesses, are complications of upper respiratory infections in children [1] or, more rarely, of a cervical injury. The primary pathogens are from the oropharyngeal and nasopharyngeal flora and are predominantly anaerobic or aero-anaerobic bacteria; the predominant groups are pyogenic streptococci and staphylococci [2][3][4]. The most common presentation of these abscesses is by torticollis with fever. These abscesses may be life-threatening in exceptional cases, causing airway obstruction, septic thrombophlebitis, or mediastinitis [5].
Before the 1990s, surgeons routinely operated on those patients presenting with such abscesses [5,6]. Nowadays, initial medical management is preferred by a large number of teams, surgery being a second-line treatment following the failure of antibiotic therapy [7][8][9][10]. Management algorithms based on the clinical evaluation and the size of the abscess on imaging have been constructed [7,11].
There needs to be a consensus on the choice of antibiotic regime. In our institutions, the preferred antibiotic therapy protocol in pediatrics is the combination of intravenous cephalosporin (cefotaxime) and anti-staphylococcal (rifampicin).
The main objective of this study was to determine whether the combination of cefotaxime and rifampicin is effective in avoiding surgery for managing deep cervical abscesses in children. The secondary objective was to determine factors associated with the need for surgery.

Materials and methods
This study was approved by the institutional ethics committee of our institution (#PADS20-260). According to French Law, no patient or parent objected to the use of the data for this research.
All charts of patients under the age of 18 years presenting with PPA and RPA over the period 2010-2020 in the Pediatric Otorhinolaryngology Department of a tertiary-care University Hospital and the Otorhinolaryngology Department of another regional public hospital, both following the same management protocol, have been included. Patients presenting with other neck abscesses, such as uncomplicated peritonsillar abscesses and lateral cervical abscesses were excluded.
Demographic data, medical history, pre-hospitalization course and treatments, blood test profile, imaging, clinical signs, suspected etiology, whether referred by a physician, transferred from another center, consultation time, surgical status, medical treatment, and the presence of complications such as transfer to intensive care, intubation or tracheotomy, cervical traction for residual torticollis were collected.

Statistical analyses
Continuous quantitative variables were analyzed using median [InterQuartile Range], and categorical variables were presented as proportions. Patients were compared by their operated and non-operated status in the univariate analysis, with the Mann-Whitney and Fisher exact tests for continuous and categorical variables, respectively.
To estimate the size of abscesses predictive of surgical treatment, a Receiver Operator Characteristics analysis (ROC) was used. The area under the curve and its 95% confidence interval (CI) were estimated, and the optimal threshold predictive of surgical treatment by abscess size was selected.
To assess the efficacy of the cefotaxime-rifampicin protocol in avoiding surgery, two types of analyses were performed: a survival analysis by the Kaplan-Meier method and a Hazard Ratio (HR) analysis by the Cox model.
All analyses were two-sided, and p < 0.05 was considered statistically significant. Statistical analyses were performed with R studio software, including R version 3.6.3 (RStudio, PBC, Boston, MA). Figures were performed using the Gplot 2 extension.

Medical and surgical management
The cefotaxime-rifampicin intravenous antibiotic therapy protocol was developed jointly with our University Hospital's department of pediatric infectious diseases and administered for five days. The dose was 150-200 mg/kg/day for cefotaxime and 20-30 mg/kg/day for rifampicin. This protocol was the preferred antibiotic regimen for the study period, covering both usual oropharyngeal bacteria and staphylococcal species. It was followed by ten days of oral intake of amoxicillin-clavulanic acid (in the absence of allergy to penicillin) or adapted to the bacteriological analysis when available.
In both departments, a strategy of primary medical treatment is the rule unless the patient presents with initial dyspnea or complication. The secondary criterion for surgical intervention is the lack of clinical improvement or worsening of the general condition despite intravenous antibiotic therapy for two days. The preferred surgical approach is transoral; the cervical approach is limited to abscesses lateral to the neck's vascular axis.

Population
After reading the data of all charts over the ten years and applying the exclusion criteria, 106 records were included. The main characteristics of the patients are summarized in Table 1, and the age distribution is in Fig. 1.

Initial evaluation
Fever and torticollis were almost constant; other clinical signs were less frequent ( Table 1). The suspected original cause of the cervical abscess was tonsillitis in 54 (50.9%) patients, nasopharyngitis in 29 (27.4%), acute otitis media in 6 (5.7%), and no etiological factor were discernable in 17 (16%) cases. The median time between the first symptoms and the initial medical consultation was 4 [3;7] days.
All patients had an initial inflammatory syndrome. Contrast-enhanced CT scanning was performed in 104 patients (98.1%), cervical ultrasounds in one, and no imaging in one.
The two patients without initial CT scanning were managed in another center and underwent CT scanning after the failure of a dual-antibiotic therapy with ceftriaxone-metronidazole. 15 (14.2%) patients underwent a secondary CT scan; 6/15 patients were operated on after this second imaging examination.
On the initial imaging, i.e., the CT and the ultrasounds, a collection was present in 94 (89.6%) patients. In the other 11 (10.4%) patients, cellulitis of the deep cervical spaces was described (Fig. 2). For 78 patients (73.6%), at least one measurement on one axis of the collection was available on the radiological report, and for 70 patients (66%), a second measurement on a second axis was available. 31.1% (33/106) of the patients had received antibiotic therapy before hospitalization. 50% (53/106) had not received any antibiotic therapy and received the cefotaxime-rifampicin protocol as the first-line treatment. The prior use of nonsteroidal anti-inflammatory drugs (NSAIDs) was reported for 40 patients (37.7%).
No side effects, anaphylactic reaction, or disturbance of hepatic or renal function were observed in the population treated by the cefotaxime-rifampicin protocol and other antibiotic therapies.
Using a univariate analysis, the median length of hospitalization was 5 days [5;5] in the non-operated patients vs. 7 days [5;8] in the operated patients (p < 0.001). The duration of medical treatment did not differ in operated and nonoperated patients, with a median of 15 days [15;15] (min 2 -max 20, p = 0.187).
Fifty-nine patients had blood cultures taken on arrival (55.7%). In only 7 was an organism identified (3 Streptococcus pyogenes, 1 Streptococcus conselatus, 1 Staphylococcus hominis, 2 Staphylococcus epidermidis). Twenty-two local samples were taken: 21 from the puncture or drainage fluid and one pharyngeal sample identifying a group A Streptococcus. 12/21 samples taken in the operating room revealed one or more pathogens (4 Streptococcus, 3 Staphylococcus aureus, 1 of which was resistant to methicillin; 2 sensitive Haemophilus; 4 polymorphic flora).

Prognostic factors in medical evaluation
No difference could be identified between the patients treated by medical treatment and those who required surgery (Table 1). NSAIDs were not significantly associated with increased risk of surgery: 57.1% of operated patients vs. 32.9% of non-operated patients (p = 0.072).
The ROC analysis of the collection size estimated that the threshold of 32 mm gave an optimal balance of sensitivity and specificity (Fig. 3). 42.9% (9/21) of abscesses larger Fig. 1 Age distribution of the patients. 76.4% of the children were younger than five years of age Fig. 2 CT scans of deep neck abscesses. A: parapharyngeal abscess originating from rhinopharyngitis. B: parapharyngeal abscess originating from a peritonsillar abscess. C: retropharyngeal abscess originating from rhinopharyngitis than 32 mm were drained at the beginning of management in our institution, with a mean central axis measured at 42 mm (min 33 -max 50). In multivariate analysis adjusted for age and sex, a size of 32 mm at hospitalization was significantly associated with a higher surgery rate with an HR of 8.5 (95% CI 2.96 to 24.2).

Cefotaxime-rifampicin protocol evaluation
Fifty-three patients received the cefotaxime-rifampicin protocol as first-line treatment, and 53 received a different one. There were no statistically significant differences in age, gender, and abscess location or size between the two groups ( Table 2). Using univariate analysis, patients receiving the cefotaxime-rifampicin protocol as the first-line treatment required surgery less frequently than patients receiving other antibiotic protocols: 7.5% (4/53) versus 32.1% (17/53), p = 0.0025. This limitation of the need for surgery was validated by a Kaplan-Meier survival curve (p = 0.0025, Fig. 4) and a Cox model analysis adjusted for age and abscess size with an HR at 0.21 (95% CI 0.047 to 0.93, Fig. 5).
This good outcome was not demonstrated when the protocol was instituted after the failure of a different antibiotic therapy (p = 0.520): 25.9% of the patients underwent surgery (7/27) if this protocol was initiated after the failure of other antibiotic treatment, versus 38.5% (10/26) in the absence of this protocol.

Discussion
This retrospective study evaluated the course of retropharyngeal and parapharyngeal abscesses in the function of the initial treatment. Globally, 19.8% of the patients required surgery. Of the 53 patients receiving the cefotaxime-rifampicin protocol as first-line treatment, 7.5% underwent surgery. In contrast, of the 53 patients who received an alternative protocol, 32.1% were operated on (p = 0.0025), corresponding to an HR of 0.21 (95% CI 0.047 to 0.93). The cefotaximerifampicin protocol introduced at first intention was a promising therapeutic strategy. The choice of this dual-antibiotic therapy aims to cover the bacteria frequently found in the naso-and oropharynx, such as methicillin-susceptible staphylococci, streptococci and anaerobic organisms using cefotaxime [2,12], and to enlarge the spectrum to methicillin-resistant staphylococcus aureus (MRSA) using rifampicin. This attitude is supported  . 4 Result of the survival analysis by the Kaplan-Meier method. The survival analysis started at the time of diagnosis; the endpoint was defined either by the date of surgery or by discharge without surgery. The blue curve corresponds to the population that received the cefotaxime-rifampicin pro-tocol from the beginning of the management, and the red curve to the group that received a different antibiotic therapy. A log-rank test compared the survival curves by Fellner et al. [13], who found that staphylococcus aureus and streptococci are the main bacteria isolated in pediatric neck abscesses. Cefotaxime is a third-generation cephalosporin with an extended spectrum and a resistance profile similar to ceftriaxone. Both are effective on methicillinsusceptible staphylococci [14], streptococci, and anaerobic organisms, and their clinical efficacy is the same in treating severe infections [15]. According to Panosetti, the effectiveness of ceftriaxone is equivalent to the amoxicillin-clavulanic acid combination on phlegmonous and abscess-forming ENT infections [16]. Third-generation cephalosporins are given because they penetrate tissues, are well tolerated even in neonates and infants [17], and provide the patient prolonged, broad-spectrum coverage until the patient begins outpatient oral antibiotic therapy. In the study by Fellner et al. [13], of the staphylococcus aureus isolates, only 1 (8%) was methicillin-resistant; but 4 (31%) were clindamycin-resistant isolates with rare multiresistant. In Europe in 2017, 5.7% of staphylococcus aureus were resistant to rifampicin, which remains one of the standards of care, even though resistance to rifampicin may be a burden in the future [18]. In a study conducted in South Korea over the 2007-2020 period, rifampicin-resistant isolates were found in 6.3% of the patients with MRSA bacteremia, and rifampicin-resistant MRSA isolates were more likely to have multidrug resistance, and a higher vancomycin minimum inhibitory concentration relative to the rifampicin-susceptible isolates [19]. In case of infection with rifampicin-resistant MRSA, antibiotic therapy should be discussed on a case-by-case basis and adapted to the antibiogram. Our results also support a better efficacy of rifampicin than clindamycin. Only 7.5% of the patients who had the cefotaxime-rifampicin protocol as the first-line treatment required surgery, which is better than the results usually reported with clindamycin: the reported failure rates are of 18% for the combination of ceftriaxone and clindamycin [7], 25% when using Clindamycin alone [20] and 68.5% when using a combination of cefuroxime and metronidazole or clindamycin [21]. The combination of cefotaxime and rifampicin can be used in most children because of the low risk of cephalosporin allergy and the low incidence of side effects for both antibiotics [22,23]. It allows us to offer a unique, well-tolerated, and effective protocol for all patients, from infants to adolescents. The protocol duration is five days, leading to hospitalization of the same duration. For some teams, an earlier return home at three days seems possible [8]. Ambulant follow-up of the patients in this critical period is also feasible. The advantage of a medical protocol is the reduction of the anesthetic risk and the absence of cervical scar and postoperative care. Conversely, choosing a medical strategy does not allow for systematic bacteriology, which is an obstacle to modifying the duration of antibiotic therapy and its spectrum in the future.
To avoid surgical drainage, bedside drainage by needle puncture under local anesthesia may be helpful if medical treatment fails; if pus is found [7], it allows bacteriologic analysis but with a higher risk of failure than drainage in the operating room. Therefore, surgical drainage in the operating room was preferred for our patients.
This study's median age of 4 years is close to that found in other studies and confirms that deep cervical abscesses are more common in young children. [1,2,8] The decreased risk of deep cervical abscesses in older children seems to be related to the regression of the lymph nodes in the retropharyngeal space around the age of 8 years [24]. The higher proportion of boys treated for this condition is consistently found in studies but is not explained [2,5,7].
CT scanning is the examination of choice for suspected deep cervical abscesses [25]. It allows the diagnosis to be established, the collection to be located, the measurements to be assessed, complications to be looked for, and the surgical procedure to be guided if necessary. It is also helpful to determine the presence or absence of pus during surgery based on the peripheral enhancement of the abscess [7]. A second CT scan is not routinely required in pediatric cases, where clinical surveillance is, in most cases, sufficient. However, a second CT scan is still necessary in the case of failure of medical treatment. In our series, 40% (6/15) of the indications for surgery were made following this second CT scan. Beyond the initial diagnosis, the CT scan can be used to correct uncertain diagnoses. In this series, three diagnoses of parapharyngeal abscesses were made following a second CT scan after the failure of orthopedic treatment for non-febrile torticollis. Three diagnoses of Kawasaki disease associated with parapharyngeal abscesses were made when fever persisted after six days of antibiotic therapy with the complete disappearance of the abscesses on the follow-up scan. Two CT scans were also needed to diagnose the recurrence of abscesses 48 h after stopping the antibiotic treatment in the face of the reappearance of torticollis.
In our study, a threshold of 32 mm was associated with a higher risk of surgical drainage of the purulent collection. This measure is only indicative, and the link of causality cannot be established: the retrospective methodology does not allow it to be used as a criterion for surgical decisions. Indeed, a specific study should be conducted using three dimensions or volume measurements, with a reading by several radiologists, to assess a precise threshold leading to an indication of surgical management. As it was not the primary objective of this study, this specific analysis was not performed, and the measurements were only taken from the reports. Hence only 78/94 patients were included in the calculation. However, extended cervical collections in a child should encourage the clinician to be vigilant. According to some authors, a threshold of 20 mm should guide the surgeon toward surgery, even without clinical or radiological evidence of complications [7]. For others, an abscess > 25 mm should make the surgeon more vigilant about the need for surgery [5]. Given our results regarding the efficacy of the cefotaxime-rifampicin protocol, it seems reasonable to institute medical treatment for 48 h before deciding on surgical drainage in the absence of clinical or radiological signs of complications, regardless of the size of the abscess.
This study does not illustrate an increased risk of failure of medical treatment when NSAIDs are used. The adverse role of NSAIDs is strongly suspected in developing serious infections [26][27][28]. NSAIDs may mask symptoms and allow the abscess to develop at a low level or aggravate the conditions. A longitudinal study, based on a cohort of more than 120 general practitioners in a sentinel network, showed that NSAIDs were prescribed in 22% of cases of pharyngitis, with an increased risk of peritonsillar abscess [27]. However, this link remains controversial and challenging to demonstrate. There are several pitfalls in studying the excess risk induced by NSAID use. There is a recruitment bias: patients treated in hospitals are more severe than those treated in the community, and self-medication with NSAIDs is also difficult to evaluate.
In contrast to NSAIDs, corticosteroids may improve the success rate of medical treatments and decrease the need for surgery. Goenka et al. found a lower drainage rate in the corticosteroid group (odds ratio: 0.28; 95% CI 0.22-0.36) on a massive cohort including 1677 patients without corticosteroids and 582 with corticosteroids on initial management [29]. Tansey et al. found a lower rate of surgical drainage (36%) in the dexamethasone group and compared with the non-dexamethasone group (53%, p = 0.043) on 153 patients mixing all cervical abscesses, including peritonsillar and lateral abscesses [30]. Villanueva-Fernandez et al. successfully treated 30 children, with 11 retropharyngeal and 19 parapharyngeal abscesses, using intravenous amoxicillin/clavulanic acid and corticosteroids, and also recommend reserving the surgical drainage for cases with a complication associated such as an airway compromise, lack of response to antibiotic therapy, and immunocompromised patients [31]. As these data on corticosteroids are recent, they may lead us to reconsider our management shortly. In our series, 26 patients received corticosteroids if they improved after 48 h of antibiotic therapy. Still, the data in the charts were not precise enough to perform statistical analysis and draw conclusions. This study has certain limitations. First, it is a retrospective study with biases related to data collection. However, the presence of computerized medical charts with the coding of the diagnosis strongly limits the loss of data. Second, to improve the accuracy of the collections' dimensions and validate their measurement, a repeatability and reproducibility study should be performed, which was not conducted here. Working on volumes rather than 2D dimensions may also improve the accuracy of the measurements on imaging. Third, validating the possible superiority of the Cefotaxime-rifampicin protocol over the others would require a prospective superiority study with a control group, which is a line of future work. Fourth, this study's results are valid for our bacterial ecology. Other combinations of antibiotics may be more relevant elsewhere, depending on the local bacterial ecology.

Conclusion
The choice of an initial medical treatment seems to be a good strategy for managing non-complicated deep cervical abscesses in children, regardless of the abscess size. The cefotaxime-rifampicin protocol appears to limit the need for surgery more effectively than other antibiotic protocols previously studied.