Management and Outcomes of Spontaneous Pneumomediastinum in Children

INTRODUCTION

Spontaneous pneumomediastinum is defined as the presence of air in the mediastinum, not precipitated by trauma or iatrogenic causes. It is primarily caused by forced exhalation against a closed glottis [1,2] and is further categorized as primary or secondary. Primary spontaneous pneumomediastinum occurs in otherwise healthy children in the absence of pulmonary conditions. Patients frequently present with complaints of chest pain, cough or discomfort with swallowing. [2,4] The majority of patients have no identifiable triggering event although activities such as vomiting, coughing, choking, athletic exertion and huffing are often described. [2] Adolescent males are most commonly affected, however cases of spontaneous pneumomediastinum have been described in children as young as five months. [2,4,3] Secondary spontaneous pneumomediastinum is associated with any predisposing pulmonary pathologies such as asthma, viral respiratory infections and pneumonia. [4,5] Secondary pneumomediastinum caused by spontaneous esophageal perforation has also been reported, albeit rarely, in the pediatric literature. [5,6,1] Differences in outcomes between primary and secondary pneumomediastinum in children have not been clearly established. There is also no consensus on the management of pediatric patients presenting to the emergency department (ED) with spontaneous pneumomediastinum, primary or secondary. Many of these patients are admitted for observation and undergo additional diagnostic studies. The objective of this study was to characterize the management of spontaneous pediatric pneumomediastinum with a focus on the diagnostic yield of additional studies and the outcomes of patients who are admitted and those who are discharged.

MATERIALS AND METHODS

Study Design

This study was approved by the Institutional Review Board of the University of Pittsburgh. As this was a retrospective study, consent for participation was waived by the Institutional Review Board. Spontaneous pneumomediastinum was defined as the presence of pneumomediastinum in the absence of recent trauma or surgical intervention. Three subgroups were defined. First, primary spontaneous pneumomediastinum was defined as spontaneous pneumomediastinum in well children without predisposing lung pathology, such as asthma, pneumonia, or upper respiratory infections and with no history of vomiting. Second, children who had a recent history of vomiting or choking prior to symptom onset were categorized as secondary spontaneous gastrointestinal-associated pneumomediastinum. We chose to analyze these patients separately, as in rare instances, a perforation of the esophagus known as Boerhaave syndrome may be associated with retching or vomiting. [5] Third, secondary spontaneous respiratory-associated pneumomediastinum was defined as spontaneous pneumomediastinum in the presence of asthma exacerbation, viral respiratory infection or pneumonia. Asthma exacerbation was defined as an episode of wheezing in a patient with history of asthma or documentation in the clinical note that the patient was experiencing an asthma exacerbation. Hypoxia was defined as an oxygen saturation less than 94%. Patients who developed cough or respiratory symptoms following a choking episode or vomiting were classified as secondary gastrointestinal-associated pneumomediastinum. Those who experienced vomiting in the setting of respiratory infection were classified as secondary respiratory-associated pneumomediastinum

Patient Population

To maximize identification of patients with pneumomediastinum, we searched two separate databases: a hospital database containing ICD codes and a radiological database, both for the period 1/1/2009 to 2/3/2015. This time range correlated with availability of complete data from the electronic health record and the radiology database. Patients between the ages of 0 and 18 years and presenting to a single tertiary pediatric ED with the ICD 9 codes 518.1 (pneumomediastinum or interstitial emphysema), 770.2 (pneumomediastinum in a newborn) and 958.7 (subcutaneous emphysema) were identified. Children with pneumomediastinum in the setting of vomiting or choking are often suspected of esophageal perforation, and for this reason 530.4 (esophageal perforation) was also included. Additionally, we performed a keyword search of our hospital’s radiological database. Any radiographic study interpreted by the department of radiology during the study period containing the term “pneumomediastinum” was identified. Charts identified using both search strategies were reviewed. Patients were excluded if they had a history of recent trauma or surgical intervention, had a tracheostomy or history of airway surgery or cardiothoracic surgery, were known to have lung or mediastinal cancer, had a diagnosis of cystic fibrosis, required critical care on presentation, or received positive pressure ventilation immediately prior to diagnosis.

Study Protocol

The following data were collected: presenting symptoms, inciting events, physical exam findings, initial oxygen saturation, diagnostic testing, medical interventions, disposition, outcomes and readmission. Charts were also reviewed for follow up visits, radiological tests or surgical interventions within six months of initial presentation.

We separately analyzed patients in the three subgroup categories: primary spontaneous pneumomediastinum, secondary spontaneous gastrointestinal-associated pneumomediastinum and secondary spontaneous respiratory-associated pneumomediastinum.

Statistical analysis

Data were analyzed using statistical package Stata/SE 15.0, College Station, TX. Analysis of the data was performed using descriptive statistics. Data with normal distribution are presented as mean standard deviation and data with nonparametric distribution are reported as median with interquartile ranges. Categorical data is reported as percentages. Chi-squared analysis or Fisher’s exact test were performed for categorical data. Continuous variables were analyzed using ANOVA with Bonferroni correction for post hoc analysis of significant findings. Using primary spontaneous pneumomediastinum as a reference group, odds ratios and mean differences are reported as well.

RESULTS

Patient Characteristics

Figure 1 presents the identification of the 183 patients included in our study.

Figure 1.

Patients were identified using two separate databases. ICD-9 codes identified 252 potential patients and 136 patients who met inclusion criteria. A keyword search of a radiologic database identified 1881 charts with radiology study interpretations that used the word “pneumomediastinum.” From this, 157 patients met inclusion criteria. There were 110 duplicate patients between the two databases and 183 patients were included for analysis.

Table 1 describes the demographics and presenting symptoms of patients grouped by pneumomediastinum type: 1) primary spontaneous pneumomediastinum, 2) secondary spontaneous gastrointestinal-associated pneumomediastinum and 3) secondary spontaneous respiratory-associated pneumomediastinum. The mean age was 12.8 ± 4.7 years, with 68% males.

Table 1.

Demographics and Presentation of Patients with Pneumomediastinum by Group

	Primary Spontaneous	Gastrointestinal - Associated	Respiratory -Associated	p-value	Odds Ratio/Mean Difference (95% Cl)
					Gastrointestinal vs. Spontaneous	Respiratory vs. Spontaneous
Total	64	31	88
Age	15.4 (2.2)	14.4 (4.7)	10.5 (5.1)	<0.001	−0.93 (−2.3, 0.5)	−4.92 (−6.3, −3.6)
Male	67% (43)	68% (21)	68% (60)	0.99
Activity				<0.001
Yes	36% (23)	90% (28)	18% (16)		26.78 (3.4,214.1)	2.55 (0.8, 7.7)
No	34% (22)	3% (1)	7% (6)
Not recorded	30% (19)	6% (2)	75% (66)		0.16(0.1,0.7)	6.80 (3.3, 13.9)
Chest pain				<0.001
Yes	90% (58)	58% (18)	50% (44)		0.09 (0.0, 0.4)	0.09 (0.0, 0.3)
No	5% (3)	32% (10)	28% (25)
Not recorded	5% (3)	10% (3)	22% (19)		2.20 (0.4, 11.5)	5.50(1.6, 19.6)
Odynophagia				<0.001
Yes	31% (20)	35% (11)	7% (6)		1.65(0.5, 5.4)	0.20 (0.1, 0.6)
No	28% (18)	19% (6)	31% (27)
Not recorded	41% (26)	45% (14)	62% (55)		1.62(0.5, 5.0)	1.41 (0.7, 3.0)
Cough
Yes	2% (1)	35% (11)	94% (83)
No	77% (49)	55% (17)	2% (2)
Not recorded	22% (14)	10% (3)	3% (3)
Vomiting
Yes	0% (0)	55% (17)	17% (15)
No	73% (47)	39% (12)	59% (52)
Not recorded	27% (17)	6% (2)	24% (21)

Of the 183 patients, 96 (52%) had documentation of the activity or setting in which symptoms first developed. Of these, 29 (16%) were at rest or waking from sleep when symptoms started. Reported inciting events included participating in an athletic activity (n=16), eating or drinking (n=14), and vomiting (n=13) or coughing (n=10) immediately prior to symptom onset. A small number reported dusting or sweeping (n=3) at the time of symptoms. The remaining ten patients reported an assortment of activities including riding a roller coaster, screaming loudly, holding in a sneeze, leaning forward, or playing games at camp. Associated symptoms were chest pain 120 (66%), cough 95 (52%), odynophagia 37 (20%), and vomiting 32 (17%). Age, activity preceding symptoms and symptoms of chest pain and odynophagia differed significantly by subgroups (Table 1).

Most patients (108, 59%) had no significant past medical history. A history of asthma was recorded in 62 (34%) patients. Of patients with a history of asthma, 42 of 62 (68%) were noted to be wheezing at the time of presentation. In all patients, a viral upper respiratory infection was diagnosed in 38 (21%) and 24 (13%) were diagnosed with a bacterial pneumonia. Other medical comorbidities included diabetes mellitus (4), Crohn’s disease (1), low body mass index and disordered eating (2), severe constipation (1), gastroesophageal reflux (2), thrombocytopenia with absent radius syndrome (1), hypothyroidism (1), and human immunodeficiency virus (1). All four of the patients with diabetes presented in diabetic ketoacidosis. Two of these patients had vomiting prior to the onset of symptoms. The second two patients with diabetic ketoacidosis denied any vomiting and both described a sudden onset of chest pain while at rest as the primary reason for seeking medical care. The patient with severe constipation presented to the ED for abdominal pain and fecal impaction and this patient’s pneumomediastinum was an incidental finding on abdominal CT.

Management and Outcomes

The diagnosis of pneumomediastinum was established by chest radiograph (CXR) in 164 (90%) patients, computed tomography (CT) of the chest in 15 (8%), neck CT in one patient, neck radiograph in 1 patient and by incidental abdominal CT finding in 2 patients. After initial diagnosis, 137 (75%) of patients underwent further studies. Subsequent CXR was performed in 99 (54%) patients, with 24 (13%) receiving 2 or more repeat CXRs. After the initial diagnosis of pneumomediastinum, Chest CT was performed in 53 patients (29%), esophagram in 45 (25%) patients and 13 (7%) patients underwent diagnostic laryngoscopy. Other studies included echocardiography (2), CT of the neck (7), CT of the abdomen (2) Doppler studies due to incidental findings on CT (1), and esophagogastroduodenoscopy (1).

The management of pneumomediastinum for each of the three categories is presented in Table 2. Patients who presented with secondary spontaneous gastrointestinal-associated pneumomediastinum had significantly more radiological studies (p = 0.002) and were more likely to undergo esophagram (p = 0.002) and other additional studies including CTs of the neck and abdomen (p<0.0001). Despite the differences in rates of additional investigations, no study in any patient yielded additional findings or diagnoses. No surgical interventions were performed. An oxygen saturation <95% was reported in 25 (14%) patients, all of whom were diagnosed with secondary spontaneous respiratory-associated pneumomediastinum and concurrent respiratory infection or asthma. None of the patients with primary spontaneous pneumomediastinum or secondary spontaneous gastrointestinal-associated pneumomediastinum had a reported oxygen saturation <95%. Overall, oxygen was administered to 66 (36%) patients; 47 of these 66 (71%) patients were classified as secondary spontaneous respiratory-associated pneumomediastinum. Oxygen was administered to 15(23%) patients with primary spontaneous pneumomediastinum and 4 (13%) patients with secondary spontaneous gastrointestinal-associated pneumomediastinum.

Table 2.

Management of Patients with Pneumomediastinum by Group

	All	Primary Spontaneous	Gastrointestinal -Associated	Respiratory -Associated	p-value	Odds Ratio/Mean Difference (95% Cl)
						Gastrointestinal vs. Spontaneous	Respiratory vs. Spontaneous
Total patients	183	64	31	88
Total studies/patient	2.6 (1.4)	2.5 (1.3)	3.6 (1.8)	2.3 (1.3)	<0.001	1.06 (0.4, 1.7)	−0.22 (−0.6,0.2)
NXR/CXR	1.8 (1.0)	1.6 (0.8)	2(1.0)	1.8 (1.0)	0.15
CT chest	37% (68)	48% (31)	45% (14)	26% (23)	0.01	0.88 (0.4, 2.1)	0.38 (0.2, 0.8)
Esophagram	25% (45)	27% (17)	58% (18)	11% (10)	<0.001	3.75(1.5,9.3)	0.35 (0.2, 0.8)
Laryngoscopy	7% (13)	8% (5)	16% (5)	3% (3)	0.03	2.87 (0.9, 9.3)	0.42 (0.1, 1.8)
Other studies	7% (13)	5% (3)	23% (7)	3% (3)	0.001	6.57(1.7, 26.1)	0.72 (0.1, 3.7)
Oxygen saturation	98 (96–100)	99 (98–100)	99 (98–100)	96 (94–99)	<0.001	1.02 (0.7, 1.5)	0.52 (0.4, 0.7)
Oxygen given	36% (66)	23% (15)	13% (4)	53% (47)	<0.001	0.80 (0.3, 2.2)	3.74(1.8,7.7)
Admitted	78% (142)	63% (40)	87% (27)	85% (75)	0.002	4.05(1.3, 13.0)	3.46(1.6, 7.5)
Admit LOS	27 (18.4–45.6)	24 (14.2–33.1)	36 (18.3–58.4)	27 (20.0–47.9)	0.06
Discharge LOS	3.5 (2.8–4.2)	3.4 (2.6–4.0)	4.5 (3.7–6.9)	3.9 (3.2–4.2)	0.03	1.89 (0.5, 3.3)	0.22 (−0.6, 1.0)
Transferred	56% (103)	64% (41)	74% (23)	44% (39)	0.01	1.61 (0.6, 4.2)	0.45 (0.2, 0.9)
Admitted	81% (83)	66% (27)	91% (21)	90% (35)	0.001	3.88(1.2, 12.5)	3.98(1.8,8.9)
Admit LOS	27 (16.7−38.6)	19 (12.3−30.1)	36 (17.4−58.6)	27 (19.5−44.4)	0.06
Discharge LOS	3.4 (2.7−4.2)	3.4 (2.5−4.1)	6.3 (4.0−8.7)	3.0 (2.4−4.0)	0.07
96 hour return	3% (6)	5% (3)	0	3% (3)	0.56
Readmission	2% (3)	5% (2)	0	1% (1)	0.76

Of the 183 patients, 103 (56%) were initially evaluated at a referring ED and transferred to the tertiary children’s hospital ED. All patients evaluated at the pediatric ED were either admitted to the hospital, 142 (78%), or discharged. None of the patients evaluated in the pediatric ED were transferred to another facility. The median length of stay in all admitted patients was 27 hours (18.4–45.6). The median length of stay in those discharged was 3.4 hours (2.8–4.2). Patients with primary spontaneous pneumomediastinum had a higher rate of discharge from the ED. Duration of admission was not significantly different between the three groups. No patients required intubation or underwent additional intervention beyond investigational studies.

Six patients returned to the ED within one week of discharge, two of these patients had been discharged directly from the ED. One patient was called back to the ED when pneumomediastinum was identified by the radiologist after initial discharge from the ED. A repeat CXR demonstrated resolving pneumomediastinum and the patient was then discharged from the ED a second time. The other patient returned due to persistent chest pain. Repeat CXR demonstrated stable pneumomediastinum and the patient was again discharged from the ED. Of the four patients who were initially admitted and returned to the ED within 96 hours after hospital discharge, three returned because of persistent chest pain. Two were found to have stable or no pneumomediastinum on repeat CXR and were discharged from the ED. One of these patients had stable findings on CXR and was readmitted for pain control and subsequently discharged. One patient with asthma and persistent cough and wheeze returned for increased subcutaneous emphysema. This patient was readmitted for asthma exacerbation and worsening pneumomediastinum. He did develop an apical pneumothorax that was managed conservatively and subsequent CXR demonstrated resolving pneumothorax and pneumomediastinum.

A total of 78 (42%) patients had follow-up evaluations in an outpatient setting or were seen in the ED within a six-month period. Of these, 23 (29%) had documented normal exams and normal follow-up CXRs. An additional 49 (55%) had normal physical exams in follow-up and no repeat imaging. Three (4%) patients had outpatient CXRs that were normal, and a visit or physical exam was not documented in the electronic health record. Two (3%) patients had normal follow-up physical exams and a CXR with improving pneumomediastinum. Two (3%) patients were evaluated in the ED for an asthma exacerbation with wheezing noted on exam and no repeat imaging. One (1%) patient followed up by telephone and the family reported symptom resolution.

DISCUSSION

To our knowledge, this is the largest study describing management and outcomes of children with spontaneous pneumomediastinum. In our institution, management is varied, with the majority of patients undergoing multiple imaging studies and hospital admission. The additional studies did not appear to change the clinical management in our patient population and did not lead to medical interventions.

Pediatric spontaneous pneumomediastinum is generally regarded as a self-limited and relatively benign process. Pneumomediastinum is suspected to occur after rupture of alveoli following exhaling against a closed glottis as may be experienced with coughing, vomiting, or strenuous activity. [7,1] Many of our patients endorsed inciting events likely associated with Valsalva maneuver. Similar to prior descriptions of pneumomediastinum presentation, our patients presented with complaints of chest pain and odynophagia.

In patients presenting with primary spontaneous pneumomediastinum, chest pain was seen more frequently than in the other two groups. Patients with secondary spontaneous respiratory-associated pneumomediastinum were notably younger and less likely to report a significant event or activity that occurred prior to symptom onset. Additionally, the only patients in this study who presented with an oxygen saturation <95% were classified as secondary respiratory-associated pneumomediastinum and had a concurrent diagnosis of respiratory infection or asthma exacerbation.

A majority of patients in our population underwent further evaluation, after initial diagnosis of pneumomediastinum. The studies included multiple repeat chest radiographs, radiographs of the neck and abdomen, CTs of the neck, chest and/or abdomen, esophagrams, direct laryngoscopy and bronchoscopy, echocardiography, and an upper extremity Doppler study. More than half of all patients had two or more chest radiographs, and surprisingly, more than one-third had CT imaging. Patients with secondary gastrointestinal-associated pneumomediastinum underwent the most diagnostic studies per patient compared to the other two groups. Interestingly, there was no difference in the frequency of CT evaluations among the three groups. In all reviewed cases, CT imaging in the setting of pneumomediastinum did not affect clinical care or outcomes. As exposure to radiation in the pediatric population is of particular concern, based on our results CT imaging is not routinely necessary in an otherwise well appearing pediatric patient with spontaneous pneumomediastinum.

A majority of patients in our study were admitted to the hospital. No specific therapeutic interventions and no complications were observed in any patient with primary spontaneous pneumomediastinum, admitted or discharged. Overall, only one of the six patients that returned to the ED after discharge had developed a worsening pneumomediastinum and small pneumothorax. This patient had secondary respiratory-associated pneumomediastinum and was admitted with an asthma exacerbation upon initial presentation to the ED. On the return ED visit, he reported persistent respiratory symptoms related to asthma. He did not require an intervention during the subsequent hospitalization.

Our results are similar to previous publications presenting the management of spontaneous pneumomediastinum, both primary and secondary respiratory-associated pneumomediastinum. In a retrospective study of 96 children with spontaneous pneumomediastinum [8] a majority (91 patients) were admitted to the hospital, 20% to the intensive care unit. None of these patients had progression of symptoms, none developed pneumothorax and no patients were intubated. Children with severe asthma were excluded and the authors describe one such patient has having a “dire” presentation, thus necessitating different management than that required by a well-appearing child with vague complaints of chest pain [8]. A study of 28 patients with asthma and pneumomediastinum demonstrated similar clinical courses of asthma exacerbations in children with pneumomediastinum compared to children experiencing asthma attacks without pneumomediastinum. The majority of children in this study had improving findings on subsequent CXR and of the few who had worsening pneumomediastinum on CXR, none progressed to pneumothorax. [7] In another review of spontaneous pediatric pneumomediastinum in 129 patients, [9] two patients developed worsening respiratory distress after ED presentation and were intubated. The authors note that in both cases the respiratory failure was secondary to respiratory infection and not worsening pneumomediastinum. Additionally, two patients in their study developed pneumothorax, which, similar to our patient who developed a small pneumothorax, did not require tube thoracostomy.

An important subset of patients includes those presenting with pneumomediastinum after emesis or choking, described in our series specifically as secondary spontaneous gastrointestinal-associated pneumomediastinum. In rare instances, the presence of air in the mediastinum may be caused by a perforation of the esophagus. Classically, this is known as Boerhaave syndrome and is associated with retching or vomiting. [5] Descriptions of spontaneous esophageal perforation causing pneumomediastinum in the pediatric literature are rare and limited to case reports. [5,6,1] In these patients, hypothetically, pneumomediastinum may be caused by esophageal rupture as opposed to alveolar rupture. It can be challenging to differentiate between spontaneous pneumomediastinum and microscopic esophageal perforation in the clinical setting. In a study of both adult and pediatric patients, Bakhos et al, compared patients with primary pneumomediastinum to those with confirmed esophageal rupture or Boerhaave syndrome. [6] Those with Boerhaave syndrome were found to be older with a mean age of 58 years old and were more likely to be tachycardic and have pleural effusion on chest x-ray. A 2006 review by Antonis, identified 26 pediatric patients with esophageal rupture in case reports and reviews. [5] The vast majority were neonates. Older children presented with pleural effusion on CXR or were described as “gravely ill.” Three case reports of adolescent patients describe the finding of mediastinal air discovered after vomiting that were treated conservatively. [10,11,12] Of note, an esophagram was not documented in two of the cases and was negative in the third. Without true confirmation of esophageal rupture, it is likely that these patients are more similar to the patients with spontaneous gastrointestinal-associated pneumomediastinum in our study than true Boerhaave syndrome. Most recently, a retrospective study by Richer and Sanchez identified 45 pediatric patients who underwent esophagram due to concern for esophageal rupture, 27 of whom were diagnosed with spontaneous pneumomediastinum [13]. In their study, esophageal studies were also negative. They concluded that esophagrams are not indicated for patients presenting with spontaneous pneumomediastinum. The findings in our study, albeit a small sample size, support this conclusion as well. None of our patients with secondary spontaneous gastrointestinal-associated pneumomediastinum had significant findings on subsequent esophagram or chest CT. Given that spontaneous pneumomediastinum occurs in otherwise healthy and well-appearing patients and has a relatively benign course, it may be considered a separate entity from Boerhaave syndrome. As in our patients, a diagnosis of secondary spontaneous gastrointestinal-associated pneumomediastinum may be more appropriate in these cases.

Based on our findings and those of prior studies, we propose a management approach for patients who are found to have pneumomediastinum on chest CXR and are otherwise well appearing and have normal vital signs. In patients with no history of trauma, respiratory infection, asthma exacerbation or vomiting and in whom the etiology of the pneumomediastinum is suspected to be truly primary spontaneous, a thorough history and exam should be completed. In patients with no respiratory distress, normal vital signs, normal oxygen saturation, and adequate pain control, we recommend observation in the ED for 2 to 4 hours. This is based on the average length of stay for discharged patients in our study, which was 3.7 hours. For patients who remain stable, discharge with primary care physician follow-up the next day could reasonably be considered. Given the relatively benign nature of primary spontaneous pneumomediastinum for many children, a shorter period of observation could be appropriate but further data is needed to support this. If the clinician elects to admit the patient for observation, observation alone should be sufficient, and in the absence of clinical deterioration, additional studies may not be indicated.

In patients with gastrointestinal-associated pneumomediastinum, additional imaging, including CT of the chest and neck, esophagrams and repeat roentograms are unlikely to yield additional information. Specifically, well appearing patients with secondary spontaneous gastrointestinal-associated pneumomediastinum and no evidence of pleural effusion or respiratory distress are unlikely to have fulminant Boerhaave syndrome. In these patients, symptomatic treatment and observation, either in the ED or briefly as an inpatient, is likely sufficient.

Similarly, in secondary spontaneous respiratory-associated pneumomediastinum, additional imaging does not typically yield additional information. In these patients, clinical management according to symptoms and clinical presentation should take precedence. We recommend the use of CXR as indicated by the underlying respiratory condition. CT chest should be obtained if indicated for evaluation of the underlying respiratory diagnoses but is not recommended for further examination of pneumomediastinum. In our study, as in others, one patient with secondary respiratory-associated pneumomediastinum did develop worsening respiratory symptoms related to their underlying respiratory diagnosis. In these patients, there is insufficient data to recommend for or against admission for observation purposes.

LIMITATIONS

There are limitations to this study. Data was collected retrospectively and subject to the inherent omissions of medical documentation. Identifying patients by ICD-9 code is efficient but there is potential for missing patients with additional or worse diagnoses and thus making the results difficult to interpret with certainty. We attempted to overcome this with a keyword search of the radiology database. There are limitations in this strategy as well. We may have missed patients for whom the term “pneumomediastinum” was spelled in an unconventional manner or a different term used to describe the phenomenon. Data for this study was collected from a single institution. Our hospital is the only pediatric hospital in a large catchment area, serving western Pennsylvania and parts of West Virginia, Ohio and New York. For this reason, pediatric patients who present for emergency care often present to our facility initially or are transferred. A significant proportion of the patients in this study were transferred to the pediatric hospital from a referring ED. Medical management performed at referring institutions was reported in this study but LOS only reflects time spent at the pediatric hospital. This study does not account for the time spent under medical observation at the initial ED or in transport. Finally, many of the admitted patients had secondary diagnoses, including asthma exacerbation, pneumonia and diabetic ketoacidosis, that likely influenced the care team’s decision to admit, the patient’s hospital length of stay and potentially, additional radiographic studies. Given the retrospective nature of this study, it is difficult to discern the specific clinical concerns that prompted the managing team to pursue additional imaging and it is challenging to delineate which diagnoses impacted length of stay the most. However, in all of the secondary diagnoses, repeat CXR, chest and neck CT are not commonly indicated.

CONCLUSION

Our data suggest that patients with spontaneous pneumomediastinum who are clinically well appearing can be managed conservatively with clinical observation, avoiding exposure to radiation and invasive procedures. In patients with primary spontaneous pneumomediastinum, symptomatic care and ED observation with close outpatient follow up is appropriate.