Necrotising pneumonia (NP) is a severe complication of community-acquired pneumonia characterised by liquefaction and cavitation of lung tissue. The present study describes the epidemiology, aetiology, management and outcomes of children hospitalised with NP over a 15-yr period.
A retrospective observational study of NP cases was conducted from January 1990 to February 2005 analysing clinical presentation, laboratory data, hospital course and long-term follow-up.
A total of 80 NP cases were identified, with the number of detected cases increasing from 12, in the period 1993–1996, to 40 in the period 2001–2004. In total, 69 (86%) cases had pleural effusion with a low pH (mean 7.08) and 38 (48%) patients had positive cultures, with Streptococcus pneumoniae as the predominant organism. Recently, other organisms, most notably methicillin-resistant Staphylococcus aureus, emerged. Patients had prolonged hospitalisations (median 12 days). A total of 69 patients required pleural interventions and those receiving chest drainage alone had similar outcomes to those managed surgically. All patients had full clinical resolution within 2 months of presentation.
Necrotising pneumonia has increasingly been identified as a complication of paediatric pneumonia. Streptococcus pneumoniae remains the predominant organism, but since 2002, different bacteria have been isolated and the age range of cases has broadened. Despite the serious morbidity, massive parenchymal damage and prolonged hospitalisations, long-term outcome following necrotising pneumonia is excellent.
Complications of community-acquired pneumonia in children include pleural effusion, empyema, lung abscess, pneumatocele and necrotising pneumonia (NP). NP, also termed cavitary pneumonia or cavitatory necrosis, has been associated with poor clinical outcomes in adults and when initially described, this complication was thought to be extremely rare in children. The first case series of NP including four children was published in 1994 1 and subsequently there have been several reported small case series of paediatric NP 2–6 including a recent report on epidemiological observations 7.
Even though the diagnosis of NP can be suspected by plain chest radiography, the frequent presence of a dense lobar consolidation and pleural effusion may obscure proper definition and chest computed tomography (CT) scanning is needed for a more definitive diagnosis. Radiographic criteria for NP include the loss of normal pulmonary parenchymal architecture and the presence of areas of decreased parenchymal enhancement, representing liquefaction, that are progressively replaced by multiple small air or fluid filled cavities 1, 8. The pathophysiology of NP is thought to be one of massive pulmonary gangrene, tissue liquefaction and necrosis 9 but the precise pathways leading to this massive damage have not been well established. The radiological diagnosis of NP has been correlated with such pathological findings 10. Prior reports have focused on NP caused by Streptococcus pneumoniae 11, although other bacterial organisms, including Staphylococcus aureus and Mycoplasma pneumoniae, have reportedly led to NP 12, 13. The increasing number of case reports of NP has coincided with several studies of childhood pneumonia, which report trends of increasing incidence of complicated pneumonia with parapneumonic effusions 14, 15. While mention is made of NP in these studies, the related role of NP to this changing trend has not been analysed.
The purpose of the present, retrospective, observational study was to review the cases of NP among children hospitalised at the Children's Hospital, Boston (Harvard Medical School, Boston, MA, USA) over a 15-yr period and describe: 1) the epidemiology, causative organisms and key clinical and laboratory characteristics of children hospitalised with NP; 2) the management strategies and presence of complications in children with NP; and 3) report on the long-term clinical outcomes for children with NP.
Cases of NP among patients hospitalised at Children's Hospital Boston from January, 1990 to February, 2005 were identified retrospectively using an electronic database of the Dept of Radiology. All cases were identified by searching for the term “necrotising” in reports of thoracic CT scans of hospitalised patients and reports with terms such as “no evidence of necrotising pneumonia” were excluded. CTs of all cases were reviewed and included for analysis when the CT revealed segments of lung showing multiple areas of lung parenchyma containing air, air and fluid, or nonenhancing fluid surrounded by contrast enhancing lung parenchyma without a defined rim of enhancement 8. Cases of solitary cavitation surrounded by a well-defined enhancing rim of lung were excluded, since this could represent a lung abscess.
A retrospective hospital chart review was conducted using a standardised data collection form. Patients with nosocomial pneumonia and those with pre-existing lung or cardiac disease were excluded from subsequent analysis. For each patient, demographics, laboratory results, microbiological culture data and clinical information originally obtained at the time of admission, during the admission and at any subsequent outpatient follow-up visits at the institution, were recorded. For the present study, a febrile day was defined as any 24-h period during which the patient had a recorded temperature ≥38°C. Hypoxia was defined as any recorded oxygen saturation of <90% by pulse oximetry, measured on room air. The study was approved by the Children's Hospital Boston Institutional Review Board.
Demographic, clinical and laboratory variables were summarised by standard descriptive statistics. Comparisons between groups based on clinical presentation, interventions and complications were performed using paired t-tests where means are reported, Wilcoxon tests where medians are reported and Chi-squared tests where proportions are reported. Two-sided p-values <0.05 were considered to be statistically significant.
A total of 80 cases of community-acquired NP were identified during the 15-yr study period. There were no cases identified in the period 1990–1993 and the number of detected cases increased over time from 3 cases·yr−1 in 1993–1996 to 14 cases·yr−1 in 2003–2004 (fig. 1⇓).
Clinical presentation and radiographical findings
The median age at presentation was 3.6 yrs (interquartile range (IQR) 0.25–19 yrs; table 1⇓). The majority of patients had no significant prior medical history; only 14 (18%) patients had a reported history of asthma or wheezing and nine (11%) had a reported history of recurrent otitis media. With the exception of two cases, these children had no known underlying immunological diseases. One patient had a known underlying immune deficiency (Schwachman–Diamond Syndrome) and another patient was diagnosed with chronic granulomatous disease subsequent to presentation with NP. Figure 2⇓ shows representative CT images from one patient, a 4-yr-old previously healthy female, and demonstrates the radiographic criteria used in identifying NP, as well as the time-course of the disease.
Almost all patients had fever (96%) and cough (84%), with a mean onset of symptoms 9 days prior to the hospitalisation. Other constitutional symptoms such as vomiting, abdominal pain and chest pain were only reported by a minority of patients. A total of 54 (74%) patients received at least one dose of oral antibiotics prior to their hospitalisation. For these patients, the mean duration (range) of pre-admission antibiotics was 3 (0–24) days.
Laboratory and microbiological characteristics
The predominant laboratory features (table 2⇓) were leukocytosis (mean white blood cell count 18.4×103 cells·μL−1), bandaemia (mean 9%), anaemia (mean serum haemoglobin 10.4 mg·dL−1) and hypoalbuminaemia (mean serum albumin 2.0 mg·dL−1). An appreciable pleural effusion on radiograph or CT scan was seen in 69 (86%) patients. Pleural fluid analysis revealed a low pH (mean 7.08), a low glucose (median 10.0 mg·dL−1) and a high cell count (median 9,600) with neutrophil predominance. Positive microbiological identification was obtained in 38 (48%) cases (fig. 1⇑). A total of 19 cases had positive pleural fluid cultures and five had positive pleural fluid latex agglutination studies for Pneumococcal antigen. Other aetiological diagnoses were based on positive blood or sputum cultures. Receiving any antibiotics pre-admission was not associated with a reduction of positive microbiological identification (48 versus 56%; p = 0.5). S. pneumoniae was identified in 18 (22%) cases (13 by positive culture and five via positive pleural latex agglutination studies). The diagnoses based on latex agglutination all had negative pleural fluid cultures. The absolute number of Pneumococcus spp. cases per year was similar throughout the study period. Review of the antibiotic susceptibility patterns of the positive Pneumococcal cultures did not reveal any penicillin-resistant Pneumococcus spp.
Since 2000, the incidence of other causative organisms, including methicillin-sensitive S. aureus, methicillin-resistant S. aureus (MRSA), Fusobacterium, Pseudomonas aeruginosa and Streptococcal sp. such as S. milleri, has increased. All cases of MRSA were detected after 2003.
Outcomes during hospitalisation
The median length (IQR 9–17) of hospitalisation was 12 days, ranging from 3–84 days (table 3⇓). During the hospitalisation, the median duration of fever was 6 days (IQR 3–9) ranging from 1–28 days. There were no statistically significant differences in the median number of febrile days between patients based on antibiotic pre-treatment (5 versus 7 days; p = 0.18) or those presenting with pleural effusion (6 versus 4.5 days; p = 0.24). There were no deaths in the cohort. Different types of antibiotic regimens were used as initial empiric treatment: including penicillins, cephalosporins, vancomycin and clindamycin, and in cases with positive cultures, antibiotic regimens were tailored to specific organism susceptibilities. All patients were treated with prolonged courses of antibiotics (median (range) 27 (3–95) days).
Management of patients presenting with pleural effusion
Of the 69 patients presenting with pleural effusions, 47 (68%) had placement of a chest tube or pigtail catheter for pleural drainage, 16 (23%) underwent a combined surgical intervention with chest tube placement and only six (9%) patients underwent thoracentesis alone (table 3⇑). In those receiving surgical intervention, 12 (17%) underwent video-assisted thoracoscopic surgery, three (4%) had an open thoracotomy and pleural decortication and one (1%) patient had a wedge resection of the affected parenchymal lobe. The median duration of pleural drainage was 6 days (IQR 5–10, range 1–52 days) and did not differ based on the type of intervention (table 4⇓). The number of febrile days also did not differ based on the type of intervention (p = 0.65); however, the median length of stay was significantly longer in patients requiring chest drainage only or surgery (14 and 14.5 days, respectively) compared with patients having only a thoracentesis (11 days) or those without pleural effusion (6 days; p = 0.0002).
Of the patients with pleural drainage, 10 (15%) developed a bronchopleural fistula (BPF) defined by persistent air leaks noted in the medical record lasting >24 h. Patients who developed BPF were older (median age 6.0 versus 3.5 yrs; p = 0.06) than patients without BPF. The presence of a BPF was associated with a longer time of pleural drainage (median 14 versus 6 days; p = 0.0007) and a longer length of stay (median 19 versus 13 days; p = 0.01). All of the BPF cases were managed conservatively without operative repair.
A total of eight (10%) patients required readmission (six due to persistent fever) within 2 weeks following their initial hospital discharge. Three patients developed a small pneumothorax, none required intervention. Of the total cohort, 64 patients were seen post-discharge in the paediatric pulmonary clinic at the Children’s Hospital (median follow-up time 6 months), all with complete clinical resolution of symptoms reported within 2 months of discharge. Of these, 12 patients had pulmonary function testing (PFT) performed; eight (67%) had normal PFTs, three patients had a mild obstructive defect and one had a mild restrictive defect. Follow-up imaging studies included chest radiographs and, in a few cases, chest CT scans (fig. 3⇓), all of which showed marked improvement within 6 months with near normalisation of pulmonary parenchymal structures.
In the present study, 80 paediatric cases of NP presenting over 15 yrs at a single institution have been identified. With the exception of two cases, these children had no significant underlying comorbid conditions, such as immunological disease or a history of prior infections. Their course was associated with significant radiographical evidence of lung damage and prolonged hospitalisations. The detection of NP in children increased over time. S. pneumoniae was the most commonly identified organism but other causative bacteria were cultured. Association with empyema was common. Despite the significant short-term morbidity, both clinical and radiographical resolution has been observed over time.
The presence of NP as a unique entity within the spectrum of complicated pneumonia has been previously analysed in a few small series of patients 1–4, with the largest series at one institution including 17 patients 2. Using a retrospective case definition for NP based on CT diagnoses, 80 cases were identified. Even accounting for the possibility of under-reported cases due to reliance on a CT definition for diagnosis, the present series represents the largest group of paediatric NP cases systematically described to date.
The age range and demographic features of the present patients are similar to other published studies of children with complicated pneumonia 14–16. Most children had developed symptoms of fever and cough for ≤1 week prior to their hospitalisation, indicating that the parenchymal damage in NP occurs rapidly. Transition from liquefaction to cavitation within 48 h has been observed via CT scan. Although the retrospective nature of the present analysis results in missing laboratory data for some of the cases, the abnormal laboratory findings of an elevated white blood cell count and low haemoglobin that were observed in many of the patients, have been previously reported in children with complicated pneumonia and NP 2, 16. A markedly decreased serum albumin level was found in many patients, potentially secondary to loss of protein into the affected lung parenchyma and pleural fluid or possibly due to protein losing enteropathy 17. The pleural fluid characteristics associated with NP in the present study, particularly the low pH, have also been associated in other studies with increased complications 18–20. The majority of the present patients had a persistent fever and, given the frequently sterile cultures and predominantly sensitive organisms, the present authors assume that the reason for the fever is not related to the lack of effectiveness of eradication of bacteria but rather to the presence of pyrogenic products of inflammation and tissue destruction 20.
Over the 15 yrs covered by the present study, an increase in the absolute numbers of NP diagnosed in the Children’s Hospital was observed. The rise is likely due to a combination of increased recognition of NP as a specific entity and heightened detection resulting from utilisation of CT scans in the evaluation of children with complicated pneumonia. There was a consistent increase in CT scans performed at the present authors’ institution starting in the early 1990s and preceding the detection of the first cases of NP. Since CT scans are the standard mode for diagnosis of NP, the increased use of CT would increase detection of NP cases. During the same time period, there was not a similar increase in the number of admissions for pneumonia to the institution. It lends itself to reason, therefore, that the observed increase of NP detection parallels the increases observed in complicated parapneumonic effusions in recent years 7, 14, 15.
The increase in NP cases over time may also be due to a changing spectrum of causative organisms, as has been postulated as the underlying cause of the rise in complicated pneumonia in general. Pneumococcus spp. was the predominant organism detected in the present study, similar to most previous reports of NP 1–4. Due to the retrospective nature of the present study, it was impossible to identify the specific Pneumococcal spp. serotypes in the patients. Thus, the present authors were unable to replicate the analysis performed by Tan et al. 14 and comment on secular changes of particular strains and their relationship to the introduction of the pneumococcal vaccine, as a possible contributor to NP. However, an increase in the variety of pathogens in these patients was observed. A number of cases were caused by other Streptococcal spp. and Staphylococcal spp., which have been previously reported in individual cases to cause NP 21, 22. Several cases were also found to be caused by Fusobacterium spp. and P. aeruginosa, two bacterial species that have not been previously associated with childhood NP. Finally, three cases were caused by MRSA in 2003 and 2004. MRSA has been reported in one other case report of childhood NP 23 and as an emerging organism in childhood empyema 15. The overall impact of the expanding range of causative organisms on both short- and long-term outcomes in NP warrants further study and 42 of the present cases had no identified microbiological cause. None of the patients with positive culture results had received >3 days of antibiotic pre-treatment. A likely explanation is that the infection triggered a severe inflammation that resulted in the development of NP but the antibiotics given prior to the present intervention were sufficient to sterilise the pleural fluid. However, an important alternative possibility is that the culture negative cases were caused by organisms such as anaerobes, viruses, Mycoplasma pneumoniae or Chlamydia pneumoniae, which have been reported to cause NP in children, though were not part of the routine microbiological investigations in this series 12, 13, 24. Such testing, using for instance PCR for virus or Mycoplasma spp. detection, could potentially increase the diagnostic yield in NP.
It is not possible to retrospectively perform rigorous evaluation of management strategies in NP, particularly because the frequent coexistence of empyema prevents uncoupling of the two conditions and separating the analyses of their respective contribution to overall morbidity. A difference in the length of pleural fluid drainage or length of fever was not observed, however, when comparing patients who had surgical intervention with those that were conservatively managed with nonsurgical chest drainage. The present authors propose that NP should be recognised as a distinct element of complicated pneumonia and, in particular, its presence should be considered as a separate complicating feature from the often accompanying pleural effusion. Diagnosis of NP requires CT imaging, and thus, such characterisation in patients with complicated pneumonia and continued symptoms despite appropriate medical therapy is recommended. The lack of recognition of NP can lead to erroneous diagnoses with serious errors in management. Based on the universally favourable outcomes with conservative management, the present authors do not agree that surgical resection for the treatment of NP is necessary, as has been implied in recent publications 25–27, and are concerned about the use of terminology, such as ‘patients required’ lung resection 14 and ‘destroyed lung’ 26. The role of such interventions needs to be evaluated in prospective management studies of NP.
Children with NP have increased risk for developing a BPF. In a prior study of nine NP cases, five developed a BPF that was thought likely to be due to the friability of the inflamed pleura abutting the necrotised lung 28. All patients who developed BPF in the present series had a chest drain in place for >7 days, raising the possibility that length of chest tube drainage is a risk factor for development of BPF. The present observations are, however, too few for a meaningful analysis.
The long-term outcomes for patients seen in follow-up have been good. All patients had resolution of clinical symptoms within 2 months of their hospitalisation. Although many children were too young for PFT, the children who had follow-up spirometry had essentially normal results. Performing PFT in these children could aid in further characterising any residual effects of NP over time. Follow-up chest radiographs and, in a few cases, CT scans have shown almost complete normalisation of pulmonary parenchyma within several months of hospitalisation (fig. 3⇑). Overall, this pattern of improvement suggests that the lung damage caused by NP in children is transient. The radiographic resolution observed in the present cases confirms earlier similar reports in smaller cases series of NP 29. Thus, when NP is diagnosed it should be recognised as a severe, yet, self-limiting and reversible disease.
In conclusion, necrotising pneumonia should be recognised as an increasingly detected complication of paediatric community-acquired pneumonia that is distinct from pleural effusion and empyema. Even though Pneumococcus spp. remains the most common causative bacterial pathogen, necrotising pneumonia in children can be caused by a variety of organisms including methicillin-resistant Staphylococcus aureus. Conservative management of necrotising pneumonia with antibiotics and chest drainage for pleural effusions results in good outcomes and there is no indication that surgical resection is necessary for the proper treatment of necrotising pneumonia. Despite the short-term severe morbidity of childhood necrotising pneumonia, long-term clinical outcomes are excellent with minimal resultant sequelae.
Some of the results of this study have been previously reported in the form of an abstract at the 7th International Conference on Paediatric Pulmonology (CIPP VII; Montreal, Canada) July 2006.
Statement of interest
The authors would like to thank R. Malley and K. McIntosh (both Division of Infectious Diseases, Childrens Hospital Boston, Harvard Medical School, Boston, MA, USA) for critical review of the manuscript.
- Received August 3, 2007.
- Accepted January 10, 2008.
- © ERS Journals Ltd