Empyema associated with community-acquired pneumonia: A Pediatric Investigator's Collaborative Network on Infections in Canada (PICNIC) study

Although morbidity and mortality associated with community-acquired pneumonia in childhood has declined markedly in the last 60 years [1], parapneumonic empyema complicating pneumonia continues to be an important problem in North America and Europe [2–13]. The diagnostic and therapeutic approach to empyema in children is not standardized and available guidelines are largely based on consensus opinion because of inadequate evidence supporting various therapeutic interventions[14]. We report a multi-center study of parapneumonic empyema in eight Canadian children's hospitals over a three-year period to describe the epidemiology of this serious childhood infection, including microbiologic etiology and clinical presentation and management across centers. Our intent was to determine if management variation occurred across centers, and to describe the burden of illness associated with various pathogens before the anticipated widespread introduction of universal infant pneumococcal vaccine programs.

Participating investigators from the Pediatric Investigators Collaborative Network on Infections in Canada (PICNIC) conducted the study at eight university-affiliated pediatric acute care facilities across Canada.

Possible cases were identified at each hospital from a search of health records databases to identify children under 18 years of age, admitted to hospital for at least 24 hours 1 January 2000 to 31 December 2003, and who had a diagnosis of empyema (International Classification of Diseases Clinical Modification-9 (ICD-9) code 510 or ICD-10 code J869 (Empyema).

The health record of each possible case was reviewed and included in the study if they had at least one of the following criteria: 1) thoracentesis with microbial growth from pleural fluid or 2) thoracentesis with no growth on culture of pleural fluid but elevated protein, or cell count (normal and abnormal reference values as determined by the clinical laboratory at each center) 3) ultrasound or other diagnostic imaging evidence of pleural fluid assessed by the radiologist as empyema or 4) diagnosis at time of thoracic surgery. Because our intent was to focus on empyema associated with community acquired respiratory tract infections, children with empyema associated with other illnesses such as chest trauma, thoracic surgery, or esophageal rupture were excluded.

A standardized case report form, with a data dictionary, was developed to collect data on demographic, clinical, diagnostic, treatment and outcome features of all cases. If available, ethnicity was described including whether the child was aboriginal (e.g. Indian, Inuit and Métis peoples of Canada). Tachypnea was defined by age: if < 12 months of age, > 75 breaths/minute; if > 12 months and < 10 years old, > 40 breaths/minute, if > 10 years old then > 30 breaths/minute. An elevated temperature was defined as temperature > 38.5°C. Death within 2 weeks of admission was noted. Data was entered at the coordinating site and analyzed using the software program SAS Version 8.2 (SAS Institute Inc., Cary, North Carolina).

We determined if bacterial pathogens were identified from sterile site samples including blood, pleural fluid and lung tissue at the time the child was diagnosed. For cases where Streptococcus pneumoniae was identified, the isolates were sent for serotyping at the National Centre for Streptococcus in Edmonton, Alberta as part of the Canadian Paediatric Society/Public Health Agency of Canada Immunization Monitoring Program Active (IMPACT) surveillance system[15]. Isolates were serotyped using the Quellung reaction, as previously described [15, 16].

The study was approved by the institutional review board at each site.

Parapneumonic empyema occurs most commonly in previously healthy children under five years of age and is associated with considerable morbidity and need for institutional care. The majority of children in this study at all centers required pain medication (78%), supplemental oxygen (77%) and had placement of a chest drain (75%). Over half were transferred from their home communities to a tertiary care facility for relatively prolonged admissions, although in most children the length of stay was 9 days or less. Empyema appears to exact a disproportionate burden in aboriginal children, who comprised 18% of these patients, despite representing less than 3.3% percent of the overall Canadian population (census 2001, Statistics Canada). Over 75% of aboriginal children were transferred from their home community for acute care. Although empyema is associated with considerable morbidity and lengthy hospital stays, only one death occurred within two weeks of diagnosis. We were not able to look at long term pulmonary function but previous studies have shown that restrictive lung function patterns are seen in up to 90% of children with parapneumonic empyema three months after discharge, but normal lung function is present at one year follow-up[17].

Marked variation in patient management was observed in this multi-center study. Consultation to pulmonology, infectious diseases, and surgical services appears to be routine in all centers, but diagnostic and therapeutic interventions vary. As has been noted in a recent review by the British Thoracic Society, wide practice variation in the management of empyema is likely due to the lack of good evidence available to inform best management, which results in treatment based on physician experience and local bias[14]. Of over 50 recommendations considered in that review, only four had high quality evidence [14]. The only randomized controlled clinical trials of therapy for pediatric empyema have examined intrapleural installation of fibrinolytics compared to video-assisted thorascopy or installation of normal saline [18–20]. A number of case series suggest that this therapy is associated with increased pleural drainage and avoidance of surgery [14] but the evidence from the three randomized trials is conflicting. We cannot make conclusions about the efficacy of different interventions in this non-randomized observational study, but it is nonetheless interesting to note that no difference in length of stay was observed in children with and without chest tube placement, and that time to chest tube placement did not appear to alter length of stay. Development of consensus guidelines or critical care pathways by inter-disciplinary multicenter group could improve standardization of care until better quality data to guide decision making is available.

The most common organism isolated from parapneumonic effusions during this three-year period was S. pneumoniae, followed by S. pyogenes, Staphylococcus aureus and S. milleri. In contrast with most studies, we found S. pyogenes as important a cause of empyema as S. aureus [21]. Based on available serotyping data from this study some empyema is potentially vaccine-preventable by heptavalent pneumococcal conjugate vaccines (PCV7). In the large randomized controlled efficacy trial of PCV7 vaccine a 17% efficacy in the prevention of a positive chest radiograph was observed in the intent-to-treat group as was a 70% decrease in consolidative pneumonia (consolidation on CXR 2.5 cm or greater)[10, 22–24]. Post-licensure surveillance indicates that invasive pneumococcal disease among children in the US has declined following universal vaccination [25]. PCV7 infant vaccine programs were not yet in place in most provinces in Canada at the beginning of the study period. In two of the six provinces infant immunization programs began 3 and 16 months, respectively, before the study ended. Thus in one province the vaccine could have prevented S. pneumoniae associated empyema in infants 6–18 months of age who would have been eligible for the study. At the time of writing almost all jurisdictions have implemented infant conjugate pneumococcal vaccine programs. The observation of an increasing incidence of empyema due to pneumococcus in some jurisdictions with universal vaccine programs, particularly due to strains not covered in the vaccine [2], is concerning, and highlights the importance of surveillance to detect changes in the epidemiology of disease following the implementation of vaccine programs. We plan to repeat this study following the more widespread introduction of pneumococcal vaccine in Canada.

Children with S. aureus-associated infection had longer length of stay than children with S. pneumoniae or S. pyogenes. Methicillin-resistant S. aureus (MRSA)infection was not common in Canadian children during the years of this study[26], and we did not identify this organism in our population.

Definitive microbiologic diagnosis was obtained in only 32% of children in this study and so most children were exposed to unnecessarily broad-spectrum antibiotics for many weeks. S. pneumoniae has been reported in over 70% of culture-negative empyema fluids[27] when molecular diagnosis is systematically used, and increased detection with 16SrDNA PCR [28]. This testing should become standard practice for care of children with culture-negative empyema in order follow the effect of universal vaccine programs on the serotype epidemiology of invasive disease. If detection of antibiotic resistant genes is included in the molecular testing, use of broad spectrum antimicrobials may be avoided. In the United Kingdom an enhanced surveillance program for pediatric empyema is planned for England and Wales which will include molecular diagnosis of culture-negative cases [5]. Given the evidence that S. pneumoniae causes most empyema in children in this study, and that a larger percentage may be found when rigorous diagnosis is pursued[27, 28], we recommend more systematic use of molecular techniques to determine microbiologic etiology of empyema. These techniques should also be considered in the implementation of surveillance programs for invasive bacterial disease so that vaccine and non-vaccine serotypes of S. pneumoniae can be documented following introduction of childhood vaccine.

One limitation of this study is that it was not population-based and thus the incidence of parapneumonic empyema for the whole country cannot be determined from this data. However, each hospital is the only referral center for pediatric care in each region and therefore we likely captured all cases in the participating sites. Secondly, we did not define a specific pleural fluid white blood cell count required for the diagnosis of empyema, rather allowing each center to define normal/abnormal fluid obtained by thoracentesis. The condition of empyema has been used variably²¹, but usually denotes a complicated effusion associated with an adjacent pneumonic process, which is distinguished from a pleural effusion, and from an uncomplicated parapneumonic effusion associated with pneumonia. We expect that the search strategy for this review, extraction of all records with ICD-9 CODE 510 or the ICD-10 CODE J869 (empyema), would have minimized this misclassification error. As well, each patient was seen by infectious disease, surgical and respiratory specialist physicians who agreed with the diagnosis of empyema. However, the lack of data on the white blood cell count in the pleural fluid could mean that uncomplicated parapneumonic effusions could have been included in our sample. This misclassification would have increased the number of cases.

Empyema associated with community-acquired pneumonia is an important cause of morbidity in Canadian children. Previously healthy children under five years of age are most commonly affected, and usually present in the winter months with tachypnea, fever, cough, and abnormal air entry on auscultation associated with dullness to percussion. The median length of hospital stay is 9 days; S. aureus infection is associated with longer hospitalization than S. pneumoniae or S. pyogenes. The most common infecting organism is S. pneumoniae. Considerable variation in use of chest tubes, thoracentesis and thorascopic lysis is observed across settings.