Objective. To describe the clinical characteristics of complicated parapneumonic effusions (CPE) in children caused by Streptococcus pneumoniae nonsusceptible to penicillin (PCN-N) and compare their clinical outcome with CPE caused by penicillin-susceptible (PCN-S) organisms.
Design. Children with parapneumonic effusions were identified retrospectively between July 1992 and June 1996. Charts of patients with CPE were reviewed for data obtained at the time of hospital admission. In addition, outpatient charts and/or the families of children with CPE caused by PCN-N S pneumoniae were reviewed to identify specific risk factors associated with PCN-N organisms.
Results. Sixty-four cases of CPE were identified during the 4-year period. In 26 cases a bacterial pathogen was recovered, withS pneumoniae accounting for 23 of these isolates (88%). Of the 23 S pneumoniae cases, 17 were PCN-S and 6 cases were nonsusceptible. Complicated parapneumonic effusions caused by PCN-NS pneumoniae occurred in significantly younger patients than CPE that were PCN-S (2.1 years vs 7.9 years). Nonsusceptible effusions also had a higher incidence of bacteremia than PCN-S effusions (100% vs 29%). There were no significant differences between the two groups for duration of chest tube drainage, febrile days, oxygen use, and hospital stay.
Conclusion. CPE caused by PCN-N S pneumoniaeis associated with a younger age and higher rate of bacteremia than CPE caused by PCN-S strains. However, there were no significant differences in outcome measures between patients infected with susceptible or nonsusceptible organisms.
- CPE =
- complicated parapneumonic effusions •
- PCN-N =
- penicillin-nonsusceptible •
- PCN-S =
- penicillin-susceptible •
- CT =
- computed tomography •
- MIC =
- minimum inhibitory concentration
The etiology of complicated parapneumonic effusions (CPE) in children seems to have come full circle in the United States during the past 60 years. In the preantibiotic era, Streptococcus pneumoniae was recovered in nearly two-thirds of all CPE.1 2 With the introduction of sulfonamides and penicillin came a significant decrease in the incidence of complications from pneumonia, including CPE.2 The number of cases of CPE caused by S pneumoniae declined, accompanied by an increase in the number of cases caused by Staphylococcus aureus, and by the late 1950s S aureus accounted for up to 92% of CPE in children.2-5 After the introduction of penicillinase-resistant penicillins in 1962 and the ensuing development of other antimicrobials effective against S aureus, the incidence of S aureus pneumonia and CPE decreased.6 7 Subsequent reports identifiedHaemophilus influenzae and S pneumoniae more frequently, but there was no predominant pathogen causing these effusions as in previous years.7-10 However, recent reports now describe S pneumoniae as again the principal bacterial pathogen in CPE, accounting for more than 70% of identified cases in our institution during the past decade.11 12
The first clinically significant case of S pneumoniaenonsusceptible to penicillin (PCN-N) was reported from Australia in 1967,13 and PCN-N strains are now recovered in many parts of the world. In the United States, the incidence of PCN-N has increased since the mid-1980s,14 and a recently published collaborative study from nine children's hospitals reported that from 1993 to 1994, 13.5% of invasive S pneumoniae isolates were PCN-N, while in the last 3 months of 1995, 27.2% were PCN-N.15 Despite the increased prevalence of PCN-N S pneumoniae, we are unaware of any reports describing the clinical characteristics of nonsusceptible strains in complicated parapneumonic effusions. In this study, we describe 6 children with PCN-N pneumococcal CPE and compare their clinical outcomes with CPE caused by penicillin-susceptible (PCN-S) strains.
MATERIALS AND METHODS
Cases involving a diagnosis of pleural effusions or empyema from July 1992 through June 1996 were retrospectively identified from the ICD-9 (International Classification of Diseases, 9th edition) codes in the Children's Hospital Medical Center medical records department. Patients were classified as CPE if they met one of the following criteria: 1) visualization of gross pus on thoracentesis or chest tube placement; 2) visualization of a pleural rind on thoracoscopy; 3) loculated effusions reported on chest radiograph or computed tomograph (CT); 4) bacterial pathogen(s) identified from pleural fluid; 5) pleural fluid with a lactate dehydrogenase >1000 IU/L and glucose <40 mg/dL.16
Pleural fluid aspirates and blood were collected and cultured on all patients by standard techniques.17 Anaerobic cultures were performed when requested. Antibiotic susceptibility forS pneumoniae was performed by the methods of the National Committee for Clinical Laboratory Standards, using cation-adjusted Mueller-Hinton broth plus 2% to 5% volume per volume lysed horse blood.18 19 Antimicrobial susceptibility was defined with use of the following minimum inhibitory concentration (MIC) breakpoints: penicillin ≤0.06 μg/mL; chloramphenicol ≤4 μg/mL; cefuroxime ≤0.5 μg/mL; erythromycin ≤0.5 μg/mL; cefotaxime ≤0.5 μg/mL; ceftriaxone ≤0.5 μg/mL.18 Pneumococcal isolates were defined as PCN-N if the MIC was >0.1 μg/mL.19Pneumococci were stored in sheep blood at −70°C and later recultured and then shipped to the Centers for Disease Control and Prevention for serotyping on the basis of capsular swelling with type-specific antisera (Quellung reaction).20
Patients identified with S pneumoniae were grouped into categories of either PCN-N or PCN-S. Signs and symptoms at the time of hospital presentation were recorded for both groups of patients and included temperature, room air oxygen saturation, complete peripheral white blood cell count, physical examination, chest radiographs, CT scans, and ultrasonographic evaluations. Additional factors recorded were patient demographics, admission into the intensive care unit within 24 hours of hospital presentation, and antibiotic treatment. Patients with PCN-N CPE were further studied with phone calls to parents and review of outpatient records to identify enrollment in day care or preschool and history of antibiotic use before hospitalization.
Outcome measures were compared between PCN-N and PCN-S patients and included duration of oxygen requirement (defined as the number of days until the patient was off oxygen therapy for 24 continuous hours), duration of fever (defined as the number of days until the patient was 24 hours without an axillary temperature >100.5° F), and length of hospital stay. Surgical and medical interventions were also recorded including placement and duration of chest tubes, administration of intrapleural urokinase, thoracoscopy, and lobectomy. Indications for chest tube placement included imminent respiratory failure, mediastinal shift, or large pleural effusions. Criteria for chest tube removal included the following findings: <50 mL of drainage during 24 hours, absence of bronchopleural fistula, and reexpansion of the lung as evidenced by chest radiography. Thoracoscopy or intrapleural urokinase was performed in children who had persistent fever and respiratory difficulties with prolonged chest tube drainage or a pleural rind prohibiting lung reexpansion or chest tube drainage. Lobectomy was indicated if the child had radiographic evidence of significant lobar necrosis or a persistent air leak from a bronchopulmonary fistula.
Data on the patient demographics and clinical and laboratory parameters were entered into the statistical program Epi-Info 5.1 and analyzed by univariate methods. Comparisons between PCN-N and PCN-S cases were calculated using Student's unpaired t test or χ2 analysis.21
Microbiology of Complicated Effusions
A total of 64 cases of CPE were identified and all were secondary to community-acquired pneumonia. Thirty-five cases were sterile with 3 cases regarded as contaminants (2 cases of Streptococcus viridans and 1 case with one colony each of coagulase-negative Staphylococcus and Oligella species). There were 26 cases in which a bacterial pathogen was recovered, with S pneumoniaeidentified in 23 cases (88%). The additional 3 isolates were group A streptococcus, S aureus, and 1 polymicrobial case. Of the 23S pneumoniae cases, 6 were PCN-N (26%).
Patient Characteristics of Pneumococcal Effusions
The average age for the PCN-S effusions was 7.9 years (range, 0.5 to 14 years) whereas children with PCN-N CPE were significantly younger at 2.1 years (range, 0.5 to 4.5 years; P < .01). There were no significant gender or racial differences between both groups of patients. None of the patients in either group had an underlying immune deficiency or any immunocompromising conditions. All patients in both groups had prodromal respiratory symptoms and fever before hospital admission, but there were no unique symptom complexes or duration of symptoms that differed between the groups (data not presented). No significant differences were found in temperature, oxygen saturation, or peripheral white blood cell count obtained at hospital admission. There were no differences in the percentage of patients admitted or transferred to the intensive care unit within 24 hours of hospital admission. On radiographic imaging multilobar pneumonia was identified in 53% and 50% of the PCN-S and PCN-N groups, respectively. The majority of effusions were unilateral, with 12% and 17% of patients in the PCN-S and PCN-N groups, respectively, noted to have bilateral effusions on chest radiograph (Table 1).
Characteristics of PCN-N Patients
The individual MICs and serotypes for each patient with PCN-NS pneumoniae are presented in Table2. Four of the 6 cases were identified between December 1994 and December 1995. Two patients had an isolate that was nonsusceptible to erythromycin, with one isolate nonsusceptible to both erythromycin and cefuroxime. Four patients were attending either day care or preschool at the time of their illness. Two patients reported antibiotic use within 3 months of their illness, with both patients receiving oral cephalosporins within 1 month of admission. All patients were initially treated with intravenous antibiotics, although none of the patients were treated with the same regimen. After identification of the PCN-N isolates, 2 patients were treated with single antibiotic regimens (ceftriaxone and cefuroxime) and 4 patients were treated with dual therapy including vancomycin and cefotaxime; cefotaxime and rifampin; vancomycin and clindamycin; and cefuroxime and chloramphenicol.
All patients had blood cultures drawn on hospital admission. All PCN-N patients were bacteremic whereas 5 patients had a positive blood culture in the PCN-S group (P < .05; Table3). All but 1 patient in each group had a chest tube placed. Patients in the PCN-N group had a higher frequency of surgical and medical intervention to lyse loculated effusions than the PCN-S patients although the difference did not reach statistical significance (P = .051). One patient in each group underwent a lobectomy. PCN-N patients had a longer duration of chest tube drainage and a longer hospitalization than PCN-S patients although these parameters did not reach statistical significance. There were no differences between the groups in the average duration of oxygen therapy or number of days before patients were afebrile. All patients survived. To assess if differences in outcome measures were influenced by the older age of the PCN-S patients, we compared the PCN-N group with age-matched PCN-S patients. Six children from PCN-S group (average age, 3.0 years; range, 0.5 to 4 years) were compared with the 6 PCN-N patients. There were no significant differences in patient characteristics or presenting features. One child from the PCN-S subgroup was bacteremic (P < .05). Five patients had a chest tube placed. Further surgical or medical intervention occurred in 1 child who had a lobectomy. There were no significant differences between the two groups for duration of chest tube drainage, oxygen therapy, fever, or length of admission.
S pneumoniae is the most frequent bacterial cause of pneumonia in infants and children. With the high incidence of PCN-N pneumococcal carriage in children under 2 years of age,22clinicians can expect to recover PCN-N more frequently in children with complicated parapneumonic effusions. Two important questions for clinicians treating children with CPE are: 1) What are the risk factors that increase a patient's likelihood of harboring a nonsusceptible organism? 2) Are nonsusceptible organisms more virulent than susceptible strains? In our study there were no unique prodromal symptom complexes or objective clinical signs upon hospital presentation that distinguished the nonsusceptible pneumococcal effusions from the susceptible effusions. However, PCN-N CPE were identified in a significantly younger patient population than effusions caused by PCN-S strains. The association of younger children with PCN-N pneumococci has been reported in otitis media and bacteremia, and is believed to be a significant risk factor, especially when associated with day care attendance and antibiotic use within the previous 3 months.23 24 The description of PCN-N pneumococcal outbreaks in child care facilities supports the hypothesis that some of the organisms contributing to the increased prevalence of PCN-N S pneumoniae in the United States originate from children attending day care.25 26 In these facilities children are in close contact with large numbers of other children, many of whom receive frequent antibiotics that can foster selective growth of nonsusceptible organisms already present. In our series only 2 children had taken antibiotics within 3 months of hospital admission. However, 4 of the 6 children were attending day care or preschool at the time of their illness, and it is possible that some of these children were colonized or infected with PCN-N pneumococcal strains obtained at these facilities.
For noncentral nervous system disease, previous pediatric case series have demonstrated that the outcomes of PCN-N pneumococcal infections were similar to PCN-S infections.27 28 In meningitis, however, the poorer penetration of some antibiotics into the cerebrospinal fluid makes it difficult to achieve predictably effective drug concentrations against PCN-N strains, and treatment failures and relapses have been reported with penicillin, chloramphenicol, and extended-spectrum cephalosporins.29 30 Antibiotic penetration to the pleura and into infected pleural fluid is generally good for most antibiotics.31 However, in our series all of the PCN-N patients had fibrotic loculations within the pleural space, thus creating the potential for poorer antibiotic penetration and reduced bactericidal concentrations against nonsusceptible organisms within these loculations. When analyzing objective outcome measures between the PCN-N and PCN-S patients, morbidity was high between both groups of patients but there were no significant differences in the duration of oxygen requirement, fever, chest tube drainage, or hospital stay. These findings are similar to results from South Africa, where there were no significant differences in the duration of fever, respiratory distress, or oxygen requirement in children with pneumonia caused by PCN-N or PCN-S organisms.28 In our study comparisons between the two groups were complicated by the higher rate of thoracoscopy and intrapleural urokinase in the PCN-N patients. It was not clear if the more aggressive treatment of the PCN-N patients was secondary to a higher severity of illness for these patients during their hospital course or, rather, reflected a clinical bias toward patients with a known nonsusceptible organism.
In this report all 6 patients with PCN-N CPE were bacteremic, which was significantly higher than the PCN-S effusions. This difference remained significant after correcting for the disparities in ages between the two groups of patients. Previous studies in childhood empyema have demonstrated a higher mortality rate in bacteremic patients.2 In contrast to our findings, adult patients with bacteremic pneumonias attributable to PCN-N S pneumoniaewere more likely to have an underlying disease, be in critical condition upon hospital presentation, and have a higher mortality rate when compared with bacteremic pneumonias attributable to PCN-S organisms.32 Children from South Africa with bacteremic pneumococcal pneumonia caused by PCN-N S pneumoniae also demonstrated a higher incidence of underlying disease with human immunodeficiency virus infection when compared with susceptible infections. In the South African study, there were children in both the susceptible and nonsusceptible groups who did not respond to antibiotics, and the severity of illness upon hospital presentation had a more direct association with outcome.28 Taken together, these reports suggest that the severity of illness at the start of therapy and the presence or absence of underlying disorders are more important in predicting the response to therapy than the susceptibility of the organism to penicillin.
In our study 88% of the cases of CPE with an identified bacterial pathogen were pneumococcal. The high incidence of pneumococcal CPE in our study and in other recent reports contrasts with many studies from the 1970s and 1980s that identified S aureus and H influenzae as the most frequent bacterial isolates in CPE.7-10 The precise reasons for the apparent decline in the proportion of S aureus and H influenzae CPE are unknown. The H influenzae vaccine may contribute largely to the decline of H influenzae CPE as H influenzae meningitis and bacteremia have also declined since the vaccine's introduction in 1990.12 33 The decrease inS aureus as a causative organism in CPE is more speculative, but may reflect earlier sterilization of staphylococcal pneumonias and effusions with the availability of broad-spectrum antibiotics.
In summary, we describe 6 cases of patients with CPE caused by PCN-NS pneumoniae. These cases occurred in a significantly younger population of children than effusions that were PCN-S. PCN-N CPE had a significantly higher incidence of bacteremia than PCN-S effusions, but there were no objective differences in outcome measures between both groups of patients. These outcome measures should be interpreted cautiously until prospective studies are published as this was a retrospective analysis of clinical features of patients with a variety of treatment regimens. Nevertheless, this reports emphasizes that clinicians treating patients with complicated parapneumonic effusions must maintain a high index of suspicion for S pneumoniae nonsusceptible to penicillin, especially in younger children.
The authors thank the Medical Records Department of Children's Hospital Medical Center for its assistance with chart reviews; Philip Khoury, MS, for statistical consultation; Barbara Chini, MD, and Christopher Harris, MD, for review of the manuscript; and the Centers for Disease Control and Prevention for serologic typing of the pneumococcal strains.
- Received July 14, 1997.
- Accepted September 23, 1997.
Reprint requests to (W.D.H.) Department of Pulmonary Medicine, Children's Hospital Medical Center, 3333 Burnett Ave, Cincinnati, OH 45229-3039.
- Ravitch MM,
- Fein R
- McLaughlin FJ,
- Goldman DA,
- Rosenbaum DM,
- Harris GBC,
- Schuster SR,
- Strieder DJ
- Brook I
- ↵Kaplan SL, US Pediatric Multicenter Pneumococcal Surveillance Group. Surveillance of pneumococcal infections in children. In: Program and Abstracts of the 36th Interscience Conference on Antimicrobial Agents and Chemotherapy (New Orleans, LA). Washington, DC: American Society for Microbiology; 1996. Abstract K21
- ↵Light RW. Pleural Diseases. Baltimore, MD: Williams & Wilkins; 1995:142
- ↵Facklam RR, Washington JA. Streptococcus and related catalase-negative gram-positive cocci. In: Balows A, Hausler WJ Jr, Herrmann KL, Isenberg HD, Shademy HJ, eds. Manual of Clinical Microbiology. 5th ed. Washington, DC: American Society for Microbiology; 1991:238–257
- ↵National Committee for Clinical Laboratory Standards. Methods for Dilution Antimicrobial Susceptibility Test for Bacteria That Grow Aerobically: Approved Standard. 3rd ed. NCCLS document M7-A3. Villanova, PA: National Committee for Clinical Laboratory Standards; 1993;13(25)
- ↵National Committee for Clinical Laboratory Standards. Minimum Inhibitory Concentration (MIC) Interpretive Standards (μg/mL) for Streptococcus pneumoniae: Approved Standard. NCCLS document M100-S5; M7-A3. Villanova, PA: National Committee for Clinical Laboratory Standards; 1994;14(16)
- Plouffe JF
- ↵Dean JA, Dean AG, Brendel KA. Epi-Info version 5: Word Processing, Database and Statistics System for Epidemiology on Microcomputers. Atlanta, GA: Centers for Disease Control; 1991
- Tan TQ,
- Mason EO,
- Kaplan SL
- Copyright © 1998 American Academy of Pediatrics