Information on patients younger than 15 years old admitted between 2005 and 2013 inclusively, with a discharge diagnosis of pleural effusion or empyema, was retrieved from our hospital electronic database. The clinical records of these patients were retrospectively reviewed to select those with pleural effusion associated with a diagnosis of community-acquired pneumonia. Patients with an effusion associated with tuberculosis or other diseases were excluded. Parapneumonic effusions were considered to be CPI if the pleural space was bigger than 1cm as measured in plain chest roentgenogram and loculations or debris were observed by pleural ultrasound. Only patients with CPI were selected for analysis.

Patients were admitted from the emergency department of our hospital or transferred from other hospitals. Two groups of patients are compared: those admitted between 2005 and 2009 (group 1) were managed according to the British Thoracic Society guidelines for the management of pleural infection in children,5 and those admitted between 2010 and 2013 (group 2) were treated according to the clinical assessment of the attending pediatricians. The clinical findings driving a decision to drain the empyema in patients of group 2 included increased effort of breathing and severe or persistent septic condition.9,10 Persistent fever was not considered a reason to drain the effusion, because it has been reported to be long-standing irrespective of whether they are drained or not.9 Chest tube insertion was usually performed by an interventional radiologist assisted by an anesthetist. The child was maintained under deep sedation and a small-bore pigtail catheter was inserted, with ultrasound guidance, using the Seldinger technique. Less frequently, a chest tube was inserted by a pediatric surgeon, more often for children admitted to the intensive care unit (ICU). Thoracentesis and computed tomography scan were usually not performed. All other decisions (chest tube removal, length of antibiotic treatment, discharge from hospital, etc.) were taken according to the best clinical judgment of the attending physicians.

The clinical files of the patients selected were thoroughly reviewed and data were extracted for each patient, including sex, age, length of the hospital stay (LHS), comorbidities, signs and symptoms, duration of fever, blood and pleural fluid laboratory and bacteriological results, antibiotics administered, length of antibiotic treatment, oxygen therapy, admission to ICU, drain tube insertion, surgical interventions and complications. Pneumococcal vaccination status could not be analyzed due to inconsistent registration.

The primary outcome measures for comparison between groups 1 and 2 were (1) the proportion of children with a chest tube inserted in our hospital and (2) the LHS. Secondary outcome measures for analysis were: treatments employed, length of antibiotic treatment and duration of fever. Length of primary fever was computed as the sum of number of days of fever before admission and the time to defervescence, considered as the time (in days) elapsed between admission and the first day with a temperature under 38°C for 24h. Fever reappeared in some patients after defervescence. In such cases, the total number of days with fever during admission and the total number of days from admission to complete disappearance of fever were calculated.

Data were transferred to SPSS (v17.0 SPSS, Chicago, IL, USA). Descriptive analysis included frequencies and percentages for categorical variables, means, standard deviation (SD) and range for numerical variables with a normal distribution, and medians, interquartile range (IQR, 25th and 75th percentiles) and range for numerical variables with a non-normal distribution. Categorical variables were compared using the Fisher exact test or the chi-squared test for trend. Continuous variables were compared by means of t-test (with normal distribution) or Mann–Whitney U test (with non-normal distribution). Reported P values were two-tailed, and the level of significance was set at 0.05.

The study was approved by the Clinical Research Ethics Committee of our hospital.

Two hundred children were discharged from our hospital between 2005 and 2013 with a diagnosis of pleural effusion. Twenty-four of them were not of infective origin and five were diagnosed with tuberculosis. The remaining 171 children had a parapneumonic pleural effusion and were selected for detailed review of their clinical records. Sixty-two had small simple effusions and were not considered to need a drainage procedure. The other 109 had large, complicated effusions, deemed to need drainage according to the main guidelines. These 109 patients were considered to have CPI and were the subject of our study. Sixty-four children were admitted in 2005–2009 (12.8/year) and 45 in 2010–2013 (11.3/year).

The main characteristics of patients on admission are depicted in Table 1. A history of asthma or recurrent wheezing was present in 10% of patients, a frequency very close to that of the general pediatric population in our country.11 One patient had cerebral palsy, two had epilepsy and another two had Down's syndrome. Two children had lesions of current chickenpox. Most of the patients were previously healthy or had other problems unrelated to the development of pneumonia or a pleural effusion.

All patients had fever at some stage of the disease. Before admission, two patients had low-grade fever (less than 38°C) for only one day. The other 107 had fever (38°C or over) ranging from 1 to 10 days. The frequency of cough, respiratory distress and chest pain is shown in Table 1. Other common signs and symptoms were (n): vomiting (28), abdominal pain (23), diarrhea (13), headache (6) and febrile seizure (2).

In patients drained, pleural fluid median values (IQR) were: protein 4g/L (4–5), glucose 10mg/dL (2–31), leukocyte count 3.4×109/L (0.6–22.8). Bacterial cultures were positive in 30 patients (28%): Streptococcus pneumoniae was recovered from 17 patients (16%: 9 blood, 8 pleural fluid), Staphylococcus epidermidis from 4 (3 blood, 1 pleural fluid), Streptococcus pyogenes from 3 (1 blood and 3 pleural fluid), Staphylococcus aureus from 2 (1 blood, 1 pleural fluid) and other bacteria from 4 patients.

The treatments are compared in Table 2. Patients usually received more than one antibiotic, including oral antibiotics before and/or after hospitalization. Antibiotics were administered for a period of 7–34 days, but the mean was 2 days longer in the second group; the intravenous route was used for 0–26 days. Fifty-four patients needed oxygen for a median of 4 days (IQR 3–7 days, range 1–15 days). Thirteen patients were admitted to the pediatric ICU for a mean of 8.3 days (SD 6.9 days, range 1–23 days), but only two required assisted ventilation.

A chest tube was placed in our hospital in 74 patients (68%): 83% of patients in group 1 and in 47% in group 2, and this was the most noticeable difference in their treatment (Table 2). Urokinase was administered in 85% of those drained (91% in group 1 and 71% in group 2, P=0.07). There was no significant trend in the percentage of patients drained between 2005 and 2009 (P=0.91), but a progressive reduction of chest tube placement was found between 2009 and 2013 after the change in treatment policy for empyema (P=0.006) (Fig. 1). Thoracoscopy was performed in just one patient, due to treatment failure after 9 days with a chest tube. Two patients admitted to our center in 2009 had been drained before being transferred from their hospital. Exclusion of these two patients made no difference in the comparisons of treatments and outcomes.

Figure 1.

Percentage of children with complicated pleural infection managed with a chest tube in our hospital per year.

(0.06MB).

The main outcomes are depicted in Table 3. A 1-year-old girl transferred to our center in 2005 had a chest tube inserted, but was transferred the next day to another specialized center due to a pericardial effusion, where she was treated and recovered. This patient was excluded from the analysis of duration of antibiotics and outcomes. LHS (mean 11.8 days, SD 6.1, range 3–32 days) did not differ significantly between the two groups, even when the two patients drained in another hospital were excluded. Primary fever lasted 11.6 days (SD 5.0 days, range 1–25 days) and the time to defervescence after admission ranged from 0 to 21 days. Fever reappeared after defervescence during hospital stay in 22 patients after an apyretic period of 1–8 days, and the new fever lasted 1–12 days. There were no differences between groups in the total number of days with fever during admission (mean 7.7 days) and the total number of days from admission to complete disappearance of fever (mean 8.4 days). Only one patient presented pneumothorax before chest tube insertion. The mean LHS was longer in children with pneumothorax (18.0 vs. 10.6 days, P<0.001), even when only drained patients were analyzed (18.0 vs. 12.1 days, P<0.001). Although long-term outcomes were not systematically registered, all patients recovered and no permanent damage was observed on follow-up. Fig. 2 shows the initial and last chest X-ray of two representative patients with very large undrained CPI.

Figure 2.

Initial and last chest X-ray of two representative patients with a very large complicated pleural effusion not drained. The time elapsed between the two images of each patient was (as usual in most of our cases) about 3 months.

(0.28MB).

Patients who were drained were similar to those who were not drained in their main characteristics on admission, but a right effusion was more commonly drained than a left one (80% vs. 59%, P=0.02). There were no differences in the antibiotics employed and the total length of antibiotic treatment, but drained patients had a longer duration of intravenous antibiotics (12.5 vs. 9.8 days, P=0.004), and more frequently required oxygen (59% vs. 29%, P=0.004) and ICU admission (18% vs. 0%, P=0.009). LHS was longer in drained patients (13.5 vs. 8.1 days, P<0.001), even if patients with pneumothorax were excluded (11.1 vs. 8.1 days, P<0.001).