THE JOURNAL OF PEDIATRICS • www.jpeds.

com ORIGINAL
ARTICLES
Dexamethasone for Parapneumonic Pleural Effusion: A Randomized,
Double-Blind, Clinical Trial
Alfredo Tagarro, PhD, MD1,2, Enrique Otheo, MD3,4, Fernando Baquero-Artigao, MD5, María-Luisa Navarro, PhD, MD6,
Rosa Velasco, MD7, Marta Ruiz, MD8, María Penín, MD9, David Moreno, PhD, MD10, Pablo Rojo, PhD, MD11, and
Rosario Madero, PhD,12 on behalf of the CORTEEC Study Group*

Objective To assess whether dexamethasone (DXM) decreases the time to recovery in patients with parapneumonic
pleural effusion.
Study design This was a multicenter, randomized, double blind, parallel-group, placebo-controlled clinical trial
of 60 children, ranging in age from 1 month to 14 years, with community-acquired pneumonia (CAP) and pleural
effusion. Patients received either intravenous DXM (0.25 mg/kg/dose) or placebo every 6 hours over a period of
48 hours, along with antibiotics. The primary endpoint was the time to recovery in hours, defined objectively. We
also evaluated complications and adverse events.
Results Among the 60 randomized patients (mean age, 4.7 years; 58% female), 57 (95%) completed the study.
Compared with placebo recipients, the patients receiving DXM had a shorter time to recovery, after adjustment by
severity group and stratification by center (hazard ratio, 1.95; 95% CI, 1.10-3.45; P = .021). The median time to
recovery for patients receiving DXM was 68 hours (2.8 days) shorter than patients receiving placebo (109 hours
vs 177 hours; P = .037). In exploratory subgroup analysis, the median time to recovery for patients with simple
effusion receiving DXM was 76 hours (3.1 days) shorter than for patients with simple effusion receiving placebo
(P = .017). The median time to recovery for patients with complicated effusion receiving DXM was 14 hours (0.5
days) shorter than for patients with complicated effusion receiving placebo (P = .66). The difference in the effect of
DXM in the 2 severity groups was not statistically significant (P = .138 for interaction). There were no significant
differences in complications or adverse events attributable to the study drugs, except for hyperglycemia.
Conclusion In this trial, DXM seemed to be a safe and effective adjunctive therapy for parapneumonic pleural
effusion. (J Pediatr 2017;185:117-23).
Trial registration ClinicalTrials.gov: NCT01261546.

arapneumonic effusion complicates 2%-13% of cases of hospitalized pneumonia, leading to long hospitalizations.1-5 To

P our knowledge, corticosteroids have not been the subject of trials for parapneumonic effusion. Corticosteroids block
inflammatory cytokine genes that are key factors in the first, exudative stage of pleural effusion.6-8 Although this is con-
troversial, some trials have suggested that dexamethasone (DXM) may be useful
in meningitis, septic arthritis, and adults with community-acquired pneumonia
(CAP).9-13
We hypothesized that the concomitant treatment of antimicrobials and early From the 1Department of Pediatrics, University Hospital
administration of DXM would prove beneficial in parapneumonic pleural effu- Infanta Sofía, San Sebastián de los Reyes; 2Biomedical
School, Universidad Europea, Madrid; 3Department of
sion. Positive results would be seen in the early phases, before an excessive in- Pediatrics, University Hospital Ramón y Cajal; 4Medical
School, Universidad de Alcalá; 5Pediatrics, Tropical and
flammatory response developed. Our aim in this study was to investigate whether Infectious Diseases Department, University Hospital La
the concomitant treatment of DXM (0.25 mg/kg/dose every 6 hours for 48 hours) Paz; 6Pediatric Infectious Diseases Unit, University
Hospital Gregorio Marañón, Madrid, Spain; 7Department
with antibiotic therapy decreased the time to recovery of parapneumonic pleural of Pediatrics, Toledo University Hospital, Toledo, Castilla-
La Mancha, Spain; 8Department of Pediatrics, Getafe
effusion compared with placebo. University Hospital, Getafe; 9Department of Pediatrics,
University Hospital Príncipe de Asturias, Alcalá de
Henares, Madrid, Spain; 10Department of Pediatrics,
University Hospital Carlos Haya, Málaga, Andalucía,
Spain; 11Pediatric Infectious Diseases Unit, University
Hospital 12 de Octubre; and 12Biostatistics Unit,
Methods University Hospital La Paz, Madrid, Spain
*A list of additional members of the CORTEEC Study
Group is available at www.jpeds.com (Appendix 1).
Corticoids for Pleural Effusion and Empyema (CORTEEC) was a multicenter,
Funded by the Spanish Ministry of Health (EC10-014) and
double-blind, parallel-group, placebo-controlled clinical trial (ClinicalTrials.gov: the Sociedad de Pediatría de Madrid y Castilla La
Mancha (2011). Kern Pharma, Inc, Barcelona, Spain,
supplied the drugs (dexamethasone and saline) and
randomization. They had no role in the study design, data
collection and analysis, or drafting of the manuscript. The
authors declare no conflicts of interest.
CAP Community-acquired pneumonia PCV13 Pneumococcal conjugate vaccine
CRP C-reactive protein SaO2 Oxygen saturation 0022-3476/$ - see front matter. © 2017 Elsevier Inc. All rights
DXM Dexamethasone VATS Video-assisted thoracic surgery reserved.
http://dx.doi.org10.1016/j.jpeds.2017.02.043

117
THE JOURNAL OF PEDIATRICS • www.jpeds.com
NCT01261546). The aim of this trial was to provide informa-
tion on the safety and potential for effectiveness of DXM over
placebo. The study was conducted in Spain at 9 urban
university-affiliated public hospitals over a period of 55 months.
The study was conducted in accordance with the Declaration
of Helsinki and Good Clinical Practice principles. The study
was approved by the Institutional Review Board of each of the
9 hospitals.
Eligible participants were hospitalized children aged 1 month
to 14 years with CAP and pleural effusion. CAP was defined
as fever >38 °C, cough, and parenchymal infiltrate on chest ra-
diography. All patients underwent chest radiography and tho-
racic ultrasound. Written informed consent was obtained from
a parent of each child. Exclusion criteria were proven aller-
gies to any study drugs or treatment with corticosteroids, im-
munodeficiency, any concomitant disease likely to worsen with
corticosteroid treatment and any condition that prevented par-
ticipation in the study.
Randomization and Masking
Randomization was stratified by center and severity of disease
(ie, complicated or simple effusion). Eight centers had planned
to recruit 7 patients each. Patients were classified as having com-
plicated effusion or simple effusion. Complicated effusion was
defined as pH <7.2 or loculations or septations on ultra-
sound, or bacteria on Gram stain. Simple effusion was defined
as effusion not fulfilling any of those criteria (Table I; avail-
able at www.jpeds.com). Patients of each center were as-
signed in a 1:1 ratio to each treatment group using a computer-
generated randomization scheme performed the manufacturer
of the study drug (Kern Pharma, Barcelona, Spain).
The participants, care providers, data manager, and inves-
tigators were all blinded until the end of the study. Kern Pharma
provided prepackaged boxes for each patient containing either
15 transparent ampules of the study drug, 4 mg/mL DXM phos-
phate (in 1-mL ampules), or 15 ampules of the placebo, 0.9%
NaCl (in 1-mL ampules). The placebo was identical in ap-
pearance, color, size, and packaging to DXM. Care providers
and researchers were blinded to the content of the ampules.
The boxes were numbered consecutively for each hospital, se-
verity stratum, and patient, according to the randomization
scheme.
Twenty-eight months into the study, with 36 patients re-
cruited, 2 unexpected severe adverse events occurred (1 death
and 1 need for insulin). After receipt of permission from the
Institutional Review Boards and the Spanish Drug Agency, a
masked interim analysis was performed to test the safety and
futility of the trial. The Spanish Drug Agency reviewed and
analyzed the results of the analysis. The trial was deemed safe
and not futile, and so the trial continued. Competitive recruit-
ment was allowed for up to 16 patients per center, and 1 new
center joined (Table II; available at www.jpeds.com). No further
changes in methods were made after the start of the trial.
Procedures
Patients were scheduled for 8 intravenous doses of DXM,
0.25 mg/kg every 6 hours or the corresponding volume of
118
Volume 185
placebo over 48 hours. The first dose was administered within
12 hours after the diagnosis of pleural effusion. DXM or placebo
was administered immediately after the first dose of cefotaxime,
at a dose of 50 mg/kg/6 hours. Ranitidine (5 mg/kg/day in-
travenously in 2 doses over 48 hours) was administered con-
comitantly to prevent gastric bleeding. Cefotaxime was
continued until 48 hours after the patient was afebrile, then
switched to amoxicillin-clavulanate (80 mg/kg/day) to com-
plete the 15-day antimicrobial therapy.
Patients with simple effusion received only medical treatment.
Diagnostic thoracentesis was recommended if effusion was
>10 mm on the ultrasound image. If biochemical data indi-
cating complicated effusion were found, appropriate drain-
age was recommended. The recommended management for
complicated effusion was medical treatment plus pleural drain-
age and fibrinolytics, or video-assisted thoracic surgery (VATS).
Patients with an immediate need for thoracentesis, drainage,
or VATS were included in the protocol. Nevertheless, a con-
servative approach without drainage was permitted at the dis-
cretion of the clinician, given that some studies have shown
that pleural drainage is not always necessary. Blood tests (ie,
hemoglobin, white blood cell count, and C-reactive protein
[CRP]) were performed at the time of diagnosis. Ultrasound
and blood tests were repeated at least 48 hours after recruit-
ment, to analyze changes.
Outcomes
The primary outcome was time to recovery, measured in hours.
Recovery criteria were continuous ambient oxygen satura-
tion (SaO2) >92%, continuous temperature <37°C, no respi-
ratory distress, end of invasive procedures, pneumonia in
resolution, and oral feeding. The presence of all criteria were
necessary to classify a patient as recovered. Time to recovery
was calculated from the first dose of trial treatment until the
time when all recovery criteria were fulfilled; for fever, this was
the first hour of absence of fever (Table III; available at
www.jpeds.com).
The secondary outcomes measured were safety, such as com-
plications of disease from the moment of hospitalization until
day 30 after discharge (Table IV; available at www.jpeds.com),
and adverse events attributable to corticosteroids during hos-
pitalization (Table V). Other secondary endpoints included the
progression of simple effusion to complicated effusion requir-
ing chest drainage, decreased CRP level, and decreased effu-
sion during days 1-3.
Statistical Analyses
A sample size of 56 patients (28 patients per group) were needed
to detect a reduction in time to recovery of ≥24 hours (as-
suming an SD of 31 hours, 80% power, a 2-sided a level of
5%, and a 10% dropout rate). The SD was obtained from a
small observational pilot study, where we observed an SD of
31 hours in time to recovery in children treated with steroids
who had a pleural effusion.
We performed an intention-to-treat analysis. We assessed
differences in time to recovery based on treatment group. We
managed the missing outcome of the patient who died by
Tagarro et al
June 2017 ORIGINAL ARTICLES

Table V. Prespecified adverse events attributable to Table VI. Baseline patient characteristics, according to
steroids treatment group
Adverse events DXM Placebo
Variables (n = 30) (n = 30)
Hyperglycemia >126 mg/dL*
Mild (126-140 mg/dL) Age, y, mean ± SD 4.6 ± 4.2 4.8 ± 5.5
Moderate (140-200 mg/dL) Sex, n (%)
Severe (>200 mg/dL) Males 13 (43) 12 (40)
Need of insulin Females 17 (56) 18 (60)
Upper gastrointestinal bleeding Underlying disease, n (%) 13 (43) 11 (36)
Anemia (decreased hemoglobin) Asthma 4 (13) 5 (16)
Mild (DHb <1 g/L from day 1 to day 3) Overweight 3 (10) 5 (16)
Moderate (DHb 1-3 g/L from day 1 to day 3) Neurologic disease 1 (3) 1 (3)
Severe (DHb >3 g/L from day 1 to day 3) Celiac disease 1 (1) 0 (0)
Transfusion Atopy 3 (10) 1 (3)
Oropharyngeal candidiasis Previous antibiotics, n (%) 9 (30) 9 (30)
Allergic reaction, rash Oral antibiotics 4 (13) 3 (10)
Other Intravenous antibiotics 8 (26) 7 (23)
Daycare or school, n (%) 29 (96) 24 (80)
Glycemia was checked once a day on days 1, 2, and 3 of treatment. ≥3 doses of PCV7, n (%) 14 (46) 19 (63)
*One measurement >126 mg/dL was sufficient to label the event as adverse event. ≥3 doses of PCV13, n (%) 4 (13) 3 (10)
Influenza immunization, n (%) 1 (3) 3 (10)
H influenzae type B immunization, n (%) 29 (96) 29 (96)
excluding that patient from the analysis of median time to event Duration of symptoms before randomization, 3.7 ± 2.5 4.0 ± 2.5
d, mean ±SD
and including the patient as a censored value in the survival Temperature, °C, median (IQR) 39 (1) 39 (1.4)
analysis. We observed differences in time to recovery accord- SaO2 <92%, n (%) 2 (6.6) 2 (6.6)
ing to severity with the Kaplan-Meier survival estimates, cen- Systolic arterial pressure, mm Hg, mean ± SD 107 ± 12 107 ± 12
Diastolic arterial pressure, mm Hg, mean ± SD 59 ± 10 66 ± 7
soring data for patients who died before the end of the study. Effusion amount, n (%)
Between-group differences in survival curves were assessed using <1 cm 18 (60) 16 (53)
the log-rank test. The magnitude of the treatment effect was >1 cm to 1/3 of hemithorax 9 (30) 12 (40)
>1/3 of hemithorax 3 (10) 2 (6)
estimated using the Cox proportional hazard regression model Distance from chest wall, cm, median (IQR) 0.8 (0.8) 1.2 (1.5)
with a single explanatory variable (treatment). Hazard ratios Thoracentesis at entry, n (%) 9 (30) 14 (46)
(HRs) and corresponding 2-sided 95% CIs were generated. In VATS, n (%)
At entry (<24 h) 2 (6) 0 (0)
addition, we performed a Cox regression analysis of the treat- Delayed (>24 h) 2 (6) 1 (3)
ment effect on time to recovery, adjusted for severity and strati- Chest tube, n (%) 9 (30) 12 (40)
fied by center. An exploratory test for interaction was performed Confirmed etiology*, n (%)
Typical bacteria identified† 8 (26) 14 (46)
to test for statistical significance in the difference in treat- S pneumoniae 5 (16) 11 (36)
ment effect across strata. An adjusted Cox model adjusted by S pyogenes 3 (10) 1 (3)
baseline features was performed. Continuous quantitative vari- S aureus 0 (0) 1 (3)
Anaerobic, gram-negative bacilli 0 (0) 1 (3)
ables are shown as mean and SD or median and IQR, depend- Unknown viral, other‡ 22 (74) 16 (54)
ing on the Kolmogorov-Smirnov test. Categorical variables are M pneumoniae 4 (13) 1 (3)
presented with frequency distributions. We assessed the dif- C pneumoniae 1 (3) 0 (0)
Tuberculosis 0 (0) 1 (3)
ferences in the safety categorical variables with the c2 test and Viral (influenza, adenovirus, metapneumovirus) 3 (10) 1 (3)
Fisher exact test when the number of events was <5. All sta- Not established 17 (56) 13 (43)
tistical tests were 2-sided. Relative risk with 95% CI was cal- *Only 8 patients (13%) underwent an extensive workup for pneumonia. Four of these patients
culated for the outcome. The significant value for the primary were found to have infections for multiple microorganisms. This workup included blood cul-
comparison was P < .05. SPSS version 21 (IBM, Armonk, New tures, pleural effusion cultures, PCR analyses of pleural fluid to detect S pneumoniae, 2 se-
rologies separated for 4 weeks to test for M pneumoniae, C pneumoniae, and L pneumoniae,
York) was used for statistical analysis. and PCR tests of nasopharyngeal secretions for a panel of respiratory viruses such as M
pneumoniae, C pneumoniae, and L pneumoniae. Only routine tests were performed for the rest
of the patients. These tests included blood culture, pleural fluid cultures, and PCR analyses of
pleural fluid to detect S pneumoniae.
Results †Typical bacteria: S pneumoniae, S aureus, and S pyogenes. P = .10.
‡May include culture-negative typical bacteria.
A total of 60 patients were randomly assigned to the DXM
(n = 30) or placebo (n = 30) group between January 1, 2011,
and May 31, 2015 (Figure 1; available at www.jpeds.com). The the placebo group. One patient, a 13-year-old female with ce-
median patient age was 4.7 years, and 35 patients (58%) were rebral palsy who had been assigned to the placebo group, died
female. A total of 24 patients (40%) had an underlying con- at 25 days after enrollment without achieving the recovery cri-
dition, including 9 with asthma (15%) and 8 with over- teria, owing to aspiration pneumonia and adult respiratory dis-
weight (13%). There were no significant differences in baseline tress syndrome. Two other patients withdrew from the trial
characteristics between the groups (Table VI). (Figure 1).
A total of 57 patients (95%) completed the protocol, in- In the multivariate Cox model, patients receiving DXM re-
cluding 29 (96%) from the DXM group and 28 (93%) from covered faster than those receiving placebo, after adjustment
Dexamethasone for Parapneumonic Pleural Effusion: A Randomized, Double-Blind, Clinical Trial 119
THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 185

(2.8 days), or 38%, shorter in the patients receiving DXM

Table VII. Time to recovery (primary endpoint)
(Table VII and Figure 2).
Characteristics Placebo DXM P value For exploratory purposes, we also studied time to recov-
Total ery according to severity group. The median time to recov-
No. of patients 30 30 ery for patients with simple effusion receiving DXM was 76
Duration, h, median (95% CI) 177 (115-238) 109 (37-180) .037*
HR (95% CI) 1 1.95 (1.10-3.45) .021† hours (3.1 days), or 61%, shorter than that for patients with
Simple effusion simple effusion receiving placebo (P = .017). The median time
No. of patients 18 18 to recovery for patients with complicated effusion receiving
Duration, h, median (95% CI) 124 (49-198) 48 (35-60) .017
Complicated effusion DXM was 14 hours (0.5 days), or 6%, shorter than that for
No. of patients 12 12 patients with complicated effusion receiving placebo (P = .66)
Duration, h, median (95% CI) 240 (129-350) 226 (187-265) .66 (Table VII and Figure 2). The difference in the effect of DXM
*Log-rank test. in the 2 severity groups was not statistically significant (P = .138
†Cox HR. for interaction). The adjusted HR of DXM corrected for dif-
ferences in baseline features (including previous antibiotic use)
remained statistically significant (Table VIII; available at
for severity group and stratification by center (HR, 1.95; 95% www.jpeds.com).
CI, 1.10-3.45; P = .021). The median time to recovery was 177 There were no between-group differences in mortality or
hours (95% CI, 115-238 hours) for patients receiving placebo pulmonary complications, such as pneumothorax and nec-
and 109 hours (95% CI, 37-180 hours) for those receiving rotizing pneumonia (Table IX and Table X; Table IX avail-
DXM. Therefore, the median time to recovery was 68 hours able at www.jpeds.com). Nine patients (30%) receiving DXM

Figure 2. Kaplan-Meier analysis of the effect of DXM on time to recovery. A, All enrolled patients. The patient who died was
censored on the day of death (608 hours). B, Patients with simple pleural effusion. C, Patients with complicated pleural effusion.

120 Tagarro et al
June 2017 ORIGINAL ARTICLES

2 (Figures 3 and 4, and Appendix 2 available at www.jpeds

Table X. Complications and adverse events attribut-
.com).
able to study treatment, according to severity and type
(per-protocol analysis)
DXM Placebo Discussion
Complications and (n = 30), (n = 30), Risk ratio P
adverse events n (%) n (%) (95% CI) value
In this trial, DXM treatment, along with standard antibiotic
All participants
Any complication 3 (10) 4 (13) 0.7 (0.1-3.0) .68
therapy, reduced the time to recovery of parapneumonic pleural
All-cause mortality 0 (0) 1 (3) N/A N/A effusion with no apparent increase in serious adverse events.
Pulmonary complications 3 (10) 4 (13) 0.7 (0.1-3.0) .68 DXM reduced the time to recovery by almost 3 days (ie, 38%).
Pneumothorax 2 (6) 1 (3) 2 (0.1-24) .56
Necrotizing pneumonia 2 (6) 3 (10) 0.6 (0.1-4.5) .64
These results are in line with reported benefits of corticoste-
Any adverse event attributable 22 (73) 19 (63) 1.5 (0.5-4.7) .40 roids in mice and humans with pneumonia and other
to the study drug infections.12-14 Some trials have suggested that corticosteroids
Hyperglycemia 15 (50) 6 (20) 2.5 (1.2-5.5) .02
Mild (126-140 mg/dL) 6 (20) 4 (13) 1.6 (0.4-6.4) .49
shorten the time to clinical stability when added to antibi-
Moderate (140-200 mg/dL) 7 (23) 3 (10) 2.7 (0.6-11.8) .17 otic treatment in immunocompetent adults with CAP. We chose
Severe (>200 mg/dL) 2 (6) 0 (0) N/A N/A time to recovery rather than duration of hospital stay, as the
Need for insulin 1 (3) 0 (0) N/A N/A
Upper gastrointestinal bleeding 0 (0) 0 (0) N/A N/A
primary endpoint, because duration of hospital stay can be in-
Anemia 10 (34) 16 (55) 0.4 (0.1-1.2) .12 fluenced by other factors unrelated to a patient’s health.
Mild (DHb <1 g/L from 6 (20) 7 (24) 0.8 (0.3-2.8) .75 The effect of DXM on time to recovery was not statisti-
day 1 to day 3)
Moderate (DHb 1-3 g/L 3 (10) 3 (10) 1 (0.1-5.4) 1
cally different between patients with simple effusion and those
from day 1 to day 3) with complicated effusion. Given the sample size, an interac-
Severe (DHb >3 g/L from 1 (3) 6 (20) 0.1 (0.02-1.3) .08 tion value of 0.138 suggests insufficient power to detect sta-
day 1 to day 3)
Transfusion 1 (3) 3 (10) 0.3 (0.03-3.1) .32
tistically significant differences in treatment effects across strata.
Oropharyngeal candidiasis 0 (0) 0 (0) N/A N/A The data suggest that when the liquid is already complex, the
Allergic reaction, rash 0 (0) 1 (3) N/A N/A effect is limited, however. A reduction of 14 hours in time to
Other
Local pain, agitation 1 (3) 0 (0) N/A N/A
recovery seems to have little clinical significance.
Simple effusion Neutrophil migration and the production of proin-
Children with simple 1/18 (5) 3/18 (16) 0.2 (0.02-3.1) .31 flammatory cytokines are key factors in the first, exudative stage
effusion who eventually
underwent pleural
of pleural effusion and fibroblast proliferation.6-15 Corticoste-
drainage roids block inflammatory cytokine genes. This study was strati-
fied in the belief that patients with advanced disease would
N/A, comparison not possible.
benefit less than patients in the early stages of the disease, fol-
and 12 patients (40%) receiving placebo required chest drain- lowing the typical pattern of meningitis.9,10,16-18 Larger studies
age after enrollment (risk ratio, 0.8; 95% CI, 0.4-1.4). There are needed to confirm this belief, however.
were no significant differences in adverse events attributable We also wanted to know whether DXM would decrease the
to the study drugs, except for hyperglycemia (Table IX and progression of simple effusion to complicated effusion and
Table X). The risk ratio of DXM for hyperglycemia was 2.5 thereby reduce the need for pleural drainage or VATS. More
(95% CI, 1.2-5.5; attributable risk, 0.3; 95% CI, 0.05-0.5). The patients receiving placebo than receiving DXM needed even-
number needed to harm for 1 case of hyperglycemia to occur tual pleural drainage (16% vs 5%). Unfortunately, the study
was 4 (95% CI, 1.8-14). Serious adverse events that necessi- was underpowered, and thus unable to determine statistical
tated additional treatment were rare. Only 1 patient in the DXM significance for this analysis.
group, a 5-year-old girl who developed persistent hypergly- There was no increase in severe adverse events in the DXM
cemia during and after treatment, required insulin for 7 days. group. Hyperglycemia (mostly transient and nonsevere) was
After the trial, she was found to be prediabetic, with anti- more frequent in the DXM group. Reductions in severe anemia
insulin antibodies, anti–pancreatic islet antibodies, and a patho- (3% vs 20%) and transfusion (3% vs 10%) were associated with
logical glucose curve. One patient receiving placebo had a rash DXM, but were not statistically significant. DXM may have
due to Mycoplasma pneumoniae infection. stimulated erythropoiesis.19 The addition of ranitidine to the
Severe anemia (Table IX and Table X) was less frequent in regimen was intended to prevent gastrointestinal bleeding, but
the DXM group (risk ratio, 0.1; 95% CI, 0.02-1.3; P = .08; at- may have contributed to fewer cases of anemia. A larger trial
tributable risk, −0.16; 95% CI, −0.009 to −0.3). The number could address these observations.
needed to treat to prevent 1 case of severe anemia was 6 (95% The inclusion criteria were not restrictive; thus, our results
CI, 3-102). There were no between-group differences in mod- may be generalized to most of the patients cared for at similar
erate or severe hyperglycemia, need for insulin, mild or mod- hospitals, and may be applicable to a wide range of patients.
erate anemia, transfusion, candidiasis, rash, or gastrointestinal The children received usual care apart from the study drugs.
bleeding (Table IX). Complicated effusion was drained at the discretion of the at-
The reduction of CRP and pleural effusion (using an on- tending physician. Nonetheless, our findings might not be ap-
treatment analysis) is detailed in Figures 3 and 4, and Appendix plicable to some settings that lack the resources for surgery or
Dexamethasone for Parapneumonic Pleural Effusion: A Randomized, Double-Blind, Clinical Trial 121
THE JOURNAL OF PEDIATRICS • www.jpeds.com
cefotaxime or ranitidine therapy. Findings must be interpreted
with caution in areas with differing epidemiologies and a higher
prevalence of preventable bacteria, tuberculosis, and HIV.
Methodological limitations include a lack of statistical power
for the analysis of subgroups. We had a sample size with suf-
ficient power to test our primary hypothesis, however. Allow-
ing competitive recruitment in 2013 compromised stratification
in each center and may have enabled opportunity sampling.
Without this decision, however, the sample would never have
been established in a reasonable time. Our primary outcome
did not reflect long-term or life-threatening outcomes, and
prognosis was usually good. Nevertheless, time to recovery
matters to patients and parents, and we believe that it is an
important endpoint.20
The full etiologic workup for all patients was incomplete and
unavailable for every center. In a patient with tuberculosis, the
diagnosis was confirmed after the completion of the trial treat-
ment. This patient was not excluded due to the intention-to-
treat analysis. Streptococcus pneumoniae isolation was twice as
common in the placebo group compared with the DXM group
(n = 5). The group of patients with incomplete workups could
include those with blood culture-negative S pneumoniae CAP,
other typical bacteria, atypical bacteria, virus, and mixed eti-
ologies. The high proportion of combined etiologies and the
sample size precluded a post hoc subgroup analysis accord-
ing to etiology. Pediatricians usually do not have complete etio-
logic information when planning to treat pleural infection. The
impact of corticosteroids according to etiology is of interest,
but beyond the scope of this trial. In this trial, we aimed to
test whether DXM improved outcomes when used at the time
of diagnosis of pleural effusion.
Effusions were predominantly mild, and there was only 1
confirmed infection due to Staphylococcus aureus. These issues
may affect the generalizability of our conclusions.
The dose of DXM was higher that that used in most men-
ingitis trials. The dose of DXM in meningitis clinical trials is
0.4-0.6 mg/kg/day for 2-4 days,21 for a total dose of 0.8-2.4 mg/
kg. This study used a 2 mg/kg accumulated dose, which rep-
resents an intermediate dose, with an emphasis on the first days
of treatment. Potential long-term risks of corticosteroids, such
as neurodevelopmental outcomes, were not measured. Such
outcomes are unlikely, and moreover are beyond the scope of
this trial.
In conclusion, in this trial, DXM appeared to be a safe and
effective adjunctive therapy for decreasing the time to recov-
ery in children with parapneumonic pleural infection. This trial
provides a basis for a larger and definitive trial that should be
powered to confirm the findings and determine whether DXM
performs equally or differently across the severity groups. More-
over, future trials should demonstrate the effect of DXM on
the long-term complications of parapneumonic pleural
effusion. ■
We are grateful to the children and their families who participated in
the study. We thank Javier González de Dios, PhD, Amy Chang, and
David Sanz-Rosa, PhD, for their meaningful comments on this manu-
script and editing assistance.
122
Volume 185
Submitted for publication Jul 20, 2016; last revision received Jan 30, 2017;
accepted Feb 15, 2017
Reprint requests: Alfredo Tagarro, PhD, MD, Department of Pediatrics,
Hospital Universitario Infanta Sofía, Ave de Europa 34, 28702 San Sebastián
de los Reyes, Madrid, Spain. E-mail: alfredotagarro@hotmail.com;
alfredo.tagarro@salud.madrid.org.
References
1. Jain S, Williams DJ, Arnold SR, Ampofo K, Bramley AM, Reed C, et al.
Community-acquired pneumonia requiring hospitalization among U.S.
children. N Engl J Med 2015;372:835-45.
2. Kelly MM, Shadman KA, Edmonson MB. Treatment trends and out-
comes in US hospital stays of children with empyema. Pediatr Infect Dis
J 2014;33:431-6.
3. Bradley JS, Byington CL, Shah SS, Alverson B, Carter ER, Harrison C, et al.
Executive summary: the management of community-acquired pneumonia
in infants and children older than 3 months of age: clinical practice guide-
lines by the Pediatric Infectious Diseases Society and the Infectious Dis-
eases Society of America. Clin Infect Dis 2011;53:617-30.
4. Avansino JR, Goldman B, Sawin RS, Flum DR. Primary operative versus
nonoperative therapy for pediatric empyema: a meta-analysis. Pediat-
rics 2005;115:1652-9.
5. Carter E, Waldhausen J, Zhang W, Hoffman L, Redding G. Manage-
ment of children with empyema: pleural drainage is not always neces-
sary. Pediatr Pulmonol 2010;45:475-80.
6. Barnes PJ. How corticosteroids control inflammation: Quintiles Prize
Lecture 2005. Br J Pharmacol 2006;148:245-54.
7. Rhen T, Cidlowski JA. Antiinflammatory action of glucocorticoids–new
mechanisms for old drugs. N Engl J Med 2005;353:1711-23.
8. Chapman SJ, Davies RJ. Recent advances in parapneumonic effusion and
empyema. Curr Opin Pulm Med 2004;10:299-304.
9. Lebel MH, Freij BJ, Syrogiannopoulos GA, Chrane DF, Hoyt MJ, Stewart
SM, et al. Dexamethasone therapy for bacterial meningitis. Results of two
double-blind, placebo-controlled trials. N Engl J Med 1988;319:964-71.
10. Odio CM, Faingezicht I, Paris M, Nassar M, Baltodano A, Rogers J, et al.
The beneficial effects of early dexamethasone administration in infants
and children with bacterial meningitis. N Engl J Med 1991;324:1525-31.
11. Harel L, Prais D, Bar-On E, Livni G, Hoffer V, Uziel Y, et al. Dexametha-
sone therapy for septic arthritis in children: results of a randomized double-
blind placebo-controlled study. J Pediatr Orthop 2011;31:211-5.
12. Blum CA, Nigro N, Briel M, Schuetz P, Ullmer E, Suter-Widmer I, et al.
Adjunct prednisone therapy for patients with community-acquired pneu-
monia: a multicentre, double-blind, randomized, placebo-controlled trial.
Lancet 2015;385:1511-8.
13. Meijvis SC, Hardeman H, Remmelts HH, Heijligenberg R, Rijkers GT,
van Velzen-Blad H, et al. Dexamethasone and length of hospital stay in
patients with community-acquired pneumonia: a randomized, double-
blind, placebo-controlled trial. Lancet 2011;377:2023-30.
14. Ghoneim HE, McCullers JA. Adjunctive corticosteroid therapy im-
proves lung immunopathology and survival during severe secondary pneu-
mococcal pneumonia in mice. J Infect Dis 2014;209:1459-68.
15. Kroegel C, Antony VB. Immunobiology of pleural inflammation: poten-
tial implications for pathogenesis, diagnosis and therapy. Eur Respir J
1997;10:2411-8.
16. Qazi SA, Khan MA, Mughal N, Ahmand M, Joomro B, Sakata Y, et al.
Dexamethasone and bacterial meningitis in Pakistan. Arch Dis Child
1996;75:482-8.
17. Molyneux EM, Walsh AL, Forsyth H, Tembo M, Mwenechanya J, Kayira
K, et al. Dexamethasone treatment in childhood bacterial meningitis in
Malawi: a randomized controlled trial. Lancet 2002;360:211-8.
18. McCracken GH Jr. Rich nations, poor nations, and bacterial meningitis.
Lancet 2002;360:183.
19. Falchi M, Varricchio L, Martelli F, Masiello F, Federici G, Zingariello M,
et al. Dexamethasone targeted directly to macrophages induces macro-
phage niches that promote erythroid expansion. Haematologica
2015;100:178-87.
Tagarro et al
June 2017 ORIGINAL ARTICLES

20. Guyatt GH, Oxman AD, Kunz R. GRADE guidelines: 2. Framing the ques-
tion and deciding on important outcomes. J Clin Epidemiol 2011;64:395-
400.
21. Brouwer MC, McIntyre P, Prasad K, van de Beek D. Corticosteroids for
acute bacterial meningitis. Cochrane Database Syst Rev 2015;9:
CD004405.
22. Tagarro A, Bote P, Sánchez A, Otheo E, Sanz JC, Sanz-Rosa D. Compli-
cations of pneumococcal bacteremia after thirteen-valent conjugate vaccine
withdrawal. Pediatr Infect Dis J 2016;35:1281-7.
23. Vuori-Holopainen E, Peltola H. Reappraisal of lung tap: review of an old
method for better etiologic diagnosis of childhood pneumonia. Clin Infect
Dis 2001;32:715-26.

Dexamethasone for Parapneumonic Pleural Effusion: A Randomized, Double-Blind, Clinical Trial 123
THE JOURNAL OF PEDIATRICS • www.jpeds.com
Appendix 1
The CORTEEC Study Group includes the following collabo-
rators (for recruitment and collection of data): Lorena Pérez1,
BS; María Luisa Herreros1, MD; Julia Yebra1, MD; Jana Rizo1,
MD; Ana Barrios1, MD; Alfonso Cañete1, MD, PhD; Lisette
Arguinzoniz1, MD; Francisco Gaya2, PhD; Carmen Vázquez3,
MD; Cristina Ots4, MD; Mar Santos5, MD; Jesús Saavedra5, MD,
PhD; Sara Guillén6, MD, PhD; Luis Prieto6, MD; José T. Ramos6,
MD, PhD; Carlos Vela7, MD; Alicia Berghezan7, MD; Antonio
Conejo, MD8; Patricia Paredes9, MD; Iván Bermejo9, MD;
Miriam Guizar9, MD; Diana Gutierrez9, MD; Castro Codesal9,
MD; Francisco Ramos9, MD; Carmen Izquierdo9, MD; Peña
Gomez-Herruz9, MD; María Isabel González-Tomé10, MD, PhD;
César Pérez-Caballero11, MD; Elena Álvarez11, MD; José Luis
Vázquez11, MD, PhD; Cristina Verdú12, MD; Ana Gómez-
Zamora 12 , MD; Juan-José Menéndez 12 , MD; Cristina
Schuffelmann12, MD; Raúl Borrego13, MD; Jesús Llorente14, PhD;
Alicia Fernández14, PhD; José Carmelo Albillos15, MD, PhD;
Martina Steiner16, MD; David Sanz-Rosa, PhD16 (methods and
writing) and Israel Thuissard16, BS (statistics).
Affiliations
1
Pediatrics Department, Hospital Universitario Infanta Sofía
(San Sebastián de los Reyes, Madrid, Spain)
2
Biostatistics Unit, Hospital Universitario La Paz (Madrid,
Spain)
3
Pediatrics Department, Hospital Universitario Ramón y Cajal
(Madrid, Spain)
4
Pediatrics, Tropical and Pediatric Infectious Diseases Unit,
Hospital Universitario La Paz (Madrid, Spain)
5
Pediatric Infectious Diseases Unit, Hospital Universitario
Gregorio Marañón (Madrid, Spain)
6
Pediatrics Department, Hospital Universitario de Getafe
(Madrid, Spain)
7
Pediatrics Department, Hospital Universitario de Toledo
(Toledo, Castilla-La Mancha, Spain)
8
Pediatrics Department, Hospital Universitario Carlos Haya
(Málaga, Andalucía, Spain)
9
Pediatrics Department, Hospital Universitario Príncipe de
Asturias (Madrid, Spain)
10
Pediatric Infectious Diseases Unit (PICU), Hospital
Universitario 12 de Octubre (Madrid, Spain)
11
PICU, Hospital Universitario Ramón y Cajal (Madrid,
Spain)
12
PICU, Hospital Universitario La Paz (Madrid, Spain)
13
PICU, Hospital Universitario Toledo (Toledo, Castilla-La
Mancha, Spain)
14
Pharmacy Department, Hospital Universitario Infanta Sofía
(San Sebastián de los Reyes, Madrid, Spain)
15
Radiology Department, Hospital Universitario Infanta Sofía
(San Sebastián de los Reyes, Madrid, Spain)
16
Biomedical School, Universidad Europea de Madrid
(Madrid, Spain)
123.e1
Volume 185
Appendix 2
Study Design
Setting. The study was conducted at 9 Spanish university hos-
pitals: 7 in the Madrid region (Hospitals La Paz, Gregorio
Marañón, Príncipe de Asturias, Getafe, Ramón y Cajal, Infanta
Sofia, and 12 de Octubre); 1 in Toledo, the capital of the Castilla-
La Mancha region (Complejo Hospitalario de Toledo); and 1
in Málaga, the second largest city in Andalusia (Hospital Carlos
Haya), between December 2010 and May 2015. Three of the
hospitals (Infanta Sofia, Príncipe de Asturias, and Getafe) are
secondary centers, and the others are tertiary centers with a
PICU and pediatric surgery capability.
Intervention. Pneumonia was defined as fever, cough, and
consolidation or infiltrate on chest radiography. Pleural effu-
sion was confirmed by ultrasound in all cases. Once re-
cruited, patients were assigned 1 box of masked study drug
(either 0.25 mg/kg/6 hours of intravenous DXM or match-
ing placebo). The boxes were numbered consecutively for each
hospital, severity stratum, and patient, according to the ran-
domization scheme. Each patient was assigned a code and a
box. Each box and ampulla were codified as follows: H# (# = 1
to 9, with the number corresponding to each center)—S or C
(depending on the severity stratum, with “S = simple effu-
sion” or “C = complicated effusion”)—P# (# = number of the
patient included in each stratum of that center); for example,
H1-C-P1.
Patients received the first dose of the study drug within 12
hours after diagnosis. The study drug was matched with
cefotaxime (50 mg/kg/6 hours) and administered 5 minutes
after the antibiotic. If the patient had already been admitted
and had been receiving another antibiotic, then that antimi-
crobial was switched to cefotaxime, which was then matched
with the administration of the study drug.
After the administration of antibiotics and steroids, the stan-
dard therapeutic approach was followed in each hospital. Fever
and pain management was done with paracetamol, ibupro-
fen, or dipyrone. If symptoms worsened in patients with simple
effusion, then disease management included consultation with
a pediatric surgeon, a pediatric critical care physician, or a phy-
sician able to provide chest drainage. In cases of complicated
effusion, management included daily visits by a pediatric phy-
sician with experience in pleural drainage. When SaO2 was
<92%, supplemetal oxygen via a nasal cannula was recom-
mended. Oxygen was withdrawn when SaO 2 >92% was
achieved with a fraction of inspired oxygen=0.21. The follow-
ing drugs were permitted as needed: salbutamol inhaler or
nebulizer, intravenous glucose, intravenous saline, insulin, blood
derivates, intrapleural urokinase or streptokinase, sedatives and
analgesic drugs (eg, midazolam, propofol, ketamine, fentanyl
and morphine) for invasive procedures for the diagnosis or
treatment of pleural effusion.
Tagarro et al
June 2017
Rescue antimicrobial medication was to be avoided when-
ever possible to standardize the trial as much as possible. Rescue
medication was permitted in specific cases of complicated ef-
fusion, such as in 1 case with a positive culture of S pneumoniae
which has a high resistance to penicillin, in 1 case with a sus-
picion of methicillin-resistant S aureus or complicated effu-
sion due to Streptococcus pyogenes, and in some cases in which
the attending physician considered rescue treatment neces-
sary after the first 48 hours of treatment. Rescue medication
was provided with either vancomycin (15 mg/kg every 6 hours)
or clindamycin (10 mg/kg every 8 hours).
Blood glucose tests were performed daily for 3 days, or longer
if results were abnormal. An additional ultrasound and blood
tests were performed on day 3 to assess key efficacy variables:
amount of effusion, anemia, and CRP level. Blood tests
were also meant to assess variables (eg, hemoglobin, blood
glucose) after the treatment. Each hospital provided thera-
peutic intervention until patient discharge. Patients were evalu-
ated clinically and by chest radiography (and ultrasound, if
necessary) on days +7 and +30 after discharge. Investigators
at each center recorded the data and sent the data to a data
manager, who introduced the data into an anonymous, cen-
tralized database.
Criteria for Early Termination of the Study. We decided be-
forehand that the study would be terminated if a significant
rate of adverse events occurred. We would have stopped the
study in the case of serious adverse reactions attributable to
DXM, such as losses to follow-up, patient decisions, unex-
pected serious adverse events, or death. We would have stopped
the study if any of the groups had seemed to have a worse
outcome than expected compared with the standard disease
management. For this reason, we performed a safety interim
analysis. We also would have stopped the study if recruit-
ment was <30% of the patients after the first year.
Data Storage. Data were entered into notebooks by the prin-
cipal investigators at each center. An independent monitor re-
viewed all of the notebooks. The monitor then checked that
all data entered were correct, and that informed consent was
signed by parents and children older than age 12 years. Finally,
the monitor transferred the case report data to an electronic
database designed by the Biostatistics Unit of the Hospital La
Paz with SPSS version 21 (IBM, Armonk, New York). We con-
trolled the quality of the information received by conducting
an exploratory analysis aimed at detecting conflicting, out-
of-range, or missing values. The main researcher, the monitor,
and the Biostatistics Unit of the Hospital La Paz had permis-
sion to access this database.
Confidentiality was guaranteed by identifying the patients
only by code (eg, H1-C-P1).
Secondary Endpoints. Data for the decrease in CRP levels were
available for 55 patients (91%). In the DXM group, CRP de-
creased by 159 (144) mg/L from day 1 to day 3. In the placebo
group, CRP decreased by 67 (89) mg/L from day 1 to day 3
(Figure 3). The distribution was significantly different between
Dexamethasone for Parapneumonic Pleural Effusion: A Randomized, Double-Blind, Clinical Trial
ORIGINAL ARTICLES
the 2 groups (P = .01). The reduction of CRP was greater in
the simple effusion stratum than in the complicated effusion
stratum (Figure 3).
Data for the reduction of pleural effusion were available for
54 patients (90%). Median pleural effussion decreased by 4.7
(8.8) mm from day 1 to day 3 in the DXM group, and by 5.2
(8.5) mm from day 1 to day 3 in the placebo group. The dif-
ference in the distribution was not statistically significant
between the 2 groups (Figure 4).
Only 8 patients (13%) underwent an extensive workup for
etiology. These patients also were included in a 2-center study,
initiated in 2013, on the etiology of CAP. This workup in-
cluded blood cultures, pleural effusion cultures, polymerase
chain reaction (PCR) analysis of pleural fluid to detect S
pneumoniae; 2 serologic analyses separated by 4 weeks to test
for M pneumoniae, Chlamydophyla pneumoniae, and Legionella
pneumoniae; and PCR analysis of nasopharyngeal secretions
for a panel of respiratory viruses, including M pneumoniae,
C pneumoniae, and L pneumoniae. Bronchoalveolar lavage and
lung tap were not performed on any patient.
Evidence of multiple infections was found in 4 of these 8
patients. One patient had M pneumoniae plus influenza, 1
patient had M pneumoniae plus adenovirus, 1 patient had S
pyogenes plus metapneumovirus, and 1 patient had rhinovi-
rus plus adenovirus.
Only routine tests were performed for the remaining pa-
tients. These tests included blood culture, pleural fluid cul-
tures, and PCR analyses of pleural fluid to detect S pneumoniae.
Overall, typical bacteria (ie, S pneumoniae, S aureus, and S
pyogenes) were identified in 22 patients (36%) (Table V).
On-Treatment Analysis. One patient was withdrawn from the
trial because of local pain and agitation after receiving the first
dose of DXM. One patient was withdrawn after receiving 7
doses of placebo and 2 doses of DXM by mistake.
A total of 58 patients (96%) completed the protocol, in-
cluding 29 (96%) from the DXM group and 29 (96%) from
the placebo group.
DXM was better than placebo with respect to time to re-
covery in the on-treatment analysis. The HR for recovery in
the on-treatment analysis was 1.75 (95% CI, 1.01-3.02;
P = .043). In the multivariate Cox model, the superiority of
DXM was consistent regardless of severity group (HR, 1.73;
95% CI, 1.02-2.9; P = .042).
The median time to recovery for all patients on DXM was
63 hours (2.6 days) shorter than that for all patients on placebo
(Table XI; available at www.jpeds.com).
For exploratory purposes, we also studied time to recov-
ery according to severity group. The median time to recov-
ery was 85 hours (3.1 days) shorter in patients with simple
effusion on DXM compared with patients with simple effu-
sion on placebo, and the median time to recovery was 14 hours
(0.5 days) shorter in patients with complicated effusion on
DXM compared with patients with complicated effusion on
placebo (Table XI). Although there is evidence of a longer
median time to recovery in patients with simple effusion com-
pared with patients with complicated effusion, the difference
123.e2
THE JOURNAL OF PEDIATRICS • www.jpeds.com
in the effect of DXM in the 2 severity groups was not statis-
tically significant at conventional significance levels (P = .129
for interaction).
Previous treatment with antibiotics was not a significant vari-
able in the Cox model.
Adjusted HR. DXM remained statistically significant, inde-
pendent of previous antibiotics received, underlying disease,
immunization history, age, and sex (Table VIII). Receipt of ≥3
doses of pneumococcal conjugate vaccine (PCV13) was a sig-
nificant variable in the Cox model for a better outcome (HR,
3.2; 95% CI, 1.07-10.01; P = .036). The interaction ratio with
the study treatment was P = .18. Therefore, there was no sta-
tistically significant difference in the effect of DXM between
children who received <3 doses of PCV13 and those who re-
ceived ≥3 doses of PCV13.
Confirmed bacterial etiology was a significant variable in
the Cox model for a worse outcome (HR, 0.46; 95% CI,
Figure 1. Enrollment, randomization, and follow-up in the trial.
123.e3
Volume 185
0.23-0.91; P = .026). The interaction ratio was P = .89. There-
fore, there was no statistically significant difference in the effect
of DXM in patients with confirmed bacterial infection.
CRP levels decreased significantly after DXM treatment, re-
flecting the anti-inflammatory effect of the steroid. High levels
of CRP have been associated with complications elsewhere.22
The benefits of DXM based on etiology remain to be seen.
In this trial, the etiologic workup was insufficient to allow us
to draw any conclusions. Even in those patients who under-
went an extensive workup, we could not be certain of any pos-
sible etiology, because bronchoalveloar lavage and lung tap were
not performed.23 The high proportion of culture-negative and
combined infections in CAP makes this issue difficult to
address.22 Clinical trials studying the benefits of DXM based
on etiology would have to include several subgroup analyses,
given the number of potential combined infections, making
it difficult to recruit a sufficient number of patients to achieve
statistical significance.
Tagarro et al
June 2017 ORIGINAL ARTICLES

Figure 3. Difference of C-reactive protein from day 1 to day 3 according to treatment group—placebo and dexamethasone
(DXM)—and stratified according to severity group (simple effusion complicated effusion). ***Significant difference.

Figure 4. Difference of pleural effusion from day 1 to day 3 according to treatment group—placebo and dexamethasone (DXM)—and
stratified according to severity group—simple effusion (SE) and complicated effusion (CE).

Dexamethasone for Parapneumonic Pleural Effusion: A Randomized, Double-Blind, Clinical Trial 123.e4
THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 185

Table I. Criteria for severity stratification3 Table III. Recovery criteria

Criteria Complicated effusion Simple effusion Recovery criteria
pH of pleural fluid <7.2 ≥7.2 Temperature <37 °C*
Echogenic features Loculations or septations Free-flowing fluid with no Ambient SaO2 >92%*
septations or loculations No respiratory distress (no tachypnea†, no retractions)
Gram Presence of bacteria No bacteria End of invasive procedures (pleural drainage, VATS)
Pneumonia in resolution‡
Fulfillment of only 1 criterion for complicated effusion was sufficient to place a patient in the Oral feeding
complicated effusion stratum. Patients with <10 mm of free fluid on the ultrasound who did
not undergo thoracentesis were classified as having simple effusion. *The last recovery criterion fulfilled was <37°C temperature in 52 patients (86%) and ambient
SaO2 >92% in 4 patients (6%).
†Tachypnea according to age was defined by World Health Organization standards.
‡Defined by a smaller lung infiltrate compared with the infiltrate in the first X-ray.

Table IV. Prespecified complications of disease

Table II. Patient recruitment according to centers and Complications
years All-cause mortality
Centers 2011 2012 2013 2014 2015 Total Pneumothorax
Necrotizing pneumonia
University Hospital Infanta 8 0 1 4 1 14 Children with initial simple effusion who eventually underwent pleural drainage
Sofía or surgery
University Hospital Ramón 3 4 0 5 0 12
y Cajal
Hospital Universitario La Paz 5 1 2 1 0 9
University Hospital Gregorio 2 1 0 0 1 4
Marañón Table VIII. Adjusted HR of DXM adjusted by basal
University Hospital Toledo 3 0 1 1 0 5
University Hospital Getafe 4 0 0 0 0 4 features
University Hospital Príncipe 1 2 1 0 1 5
Risk factors Adjusted HR (95% CI) P value
de Asturias
University Hospital Carlos 4 0 0 0 0 4 Underlying disease 1.84 (1.07-3.1) .02
Haya Previous antibiotics 1.90 (1.1-3.2) .02
University Hospital 12 de 0 0 0 3 0 3 ≥3 doses of PCV13 1.87 (1.09-3.1) .02
Octubre Age <2 y 1.85 (1.08-3.1) .02
Total 30 8 5 14 3 60 Male sex 1.95 (1.1-3.3) .01

123.e5 Tagarro et al
June 2017 ORIGINAL ARTICLES

Table IX. Complications and adverse events attributable to study treatment, according to severity and type (on-
treatment analysis)
DXM (n = 29), Risk ratio
Complications and adverse events n (%) Placebo (n = 29), n (%) (95% CI) P value
All participants
Any complication 3 (10) 4 (13) 0.7 (0.1-3.5) .68
All-cause mortality 0 (0) 1 (3) 0 (0) N/A
Pulmonary complications 3 (10) 4 (13) 0.7 (0.1-3.5) .68
Pneumothorax 2 (6) 1 (3) 2 (0.2-20) N/A
Necrotizing pneumonia 2 (6) 3 (10)
Any adverse event attributable to study drug 21 (72) 19 (65) 1.3 (0.4-4.2) .50
Hyperglycemia 15 (51) 6 (20) 2.5 (1.2-5.5) .01
Mild (126-140 mg/L) 6 (20) 4 (13) 1.6 (0.4-6.5) .48
Moderate (140-200 mg/dL) 7 (24) 3 (10) 2.5 (0.6-11.9) .17
Severe (>200 mg/dL) 2 (6) 0 (0) N/A N/A
Need of insulin 1 (3) 0 (0) N/A N/A
Upper gastrointestinal bleeding 0 (0) 0 (0) N/A N/A
Anemia 10 (34) 16 (55) 0.4 (0.1-1.2) .11
Mild (DHb <1 g/L from day 1 to day 3) 6 (20) 7 (24) 0.8 (0.2-2.8) .75
Moderate (DHb 1-3 g/L from day 1 to day 3) 3 (10) 3 (3) 1 (0.1-5.4) 1
Severe (DHb >3 g/L from day 1 to day 3) 1 (3) 6 (20) 0.1 (0.01-1.2) .07
Transfusion 1 (3) 3 (10) 0.3 (0.03-3.1) .32
Oropharyngeal candidiasis 0 (0) 0 (0) N/A N/A
Allergic reaction, rash 0 (0) 1 (3) N/A N/A
Other
Local pain, agitation 1 (3) 0 (0) N/A N/A
Simple effusion
Children with simple effusion who eventually underwent pleural drainage 1/18 (5) 3/18 (16) 0.2 (0.03-3.1) .31

N/A, comparison not possible.

MD
PhD 26 13 21
21 13 21 13
26 18
Lorena Pérez, BS

25 27
211328 19
2113
, Iván Bermejo, MD
, Alicia Fernández, PhD
13 1321 13 2114
28 262128
, María Luisa Herreros, MD

20
1522 1416 2317
, Miriam Guizar, MD
Hospital Universitario Gregorio Marañón, Madrid;
27
Intensive Care Unit, Hospital Universitario La Paz, Madrid;
, Julia Yebra, MD

21
, Jana Rizo, MD
, Diana Gutierrez, MD
, José Carmelo Albillos, MD, PhD
, Ana Barrios, MD
, Castro Codesal, MD
, Martina Steiner, MD
Pediatrics Department, Hospital Universitario de Getafe, Madrid;
, David Sanz
, Alfonso Cañete, MD, PhD
, Francisco Ramos, MD
, and Israel Thuissard, BS
, Lisette Arguinzoniz, MD
, Carmen Izquierdo, MD
From the
, Francisco Gaya, PhD
, Peña Gomez-Herruz, MD
, Carmen Vázquez, MD
, María Isabel González-Tomé, MD, PhD
, Cristina Ots, MD
Pediatrics Department, Hospital Universitario Infanta Sofía, San Sebastián de los Reyes, Madrid;
Pediatrics Department, Hospital Universitario de Toledo, Toledo, Castilla-La Mancha;
Pediatric Intensive Care Unit, Hospital Universitario Toledo, Toledo, Castilla-La Mancha;
, Mar Santos, MD
, César Pérez-Caballero, MD
Pediatrics Department, Hospital Universitario Carlos Haya, Málaga, Andalucía;
Pharmacy Department, Hospital Universitario Infanta Sofía, San Sebastián de los Reyes, Madrid;
, Elena Álvarez, MD
17 2318
2315
, Jesús Saavedra, MD, PhD

28 22
Biostatistics Unit, Hospital Universitario La Paz, Madrid;
, Sara Guillén, MD, PhD
, José Luis Vázquez, MD, PhD , Cristina Verdú, MD
18 16 24
24
, Luis Prieto, MD 18 19
23 24
, José T. Ramos, MD, PhD
, Ana Gómez-Zamora, MD
Pediatrics Department, Hospital Universitario Ramón y Cajal, Madrid;
Pediatrics Department, Hospital Universitario Príncipe de Asturias, Madrid;
Radiology Department, Hospital Universitario Infanta Sofía, San Sebastián de los Reyes, Madrid, Spain; and
, Carlos Vela, MD
, Juan-José Menéndez, MD
, Alicia Berghezan, MD19 24 172025
24
, Cristina Schuffelmann, MD
, Antonio Conejo, MD
, Raúl Borrego, MD
Pediatrics, Tropical and Pediatric Infectious Diseases Unit, Hospital Universitario La Paz, Madrid;
Pediatric Infectious Diseases Unit, Hospital Universitario 12 de Octubre, Madrid;
Biomedical School, Universidad Europea de Madrid, Madrid
, Patricia Paredes,
, Jesús Llorente,
Pediatric Infectious Diseases Unit,
Pediatric Intensive Care Unit, Hospital Universitario Ramón y Cajal, Madrid; Pediatric

Table XI. Time to recovery (primary endpoint) in the on-

treatment analysis
Characteristics Placebo DXM Difference
Total
No. of patients 29 29
Duration, h, median (95% CI) 179 (134-223) 116 (18-213) 63
Simple effusion
No. of patients 17 17
Duration, h, median (95% CI) 133 (69-196) 48 (36-59) 85
Complicated effusion
No. of patients 11 12
Duration, h, median (95% CI) 240 (129-350) 226 (187-265) 14

Dexamethasone for Parapneumonic Pleural Effusion: A Randomized, Double-Blind, Clinical Trial 123.e6