C ommunity-acquired pneumonia (CAP) is one of the most com- younger patients.9 For this reason, we present a study based on a
mon infections in children and continues to be one of the lead- modified PIRO scale to be used on admission for stratification of
ing causes of hospital admission worldwide.1–3 children with CAP into mortality-risk group.
Mortality associated with pneumonia varies around the
world, with an annual incidence of 3.0–3.6 cases per 100 inhabit- MATERIALS AND METHODS
ants in developed countries and 7–40 cases per 100 in developing
countries.4 Even though the associated death rate has fallen in the Study Site
decades since the introduction of penicillin and vaccines (for mea- The study was conducted in the Department of Pediatrics
sles, Bordetella pertusis, Streptococcus pneumoniae and Haemo- of the Instituto de Medicina Tropical in Asuncion, Paraguay—a
philus influenzae type b), it is estimated that pneumonia and other national reference center for infectious and tropical diseases in the
associated conditions can cause approximately 18% of deaths of country.
children younger than 5 years worldwide.1–4
Study Design
This was a retrospective observational study based on a
Accepted for publication September 10, 2015. review of medical records of patients diagnosed on admission with
From the *Department of Pediatrics, Institute of Tropical Medicine, Avda. Ven-
ezuela y Florida, Asunción, Paraguay; and †Facultad de Ciencias Médicas, CAP and hospitalized in the department between January 2004 and
Universidad Nacional de Asunción, Av. Mcal. Francisco Solano Lopez, San June 2013. Only those patients younger than 15 years were included.
Lorenzo, Paraguay. Cases of CAP were defined by the presence of respiratory signs and
The authors have no funding or conflicts of interest to disclose. symptoms and a history of fever with the presence of consolida-
Presented at the 54th Interscience Conference on Antimicrobial Agents and Che-
motherapy; September 5–9, 2014; Washington, DC. Abstract G-298. tion in chest radiography (CXR) at admission in patients who had
Address for correspondence: Antonio Arbo, MD, Department of Pediatric, not been hospitalized within the previous 30 days. Patients with
Instituto de Medicina Tropical, Avda. Venezuela y Florida, Asunción,
respiratory infections of other types such as pulmonary tubercu-
Paraguay. E-mail: antonioarbo@hotmail.com.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
losis, bronchiolitis, laryngitis and laryngotracheobronchitis were
ISSN: 0891-3668/16/3504-0369 excluded. Patients were also excluded when diagnosis of pneumonia
DOI: 10.1097/INF.0000000000001018 was done more than 48 hours after hospitalization, in cases when
The Pediatric Infectious Disease Journal • Volume 35, Number 4, April 2016 www.pidj.com | 369
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Araya et al The Pediatric Infectious Disease Journal • Volume 35, Number 4, April 2016
radiographically diagnosed pneumonia resolved within 72 hours, or base,5 with modifications to include as variables within P: age <6
when a diagnosis of pneumonia was not confirmed by CXR. months, grade II or III comorbidity (malnutrition, gastroesopha-
geal reflux, asthma, cardiopathy, Down syndrome and congenital
Variables Analyzed or acquired immunodeficiency); within I: hypoxia (O2 saturation
Demographic variables, the presence of comorbidities, <90), hypotension (for age) and bacteremia; within R: multilobar
clinical signs and symptoms and lab and radiographic studies pneumonia, complicated pneumonia (pneumothorax, bullae or
were all included in the study, with outcome (survival or death) empyema) and O in case of presence of multiple organ dysfunction
treated as an independent variable. Among the demographic vari- (acute renal failure, acute respiratory distress syndrome and acute
ables were included age, sex and day-care attendance. Comor- liver failure). A point was assigned to each variable presented by the
bidities were classified by severity as grades I, II or III. Grade patient, creating thereby a scale of 0–10 to correlate with mortality.
I comorbidities included being overweight or obese and mild Discrimination of PIRO score for mortality of CAP was assessed
malnutrition or at risk of malnutrition (Z score < –1)10; grade using the area under the receiver operating characteristic (ROC)
II included congenital heart disease, asthma, Down syndrome curve. The ROC curve was plotted for each score using sensitivity
and moderate malnutrition (Z score between −2 and −3); grade and specificity values for true prediction of the mortality across the
III comorbidities included predisposing conditions such as HIV whole range of possible cut-off of predictive mortality. The area
infection, immunosuppressant treatment (for prednisone a dose under the ROC curve ranges from 0.5 to 1, with higher values indi-
of 2 mg/kg/day for more than 15 days), cancer or severe mal- cating better discrimination. The association between the modified
nutrition (Z score < –3). Malnutrition grade was determined by PIRO and mortality was studied also by stratifying patients into
weight–age ratio in patients younger than 2 years, whereas deter- different mortality-risk groups (0–2 points, 3–4 points, 5–6 points
mined by weight–height ratio in patients older than 2 years.10,11 and 7–10 points). The data were analyzed using SPSS version 11
Among those clinical variables recorded, the presence of tachyp- for Windows (SPSS Inc., Chicago, IL).
nea, hypotension, shock, consciousness level, hypoxemia and
renal or liver failure was considered. Tachypnea was defined as a RESULTS
respiratory rate >60/min in patients younger than 2 months, >50 During the study period, 860 patients younger than 15 years
breaths/min for patients aged 2–12 months and >40 breaths/min met the inclusion criteria for the study. The mean age of patients
for patients older than 12 months; hypotension was defined as a was 2.8 ± 3 years, with a range of 35 days to 15 years. One hundred
mean arterial pressure under the fifth percentile for age or less ten patients (13%) were younger than 6 months of age, 627 (73%)
than 2 standard deviation (SD) for age; shock was defined by were cases between 6 months to 5 years of age, and 123 (14%)
the requirement for antihypotensive agents for more than 4 hours cases were older than 5 years. There was no predominance of sex
after adequate fluid replacement to achieve appropriate arterial (56% males; 44% females).
pressure; impairment of consciousness was defined as Glasgow Classical symptoms of pneumonia such as fever, respira-
score ≤13 upon admission; hypoxaemia needing oxygen deliv- tory symptoms and taquipnea were observed in 79.7%, 100% and
ery was defined as arterial oxygen saturation (SpO2) less than 57.6% of cases, respectively, and 112 patients (13%) had hypox-
90% found by pulse oximetry measured while the patient was emia at admission. Underlying illnesses or coexisting conditions
breathing room air for 15 minutes; renal failure was deemed as were present in 62% of the cases (534/860). The most common
urine production <0.6 mL/kg/h with increased urea and serum underlying illnesses were genetics disorders (mainly Down syn-
creatinine. Liver failure was defined as the presence of clinical drome), congenital heart diseases, HIV infection, gastroesophageal
findings secondary to liver failure with signs of encephalopa- reflux and malnutrition (Table 1). The microorganism responsible
thy, bleeding disorders, metabolic disorders and jaundice with of the pneumonia was identified in 141 cases (16%), being the most
transaminase >5 times normal value, increased direct bilirubin common pathogens detected S. pneumoniae (in 10.4% of patients)
and altered prothrombin time. and Staphylococcus aureus (in 5.8% of patients). A total of 102
Laboratory parameters included were a complete blood patients (12%) required admission to the ICU, 77 (8.9%) required
count and blood sodium and potassium—both at admission and invasive mechanical ventilation and 56 (6.5%) died during hospi-
on day 3 of hospitalization; arterial blood gas (pH, arterial PaO2 talization (Table 1).
and PaCO2, and bicarbonate); coagulation time; liver and kidney Analysis of the demographic and clinical findings showed
function and blood culture. Bacterial pneumonia was diagnosed by that an age younger than 6 months (OR: 2.2; 95% CI: 1.14–4.2;
isolation of a respiratory pathogen in blood culture. Pneumonia was P = 0.018), presence of grades II comorbidity (OR: 4.9; 95%
defined as lobar upon radiographic detection of consolidation in 1 CI: 2.8–8.5; P < 0.001) or grade III comorbidity (OR: 6.18; 95%
lobe and multilobar when consolidation was present in more than CI: 3.4–10.9; P < 0.001), apyrexia (OR: 2.3; 95% CI: 1.3–4.1;
1 pulmonary lobe. Complicated pneumonia was defined as multi- P 0.003), tachypnea (OR: 3.2; 95% CI: 1.6–6.3; P < 0.0001), hypo-
lobar involvement on CXR or the existence of pleural empyema, tension (OR: 48.7; 95% CI: 24.8–95.6; P < 0.0001), seizures (OR:
pneumatocele or pyopneumothorax. 10.6; 95% CI: 4.9–23; P < 0.0001) or Glasgow score ≤13 (OR: 324;
95% CI: 131–805; P < 0.0001) upon admission were associated
Statistical Analysis with greater mortality (Table 2).
Discrete variables are described as counts (%) and con- When laboratory and CXR findings were assessed, it was
tinuous variables as the mean with SD. For comparative analysis, observed that a white blood cell count <4000/mm3 (OR: 6.5; 95%
χ2 test (proportion testing), Student’s t test (parametric variables) CI: 2.7–15.6; P < 0.0001), pH ≤7.2 (OR: 77; 95% CI: 35–171;
and Mann–Whitney U test (nonparametric variables) were used. P < 0.0001), HCO3 ≤15 (OR: 26.7; 95% CI: 13–54; P < 0.0001), O2
Regression analysis was done to study the relationship of the vari- saturation <90% (OR: 11; 95% CI: 6.2–19.6; P < 0.0001), multilo-
ables with mortality and odds ratio (OR) calculated with a confi- bar involvement on CXR (OR: 4.9; 95% CI: 2.7–8.9; P < 0.0001),
dence interval (CI) of 95%. Following analysis of the variables in pleural empyema (OR: 2.6; 95% CI: 1.1–6.2; P < 0.05) or pneu-
the study population, significant variables (P < 0.05) were consid- mothorax (OR: 15; 95% CI: 2.9–76.6; P < 0.01) were significantly
ered, and a point scale of risk was created that classified outcomes associated with death (Table 3). In the course of treatment, risk
as patients surviving or deceased. The PIRO scale was used as a of death was also associated with need for mechanically assisted
370 | www.pidj.com © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
The Pediatric Infectious Disease Journal • Volume 35, Number 4, April 2016 Pneumonia Severity Score
Survivors, Deceased,
Variable N = 804 N = 56 OR 95% CI P
Age groups
<6 mo 97 (12) 13 (23) 2.2 1.14–4.2 0.018
7 mo to 5 yr 560 (70) 25 (45) 0.36 0.2–0.6 <0.001
>5 to 15 yr 147 (18) 18 (32) 2.1 1.17–3.8 <0.05
Comorbidity*
Grade I 326 (41) 1 (1.8) 0.02 0.003–0.19 <0.001
Grade II 144 (18) 29 (52) 4.9 2.8–8.5 <0.001
Grade III 87 (11) 24 (43) 6.18 3.4–10.9 <0.001
Clinical variables
Apyrexia 155 (19) 20 (36) 2.3 1.3–4.1 0.003
Tachypnea 450 (56) 45 (80) 3.2 1.6–6.3 <0.001
Hypotension 23 (2.8) 33 (59) 48.7 24.8–95.6 <0.0001
Seizures 20 (2.5) 12 (21) 10.6 4.9–23 <0.0001
Glasgow ≤13 10 (1.2) 45 (80) 324 131–805 <0.0001
*Grade: I (overweight and risk of malnutrition); grade II (congenital heart disease, Down syndrome, moder-
ate malnutrition and gastroesophageal reflux); grade III (HIV, hematological cancer and severe malnutrition).
© 2015 Wolters Kluwer Health, Inc. All rights reserved. www.pidj.com | 371
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Araya et al The Pediatric Infectious Disease Journal • Volume 35, Number 4, April 2016
Survivors, Deceased,
Variable N = 804 N = 56 OR 95% CI P
Laboratory variables
Leukopenia (<4000/mm3) 20 (2.5) 8 (14) 6.5 2.7–15.6 <0.0001
Leukocytosis (>15,000/mm3) 349 (43) 6 (10.7) 0.15 0.06–0.36 <0.0001
pH < 7.2 11 (1.4) 29 (51.7) 77 35–171 <0.0001
HCO3<15 19 (2.4) 22 (39) 26.7 13–54 <0.0001
Saturated O2 <90 81 (10) 31 (55) 11 6.2–19.6 <0.0001
CXR finding
Unilobar consolidation 221 (27.5) 20 (36) 1.5 0.8–2.6 0.18
Patchy consolidation 466 (58) 7 (12.5) 0.1 0.04–0.23 <0.0001
Multilobar consolidation 76 (9.5) 19 (34) 4.9 2.7–8.9 <0.0001
Pleural effusion 41 (5) 7 (12.5) 2.6 1.1–6.2 <0.05
Pneumothorax 3 (0.4) 3 (5) 15 2.9–76.6 <0.01
Others variables
Mechanical ventilation 35 (4.4) 42 (75) 66 32.9–132 <0.0001
Hepatic impairment 8 (1) 9 (16) 19 7–51.6 <0.0001
Renal impairment 10 (1.2) 9 (16) 15.2 5.8–39 <0.0001
The PSI is a prognostic model that calculates a severity-of- in intensive care. In the same line, the Pediatrics Infectious Soci-
illness score based on 20 separate characteristics (demographic, ety and Infectious Disease Society of America recommend almost
clinical and laboratory) of the patients, including underlying similar criteria for hospitalization of children with CAP in ICU.14
comorbidities.6 The SMARTCOP scale was proposed by Australian Although several point scale are being used to classify adult
researchers. Its name is derived from the components of the scale, patients with pneumonia on admission, scale that can be used
with the S referring to systemic arterial pressure, the M to multi- in children with pneumonia are scarce. The respiratory index of
lobar involvement on radiography, A to albumin, R to respiratory severity in children (RISC) scale was published by Reed et al.15 The
rate, T for tachycardia, C to confusion (altered mental state), O to RISC scale predicts mortality in children younger than 2 years with
oxygenation and P to arterial pH.7 The value assigned to the find- respiratory tract infections, using 2 models for patients accord-
ings held in common, arterial pressure, oxygen saturation, arterial ing to the presence or absence of HIV. The variables considered in
pH and radiographic finding of involvement is indisputable. HIV-negative patients were O2 saturation <90%, presence of chest
Another greatly simplified scale, CURB-65, has been pro- indrawing, presence of wheezing, refusal to feed and weight–age Z
posed for achieving a prognosis for adults with CAP, with C refer- score. In HIV-positive patients, all the above were included, as well
ring to confusion, U to urea, R to respiratory rate and B to blood as the HIV clinical classification. They showed that a simple com-
pressure, while ‘65’ is the point at which age itself becomes a risk bination of these factors in a scoring model provides good discrimi-
factor.8 Recently, CRB-65, which drops the requirement for deter- nation and calibration for predicting mortality in young children
mining blood urea replaced CURB-65.12 with pneumonia.15 However, the RISC score study included partici-
A joint statement by the Infectious Disease Society of pants with bronchiolitis as well as children in whom the diagnosis
America and American Thoracic Society13 recommended for adults of pneumonia was done based only on clinical criteria (only 16.5%
the use of different clinical, radiographic and laboratory criteria, of the patients had alveolar consolidation on CXR).15
along with initial treatment requirements ordered by importance In this study, we have used a new scale to predict mortal-
according to the severity of findings, to guide the clinical physician ity in children with CAP. The variables used in the creation of
in determining a prognosis and need for placement of the patient the scale is an adaptation of the PIRO scale previously used for
prognostic stratification of adult patients with pneumonia.5 This
scale includes aspects of host response to the infection as well as
predisposing conditions, in recognition that immunodeficient or
chronically ill children represent a significant proportion of the
real population.
The mean age of patients in our series is similar to those
reported in other studies of CAP in children, with a predominance
of patients younger than 5 years. The high mortality (6.5%) found
in our study series may have been influenced by the high proportion
diagnosed as immunodeficient, primarily secondary to HIV infec-
tion and by the fact that our institution is a referral center.
Of the patients who died, approximately 60% showed hypo-
tension on admission, 80% required mechanically assisted venti-
lation and one-third showed multilobar involvement in radiog-
raphy. The presence of significant (grades II or III) comorbidity
and age younger than 6 months were statistically associated with
death. These findings are not surprising because conditions such
as malnutrition, cancer and HIV profoundly impact the innate and
FIGURE 1. ROC curve and discrimination of PIRO score for adaptive immune response of the host while patients in the earliest
mortality of CAP in children. stages of life have not yet achieved full development.
372 | www.pidj.com © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
The Pediatric Infectious Disease Journal • Volume 35, Number 4, April 2016 Pneumonia Severity Score
The point scale we present strongly correlates with the underlying risk factors and causative pathogens for 192 countries. J Global
mortality rates in CAP and helps to discriminate the probability Health. 2013;3:10401.
of death in children hospitalized with pneumonia. Its additional 5. Rello J, Rodriguez A, Lisboa T, et al. PIRO score for community-acquired
pneumonia: a new prediction rule for assessment of severity in intensive
advantages are simplicity and absence of sophisticated laboratory care unit patients with community-acquired pneumonia. Crit Care Med.
determinations. Although any patient with a score of 2 or less died, 2009;37:456–462.
it was noted that to the extent that increased the score on the scale, 6. Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to identify low-
at the same time increased mortality. So, in patients with a modi- risk patients with community-acquired pneumonia. N Engl J Med.
fied PIRO score of 3–4, the mortality rate was 18%, increasing sig- 1997;336:243–250.
nificantly to 83% in patients with a score of 5–6 and to 100% in 7. Charles PG, Wolfe R, Whitby M, et al; Australian Community-Acquired
patients with a modified PIRO score of 7 or higher. Pneumonia Study Collaboration. SMART-COP: a tool for predicting
the need for intensive respiratory or vasopressor support in community-
Our study has a number of limitations, including its ret- acquired pneumonia. Clin Infect Dis. 2008;47:375–384.
rospective design and the fact that data were collected from a 8. Lim WS, van der Eerden MM, Laing R, et al. Defining community acquired
referral center. Thus, the results may not be applicable to com- pneumonia severity on presentation to hospital: an international derivation
munity healthcare settings. All selected variables were chosen and validation study. Thorax. 2003;58:377–382.
arbitrary, although all of them were significantly associated with 9. McCracken GH Jr. Diagnosis and management of pneumonia in children.
mortality. Several other variables (for instance, seizure and age Pediatr Infect Dis J. 2000;19:924–928.
older than 5 years) that were significantly associated with mor- 10. WHO Multicentre Growth Reference Study Group. WHO Child
tality were not included because when they were taken in, we Growth Standards: Length/Height-for-age, Weight-for-age, Weight-for-
length, Weight-for-height and Body Mass Index-for-age: Methods and
did not observe an improvement of the discriminatory power in Development. Geneva, World Health Organization; 2006. Available at:
the PIRO score. http://www.who.int/childgrowth/standards/technical_report/en/index.html.
Prospective studies applying the present modified PIRO Accessed January 6, 2016.
scale in children with CAP are needed to confirm these results. The 11. Nilesh M. Mehta NM, Corkins MR, et al. Defining pediatric malnutrition:
broad predictive value of mortality of the scale presented here could paradigm shift toward etiology-related definitions. JPEN J Parent Enter
Nutr. 2013;37:460–481.
provide a useful tool for identification of patients with a greater risk
12. Macfarlane JT, Boldy D. 2004 update of BTS pneumonia guidelines: what’s
of death and guide a more effective early treatment. new? Thorax. 2004;59:364–366.
13. Mandell LA, Wunderink RG, Anzueto A, et al; Infectious Diseases Society
REFERENCES of America; American Thoracic Society. Infectious Diseases Society of
America/American Thoracic Society consensus guidelines on the manage-
1. Rudan I, Tomaskovic L, Boschi-Pinto C, et al; WHO Child Health
ment of community-acquired pneumonia in adults. Clin Infect Dis. 2007;44
Epidemiology Reference Group. Global estimate of the incidence of clini-
(Suppl 2):S27–S72.
cal pneumonia among children under five years of age. Bull World Health
Organ. 2004;82:895–903. 14. Bradley JS, Byington CL, Shah SS, et al. Executive summary: the man-
agement of community-acquired pneumonia in infants and children older
2. Rudan I, Boschi-Pinto C, Biloglav Z, et al. Epidemiology and etiology of than 3 months of age: clinical practice guidelines by the Pediatric Infectious
childhood pneumonia. Bull World Health Organ. 2008;86:408–416. Diseases Society and the Infectious Diseases Society of America. Clin Infect
3. Walker CL, Rudan I, Liu L, et al. Global burden of childhood pneumonia Dis. 2011;53:617–630.
and diarrhoea. Lancet. 2013;381:1405–1416. 15. Reed C, Madhi SA, Klugman KP, et al. Development of the Respiratory
4. Rudan I, O’Brien KL, Nair H, et al. Epidemiology and etiology of childhood Index of Severity in Children (RISC) Score among young children with res-
pneumonia in 2010: estimates of incidence, severe morbidity, mortality, piratory infections in South Africa. PLoS One. 2012;7:e27793.
© 2015 Wolters Kluwer Health, Inc. All rights reserved. www.pidj.com | 373
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.