National Hospitalization Trends for Pediatric

Pneumonia and Associated Complications

WHATS KNOWN ON THIS SUBJECT: The incidence of IPD has AUTHORS: Grace E. Lee, MD,a,b,c Scott A. Lorch, MD,
decreased since widespread use of PCV7. Postlicensure MSCE,c,d,e,f Seth Shefer-Collins, MPH,b Matthew P.
epidemiological studies revealed conicting data about changes Kronman, MD,b,f and Samir S. Shah, MD, MSCEa,b,c,e,f
in CAP and CAP-associated complications. Previous studies Divisions of aGeneral Pediatrics, bInfectious Diseases, and
dNeonataology, Childrens Hospital of Philadelphia, Philadelphia,
focused on empyema and typically did not include school-aged
Pennsylvania; and Departments ofcPediatrics and eBiostatistics
children or adolescents. and Epidemiology and fCenter for Clinical Epidemiology and
Biostatistics, School of Medicine, University of Pennsylvania,
WHAT THIS STUDY ADDS: With nationally representative data, we Philadelphia, Pennsylvania
found that rates of CAP discharges remained relatively KEY WORDS
unchanged overall. Systemic complication rates were greatest pneumonia, pleural empyema, epidemiology, heptavalent
for infants. Local complication rates increased in all age groups pneumococcal conjugate vaccine
and were highest among 1- to 5-year-old children. ABBREVIATIONS
CAP community-acquired pneumonia
PCV7 heptavalent pneumococcal conjugate vaccine
PPV2323-valent polysaccharide pneumococcal vaccine
IPDinvasive pneumococcal disease

abstract KIDKids Inpatient Database

ICD-9-CMInternational Classication of Diseases, Ninth
Revision, Clinical Modication
OBJECTIVE: To determine current rates of and trends in hospitaliza- CI condence interval
tions for community-acquired pneumonia (CAP) and CAP-associated The content of this article is solely the responsibility of the
complications among children. authors and does not necessarily represent the ofcial views of
METHODS: We performed a cross-sectional, retrospective, cohort the National Institutes of Health.

study by using the 1997, 2000, 2003, and 2006 Kids Inpatient Database. www.pediatrics.org/cgi/doi/10.1542/peds.2009-3109

National estimates for CAP and CAP-associated local and systemic com- doi:10.1542/peds.2009-3109
plication rates were calculated for children 18 years of age. Patients Accepted for publication May 11, 2010
with comorbid conditions or in-hospital birth status were excluded. Address correspondence to Samir S. Shah, MD, MSCE, Childrens
Percentage changes were calculated by using 1997 (before heptava- Hospital of Philadelphia, Division of Infectious Diseases, Room
1526, North Campus, 34th Street and Civic Center Boulevard,
lent pneumococcal conjugate vaccine [PCV7]) and 2006 (after PCV7) Philadelphia, PA 19104. E-mail: shahs@email.chop.edu
data. PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).
RESULTS: There were a total of 619 102 CAP discharges for 1997, 2000, Copyright 2010 by the American Academy of Pediatrics
2003, and 2006, after application of inclusion and exclusion criteria. FINANCIAL DISCLOSURE: The authors have indicated they have
Overall rates of CAP discharges did not change substantially between no nancial relationships relevant to this article to disclose.
1997 and 2006, but stratication according to age revealed a 22% de- Funded by the National Institutes of Health (NIH).
crease for children 1 year of age, minimal change for children 1 to 5
years of age, and increases for children 6 to 12 years (22%) and 13
years (41%) of age. Systemic complication rates were highest among
children 1 year of age but decreased by 36%. In all other age groups,
systemic complication rates remained stable. Local complication rates
increased 78% overall. Children 1 to 5 years of age had the highest local
complication rates.
CONCLUSIONS: After the introduction of PCV7 in 2000, rates of CAP-
associated systemic complications decreased only for children 1
year of age. Rates of pediatric CAP-associated local complications are
increasing in all age groups. Pediatrics 2010;126:204213

204 LEE et al
 ARTICLES

Streptococcus pneumoniae is the studies evaluating the impact of PCV7 talization because of CAP in 1997, 2000,
most-commonly identied bacterial introduction on the severity of illness 2003, or 2006.
cause of community-acquired pneu- in children hospitalized with CAP. We
monia (CAP) in children. A heptavalent conducted a retrospective cross- Denition of Pneumonia
pneumococcal conjugate vaccine sectional study by using a national da- By using a previously validated algo-
(PCV7) was licensed in the United tabase to determine the rates of rithm, patients were considered to
States in February 2000 and subse- hospitalizations with CAP and CAP- have CAP if they met 1 of 2 criteria, that
quently was added to the routine child- associated complications in otherwise is, (1) an International Classication of
hood vaccination schedule. Since then, healthy children in the United States Diseases, Ninth Revision, Clinical Mod-
overall rates of invasive pneumococcal and to describe changes in rates, if ication (ICD-9-CM), primary diagnosis
disease (IPD) (ie, bacteremia and men- any, after the introduction of PCV7. code indicating pneumonia (codes
ingitis) have decreased for both chil- 480 483 and 485 486), empyema
dren17 and adults,810 largely because METHODS (code 510), or pleurisy (codes 511.0,
of signicant reductions in the burden 511.1, and 511.9) or (2) a primary diag-
Study Design and Data Source
of disease caused by vaccine serotype nosis of a pneumonia-related symp-
isolates. However, reductions in the in- We performed a cross-sectional analy- tom (eg, cough, fever, or tachypnea)
cidence of pediatric CAP seem to be sis of pediatric hospitalizations in the (see Appendix 1 for ICD-9-CM codes)
less dramatic and have been limited to United States by using the 1997, 2000, and a secondary diagnosis of pneumo-
young children. In prelicensure, ran- 2003, and 2006 Kids Inpatient Data- nia, empyema, or pleurisy.22
domized, controlled trials, the risk of base (KID). The KID is part of the Health-
care Cost and Utilization Project spon- Exclusion Criteria
radiographically conrmed pneumo-
nia was 20% lower for PCV7 recipi- sored by the Agency for Healthcare Patients with the following comorbid
ents 2 years of age, compared with Research and Quality. It is the only data conditions were excluded, because
nonrecipients.11 Postlicensure epide- set on hospital use and outcomes that these comorbidities are characterized
miological studies revealed decreases was designed specically to study chil- by risk factors not reective of the gen-
in all-cause pneumonia incidence drens use of hospital services in the eral pediatric population: acquired or
United States The KID samples pediat- congenital immunologic disorders,
rates of 39% to 52% among children
ric discharges from all community malignancies, collagen vascular dis-
2 years of age12,13 but no changes for
nonrehabilitation hospitals (including ease, sickle cell disease, cystic bro-
older children.13,14
academic medical centers) in states sis, organ transplant, congenital heart
The impact of PCV7 vaccination on the defects, and heart failure (Appendix 1).
participating in the Healthcare Cost
severity of pediatric CAP is less clear. Cases identied as in-hospital births
and Utilization Project, across pediat-
Although vaccination with PCV7 has re- were excluded, to minimize the inclu-
ric discharge type and hospital char-
duced the incidence of IPD in children, sion of perinatally acquired and neo-
acteristics, by using a complex strati-
several authors reported regional in- natal nosocomial infections. Patients
cation system. Discharge-level weights
creases in rates of pediatric empyema, with a secondary diagnosis code indi-
assigned to discharges within the stra-
a CAP-associated complication, after cating trauma also were excluded, be-
tum permit calculations of national es-
widespread PCV7 uptake.1517 Studies
timates. Each data set contains 3 cause a diagnosis of pneumonia in this
examining national trends in population likely reects a nosocomial
million discharges (unweighted), and
pneumonia-associated complications cause (Appendix 1). CAP-associated
data sets have been released every 3
also focused solely on empyema18,19 complications were identied by using
years, beginning in 1997. The 2006 KID
and were limited to infants and ICD-9-CM diagnosis and procedure
is the most-recent data set available
preschool-aged children.18 Among codes (Appendix 2). Complications
and contains hospital administrative
adults hospitalized with CAP, previous were classied as local (empyema,
data from 38 states, representing
recipients of a 23-valent polysaccha- lung abscess, necrotizing pneumonia,
88.8% of the estimated US population.21
ride pneumococcal vaccine (PPV23) or bronchopleural stula), systemic
had lower all-cause mortality rates Study Participants (acute respiratory failure, sepsis, ex-
and risks of respiratory failure, sepsis tracorporeal membrane oxygenation,
syndrome, and cardiac arrest, com- Inclusion Criteria or hemolytic uremic syndrome), or
pared with vaccine nonrecipients.20 To Patients 18 years of age were eligi- metastatic (meningitis, central ner-
our knowledge, there have been no ble for inclusion if they required hospi- vous system abscess, mastoiditis,

PEDIATRICS Volume 126, Number 2, August 2010 205

TABLE 1 Characteristics of Patients With CAP Discharges in 19972006 (Stata Corp, College Station, TX). This
Year n (%) study was considered exempt from the
1997 (N 148 702) 2000 (N 157 847) 2003 (N 157 743) 2006 (N 156 810) need for review by the institutional re-
Race view board of the Childrens Hospital of
White 56 348 (38) 68 643 (44) 54 903 (35) 56 108 (36) Philadelphia.
Black 22 864 (15) 22 580 (14) 17 960 (11) 18 800 (12)
Other 22 203 (15) 38 448 (24) 39 138 (25) 40 803 (26)
Missing data 47 287 (32) 28 175 (18) 45 588 (29) 41 099 (26) RESULTS
Age
1 y 43 851 (29) 44 470 (28) 37 798 (24) 37 705 (24) Study Sample
15 y 75 033 (50) 76 385 (48) 77 530 (49) 79 519 (51) The 1997, 2000, 2003, and 2006 KID con-
612 y 19 372 (13) 21 403 (14) 23 126 (15) 23 494 (15)
1318 y 10 446 (7) 15 589 (9) 19 289 (12) 16 092 (10) tained a total of 28 916 332 weighted
Male 83 291 (56) 87 891 (56) 86 034 (55) 85 508 (55) pediatric discharges. Of those, 619 102
Died 334 (0.22) 394 (0.25) 270 (0.17) 193 (0.12) cases (2.1%) remained after inclusion
and exclusion criteria for CAP were
applied.
pericarditis, endocarditis, osteomyeli- race data were included as a separate
tis, or septic arthritis). variable, to preserve the integrity of Subject Characteristics
our estimates. Ratios of CAP discharge Table 1 presents subject character-
Statistical Analyses
and CAP-associated complication istics according to year. Age cate-
Cases were characterized on the basis rates for black and white study partic- gory proportions remained stable
of age, race, gender, presence and ipants were calculated by using same- over time, with children 1 to 5 years
type of complication, and discharge year data. of age representing the largest age
status (in-hospital death). Analyses
Categorical variable results were sum- group. There was a slight predomi-
were subsequently stratied accord-
marized as frequencies and propor- nance of male participants in all
ing to age and race. Age groups were
years. White children composed the
dened as 1 year (infant), 1 to 5 tions. Rate calculations were performed
by using weighted observations as nu- largest represented racial group
years (preschool age), 6 to 12 years
(35% 44%) in all years; race data
(school age), or 13 years (adoles- merators and annual, age-specic popu-
lation estimates obtained from the US were missing for 18% to 32% of dis-
cent), to capture differences in CAP
Census Bureau as denominators.23 charges. Mortality rates were low at
and CAP-associated complication
Point estimates with 95% condence in- 0.25% for all pediatric CAP dis-
rates related to age-inuenced factors
charges in each sample year.
such as PCV7 vaccination status, CAP tervals (CIs) were calculated by using
microbiologic cause, and age-based Taylor series estimates. Percentage
Rates of CAP
sociological factors. changes for CAP and CAP-associated
Race was recorded as a single variable complications were calculated by us- Overall
(white, black, other, or missing data). ing 1997 (pre-PCV7 period) and 2006 The rates of CAP discharges according
Rate estimates for each race category (post-PCV7 period) rates. Calculations to year are shown in Table 2. Overall
were calculated; cases with missing were performed by using Stata 10 rates of pediatric CAP discharges

TABLE 2 Rates of CAP and Associated Complications in 19972006

1997 2000 2003 2006 Change, 1997
vs 2006, %
CAP discharges, estimate (95% CI), cases per 100 000 199.1 (198.1200.1) 207.6 (206.6208.6) 204.3 (203.1205.3) 201.2 (200.2202.2) 1.1
Any complication
Rate, estimate (95% CI), cases per 100 000 11.8 (11.612.1) 14.6 (14.314.8) 15.8 (15.315.8) 15.1 (14.815.3) 28.0
Proportion of CAP cases, % 5.9 7.0 7.7 7.5
Local complications
Rate, estimate (95% CI), cases per 100 000 5.4 (5.25.6) 7.4 (7.27.6) 8.9 (8.69.0) 9.6 (9.49.9) 77.8
Proportion of CAP cases, % 2.7 3.6 4.4 4.8
Systemic complications
Rate, estimate (95% CI), cases per 100 000 6.8 (6.67.0) 7.7 (7.57.9) 7.5 (7.37.7) 6.2 (6.06.3) 8.8
Proportion of CAP cases, % 3.4 3.7 3.7 3.1
Rates are reported as cases per 100 000 age-specic US population.

206 LEE et al
 ARTICLES

TABLE 3 Rates of CAP in 19972006, Stratied According to Age

Age 1997 2000 2003 2006 Change,
1997 vs
n (%) Rate, Estimate (95% n (%) Rate, Estimate (95% n (%) Rate, Estimate (95% n (%) Rate, Estimate (95%
2006, %
CI), Cases per CI), Cases per CI), Cases per CI), Cases per
100 000 100 000 100 000 100 000
1 y 43 851 (30) 1169.0 (1158.21179.9) 44 691 (28) 1159.1 (1148.41169.8) 37 798 (24) 937.6 (928.2947.0) 37 705 (24) 912.9 (903.8922.1) 21.9
15 y 75 033 (50) 383.1 (380.3385.8) 76 775 (48) 397.7 (394.9400.5) 77 530 (49) 395.6 (392.8398.4) 79 519 (51) 390.4 (387.7393.2) 1.9
612 y 19 372 (13) 69.3 (68.370.2) 21 531 (14) 74.1 (73.175.1) 23 126 (15) 80.9 (79.982.0) 23 494 (15) 84.5 (83.485.6) 21.9
1318 y 10 446 (7) 44.7 (43.845.5) 15 663 (10) 64.7 (63.665.7) 19 289 (12) 77.1 (76.178.2) 16 092 (10) 62.8 (61.963.8) 40.5
Rates are reported as cases per 100 000 age-specic US population.

peaked in 2000 and then returned to According to Race complications increased by 77.8% (5.4
pre-PCV7 levels. Rates of CAP discharges for black chil- and 9.6 cases per 100 000 population,
According to Age dren were greater than those for white respectively). Empyema accounted for
children in all years studied (Table 4). 97% of all local complications. The
The rates of CAP discharges varied in-
However, this difference decreased systemic complication rate decreased
versely with age, with the highest rates
over time, from a rate ratio of 1.98 in by 8.8% (6.8 and 6.2 cases per 100 000,
occurring for children 1 year of age.
The rates of CAP discharges for chil- 1997 to a rate ratio of 1.59 in 2006. respectively) (Table 2). The proportion
dren 1 year of age decreased by of discharges with any associated
Rates of CAP-Associated complication increased from 5.9% to
21.9% between 1997 and 2006, with Complications
90% of the decrease occurring by 2003. 7.5%, whereas the proportion with lo-
There were minimal interval changes Overall cal complications increased from 2.7%
in rates of CAP discharges for children Between 1997 and 2006, the rate of dis- to 4.8%. The proportion with systemic
1 to 5 years of age, whereas rates in- charges with any CAP-associated com- complications remained relatively sta-
creased by 21.9% for children 6 to 12 plication increased by 28% (11.8 and ble at 3.1% to 3.7%. In 1997, 2000, 2003,
years of age and by 40.5% for children 15.1 cases per 100 000 population, re- and 2006, there were an estimated 75,
13 years of age (Table 3). spectively), whereas the rate of local 100, 72, and 98 discharges, respec-

TABLE 4 Rates and Rate Ratios of CAP and CAP-Associated Complications in 19972006, Stratied According to Race
Year 1997 2000 2003 2006 Change,
1997 vs
n (%) Rate, Estimate (95% n (%) Rate, Estimate (95% n (%) Rate, Estimate (95% n (%) Rate, Estimate (95%
2006, %
CI), Cases per CI), Cases per CI), Cases per CI), Cases per
100 000 100 000 100 000 100 000
CAP
White 56 348 (38) 96.0 (95.296.8) 68 975 (43) 115.5 (114.7116.4) 54 903 (35) 91.3 (90.592.1) 56 108 (36) 92.7 (92.093.5) 3.4
Black 22 864 (15) 190.0 (187.5192.5) 22 694 (14) 182.6 (180.3185.0) 17 960 (11) 142.6 (140.6144.7) 18 800 (12) 147.6 (145.5149.8) 22.3
Black/white rate 1.98 (1.942.01) 1.58 (1.561.60) 1.56 (1.541.59) 1.59 (1.561.62)
ratio
Any complication
White 3567 (40) 6.1 (5.96.3) 4757 (43) 8.0 (7.78.2) 4578 (38) 7.6 (7.47.8) 4609 (39) 7.6 (7.47.8) 24.6
Black 1108 (13) 9.2 (8.79.8) 1394 (13) 11.2 (10.611.8) 1207 (10) 9.6 (9.110.1) 1356 (12) 10.7 (10.111.2) 16.3
Black/white rate 1.51 (1.421.62) 1.40 (1.331.49) 1.26 (1.181.34) 1.40 (1.311.49)
ratio
Local complications
White 1831 (46) 3.1 (3.03.3) 2771 (49) 4.6 (4.54.8) 2839 (42) 4.7 (4.64.9) 3100 (41) 5.1 (4.95.3) 64.5
Black 491 (12) 4.1 (3.74.5) 719 (13) 5.8 (5.46.2) 714 (11) 5.7 (5.36.1) 871 (12) 6.8 (6.47.3) 65.9
Black/white rate 1.32 (1.181.45) 1.26 (1.151.35) 1.21 (1.111.30) 1.33 (1.241.44)
ratio
Systemic
complications
White 1854 (37) 3.2 (3.03.3) 2175 (37) 3.6 (3.53.8) 1959 (34) 3.3 (3.13.4) 1750 (36) 2.9 (2.83.0) 9.4
Black 650 (13) 5.4 (5.05.8) 730 (12) 5.9 (5.56.3) 553 (10) 4.4 (4.04.8) 559 (12) 4.4 (4.04.8) 18.5
Black/white rate 1.69 (1.561.87) 1.61 (1.481.75) 1.35 (1.221.48) 1.52 (1.381.67)
ratio
Rates are reported as cases per 100 000 US population 18 years of age.

PEDIATRICS Volume 126, Number 2, August 2010 207

tively, with CAP-associated metastatic complication increased 31.5%, from 35.5% decrease. Rates remained sta-
complications. There were so few met- 19.7 to 25.9 cases per 100 000. Among ble among the other age groups.
astatic complications as to preclude children 6 to 12 years of age, the rate of Rates of CAP-associated local compli-
us from presenting meaningful rates any complication increased 44.4%, cations increased in all age groups. Al-
or subset analyses. from 5.4 to 7.8 cases per 100 000. Chil- though rates of local complications
dren 13 years of age had the largest were highest among children 1 to 5
According to Age
percentage rate increase, with a 67.2% years of age for all years studied, chil-
Between 1997 and 2006, rates of any
CAP-associated complication among
increase between 1997 and 2006 (from dren 13 years of age had the largest
6.1 to 10.2 cases per 100 000). percentage increase in rates over
children 1 year of age decreased
(Fig 1A), whereas rates in all other age Rates of CAP-associated systemic com- time. Among children 1 year of age,
groups increased (Fig 1 BD). The rate plications were highest among chil- rates increased 67.2%, from 6.7 to 11.2
of any CAP-associated complication dren 1 year of age in all years stud- cases per 100 000 (Fig 1A). Among chil-
was highest at each time point for chil- ied, whereas the lowest rates dren 1 to 5 years of age, rates in-
dren 1 year of age; however, this occurred among children 6 to 12 years creased 79.3%, from 9.2 to 16.5 cases
group experienced a 25.5% decrease of age. Between 1997 and 2006, rates of per 100 000 (Fig 1B). Among children 6
between 1997 and 2006 (from 54.9 to systemic complications in children 1 to 12 years of age, rates increased
40.9 cases per 100 000). Among chil- year of age decreased from 49.0 to 31.6 71.4%, from 3.5 to 6.0 cases per
dren 1 to 5 years of age, the rate of any cases per 100 000, representing a 100 000 (Fig 1C). Children 13 years of

FIGURE 1
Rates of hospital discharges for CAP-associated complications in 19972006, stratied according to age. A, 1 year; B, 1 to 5 years; C, 6 to 12 years; D, 13
to 18 years. Open circles indicate any complication, black squares indicate systemic complications, and black circles indicate local complications. Vertical
lines indicate 95% CIs. Rates are cases per 100 000 age-specic US population.

208 LEE et al
 ARTICLES

age experienced an 88.1% increase in associated complications than did ents of PCV7.28 Adult data suggest that
rates between 1997 and 2006 (from 4.2 white children in all years studied. pneumococcal vaccination can modify
to 7.9 cases per 100 000) (Fig 1D). Although CAP-related hospitalization the severity of illness for patients hos-
rates for the entire cohort were stable pitalized with CAP and may reduce the
According to Race occurrence of CAP-associated compli-
overall, there were differences accord-
Rates of any, systemic, and local CAP- ing to age group. Infants were the only cations.20 A plausible mechanism may
associated complications were higher age group to experience a decrease in be the reduction of concomitant pneu-
for black children than for white chil- CAP discharge rates, a nding consis- mococcal bacteremia among PPV23
dren for all years studied (Table 4). Ra- tent with other studies that showed recipients.29 Experimental model stud-
tios of rates for black and white chil- post-PCV7 reductions in all-cause ies showed that cell wall components
dren for any, systemic, and local pneumonia rates for children 2 of killed pneumococci are capable of
complications showed a downward years of age.1113,25 Although previous triggering an inammatory cascade re-
trend between 1997 and 2003, with a sponse in the host, resulting in death.30
postlicensure studies did not show
slight increase from 2003 to 2006. Be-
CAP rate changes for children 2 The reduction of pneumococcal bactere-
tween 1997 and 2006, rates of any CAP- mia may prevent the initiation of these
years of age,13,14 we report that CAP dis-
associated complication increased inammatory processes, reducing the
charge rates increased for children
24.6% for white children and 16.3% for
5 years of age. We might have been severity of illness among patients requir-
black children. Systemic complication ing hospitalization for CAP. Large de-
able to nd a difference in rates for the
rates decreased 18.5% for black chil- creases in rates of IPD, including bacte-
older age groups because of the larger
dren and 9.4% for white children. The remia, occurred among children 2
size of our cohort, compared with pre-
groups experienced similar increases years of age after PCV7 licensure,6,7,15,3135
vious studies.13,14 However, the reason
in rates of local complications (64.5% which possibly explains why, similar to
for the increase in CAP discharges is
and 65.9% for white and black chil- ndings for adult recipients of PPV23,
unclear. Pneumococcal serotype re-
dren, respectively). PCV7 may reduce the frequency of CAP-
placement has been occurring since
DISCUSSION the introduction of PCV71,57,15 and may associated systemic complications.
contribute to the increase in CAP dis- In contrast to trends in CAP-associated
We describe national changes in dis-
charge rates for older children, al- systemic complication rates, rates of
charge rates for pediatric CAP and
though data suggest that serotype re- local complications increased for all
CAP-associated complications in the
placement is more commonly seen age groups, with the highest rates oc-
pre-PCV7 and post-PCV7 periods. Since
among young children and older curring among preschool-aged chil-
the introduction of PCV7 in 2000, up-
adults.5,6,15 It also is possible that dren. In addition, the presence of any
take has been rapid, with 68% of 19- to
changes in the epidemiological fea- CAP-associated complication among
35-month-old children having received
tures of other pathogens, such as children 1 to 18 years of age was
3 doses in 2003 and 87% in 2006.24
We report that, although overall rates methicillin-resistant Staphylococcus largely attributable to local complica-
of CAP discharges remained relatively aureus26,27 or atypical organisms, tions. It is unclear, however, whether
unchanged, rates decreased for chil- rather than changes in rates of IPD, this trend can be attributed fully to the
dren 1 year of age and increased for are responsible for this trend. changing epidemiological features of
children 6 years of age. Overall rates This is the rst national study to exam- IPD after the introduction of PCV7. Two
of systemic complications were dra- ine rates of systemic CAP-associated studies reported increasing regional
matically higher for infants than for complications in the pre-PCV7 and rates of empyema in children before
any other age group, but infants were post-PCV7 eras. Rates of systemic com- PCV7 licensure,27,36 which raises the
the only age group to experience de- plications varied inversely with age, possibility that the current increase in
creases over time in this area. In con- with infants having the highest rates local complication rates is a continua-
trast, local complication rates were and children 6 years of age having tion of a previous trend. Rates of local
found to be increasing in all pediatric the lowest rates. The decrease in sys- complications also may be inuenced by
age groups and were highest among temic complication rates for the entire the increasing prevalence of community-
children 1 to 5 years of age. Race cohort was largely attributable to the acquired methicillin-resistant S aureus,
seemed to play a role, because black decrease in rates for infants and might which has become the pathogen most
children had consistently higher dis- be explained in part by the fact that commonly isolated from empyema in
charge rates of CAP and CAP- infants have been the primary recipi- several centers.26,27 The limitations of

PEDIATRICS Volume 126, Number 2, August 2010 209

standard microbiologic isolation tech- base of discharge-level data, without primary diagnosis. For example, a
niques have made etiologic and inci- clinical information beyond that cap- patient hospitalized after traumatic
dence studies of empyema problem- tured in ICD-9-CM codes. The identica- injury who develops ventilator-
atic. Bacteria are isolated infrequently tion of CAP discharges depends on the associated pneumonia is likely to have
from blood or pleural uid cultures for accuracy of ICD-9-CM coding, and mis- trauma, rather than pneumonia or a
children with empyema17,26,27 and, at coding of cases with CAP and CAP- pneumonia-related symptom, listed as
one center, culture-negative empyema associated complications is possible. the primary diagnosis. Fourth, we
accounted for most of the increase in We might have underestimated the were unable to determine the vaccina-
empyema frequency.17 Small studies rate of complications, because pa- tion status of our study population, to
using polymerase chain reaction- tients with CAP and a CAP-associated assess the efcacy of PCV7 in prevent-
based assays identied S pneumoniae metastatic complication such as os- ing CAP, and could only infer the im-
in 75% to 87.5% of culture-negative em- teomyelitis might have received a pri- pact of PCV7 introduction on the gen-
pyema cases.3740 Empyema may be an mary discharge diagnosis of osteomy- eral pediatric population. Last, the
overlooked major category of IPD in elitis rather than CAP, which would inclusion of every third year in the KID
the post-PCV7 era. have resulted in the exclusion of such might have affected our ability to inter-
patients from our study. This might ex- pret trends accurately, given year-to-
Black race is an independent risk fac-
plain the extremely low incidence of year epidemiological uctuations in in-
tor for IPD,6,41 and racial disparities
metastatic complications found in our cidence caused by factors unrelated to
among children with IPD are greatest
study. We expect that such misclassi-
for children 2 years of age.42,43 pneumococcal vaccination.
cation would not occur disproportion-
Discharge rates of CAP and CAP-
ately in one year versus the next, and CONCLUSIONS
associated complications were higher
the observed trend in CAP-associated
for black children than for white chil- Since the introduction of PCV7 in 2000,
complication rates likely reects a
dren. Our ndings suggest that the dif- rates of CAP hospitalizations have de-
true trend. Second, although ICD-9-CM
ference in rates of CAP discharges and creased for children 1 year of age
discharge diagnosis codes identify
systemic complications between black but seem to be increasing for children
CAP with high sensitivity, ICD-9-CM
and white children has decreased over 5 years of age. Rates of systemic
codes have poor sensitivity in identify-
time. Postlicensure data demon- complications have decreased for chil-
ing pneumonia caused by specic
strated a similar reduction in the ra- dren 1 year of age, but rates of local
pathogens.45 Therefore, we could not
cial disparity in rates of IPD among complications are increasing in all pe-
identify whether changes in rates of
children after the rst 2 years of wide- diatric age groups. Additional studies
CAP-associated complications were
spread PCV7 vaccination.4244 Rates of are needed to determine the underly-
attributable to specic pathogens.
CAP discharges and systemic compli- ing epidemiological factors associated
Third, patients with hospital-acquired
cations among black children have with these changes.
pneumonia might have been included,
seemed to plateau, with no additional because there is no specic ICD-9-CM
decreases after 2003. The groups expe- code for CAP. We minimized the likeli- ACKNOWLEDGMENTS
rienced similar increases in rates of hood of such misclassication by us- Drs Lee and Kronman are recipients of a
local complications over time. Addi- ing a previously validated approach to Young Investigator Award from the Aca-
tional efforts will be required to en- identifying patients with CAP.22 In this demic Pediatric Association. Dr Shah re-
sure that reductions in racial disparity approach, patients with pneumonia ceived support from the National Insti-
can be maintained as the epidemiolog- listed as a secondary diagnosis rather tute of Allergy and Infectious Diseases
ical features of IPD change. than a primary diagnosis must have a (grant K01 AI73729) and the Robert Wood
This study has several limitations. pneumonia-related symptom (eg, Johnson Foundation, through its Physi-
First, the KID is an administrative data- cough or tachypnea) listed as the cian Faculty Scholar Program.

REFERENCES
1. Messina AF, Katz-Gaynor K, Barton T, et al. children from 1999 through 2005. Pediatr conjugate vaccine. N Engl J Med. 2003;348(18):
Impact of the pneumococcal conjugate vac- Infect Dis J. 2007;26(6):461 467 17371746
cine on serotype distribution and antimi- 2. Whitney CG, Farley MM, Hadler J, et al. De- 3. Kaplan SL, Mason EO Jr, Wald ER, et al. De-
crobial resistance of invasive Streptococ- cline in invasive pneumococcal disease after crease of invasive pneumococcal infections
cus pneumoniae isolates in Dallas, TX, the introduction of protein-polysaccharide in children among 8 childrens hospitals in

210 LEE et al
 ARTICLES

the United States after the introduction of quired pneumonia in children and adults. States, 19972006. MMWR Morb Mortal
the 7-valent pneumococcal conjugate vac- Vaccine. 2008;26(38):4947 4954 Wkly Rep. 2009;58(1):1 4
cine. Pediatrics. 2004;113(3):443 449 15. Singleton RJ, Hennessy TW, Bulkow LR, et al. 26. Schultz KD, Fan LL, Pinsky J, et al. The chang-
4. Black S, Shineeld H, Baxter R, et al. Postli- Invasive pneumococcal disease caused by ing face of pleural empyemas in children:
censure surveillance for pneumococcal in- nonvaccine serotypes among Alaska native epidemiology and management. Pediatrics.
vasive disease after use of heptavalent children with high levels of 7-valent pneu- 2004;113(6):17351740
pneumococcal conjugate vaccine in North- mococcal conjugate vaccine coverage. 27. Buckingham SC, King MD, Miller ML. Inci-
ern California Kaiser Permanente. Pediatr JAMA. 2007;297(16):1784 1792 dence and etiologies of complicated parap-
Infect Dis J. 2004;23(6):485 489 16. Byington CL, Korgenski K, Daly J, Ampofo K, neumonic effusions in children, 1996 to
5. Hicks LA, Harrison LH, Flannery B, et al. Inci- Pavia A, Mason EO. Impact of the pneumo- 2001. Pediatr Infect Dis J. 2003;22(6):
dence of pneumococcal disease due to non- coccal conjugate vaccine on pneumococcal 499 504
pneumococcal conjugate vaccine (PCV7) parapneumonic empyema. Pediatr Infect 28. Advisory Committee on Immunization Prac-
serotypes in the United States during the Dis J. 2006;25(3):250 254 tices. Preventing pneumococcal disease
era of widespread PCV7 vaccination, among infants and young children: recom-
17. Hendrickson DJ, Blumberg DA, Joad JP, Jha-
1998 2004. J Infect Dis. 2007;196(9): mendations of the Advisory Committee on
war S, McDonald RJ. Five-fold increase in
1346 1354 Immunization Practices (ACIP). MMWR
pediatric parapneumonic empyema since
6. Albrich WC, Baughman W, Schmotzer B, Far- introduction of pneumococcal conjugate Recomm Rep. 2000;49(RR-9):135
ley MM. Changing characteristics of inva- vaccine. Pediatr Infect Dis J. 2008;27(11): 29. Jackson LA, Neuzil KM, Yu O, et al. Effective-
sive pneumococcal disease in Metropolitan 1030 1032 ness of pneumococcal polysaccharide vac-
Atlanta, Georgia, after introduction of a cine in older adults. N Engl J Med. 2003;
18. Grijalva CG, Nuorti JP, Zhu Y, Grifn MR. In-
7-valent pneumococcal conjugate vaccine. 348(18):17471755
creasing incidence of empyema complicat-
Clin Infect Dis. 2007;44(12):1569 1576 30. Tuomanen EI, Austrian R, Masure HR. Patho-
ing childhood community-acquired pneu-
7. Steenhoff AP, Shah SS, Ratner AJ, Patil SM, monia in the United States. Clin Infect Dis. genesis of pneumococcal infection. N Engl
McGowan KL. Emergence of vaccine-related J Med. 1995;332(19):1280 1284
2010;50(6):805 813
pneumococcal serotypes as a cause of bac- 31. Black S, France EK, Isaacman D, et al. Sur-
teremia. Clin Infect Dis. 2006;42(7):907914 19. Li ST, Tancredi DJ. Empyema hospitaliza-
tions increased in US children despite pneu- veillance for invasive pneumococcal dis-
8. Tsai CJ, Grifn MR, Nuorti JP, Grijalva CG. ease during 2000 2005 in a population of
mococcal conjugate vaccine. Pediatrics.
Changing epidemiology of pneumococcal children who received 7-valent pneumococ-
2010;125(1):26 33
meningitis after the introduction of pneu- cal conjugate vaccine. Pediatr Infect Dis J.
mococcal conjugate vaccine in the United 20. Fisman DN, Abrutyn E, Spaude KA, Kim A,
2007;26(9):771777
States. Clin Infect Dis. 2008;46(11): Kirchner C, Daley J. Prior pneumococcal
32. Whitney CG, Farley MM, Hadler J, et al. In-
1664 1672 vaccination is associated with reduced
creasing prevalence of multidrug-resistant
death, complications, and length of stay
9. Kyaw MH, Lyneld R, Schaffner W, et al. Ef- Streptococcus pneumoniae in the United
among hospitalized adults with community-
fect of introduction of the pneumococcal States. N Engl J Med. 2000;343(26):
acquired pneumonia. Clin Infect Dis. 2006;
conjugate vaccine on drug-resistant Strep- 19171924
42(8):10931101
tococcus pneumoniae. N Engl J Med. 2006; 33. Stoll ML, Rubin LG. Incidence of occult bac-
354(14):14551463 21. Agency for Healthcare Research and Qual-
teremia among highly febrile young chil-
10. Shah SS, Ratner AJ. Trends in invasive pneu- ity. HCUP Kids Inpatient Database (KID)
dren in the era of the pneumococcal conju-
mococcal disease-associated hospitaliza- Healthcare Cost and Utilization Project
gate vaccine: a study from a childrens
tions. Clin Infect Dis. 2006;42(1):e1 e5 HCUP, 1997, 2000, 2003, and 2006. Rockville,
hospital emergency department and urgent
MD: Agency for Healthcare Research and care center. Arch Pediatr Adolesc Med.
11. Black SB, Shineeld HR, Ling S, et al. Effec-
Quality; 2006 2004;158(7):671 675
tiveness of heptavalent pneumococcal con-
jugate vaccine in children younger than ve 22. Whittle J, Fine MJ, Joyce DZ, et al. 34. Herz AM, Greenhow TL, Alcantara J, et al.
years of age for prevention of pneumonia. Community-acquired pneumonia: can it be Changing epidemiology of outpatient bacte-
Pediatr Infect Dis J. 2002;21(9):810 815 dened with claims data? Am J Med Qual. remia in 3- to 36-month-old children after
1997;12(4):187193 the introduction of the heptavalent-
12. Zhou F, Kyaw MH, Shefer A, Winston CA,
Nuorti JP. Health care utilization for pneu- 23. Centers for Disease Control and Prevention. conjugated pneumococcal vaccine. Pediatr
monia in young children after routine pneu- National Center for Health Statistics Infect Dis J. 2006;25(4):293300
mococcal conjugate vaccine use in the bridged race estimates, 1990 2006. Avail- 35. Wilkinson M, Bulloch B, Smith M. Prevalence
United States. Arch Pediatr Adolesc Med. able at: http://wonder.cdc.gov/bridged- of occult bacteremia in children aged 3 to
2007;161(12):11621168 race-v2006.html. Accessed October 1, 2008 36 months presenting to the emergency de-
13. Grijalva CG, Nuorti JP, Arbogast PG, Martin 24. Centers for Disease Control and Prevention. partment with fever in the postpneumococ-
SW, Edwards KM, Grifn MR. Decline in pneu- National Immunization Survey: children cal conjugate vaccine era. Acad Emerg Med.
monia admissions after routine childhood im- (19 35 months). Available at: www.cdc.gov/ 2009;16(3):220 225
munisation with pneumococcal conjugate vaccines/stats-surv/imz-coverage.htm#nis. 36. Tan TQ, Mason EO Jr, Wald ER, et al. Clini-
vaccine in the USA: a time-series analysis. Lan- Accessed October 26, 2010 cal characteristics of children with com-
cet. 2007;369(9568):1179 1186 25. Centers for Disease Control and Prevention. plicated pneumonia caused by Strepto-
14. Nelson JC, Jackson M, Yu O, et al. Impact of Pneumonia hospitalizations among young coccus pneumoniae. Pediatrics. 2002;
the introduction of pneumococcal conju- children before and after introduction of 110(1):1 6
gate vaccine on rates of community ac- pneumococcal conjugate vaccine: United 37. Eastham KM, Freeman R, Kearns AM, et al.

PEDIATRICS Volume 126, Number 2, August 2010 211

 Clinical features, aetiology and outcome of Spain. Emerg Infect Dis. 2008;14(9): pneumococcal disease in young children af-
empyema in children in the north east of 1390 1397 ter introduction of the conjugate pneumo-
England. Thorax. 2004;59(6):522525 41. Robinson KA, Baughman W, Rothrock G, et coccal vaccine. Pediatr Infect Dis J. 2004;
38. Eltringham G, Kearns A, Freeman R, et al. al. Epidemiology of invasive Streptococcus 23(8):726 731
Culture-negative childhood empyema is pneumoniae infections in the United States, 44. Poehling KA, Talbot TR, Grifn MR, et al. In-
usually due to penicillin-sensitive Strepto- 19951998: opportunities for prevention in vasive pneumococcal disease among in-
coccus pneumoniae capsular serotype 1. the conjugate vaccine era. JAMA. 2001; fants before and after introduction of pneu-
J Clin Microbiol. 2003;41(1):521522 285(13):1729 1735 mococcal conjugate vaccine. JAMA. 2006;
39. Tarrago D, Fenoll A, Sanchez-Tatay D, et al. 42. Flannery B, Schrag S, Bennett NM, et al. Im- 295(14):1668 1674
Identication of pneumococcal serotypes pact of childhood vaccination on racial dis- 45. Guevera RE, Butler JC, Marston BJ, et al.
from culture-negative clinical specimens by parities in invasive Streptococcus pneu- Accuracy of ICD-9 CM codes in detecting
novel real-time PCR. Clin Microbiol Infect. moniae infections. JAMA. 2004;291(18): community-acquired pneumococcal
2008;14(9):828 834 21972203 pneumonia for incidence and vaccine ef-
40. Obando I, Munoz-Almagro C, Arroyo LA, et 43. Talbot TR, Poehling KA, Hartert TV, et al. Elim- cacy studies. Am J Epidemiol. 1999; 149:
al. Pediatric parapneumonic empyema, ination of racial differences in invasive 282289

APPENDIX 1 ICD-9-CM Codes Used to Identify Pneumonia-Related Symptoms, Comorbidities, and

Trauma
Symptom ICD-9-CM Codes
Pneumonia-related symptoms
Fever 780.6
Respiratory abnormality, unspecied 786.00
Shortness of breath 786.05
Tachypnea 786.06
Wheezing 786.07
Cough 786.2
Hemoptysis 786.3
Abnormal sputum 786.4
Chest pain 786.5
Precordial pain 786.51
Painful respiration 786.52
Abnormal chest sound (rales, friction, or 786.7
abnormal percussion)
Comorbidities
HIV 042
Malignancy 140.x208.x
Cystic brosis 277
Immune mechanism disorder 279, 334.8
Sickle cell disease 282.6x
Diseases of white blood cells 288
Other lung conditions 507, 517
Congenital heart defects 745747
Encounter for radiation, chemotherapy, or V42, V58.0, V48.1x
transplant
Trauma
Railway accidents E800E807
Motor vehicle trafc accidents E810E819
Motor vehicle nontrafc accidents E820E825
Other road vehicle accidents E826E829
Water transport accidents E830E838
Accidental falls E880E888
Accidents caused by re or ame E890E899
Accidents caused by submersion or E910E915
suffocation
Other accidents E916E925
Suicide, self-inicted injury E953E959
Homicide/injury inicted purposely E960E969
Legal intervention E970E977
Undetermined accident or inicted E983E989
purposely
x indicates digits 0 to 9, where applicable.

212 LEE et al
 ARTICLES

APPENDIX 2 ICD-9-CM Codes for Identication of CAP-Associated Complications

Complication Complication Description ICD-9-CM Codes
Subcategory
Local Emypema 510.x, 511.0, 511.1, 511.9
Lung abscess 513.x
Bronchopleural stula 33.42, 510.0
Necrotizing pneumonia 32.xa
Systemic Acute respiratory failure 518.8x, 799.1
SIRS/sepsis 995.9x
ECMO 39.6x
Hemolytic uremic syndrome 283.11
Metastatic Meningitis 320.0, 320.1, 320.2, 320.3, 320.8x
Central nervous system abscess 324.x
Mastoiditis 383.0x
Pericarditis 420.x
Endocarditis 421.x
Osteomyelitis 730.0x, 730.2x
Septic arthritis 711.x
ECMO indicates extracorporeal membrane oxygenation; SIRS, systemic inammatory response syndrome; x, digits 0 to 9,
where applicable.
a ICD-9-CM code for lung resection as a proxy indicator.

PEDIATRICS Volume 126, Number 2, August 2010 213