Background. Little is known about the epidemiology of invasive pneumococcal disease (IPD) after the introduction of 7-valent pneumococcal conjugate vaccine (PCV7) in Spain and other European countries.
Methods. We performed a 10-year prospective study including all children with culture-proven IPD admitted to Sant Joan de Deu Hospital, a children's center in the southern area of Barcelona, Catalonia, Spain. PCV7 was introduced in June 2001, and the current estimate of PCV7 coverage is 45%–50%.
Results. Comparing the prevaccine period (1997–2001) with the vaccine period (2002–2006), among children aged <2 years, the rate of IPD increased from 32.4 episodes per 100,000 population to 51.3 episodes per 100,000 population (an increase of 58%; 95% confidence interval, 2%–145%), and among children aged 2–4 years, the rate increased from 11.3 episodes per 100,000 population to 26.5 episodes per 100,000 population (an increase of 135%; 95% confidence interval, 31%–320%). At clinical presentation, the rate of pneumonia and/or empyema among children aged <5 years increased from 3.6 episodes per 100,000 population to 15.1 episodes per 100,000 population (an increase of 320%; 95% confidence interval, 98%–790%). These increased rates of IPD were caused by non-PCV7 serotypes, which represented 38% and 72% of infecting serotypes in the prevaccine and vaccine periods, respectively (P<.001). Penicillin resistance decreased from 48% in the prevaccine period to 27% in the vaccine period (P=.005). In the vaccine period, there was an emergence of previously established virulent clones of non-PCV7 serotypes 1 and 5. There was also an increase in the prevalence of serotypes 19A and 6A expressed with different clonal types, including Spain23F-1 and Spain6B-2.
Conclusions. Since the introduction of PCV7 for children, there has been an emergence of IPD caused by virulent clones of non-PCV7 serotypes that has been associated with significant clinical changes and a decrease in antibiotic resistance.
Streptococcus pneumoniae is a major bacterial pathogen in children and adults. The World Health Organization estimates that, annually, ∼1 million children aged <5 years die of pneumococcal pneumonia, meningitis, and/or sepsis worldwide . The incidence of invasive pneumococcal disease (IPD) varies widely depending on factors, such as age, geographical area, race, economic status, site of infection, and underlying conditions . More than 90 serotypes of S. pneumoniae have been identified, but only ∼23 serotypes produce the majority of cases of IPD .
Studies conducted in the United States that have used the 7-valent conjugate pneumococcal vaccine (PCV7), which includes serotypes 4, 6B, 9V, 14, 18C, 19F, and 23F, have proven the vaccine to be safe and efficacious in preventing IPD in children [4, 5]. In addition, other trials of the 9-valent conjugate pneumococcal vaccine, which includes serotypes 1 and 5 in addition to the serotypes in PCV7, have shown the vaccine to be efficacious in reducing mortality in Gambian children  and in reducing the incidence of lower respiratory tract infection in HIV-infected South African children . It is worth mentioning that serotype prevalence may vary among different geographical areas and may change over time in response to selective pressure caused by different factors [8–10]. The purpose of this study was to determine temporal trends in IPD, clinical presentation, and serotypes and clones of S. pneumoniae among children in Barcelona, Spain, before and after the implementation of PCV7.
Patients and Methods
We performed a prospective study comprising all children and adolescents (age, <18 years) with IPD who were admitted to Sant Joan de Deu Hospital, a 345-bed children's hospital in Barcelona, Catalonia, Spain, during a 10-year period (January 1997–December 2006). The hospital is a children's center for the southern area of Barcelona, with a pediatric population ranging from 187,451 in 1997 to 216,242 in 2006 . During this period, the mean number of annual hospital admissions was 16,227 (range, 14,749–17,201 hospital admissions). In Catalonia county, with ∼7 million population and 1.2 million persons aged <18 years, our hospital captured 17.6% of all pediatric hospitalizations in 1997 and 17.8% in 2005 (data not available for 2006) .
Patients and definitions. All patients with pneumococcal infection who have had S. pneumoniae isolated from any clinical sample were prospectively recorded at the Clinical Microbiology Department. Since 1991, several variables have been routinely registered, including hospital identification number, type of sample, and antimicrobial susceptibility. Identification numbers of all isolates from sterile fluid samples were used to review electronic medical records and registered demographic and clinical variables, including age, sex, date of hospital admission, clinical data, outcomes, and PCV7 vaccination status. Data were recorded in accordance with the guidelines of the hospital's ethical committee.
PCV7 was introduced in Spain in June 2001, and the Spanish Paediatric Academy has recommended an immunization schedule . The academy recommends PCV7 vaccination for children aged <2 years—scheduled at 2, 4, and 6 months of age, with a booster in the second year of life—and for older children at high risk of IPD. However, PCV7 is not currently subsidized by the Spanish Health Service. It has been estimated that <50% of children in Spain are currently vaccinated with PCV7 [14, 15]. We have estimated the vaccine uptake in children born after January 2002 on the basis of 2 studies of nasopharyngeal carriage performed in nonimmunocompromised children who were attended for minor surgical procedures. The rate of PCV7 uptake was 36% in 2005 (among 204 children studied) and 47% in 2007 (among 90 children studied). The vaccination status was ascertained by means of patients' records or the children's vaccination cards.
IPD was defined as the presence of clinical findings of infection together with isolation of S. pneumoniae in a blood sample, CSF sample, or any other sterile fluid sample. IPD was classified according to the International Classification of Diseases, Ninth Revision code specific for diseases caused by S. pneumoniae, including meningitis, pneumonia, parapneumonic empyema, occult bacteremia, sepsis, arthritis, peritonitis, and endophthalmitis.
Microbiological studies. Pneumococcal isolates were identified by standard microbiological methods that did not change during the study period, as reported elsewhere . Agar dilution technique was used to determine the MICs of penicillin and other antibiotics. Antibiotic susceptibilities were defined according to the 2006 break points suggested by the Clinical Laboratory Standards Institute . Isolates with intermediate- or high-level resistance were defined as nonsusceptible. Serotyping was performed by the Quellung reaction. Serotypes were classified in the 2 following groups: PCV7 serotypes, including serotypes 4, 6B, 9V, 14, 18C, 19F, and 23F, and non-PCV7 serotypes, including all other serotypes. Determination of both MICs and serotypes was performed at the National Pneumococcus Reference Centre (Majadahonda, Madrid, Spain).
Genetic characterization was performed using multilocus sequence typing . The assignment of alleles and sequence types (ST) was performed using the software at the Multi Locus Sequence Typing Web site . Analysis of ST and assignment to clonal complex were performed using the eBURST program .
Statistical analysis. The study period was divided into 2 periods: the prevaccine period (1997–2001) and the vaccine period (2002–2006). Rates of IPD, defined as the number of episodes per 100,000 population, were calculated using the annual estimates of pediatric population obtained from the Department of Statistics in Catalonia . We used the χ2 test or Fisher's exact test to compare proportions. Statistical analyses were performed using SPSS for Windows, version 14.0 (SPSS), and Epi Info, version 6.0 (Centers for Disease Control and Prevention). We calculated 95% CIs, and 2-sided P values ⩽.05 were considered to be statistically significant.
A total of 198 episodes of IPD occurred in 194 children, including 120 male patients (62%) and 74 female patients (38%), with a mean age of 3 years (range, 1 month to 17 years). Ninety-six episodes (48%) were in children aged <2 years, 61 episodes (31%) were in children aged 2–4 years, and 41 episodes (21%) were in children and adolescents aged 5–17 years. All 198 episodes of IPD involved ⩾1 culture positive for S. pneumoniae; there were 126 positive blood specimens, 40 positive pleural fluid specimens, 35 positive CSF specimens, 12 positive joint specimens, 7 positive peritoneal fluid specimens, and 1 positive aqueous humor specimen. The clinical presentation of IPD by age group is shown in table 1. Eighteen children had been vaccinated with PCV7, and all but 1 were infected with nonvaccine serotypes (one 12-month-old child received 3 doses at 2, 4, and 6 months of age and developed pneumococcal bacteremia caused by serotype 19F).
According to the criteria of the American Academy of Pediatrics , 14 of our children (accounting for 18 episodes) were at high risk of IPD, including 6 children with malignant disease who were receiving immunosuppressive therapy (7 episodes), 1 with CSF leakage (4), 2 with HIV infection (2), 2 with chronic pulmonary disease (2), 2 with chronic cardiac disease (2), and 1 with chronic renal failure (1). During the prevaccine period, 6 (11%) of 55 children were at high risk for IPD; during the vaccine period, 8 (6%) of 139 children were at high risk for IPD (P=.211)
Children were admitted to the pediatric intensive care unit for 43 (22%) of the 198 IPD episodes. Overall, there were 7 deaths; 4 (10%) of 39 cases of meningitis, 2 (4%) of 57 cases of occult bacteremia and/or sepsis, and 1 (1.2%) of 79 cases of pneumonia resulted in death.
Incidence and changes in clinical presentation. figure 1, the increased rates of IPD during the vaccine period were attributable to non-PCV7 serotypes, and the rates of IPD caused by PCV7 serotypes slightly decreased.
There was an increase in the number of blood cultures performed during the study period (from 21,589 in the prevaccine period to 26,472 in the vaccine period), but this was not statistically significant (P=.062). The rates (number of positive blood culture results per 1000 blood cultures performed) for the most important bacterial pathogens during the prevaccine and vaccine periods were as follows: for Staphylococcus aureus, the rates were 3.7 and 3.4, respectively (P=.68); for Neisseria meningitidis, the rates were 1.6 and 1.2 (P=.32); for Escherichia coli, the rates were 3.5 and 5.2 (P=.007); and for Streptococcus pneumoniae, the rates were 1.9 and 3.2 (P=.001).
Among children aged <5 years, a relevant finding at clinical presentation was the increased rate of pneumonia and/or empyema during the vaccine period, compared with the prevaccine period (3.6 episodes per 100,000 population during the prevaccine period vs. 15.1 episodes per 100,000 population during the vaccine period; an increase of 320%; 95% CI, 98%–790%). Rates of pneumonia and/or empyema for each age group are shown in table 1.
Antibiotic susceptibility, serotypes, and molecular studies. The prevalence of infection with pneumococcal strains with penicillin resistance decreased between the prevaccine and vaccine periods, from 48% (31% with intermediate resistance and 17% with full resistance) to 27% (21% and 6%; P=.005). Resistance to other antibiotics also decreased (table 2).
Overall, 29 different serotypes were identified. Of the 198 strains, 75 (38%) were PCV7 serotypes, and 123 (62%) were non-PCV7 serotypes (77 nonrelated serotypes, 42 related serotypes, and 4 nontypeable strains). Serotype distribution varied during the prevaccine and vaccine periods. During the prevaccine period, 36 (62%) of 58 isolates were vaccine serotypes, and 22 (38%) of 58 isolates were nonvaccine serotypes; during the vaccine period, 39 (28%) of 140 isolates were vaccine serotypes, and 101 (72%) of 140 isolates were nonvaccine serotypes (P<.001).
Serotype 19A was almost solely detected during the vaccine period (for 27 of 28 isolates). Among patients aged <18 years, there was a statistically significant increase in the rate of IPD caused by serotype 1 (from 0.9 episodes per 100,000 population during the prevaccine period to 2.9 episodes per 100,000 population during the vaccine period; an increase of 188%; 95% CI, 30%–536%), serotype 19A (from 0.1 episodes per 100,000 population to 2.7 episodes per 100,000 population; an increase of 2293%; 95% CI, 225%–17,510%), and serotype 7F (from 0 episodes per 100,000 population to 0.9 episodes per 100,000 population; an increase of 100%).
There were statistically significant differences (P<.001) in the clinical diagnosis according to serotype group (PCV7 serotype vs. non-PCV7 serotype). Among the 75 episodes caused by PCV7 serotypes, occult bacteremia and/or sepsis occurred most frequently (28 episodes [37%]), followed by meningitis (17 [23%]), pneumonia and/or empyema (17 [23%]), and other infections (13 episodes [18%]). On the other hand, among the 123 episodes caused by non-PCV7 serotypes, pneumonia and/or empyema occurred most frequently (62 episodes [50%]), followed by occult bacteremia and/or sepsis (29 [24%]), meningitis (22 [18%]), and other infections (10 [8%]).
Molecular analysis by multilocus sequence typing was performed for 100 strains expressing the main emergent serotypes. Overall, when comparing our data with isolates listed on the Multi Locus Sequence Typing database, there were 43 different STs, including 12 new ST multilocus sequence typing profiles and 5 new capsular switches (table 3); eBURST analysis identified 6 clonal complexes with ⩾2 STs and 27 singletons.
The most important clonal complex represented 21% of the studied pneumococci and contained 19 isolates with allelic profile ST306 (identical to antibiotic-susceptible global clone Sweden1-27; recovered in Sweden in the late 1990s) and 2 additional related clones (ST2376 and ST228). Of these 21 strains, 19 expressed serotype 1, and 2 were nontypeable strains. The first time that we documented ST306 and ST228 in our study was during the period 1999–2000, and ST2376 was detected only during the period 2005–2006. The clinical manifestation in patients infected by this clonal complex was pneumonia and/or empyema in all but 1 patient with meningitis detected in 2000. The other isolates of serotype 1 (in 7 strains recovered from 5 patients with pneumonia and/or empyema—1 with appendicitis and 1 with bacteremia) presented allelic profile ST304 and were not in the same group by eBurst analysis. The first time that we documented ST304 was in 1998.
The second most prevalent clonal complex involved 12 strains with serotype 5 that were recovered from patients with pneumonia and/or empyema (9 episodes), meningitis (2), and occult bacteremia (1). The predictor founder of the clonal complex was ST289, allelic profile of the global clone, tetracyclin-resistant columbia5-19. During the prevaccine period, only ST289 was detected, but during the vaccine period, an expansion of the other clones (ST1223 and ST245) included in the clonal complex was observed.
A high level of diversity with evidence of capsular switch within vaccine serotypes was observed in the strains of serotypes 19A and 6A, which expressed different clonal types, including Spain23F-1 (ST81) and Spain6B-2 (ST90). A singleton clone, penicillin-susceptible ST1201, was the most prevalent ST in strains of serotype 19A (in 4 of the 14 strains tested), but a multiresistant clone, ST276, was detected in another 3 strains. This clone is highly genetically related to the global clone Denmark14-32. The global clone Netherlands15B-37 (ST199), which is the most important ST among clinical isolates in the United States, was detected in 1 additional strain.
To date, several reports have shown a reduction in the overall incidence of IPD after the introduction of PCV7 [22–25], and in some studies, the rate increase has been associated with an increase of infections caused by non-PCV7 serotypes [26–28]. In our study, with current PCV7 vaccination rates of 45%–50% among children, we observed a significant increase in the rate of IPD caused by non-PCV7 serotypes and a slight reduction in the rate of IPD caused by PCV7 serotypes. Recently, an important increase in the rate of IPD caused by non-PCV7 serotypes was reported among Alaska Native children with high levels of vaccine coverage 
Several factors and limitations of the study must be considered when evaluating our results. First, changes in clinical and blood culture practices might influence the rate of IPD in our institution. However, our guidelines for evaluating children with fever have not substantially changed throughout the study period. Although we did observe a nonstatistically significant increase in the number of blood cultures performed, we think that this could not be the only explanation for the increase in the rate of IPD. It could be expected that an increase in the number of blood cultures performed would occur more frequently for minimally symptomatic children, causing an increase of mild disease, but this was not the case in our study, because we observed an increase in the rate of complicated IPD (e.g., pneumonia and/or empyema), for which the diagnostic approach is more standard.
Second, potential changes in the number of children with IPD whose cases were captured for the study may contribute to the increased rates. However, this would not be a good explanation, because the proportion of captured pediatric hospitalizations at our institution did not change over the study period.
Third, rates of IPD could be modified by the presence of epidemics of viral respiratory infection, because these viral infections may predispose to IPD . However, we could not find a correlation between the rates of IPD and the presence of epidemics of viral respiratory infection according to the number of viral respiratory tests submitted to our laboratory (data not shown).
Fourth, an important factor and limitation of this study is the low vaccine coverage (<50%) among our population. Although an increase in the rate of IPD caused by non-PCV7 serotypes may be attributable to several factors, we believe that the early phase of implementing PCV7 in our area may have contributed to this increase. The potential impact of PCV7 on serotype distribution could vary according to the serotypes circulating before implementing the vaccine and to the rates of children vaccinated.
In Spain, serotypes 1 and 5 were quite prevalent among pneumococcal strains isolated from children during the period 1990–1999 . These serotypes, particularly serotype 1, have a high potential to cause epidemic outbreaks of infection , and it could be expected that implementing a vaccine without these epidemic serotypes may enhance potential outbreaks. In our study, the increase in the rate of IPD was caused, in part, by these serotypes, and we observed a spread of 2 clones of serotype 1, ST 306 and ST 304, and the clonal complex ST289 of serotype 5. Clone ST306 has previously been associated with an outbreak of pneumococcal disease not related to vaccine use in Sweden .
Geographical differences in rates of IPD have been associated with significant variations in serotype and clonal type distribution . For example, in our region, the major clone of serotype 1 was ST306, and in the United States, the major clone of serotype 1 was ST227.
An increase in the prevalence of non-PCV7 serotypes may occur by different ways. First, it is possible that clones of non-PCV7 serotypes that were present during the prevaccine period could disseminate during the PCV7 period. This phenomenon could be the predominant cause of the increase in the prevalence of non-PCV7 serotypes in our area, with widespread existence of strains of clone ST306. Second, in our study, there was an increase in the prevalence of serotypes 19A and 6A expressed with different clonal types, including Spain23F-1 and Spain6B-2, which suggests a genetic exchange of the capsular locus (i.e., capsular switching) among vaccine and nonvaccine serotypes. Serotype 19A variants of the Spanish 23F multiresistant clone were circulating in our region during the prevaccine period . Currently, the prevalence of serotype 19A is also increasing in the United States . Serotypes 6A and 19A were the first serotypes to be associated with multiple antibiotic resistance in children .
Changes in the distribution of serotypes over time may be associated with changes in clinical types of IPD . The association of serotype 1 with empyema was described in the prepenicillin  and prevaccine eras [40, 41]. Currently, our study and other reports have shown an emergence of empyema caused by serotype 1 .
In accordance with other studies , we observed a substantial reduction in the rate of antibiotic-resistant strains causing IPD after the introduction of PCV7. This occurred because antibiotic resistance is mainly associated with PCV7 serotypes. However, recent reports have found an increase in antibiotic resistance in non-PCV7 serotypes in the PCV7 era [43, 44].
In conclusion, the results of our study must be considered to be preliminary because of the low vaccine coverage in our area, and because others factors might have contributed to these epidemiological changes. However, our study does show that the rate of IPD caused by non-PCV7 serotypes is increasing in Barcelona. This increase is associated with changes in clinical presentation, a decrease in antibiotic resistance, overgrowth of previously established virulent clones of non-PCV7 serotypes, and capsular switching among vaccine and nonvaccine serotypes. Further progress in pneumococcal vaccine development, based on conjugate vaccines that include these emergent serotypes and the protein-based vaccines that are nonserotype dependent, can be expected.
We thank Drs. Cristina Esteva, Edgar Palacín, Susana Hernandez-Bou, Carolina Polo, and Sandra Gala, for their contributions to the care of patients and/or microbiological studies; Dr. Asuncion Fenoll, for serotyping; and Dr. Angela Brueggeman, for performing clonal studies of some strains from children with empyema.
Financial support. The Spanish Pneumococcal Infection Study Network (G03–103); Ciberes, Instituto Salud Carlos III, Ministry of Health, Barcelona, Spain; GlaxoSmithKline Biological; Caja Navarra Foundation; and Sant Joan de Deu Foundation.
Potential conflicts or interest. C.M.-A. has received a travel grant from Glaxo Smith Kline Biologicals. A.G. has received a travel grant from Wyeth. J.J.G.-G. has received a travel grant from Wyeth. All other authors: no conflicts.