Risk Factors and Course of Illness Among Children With Invasive Penicillin-resistant Streptococcus pneumoniae
Objectives. To assess differences in risk factors, clinical presentation, and course of illness between children infected with penicillin-sensitive and drug-resistantStreptococcus pneumoniae (DRSP).
Design. A retrospective cohort study conducted in Uruguay and Argentina using information from a hospital-based surveillance system. Hospitalized children 5 years of age and younger who hadS pneumoniae isolated from a normally sterile site between June 1993 and October 1996 were eligible. Hospital records were linked with surveillance data. Both stratified univariate analysis and logistic regression was completed.
Results. Of the 380 children eligible for the study, 274 records (72%) were available for review. Ninety-nine children (36%) had DRSP; 46 showed intermediate susceptibility (minimum inhibitory concentration, 0.12–1.0 μg/mL) and 53 showed high-level resistance (minimum inhibitory concentration ≥2.0 μg/mL). Children with meningitis were less likely to have DRSP than those with other forms of invasive disease (relative risk = 0.5; 95% confidence interval [CI], 0.2–0.9). Risk factors associated with DRSP were use of penicillin or ampicillin in the 3 months before illness (odds ratio = 2.9; 95% CI, 1.5–5.7) and possession of private medical coverage (odds ratio = 2.4; 95% CI, 1.2–5.0). Response to therapy, including response to penicillin or ampicillin among children with nonmeningeal invasive disease, course of illness, and clinical outcome did not differ significantly between children infected with penicillin-susceptible or penicillin-resistant isolates.
Conclusion. In this study, previous use of penicillin or ampicillin and private medical coverage were associated with having DRSP. Children with nonmeningeal invasive disease responded equally well to penicillin regardless of the penicillin susceptibility of their pneumococcal isolate.
Streptococcus pneumoniae is a major cause of morbidity and mortality worldwide.1–4 It causes invasive illnesses such as meningitis, bacteremia, and pneumonia, and is also a common cause of sinusitis and otitis media.3,,5Since the late 1960s, strains of drug-resistant S pneumoniae(DRSP) have been isolated with increasing frequency around the world.6–10 Factors associated with penicillin resistance include extremes of age, previous use of antibiotics, underlying medical conditions, human immunodeficiency virus infection, hospitalization, nosocomial infection, and alcoholism.1,,2,5,610–14
The emergence of penicillin resistance has made the choice of empiric therapy for pneumococcal infections more difficult.15Invasive infections attributable to penicillin-resistant S pneumoniae may be more difficult to treat and be associated with increased financial costs, sickness, and morbidity,1,,7although some studies have suggested that DRSP is not more virulent than susceptible isolates.6,,10,12,14 The success of therapy may depend on the degree of resistance and the site of infection.12 Because children with penicillin-resistant pneumococcal meningitis have not responded to intravenous penicillin, it is recommended that meningitis caused by DRSP be treated with an alternative antibiotic.6,,10,16,17 However, patients with pneumococcal pneumonia or bacteremia caused by DRSP may respond to intravenous penicillin when the penicillin susceptibility is in the intermediate range.6,,12
A retrospective cohort study was conducted in Uruguay and Argentina among hospitalized children with invasive S pneumoniae to determine if risk factors, clinical presentation, course of illness, and disease outcome differed between children infected with penicillin-sensitive S pneumoniae and DRSP. The study was a collaborative effort between the Canadian Laboratory Center for Disease Control, the Pan American Health Organization, and the S pneumoniae Working Groups from Uruguay and Argentina. Both of these countries conducted surveillance for invasive S pneumoniae in Latin American children 5 years of age or younger as participants of Sistema Regional de Vacunas, a Pan American Health Organization-sponsored initiative.18
Children eligible for inclusion in the study were 5 years of age and younger, had S pneumoniae isolated from a normally sterile body fluid, and were admitted as inpatients to 1 of 13 participating hospitals between June 1993 and November 1995 in Argentina, and between May 1994 and October 1996 in Uruguay. Eight of the hospitals were located in Argentina, in the cities of Santa Fe, Cordoba, Mendoza, and the Federal Capital Region; and 5 hospitals were in the city of Montevideo, Uruguay. These hospitals were part of a larger laboratory-based surveillance network within these and four other Latin American countries.18 The 13 hospitals were a convenience sample of hospitals participating in the surveillance network, representing 57% of participating hospitals and 65% of the total number of isolates identified in the two countries during the study period.
Sterile site specimens included blood and cerebrospinal, pleural, peritoneal, and articular fluid. Nasopharyngeal swabs, bronchial aspirates, or specimens from bronchoalveolar lavage were excluded. When pneumococci were isolated from more than one specimen type, only one was listed, with blood considered the primary source followed by cerebrospinal fluid. The national reference laboratories within each country received all invasive S pneumoniae isolates for confirmatory testing, serotyping, antimicrobial susceptibility testing, and storage. Isolates were screened for reduced susceptibility to penicillin by the Kirby Bauer method using an oxacillin disk. Resistance was confirmed by broth microdilution testing using methods of the National Committee for Clinical Laboratory Standards.19 Penicillin antimicrobial susceptibility was classified using the following minimum inhibitory concentration (MIC) breakpoints: susceptible (≤0.06 μg/mL), intermediate susceptibility (0.12 to 1.0 μg/mL), and high-level resistance (≥2.0 μg/mL).
Basic epidemiologic information was collected on each of the children identified with invasive S pneumoniae as part of the Sistema Regional de Vacunas surveillance project.18 Additional clinical information was collected on the study participants to supplement this information by linking the surveillance data with the child's medical and laboratory records. Ten trained investigators collected data using a standardized data collection form. Two people abstracted data from each medical record, and 1 investigator collected data from every record included in the study. Information about demographics, risk factors, illness presentation, course of illness, disease outcome, and antibiotic susceptibility was collected. Risk factor information was based on parental recall, whereas clinical presentation, course of illness, and disease outcome information was based on physicians' and nurses' documentation in the child's medical record, as well as radiologic and laboratory information. Information on antibiotic susceptibility was obtained after the medical records were reviewed to blind the researchers to the child's penicillin susceptibility status. Study participants were classified as exposed or nonexposed members of the cohort based on the penicillin susceptibility of their pneumococcal isolate.
The clinical syndrome was determined from the physicians' reports and radiologic and laboratory information in the child's medical record. Pneumonia was considered the primary diagnosis if another clinical syndrome was also present, because the lower respiratory tract most likely represents the primary site of infection. All children diagnosed with pneumonia had radiologic evidence of pneumonia. Clinical status during presentation was assessed on the 2nd, 3rd, 7th, 14th, 21st, and 28th day of admission. Children were classified as having an improvement in status if the child was afebrile, clinical signs of cardiorespiratory distress had resolved, the child's responsiveness was returning to normal, or if the child was discharged from the hospital. Children were classified as deteriorated if they had worsening of clinical signs or overall condition, or a poor clinical response to therapy necessitating an alteration in therapy.
Children were classified as having responded favorably to antibiotic therapy if blood cultures became negative or if they had a documented improvement in clinical symptoms, including a diminished fever, while on the antibiotic. If a change in antibiotic therapy was based on antibiotic susceptibility results and the child's condition had not improved or deteriorated while on therapy, or if the child had been on therapy for <48 hours, then the response to therapy was deemed unassessable.
Data were entered into Epi-Info 6.04 (Centers for Disease Control and Prevention; Atlanta, GA) and analyzed using Epi-Info and Log Xact Turbo 1.1 (CYTEL Software Corporation, Cambridge, MA). Risk factor variables were based on parental recall; thus, odds ratios (OR) are reported. Clinical presentation, course of illness, and disease outcome variables were determined from medical records prospectively from the time of sampling; thus, relative risks (RR) are reported. All analyses were stratified by country of residence of the child, because country may confound the results. Stratified logistic regression was performed using the risk factor variables found to be significant in the univariate analysis. Stratified logistic regression was also completed to determine risk factors associated with response to penicillin or ampicillin therapy and factors associated with complications from invasive pneumococcal disease. A P value of <.05 was considered statistically significant.
A total of 380 children were eligible for the study and 274 records (72%) were available for review. The remaining records could not be located in the hospitals' archives; the majority of hospitals did not have computerized records systems. Within Argentina, 78% of records were located, whereas in Uruguay 64% were found (P < .05). Children whose records were missing did not have a significantly different age distribution, sex ratio, year of infection, or proportion of S pneumoniae susceptible to penicillin as children whose records were located.
Of the 274 available records, 105 (38%) were from Uruguay and 169 (62%) were from Argentina. S pneumoniae was isolated from blood in 167 children (61%), pleural fluid in 59 (22%), cerebrospinal fluid in 37 (13%), and other sterile sites in 11 children (4%). Ninety-nine children (36%) had DRSP; 46 (46%) showed intermediate susceptibility and 53 (54%) showed high-level resistance. Of the 53 isolates with high-level resistance, 36 (68%) had an MIC of 2 μg/mL and the remaining 17 had an MIC of 4 μg/mL. No isolates had an MIC of >4 μg/mL.
Epidemiologic characteristics of the 274 children are shown in Table 1. Children from Uruguay were more likely to be infected with DRSP than children from Argentina (RR = 1.6; 95% confidence interval [CI], 1.2–2.2). When children with intermediate susceptibility and high-level resistance were compared, Uruguayan children were more likely to be infected with isolates having high-level resistance than children from Argentina (RR = 2.0; 95% CI, 1.3–3.0). There was a linear association between year of infection and being infected with DRSP (χ2 test for trend 13.4;P = .0003). DRSP was inversely associated with age; however, this association failed to reach statistical significance (χ2 test for trend 3.7; P = .055). Age was not associated with the degree of penicillin resistance.
Pneumonia was the most common form of invasive pneumococcal disease among the study participants. Children with a diagnosis of meningitis were less likely to be infected with DRSP than children with other forms of invasive disease (RR = 0.5; 95% CI, 0.2–0.9). There were no associations between symptoms at presentation, including the presence of tachycardia or respiratory distress, and penicillin resistance.
A number of potential exposure variables were evaluated in the study population (Table 2). Overall, one or more of these variables were present in 69% of children. Stratified univariate analysis revealed that only two exposure variables were associated with having DRSP. Use of penicillin or ampicillin in the 3 months before illness (OR = 2.9; 95% CI, 1.5–5.7) and possession of private medical coverage (OR = 2.4; 95% CI, 1.2–5.0) remained significantly associated with DRSP in a logistic regression model. Use of antibiotics other than penicillin and ampicillin in the 3 months before illness was not associated with penicillin resistance.
Ninety-nine percent of the study participants (271 children) were treated with antibiotics; however, 9 of these children received therapy for <48 hours. Of the 262 children treated for 48 hours or longer, 18 (7%) had their therapy changed within the first 48 hours. Of the remaining 244 children, 201 (82%) were started on monotherapy—52% on a cephalosporin (1% first generation, 23% second generation, 28% third generation), 47% on penicillin or ampicillin, and 0.5% on chloramphenicol. Children infected with DRSP had a longer mean number of days on antibiotics than those with penicillin-sensitive isolates; however, this difference was not significant (19 ± 15 days vs 16 ± 13 days; P = .1). Overall, children infected with DRSP were just as likely to respond favorably to antibiotic therapy as children with penicillin-sensitive isolates (RR = 1.0; 95% CI, 0.8–1.1).
Ninety-five children were started on penicillin or ampicillin alone and were on therapy for at least 48 hours. Of these children, 36 (38%) had DRSP (12 with intermediate susceptibility and 24 with high-level resistance). Three children had pneumococcal meningitis caused by a penicillin-sensitive organism. Therefore, the analysis was limited to the 92 children with nonmeningeal invasive disease (Table 3). Sixty-five (71%) of these children responded favorably to therapy. Children infected with DRSP were just as likely to respond favorably to penicillin or ampicillin as children with penicillin-susceptible isolates, including the 87 children with pneumococcal pneumonia, when analyzed separately. When the analysis was limited to the 80 children with high-level resistant or susceptible isolates, high-level penicillin resistance had no effect on the response to penicillin or ampicillin among children with nonmeningeal invasive disease.
Multivariate logistic regression analysis was completed to determine factors associated with a favorable response to penicillin or ampicillin therapy. No factors were associated with a favorable response, including having DRSP (OR = 1.3; 95% CI, 0.4–3.9). Similarly, when limiting the analysis to children with high-level resistant and susceptible strains of S pneumoniae, penicillin resistance was not associated with response to therapy (OR = 2.6; 95% CI, 0.7–10.2).
Course of illness was assessed by determining the need for supplemental oxygen or intensive care, duration of fever, response to antibiotics, and the clinical status throughout hospitalization. There were no differences between children with penicillin-sensitive S pneumoniae and DRSP in the need for or duration of supplemental oxygenation or intensive care admission. The two groups also exhibited no differences in clinical status throughout hospitalization. Children with DRSP had a longer mean duration of fever (4.5 ± 5.4 days vs 3.5 ± 4.6 days; P = .08) and mean duration of hospitalization (17 ± 20 days vs 15 ± 12 days;P = .2); however, these differences were not significant.
One hundred fifty-two children (55%) suffered from complications secondary to S pneumoniae, and the case fatality rate was 7% (18 deaths). Complications, which were not mutually exclusive, included empyema (30%), seizure (9%), atelectasis (3%), pneumothorax or pneumatocele (3%), subdural abscess (3%), readmission to hospital with the same diagnosis within 1 month of discharge (3%), septic shock (2%), bronchopulmonary fistula (1%), pericardial effusion (1%), and cerebral infarct (1%). Logistic regression revealed that the only risk factor positively associated with complications was a diagnosis of meningitis (OR = 2.3; 95% CI, 1.1–5.0). Complications were inversely associated with having private medical coverage (OR = 0.4; 95% CI, 0.2–0.8). The only risk factor for death was the presence of meningitis (OR = 7.4; 95% CI, 2.3–24.0). DRSP was not associated with an increased risk of complications (OR = 1.0; 95% CI, 0.6–1.7) or death (OR = 1.1; 95% CI, 0.3–3.8).
Among the 274 isolates, there were 33 different capsular types. One hundred and three isolates (38%) were type 14. Univariate analysis revealed a significant association between type 14 and penicillin resistance (RR = 6.0; 95% CI, 3.7–9.7).
The recent emergence of penicillin-resistant S pneumoniae in some Latin American countries20–23 has raised concerns about the effectiveness of programs for acute lower respiratory tract infections which use penicillin for primary treatment. Our study found that children with nonmeningeal invasive pneumococcal disease caused by DRSP respond just as favorably to penicillin as children with penicillin-sensitive organisms. Also, when excluding isolates having intermediate susceptibility, we found that children with high-level resistance responded just as well to therapy as those with sensitive isolates. Although it is accepted that penicillin therapy is inappropriate for penicillin-resistant pneumococcal meningitis, the same may not be true for other forms of invasive pneumococcal disease.6,,10,14,16 Our study supports the use of penicillin therapy for nonmeningeal invasive disease caused by DRSP. Because penicillin is a relatively inexpensive antibiotic compared with alternative antimicrobials, this finding has important economic implications.
In both countries the standard dosage of penicillin routinely used to treat pneumococcal pneumonia was between 100 000 to 200 000 IU/kg/d. The serum concentration of penicillin or ampicillin achieved with standard intravenous dosages are much greater than the MICs for most penicillin-resistant strains.10,,24 It is not clear what level of penicillin resistance in children is clinically relevant for invasive infections other than meningitis. Although a prospective study in which standard antibiotic protocols are used may be able to clarify this issue further, it is likely that ongoing surveillance for penicillin resistance and investigations like ours will be critical in contributing to our understanding of this problem. Centers that continue to use penicillin as their first-line antibiotic for treatment of childhood pneumonia should have surveillance systems in place to monitor penicillin resistance among S pneumoniae isolates and clinical outcomes in the children.
Children with meningitis were significantly less likely to be infected with DRSP than children with other forms of invasive pneumococcal disease. This association is not commonly seen in the literature and may have been a spurious finding because of the large number of associations examined in this study. However, children with meningitis may have been less likely to have previously received an antimicrobial agent because of the more rapid onset of this illness or because of socioeconomic factors that may affect access to health care and the incidence of disease. This needs to be explored in future studies.
Studies have revealed that previous use of antibiotics is a risk factor for infection with penicillin-resistant pneumococci.5,,6,11,13,25 Our study supports this conclusion; however, it was solely the use of penicillin or ampicillin in the 3 months before infection that was associated with penicillin resistance, not the previous use of any antibiotic. Our study also found that children with private medical coverage were at greater risk of infection with DRSP. This association remained significant after controlling for previous use of penicillin or ampicillin. Uruguay and Argentina have both private and public medical coverage. Most of the population has public coverage, with only 17% of the children in this study covered privately. In both countries, antibiotics are available without a prescription. Although this finding is not commonly observed in the literature, it could be postulated that children from families with private medical coverage may be more likely to have seen a physician in the past and had exposure to antibiotics. This may not have been evident in the medical record review if the information was not noted in the chart. These families also may have more income available to purchase antibiotics with or without a prescription, which may have resulted in an increased risk of being infected with DRSP.
The incidence of penicillin-resistant S pneumoniae varies throughout the world, however, it is important to critically examine data before generalizing results.3,7–9 In this study children from Uruguay were more likely to be infected with DRSP than were children from Argentina. However, this was likely attributable to the linear association between the year the child was infected and infection with an isolate having penicillin resistance, because isolates from Uruguay were collected later than isolates from Argentina.
Some limitations should be noted when interpreting the results of this study. Twenty-eight percent of records were not located within the hospitals and were excluded from the study. This could have affected the results if children without records were systematically different from those included in the study. Although children whose medical records were missing did not have a different proportion of isolates resistant to penicillin, we were not able to assess whether they were more likely to be more seriously ill, because this information was not available. The penicillin susceptibility of the study participant's isolate was not known at the time of data abstraction; however, occasionally this information was written in the child's medical record. Although this may have resulted in biased data abstraction, training before data collection was meant to minimize this bias.
Our study suggests that previous use of penicillin or ampicillin and private medical coverage are associated with DRSP. However, susceptibility to penicillin does not seem to affect the severity of illness, the response to antibiotics, or the development of complications. The study also suggests that children with nonmeningeal invasive S pneumoniae respond equally well to penicillin regardless of the penicillin susceptibility of the pneumococcal isolate. This finding could have important economic implications, especially as the prevalence of penicillin resistance increases.
Partial funding for this study was obtained through the Canadian International Development Agency.
We gratefully acknowledge the assistance of Claudia Pérez from Uruguay and Ezequiel Tuduri Franco from Argentina. The suggestions of Dr Jamie Hockin in the preparation of the manuscript were also appreciated.
The Streptococcus pneumoniae Working Group members were: R. Leis, A. M. Ferrari, A. Neirac, L. Pehuffo, J. L. Pefia, H. Bakir, M. Vázquez, R. Bologna, S. Yudowsky, M. Tregnaghi, F. Ferrero, D. Logarzo, E. Q. de Aprá, L. Balbi, J. Pierini, and C. Mayoral.
- Received March 23, 1998.
- Accepted August 13, 1998.
Reprint requests to (J.S.S.) Bureau of Infectious Diseases, Laboratory Centre for Disease Control, Tunney's Pasture, 0603E1, Ottawa, Ontario, K1A OL2.
↵** The Streptococcus pneumoniae Working Group members are listed in the “Acknowledgments.”
- DRSP =
- drug-resistant Streptococcus pneumoniae •
- MIC =
- minimum inhibitory concentration •
- RR =
- relative risk •
- CI =
- confidence interval •
- OR =
- odds ratio
- Tomasz A
- Kristinsson KG
- Applelbaum PC
- Bedos J,
- Chevret S,
- Chastang C,
- Geslin P,
- Regnier B
- Meynard JL,
- Barbut F,
- Blum L,
- et al.
- Butler JC,
- Hofmann J,
- Cetron MS,
- et al.
- John CC
- DiFabio JL,
- Homma A,
- De Quadros C
- ↵National Committee for Clinical Laboratory Standards. Methods for Dilution Antimicrobial Susceptibility Tests for Bacteria That Grow Aerobically. 3rd ed. Approved Standard. M100–S5,M7–A3. Villanova, PA: NCCLS; 1994
- Rossi A,
- Ruvinsky R,
- Regueira M,
- et al.
- Hortal M,
- Algorta G,
- Bianchi I,
- et al.
- Brandileone MCC,
- Vieira VSD,
- Casagrande ST,
- et al.
- Castañeda E,
- Leal AL,
- Castillo O,
- et al.
- ↵Norris S, Nightingale CH, Mandel GL. Tables of antimicrobial agent pharmacology. In: Mandel GL, Douglas RG Jr, Bennet JE, eds. Principles and Practices of Infectious Diseases. 3rd ed. New York, NY: Churchill Livingstone; 1990:434–460
- Nava JM,
- Bella F,
- Garau J
- Copyright © 1999 American Academy of Pediatrics