To determine the prevalence ofStreptococcus pneumoniae nasopharyngeal carriage, antibiotic resistance patterns, and serotypes; to examine the variability of microbiological findings between child care centers; and to determine risk factors for antibiotic resistance.
Point prevalence survey.
Licensed child care centers in Toronto, Ontario.
Healthy children attending the centers.
Main Outcome Measures
Prevalence (simple and adjusted for clustering) of carriage, antibiotic resistance, and serotypes; multivariate analysis of risk factors for resistance.
Of 1322 children from 59 centers, 586 (44.3%) carried 599 S pneumoniae isolates. On the day of study, 129 (10.7%) of 1203 children for whom a questionnaire was completed were taking antibiotics and 336 (227.9%) had taken them in the previous month. Decreased susceptibility to penicillin was found in 102 isolates (17.0%) and 82 (13.7%) were resistant to multiple antibiotics. The most common serotypes, in order, were 6B, 23F, 6A, 19F, 14, 11A, and 19A, composing 78% of all isolates. Microbiological results from individual centers were variable, but the overall prevalence of carriage, antibiotic resistance, and serotypes was not significantly different when adjusted for effects of clustering within centers. Multiple logistic regression determined that age younger than 24 months and antibiotic use within the previous month were significant risk factors for carriage of S pneumoniae resistant to penicillin, sulfamethoxazole-trimethoprim, and erythromycin.
Efforts to reduce antibiotic use in children should be particularly directed toward young children attending child care centers. Studies of infectious diseases in child care centers should consider clustering of pathogens or factors promoting transmission within centers that may result in variability between centers.
THE RISK of invasiveStreptococcus pneumoniae infections is greatly increased in children attending child care centers.1,2 Children attending child care centers have more frequent infections than those cared for at home, or in family day homes, at least during the first 2 years of attendance.3- 5
Several recent studies have evaluated risk factors for nasopharyngeal (NP) carriage of penicillin-nonsusceptibleS pneumoniae (PNSP) in children who were well or had minor illnesses.6- 12 Multivariate analysis found that child care center attendance was a risk factor for PNSP carriage in 2 studies,8,10 but not in 2 others.9,11 Other risk factors for PNSP carriage include age younger than 2 years, recent antibiotic use, current otitis media, white race, and geographic location.7- 11 One other study identified low daily dose of a β-lactam antibiotic and long duration of antibiotic treatment as risk factors for PNSP carriage.7 Most studies were small, with fewer than 50 children carrying PNSP in all but 1, limiting the validity of multivariate analysis.13 Only 1 study evaluated risk factors for resistance to other antibiotics.11
Studies of NP carriage of S pneumoniae in child care centers have examined just 1 or a small number of child care centers.14- 16 The prevalence of carriage of PNSP may vary widely between individual child care centers, suggesting clustering of antibiotic-resistant strains within particular centers. For example, Radetsky et al17 found that 27% of children in one room of a child care center in Denver, Colo, carried PNSP, compared with 11% of other children and staff in the same center. No PNSP carriage was found in 6 other local centers.
We performed a survey to determine the prevalence of S pneumoniae carriage, antibiotic resistance patterns, serotypes, and risk factors for carriage of antibiotic-resistant S pneumoniae in children attending 59 child care centers in Toronto, Ontario. We wished to determine whether there are unique risk factors for the carriage of antibiotic-resistant S pneumoniae in children attending child care centers. We further sought to examine the variability of microbiological findings between individual child care centers and used clustered data analysis to evaluate the impact of this variability.
A cross-sectional survey was performed to determine the prevalence of nasopharyngeal carriage of S pneumoniae in children attending licensed child care centers whose enrollment included children younger than 12 months in the cities of Toronto and North York within the region of metropolitan Toronto. The survey was conducted from March 29 to July 11, 1995, and from October 3, 1995, to February 7, 1996. There was a break during the summer because of reduced attendance and staffing at child care centers. In the province of Ontario, all child care centers enrolling more than 5 children are required by law to be licensed annually. To obtain and maintain licensure, minimal requirements must be fulfilled regarding the physical layout of the center, staffing levels, and infection control practices.
All children attending participating centers were considered eligible for enrollment in the study unless they had an underlying illness that made an NP swab contraindicated (eg, hemophilia). Children who were absent because of illness on the day of the study enrolled during a second visit.
Written informed consent was obtained from the parent or guardian of each child. If a parent or guardian refused consent or was not available to provide consent, that child was excluded from the study. The study was approved by Research Ethics Board of the Hospital for Sick Children, Toronto.
A self-report questionnaire was completed by a parent or caregiver at the time of consent. The questionnaire included questions about antibiotic use at the time of the study or in the previous month (what antibiotic and for what indication), use of other medications at the time of the study (what medication and for what indication), hospitalization in the previous 6 months (and for what indication), and number of episodes of otitis media in the previous 12 months.
We estimated that around 40% of children would carry S pneumoniae and that 10% of those carrying S pneumoniae (or 4% of all participants) would have isolates with reduced susceptibility to penicillin. To obtain a 95% confidence interval (CI) of 4%±1% (ie, 10%±2.5% of isolates with reduced penicillin susceptibility),18 we estimated a total sample size of 1500 children, yielding 600 children carrying S pneumoniae and 60 isolates with reduced penicillin susceptibility.
A single NP swab was obtained from each participating child by means of a Dacron-tipped flexible wire swab (NCS Diagnostics, Etobicoke, Ontario) inserted into the posterior part of the nares. The swabs were transported at ambient temperature to the Hospital for Sick Children and plated within 6 hours.
Standard methods were used to identify S pneumoniae.19 Each swab was inoculated on 2 blood agar base plates (Quelab Laboratories Inc, Montreal, Quebec) with 5% horse blood. Each plate was incubated overnight at 35°C, 1 in 5% carbon dioxide and 1 anaerobically. Plates were read at 24 and 48 hours. When growth was observed, up to 5 α-hemolytic colonies characteristic ofS pneumoniae were subcultured to another horse blood agar plate with an optochin disk and incubated for 20 to 24 hours. All optochin-sensitive isolates (≥15-mm inhibition) were confirmed asS pneumoniae by bile solubility testing.
Isolates were stored frozen in glycerol citrate at −70°C or less and tested for antibiotic susceptibility at the Mount Sinai Hospital, Toronto, by microbroth dilution assay, with the use of cation-adjusted Mueller-Hinton broth with 2% lysed horse blood.20 Isolates were classified as susceptible, intermediate, or resistant according to National Committee for Clinical Laboratory Standards guidelines.21 Penicillin resistance was further defined as PNSP (minimum inhibitory concentration, ≥0.1 µg/mL) or penicillin-resistant S pneumoniae (minimum inhibitory concentration, ≥2 µg/mL).22 Multidrug nonsusceptibility was defined by intermediate susceptibility or resistance to 2 or more drug classes that may be used to treat S pneumoniae infections.23
Frozen isolates were sent in batches to the National Centre for Streptococcus in Edmonton, Alberta, for serotyping. Serotype determinations were made with the Quellung test with antisera obtained from the Statens Seruminstitut of Copenhagen, Denmark, with the use of the specific methods of the National Centre for Streptococcus, Edmonton.24,25
In some cases, 2 or more colonies of S pneumoniae with different morphologic appearances (morphotype) were identified from the NP swab of a single child. In cases where each morphotype from a child had a different serotype, the morphotypes were considered to be different strains. If each morphotype had the same serotype, the morphotypes were considered to be different strains only if there was a 4-fold or greater difference in minimum inhibitory concentration for 2 or more of penicillin, sulfamethoxazole-trimethoprim, erythromycin, and clindamycin.
All data were recorded in a relational database (FoxPro v2.6; Microsoft Corp, Redmond, Wash) on a desktop computer. Statistical analysis was performed with the programs SPSS (version 6.1.1; SPSS Inc, Chicago, Ill), STATA (version 5.0; Stata Corp, College Station, Tex), and Minitab Student Edition (version 8; Addison-Wesley Inc, Reading, Mass). Categorical variables were summarized by means of simple proportions with 95% CIs. Continuous variables were reported as means or medians, with ranges, SDs, or 95% CIs. Differences between groups on categorical and continuous variables were tested by the χ2 test and Student t test, respectively. For categorical variables, the association between the predictor factor and the outcome was expressed by means of the odds ratio (OR). Ninety-five percent CIs were calculated by the test-based method of Miettinen.26
Children within individual child care centers may not have been independent from each other regarding the likelihood of carriage ofS pneumoniae and antibiotic-resistant S pneumoniae. Particular strains of S pneumoniae may, for example, be more likely transmitted from child to child within a center regardless of individual risk factors. To address the effect of any clustering ofS pneumoniae within child care centers, cluster survey design methods were applied to culture results, antibiotic susceptibility, and serotype proportions.27,28 A finite population sampling approach was used, taking the overall population in eligible child care centers in the area as the target population. Two assumptions were made: that the participating centers represented a random sample from the overall pool and that children were sampled randomly within each center. Appropriate weights based on sampling probabilities were used to allow for the varying sampling fractions across centers. Variance calculations taking into account the 2-stage sampling were computed with a Taylor series approach.
Multiple logistic regression analysis was used to evaluate possible risk factors independently associated with carriage of PNSP, sulfamethoxazole-trimethoprim resistance, and erythromycin resistance. The model building strategy used was based on a 2-step process described by Hosmer and Lemeshow.29 First, univariate analyses assessed the association between plausible independent variables and the dependent variable. Second, backward stepwise logistic regression was performed, initially including all possible independent variables; then, in a series of successive steps, statistically insignificant variables were eliminated, leaving the greatest number of significant variables in the final model. The statistical significance of each independent variable in the model was evaluated by examining the relationship of the estimated coefficient (β) and its SE by means of the Wald test (significant if <.05). The coefficients were expressed as ORs with 95% CIs.
Of a total of 94 eligible child care centers, 81 were approached and 59 (73%) agreed to participate. The centers had an average (±SD) enrollment of 55±20 children (range, 20-127). In total, 1322 (43.1%) of 3070 eligible children participated. The average number of participants from each center was 22 (range, 3 to 55) and the rate of participation was 4% to 88%. The average age of participants was 34±17 months, 32% were younger than 2 years, and 91% were younger than 5 years. More than one half of the subjects were male (54%).
Reasons for refusal to participate were sought from the center directors. Ten stated that the center was too busy or short-staffed or the timing was bad because of cutbacks in government funding, 3 were never successfully contacted, 2 believed the study was too invasive, 2 had concerns about previous research studies, 2 thought their population was not appropriate, 1 had too many other research projects ongoing, 1 had recently had a fire, and 1 did not give a reason. Three centers that refused to participate had large enrollments (195, 145, and 125, respectively). The average enrollment of the others (58 children) did not differ from that of the participating centers (P=.18). Thirteen centers were not approached because the necessary sample size had been obtained. These centers were scattered throughout the study area, and their average enrollment (52 children) was not different from that of the participating centers (P=.97).
The questionnaire was returned for 1203 (91.0%) of the participants (Table 1). Antibiotics had been used by a total of 389 children (32.3%) either at the time of the study or in the previous month. Seventy-six of these children (19.5%) reported antibiotic use at the time of the study and in the previous month. Among children younger than 24 months, 15% were taking antibiotics at the time of the study compared with 9% of children aged 24 months or older (OR, 1.91; P<.01). For antibiotic use in the previous month, the proportions were 41% and 23%, respectively (OR, 2.39; P<.01).
Use of medications other than antibiotics on the study date was reported by 161 children (13.4%). Most commonly these were asthma medications (67 children [41.6%]) and cough and cold medications (36 children [22.4%]). Hospitalization in the previous 6 months was reported by 90 children (7.5%), and the stay was 3 days or less for 66 of the children (73%). The most common reasons for admission were respiratory conditions (28%) and surgery (26%). A history of at least 1 episode of otitis media in the previous 12 months was reported by 638 children (53.0%).
No data were obtained on the nonparticipants in each center. However, when subjects were categorized according to the participation rate at their center (≤25%, >25 to ≤50%, >50% to ≤75%, or >75%), there were no differences between these groups in the proportion less than 24 months old (P=.44), sex (P=.69), current antibiotic use (P=.16), antibiotic use in the previous month (P=.14), hospitalization in the previous 6 months (P=.21), frequency of ear infections in the previous 12 months (P=.30), or PNSP rate (P=.52).
Of the 1322 children tested, 586 (44.3%) had growth ofS pneumoniae from their NP swabs. Adjusted for clustering, the prevalence of carriage was 43.1%. Thirteen children's swabs yielded 2 distinct types of S pneumoniae, thus providing a total of 599 isolates. Streptococcus pneumoniae was found in 58 of 59 centers (culture-positive rate, 0% to 83%) and the median number of positive children at each center was 10 (range, 0 to 32). There were just 3 participants in the center where no S pneumoniae was found. Cultures were more often positive in children younger than 24 months (57% if younger than 24 months, 39% if 24 months or older; OR, 2.10; P<.01). Cultures were positive less often in children taking antibiotics at the time of the study (29% if taking antibiotics, 46% if not taking antibiotics; OR, 0.47;P<.01) or in the month before the study (39% if taking antibiotics, 46% if not taking antibiotics; OR, 0.74;P=.02).
One hundred two (17.0%) of 599 isolates were PNSP. These were found in 101 children from 35 (59%) of the 59 centers. Table 2 also shows the overall prevalence of isolates that were susceptible, intermediate, or resistant to penicillin, cefuroxime, cefotaxime, sulfamethoxazole-trimethoprim, erythromycin, and clindamycin as simple proportions and proportions adjusted for clustering. The number of centers with each level of resistance for each antibiotic is also shown. There was little difference between the simple and adjusted proportions for each antibiotic, although the results from individual centers were variable.
Of the 102 PNSP isolates, 82 (13.7% of all isolates) were resistant to 1 or more other antibiotic classes, and 65 isolates (10.9% of all isolates) were resistant to 2 or more other classes of antibiotics. The number (and percentage) of PNSP isolates that were also resistant to sulfamethoxazole-trimethoprim, erythromycin, cefuroxime, clindamycin, and cefotaxime was 80 (78%), 53 (52%), 44 (43%), 33 (32%), and 12 (12%), respectively. The average enrollment of child care centers attended by children with isolates resistant to any antibiotic was not different for penicillin, cefotaxime, erythromycin, clindamycin, or sulfamethoxazole-trimethoprim (P≥.19 for all comparisons). For cefuroxime there was a difference (average center enrollment of children with resistant isolates vs average enrollment of children without resistance, 49 vs 55; P<.01).
Serotyping was performed on 589 viable isolates received by the National Centre for Streptococcus. There were a total of 27 serotypes or serogroups found, with 18 isolates being nontypable (Table 3). Seven serotypes (6B, 23F, 6A, 19F, 14, 11A, and 19A) accounted for 78% of all isolates. Among 101 PNSP isolates, serotypes 6B (30.7%), 23F (25.7%), 19F (13.9%), and 19A (12.9%) composed 83% of all isolates (P<.01 compared with penicillin-susceptible isolates). The number of different serotypes found in each center varied from 1 to 9. In the 4 centers with 20 or more isolates (32, 22, 21, and 20 isolates), the number of different serotypes was 6, 9, 6, and 4, respectively.
Heptavalent conjugate vaccines currently under investigation contain capsular polysaccharides of serotypes 4, 6B, 9V, 14, 18C, 19F, and 23F.30,31 These 7 serotypes comprised 384 (65.2%) of 589 serotypes. In contrast, a population-based study of S pneumoniae infections in children from Toronto during 1995 found that the above 7 serotypes comprised 83% of all invasive isolates.32
Table 4 summarizes the univariate and multivariate logistic regression analyses comparing clinical and demographic factors between the 485 children carrying penicillin-susceptible S pneumoniae with the 101 children carrying PNSP. Multivariate analysis identified that age younger than 24 months and use of antibiotics in the month before the study were associated with PNSP carriage. There was a trend toward significance for female sex.
The same multivariate model was used to predict for carriage of sulfamethoxazole-trimethoprim–resistant and erythromycin-resistantS pneumoniae, and the same 2 risk factors were found. Age younger than 24 months (OR, 1.56; 95% CI, 1.08-2.26;P=.02) and use of antibiotics in the month before the study (OR, 1.59; 95% CI, 1.05-2.41;P=.03) were associated with carriage of sulfamethoxazole-trimethoprim–resistant S pneumoniae. Age younger than 24 months (OR, 2.46; 95% CI, 1.40-4.30; P<.01) and use of antibiotics in the month before the study (OR, 2.44; 95% CI, 1.40-4.27; P<.01) were associated with carriage of erythromycin-resistant S pneumoniae. Other factors were not significant, including female sex (OR for sulfamethoxazole-trimethoprim resistance, 1.06; 95% CI, 0.87-1.29; P=.59; and OR for erythromycin resistance, 0.90; 95% CI, 0.67-1.21;P=.48).
This study found that NP carriage of S pneumoniae is common among children attending child care centers, as are infectious illnesses and antibiotic use. This is the first study, to our knowledge, that sampled NP carriage of S pneumoniae at a large number of child care centers across a population, rather than studying a convenience sample of 1 or just a few centers. We were able to compare findings between centers and demonstrate the variability of findings from individual centers. The microbiological findings were variable between centers, but the overall prevalence of antibiotic resistance and specific serotypes was not significantly different after correcting for cluster effect, perhaps because of the large number of centers studied. Variable results between small numbers of child care centers have been previously reported. Studies in Denver and Cleveland, Ohio, found frequent asymptomatic carriage of PNSP strains within centers after cases of PNSP infection in children attending these centers were found.17,33 However, at the same time, no PNSP carriage was found at other child care centers in the same cities.17,33
There are host and environment factors that influence the transmission of infectious agents within a child care center and that will differ from center to center. Host factors include age-specific hygiene habits (eg, increased likelihood of spread of respiratory secretions in young children) and immunological or physiological factors (eg, increased likelihood of carriage of bacteria resulting from lack of immunity or lack of previous exposure).3,34 Environmental factors include the enrollment of the center (eg, with increased enrollment pathogens will be introduced to a center more frequently) and frequency of age-group mixing (eg, asymptomatic older carriers of bacteria may transmit these bacteria to younger, susceptible children if they are in frequent contact).34 Numbers of staff and their qualifications and the physical environment of child care centers also play a role in transmission.35
Conjugate polyvalent S pneumoniae vaccines are being developed with the aims of improved efficacy to prevent disease at all ages compared with the current polysaccharide vaccine and reduction of nasopharyngeal carriage to prevent transmission in healthy carriers.30,31 Heptavalent conjugate vaccines contain capsular polysaccharides of serotypes 4, 6B, 9V, 14, 18C, 19F, and 23F. These 7 serotypes cause 65% to 85% of invasive S pneumoniae disease36,37 and comprised 62% of the serotypes in this study. Thus, these new vaccines have the potential to reduce carriage of the majority of S pneumoniae in our population, including the most important serotypes causing invasive disease.
Recent antibiotic use has been described as a risk factor for carriage of antibiotic-resistant S pneumoniae.9- 11,38 Antibiotics suppress growth of antibiotic-susceptible bacteria in the nasopharynx (S pneumoniae and others), yet may allow growth of antibiotic-resistant bacteria. One recent study did not find recent antibiotic use to be a risk factor for carriage of antibiotic-resistantS pneumoniae.6 In our study, it is likely that current antibiotic use was not a risk factor for carriage of resistantS pneumoniae because of the low recovery rate of S pneumoniae from children taking antibiotics at the time of the study and the resultant low power. In addition, the number of children for whom antibiotic use both at the time of the study and in the previous month was reported was too small to analyze separately from those taking antibiotics only at the time of study and only in the past month.
Age younger than 24 months was the strongest risk factor for antibiotic-resistant S pneumoniae carriage. In part, this is because carriage of any S pneumoniae was so much more prevalent in this age group. In addition, there may be other factors for which younger age is a marker. Young children may have increased exposure to antibiotic-resistant S pneumoniae carried by others and, therefore, an increased likelihood of transmission, eg, frequent close contact in groups such as child care centers.
Sex was considered as a risk factor for antibiotic-resistant S pneumoniae carriage because of the known male preponderance of invasive S pneumoniae disease.39- 42 Although female sex approached significance as a risk factor for PNSP carriage, its magnitude was very small and unlikely to have been of clinical importance.
A strength of this study was the large number of cases of children with antibiotic-resistant S pneumoniae carriage. Studies that evaluate fewer than 10 cases with any outcome of interest for each risk factor evaluated in multivariate analysis may unpredictably overestimate or underestimate the significance of each risk factor.13 In this study there were 101 children with PNSP and 7 risk factors considered, so the ratio of cases to factors was appropriate.
There are limitations to our findings that may reduce the generalizability of our results to other child care centers or children in other care settings. The selection of participating child care centers was not randomized, several centers refused participation, and the rate of participation within each center was low. We knew little about the nonparticipating centers (apart from total enrollment) and so could not make meaningful comparisons with the participating centers. Based on informal feedback provided by the center staff and participating families, the main reason for nonparticipation was lack of awareness of the study, because not all families were informed of the study by the staff at each center. There were no differences between the centers with low to high participation rates for the factors examined in this study, suggesting that the rate of participation did not materially influence the overall result.
Children attending child care centers may have more frequent infections than children in other care settings, at least during the first 2 years of attendance.3- 5,43 The incidence of respiratory infections (including otitis media) in child care centers depends on several factors, including the type of child care arrangement, weekly hours of attendance, and cumulative duration of enrollment.3,4,43 Antibiotic use may also be more common in child care centers.44 These factors may limit the generalizability of our findings in children not attending child care centers.
Children younger than 2 years old, particularly those attending child care centers, are a population that should be targeted in efforts to reduce unnecessary antibiotic use. National and international organizations, including the World Health Organization, Centers for Disease Control and Prevention in the United States, and the Laboratory Center for Disease Control in Canada, have developed strategies to control antibiotic resistance.45- 47 Promoting the judicious use of antibiotics is an important component of these strategies.48 Targeted efforts to educate physicians, parents, and child care center staff regarding judicious antibiotic use in young children attending child care centers may be particularly worthwhile.
Accepted for publication September 22, 1998.
This study was supported by grants from Lederle Praxis Biologics Inc, Pearl River, NY, and Pfizer Canada, Kirkland, Quebec. Dr Kellner was supported by the Canadian Infectious Diseases Society Eli Lilly Research Fellowship. Dr Low was supported in part by the Canadian Bacterial Diseases Network.
The following members of the Toronto Child Care Centre Study Group contributed as authors, meeting the International Committee of Medical Journal Editors qualifications for authorship: Donald E. Low, MD, FRCPC, Anne Matlow, MD, FRCPC, Elaine E. L. Wang, MD, MSc, FRCPC, Mary Corey, PhD, Barbara Yaffe, MD, MHSc, FRCPC, Irene Kyle, MSc, and Ronald Gold, MD, MPH, FRCPC.
We thank Heather Watson, BScN, and Lisa Palermino, RN (Hospital for Sick Children), for coordinating and performing the data collection; Michael Bates and Howard Njoo, MD (Toronto), for supporting the involvement of child care centers; Margaret Roscoe, BSc (Hospital for Sick Children Microbiology Laboratory), for coordinating the microbiology studies; James Talbot, MD, PhD, and Marguerite Lovgren, BSc (National Centre for Streptococcus), for serotyping; and Rollin Brant, PhD, and Victoria Stagg, MSc (University of Calgary), for performing the clustered data analysis.
Presented as a poster at the 36th Interscience Conference on Antimicrobial Agents and Chemotherapy, New Orleans, La, September 16, 1996.
Reprints: James D. Kellner, MD, MSc, FRCPC, Division of Infectious Diseases, Alberta Children's Hospital, 1820 Richmond Rd SW, Calgary, Alberta, Canada T2T 5C7 (e-mail: email@example.com).
Editor's Note: William Carlos Williams would have loved this study. He could have achieved 1322 cultures without having to battle with 1 child to open her mouth.—Catherine D. DeAngelis, MD
Kellner JD, Ford-Jones EL, and Members of the Toronto Child Care Centre Study Group. Streptococcus pneumoniae Carriage in Children Attending 59 Canadian Child Care Centers. Arch Pediatr Adolesc Med. 1999;153(5):495-502. doi:10.1001/archpedi.153.5.495