Reduction in Antibiotic Use Among US Children, 1996–2000
Background. High rates of antibiotic prescribing to children contribute to antibiotic resistance in the community. The Centers for Disease Control and Prevention, in collaboration with other national and state level organizations, have actively promoted more judicious prescribing for children.
Objective. We sought to assess changes in the rate of antibiotic prescribing from 1996–2000 in 9 US health plans, patterns of diagnosis and treatment responsible for these trends, and changes in the use of first-line antimicrobial agents.
Design/Methods. We analyzed claims data for dispensed medications and physician visits from 9 health plans. Each provided data on 25 000 children aged 3 months to <18 years enrolled between September 1, 1995, and August 31, 2000. Antibiotic dispensings were linked with an ambulatory visit claim to assign diagnosis. Antibiotic dispensings per person-year (antibiotics/p-y) were calculated for the age groups 3 months to <3 years, 3 years to <6 years, and 6 years to <18 years. The contribution of each diagnosis to changes in the overall rate of antibiotic use was determined. Generalized linear mixed models were used to test for trend and assess differences in rates by site.
Results. From 1996–2000, antibiotic rates for children 3 months to <3 years decreased from 2.46 to 1.89 antibiotics/p-y (24%); for children 3 years to <6 years from 1.47 to 1.09 antibiotics/p-y (25%); and for children 6 to <18 years from 0.85 to 0.69 antibiotics/p-y (16%). The reduction varied among health plans from 6% to 39% for children 3 months to <3 years. A decrease in prescriptions for otitis media accounted for 59% of the total decrease, and was primarily accounted for by a decrease in the rate of diagnosis of this condition. The proportion of first-line penicillins increased from 49% to 53%, with health plans with the lowest initial rates increasing most.
Conclusions. Antibiotic prescribing decreased significantly between 1996 and 2000, concurrent with decreased frequency of diagnosis of potential bacterial infections, especially otitis media. Attention by public health and professional organizations and the news media to antibiotic resistance may have contributed to changes in diagnostic thresholds, resulting in more judicious prescribing.
Antibiotic resistance is considered a public health crisis in the United States.1–3 National estimates of nonsusceptibility to penicillin of Streptococcus pneumoniae, for example, increased from 16% in 19944 to 24% in 1998,5 with rates well over 50% reported from some child care settings.6–8 A multiagency federal action plan recently recommended decreasing unnecessary antibiotic prescribing as critical to combatting antimicrobial resistance.2 Because children have the highest rates of antibiotic use,9 the Centers for Disease Control and Prevention (CDC) has, since 1998, promoted a campaign to encourage judicious antibiotic prescribing for children, and has engaged the American Academy of Pediatrics, the American Academy of Family Physicians, state health departments, and health plans in this effort.10
Increasing antibiotic use in children from 1977 through the early 1990s is well-documented.9,11 This coincided with increases in the proportion of children in group child care12 and increases in the incidence of diagnosed otitis media,13 the most frequent condition associated with antibiotic prescriptions for children.14 Some studies suggest high rates of prescribing for viral diagnoses such as viral upper respiratory infection (URI) or bronchitis,15 but health plan-based analyses suggest that clearly inappropriate indications account for only a small minority of prescriptions written.14
A study based on the National Ambulatory Medical Care Survey (NAMCS) recently reported a substantial downward trend in antibiotic use for children, especially for otitis media.16 A separate NAMCS analysis17 showed a decrease in prescribing among children diagnosed with viral URIs from 1995–1998, but the fraction treated by pediatricians (38%) still far exceeds that reported elsewhere.14 The NAMCS monitors practice for 1-week periods in specified medical practices, selected to be representative of US medical care. Although these estimates have been extrapolated to the entire US population, NAMCS does not measure directly the patient denominator of reported rates. These office-based data do not include antibiotics prescribed without an office visit (eg, after a telephone encounter) or refills of an existing prescription.
The Health Maintenance Organization (HMO) Research Network Center for Education and Research on Therapeutics (CERTs) is a coalition of 9 geographically dispersed health plans that collaborate in research.18 Analysis of health plan enrollment files allows precise determination of population denominators, and pharmacy claims identify all antibiotics dispensed, whether prescribed by telephone, office visit, or dispensed as refills. Dispensing events after an office visit can be linked to the diagnosis given at that encounter.14 We used these data linkage capacities to study changes in antibiotic prescribing patterns for children from 1996–2000. In particular, we sought to assess the overall trends in antibiotics received, trends in use of particular antibiotic classes, and the fraction of the change in prescribing attributable to common respiratory tract illness diagnoses. We focused on patterns of antibiotic use among infants and children younger than 3 years in whom the rates of antibiotic use are highest.
We analyzed automated claims data from 9 health plans that are diverse in size, geography, and delivery system type (Table 1). For this analysis, each health plan used standard computer programs (SAS Institute, Cary, NC) to randomly select 25 000 children aged 3 months to <18 years who were enrolled in each plan, with pharmacy benefits, for a minimum of 90 days during any study year in the period between September 1, 1995, and August 31, 2000. Rates for each year were calculated from September 1 to August 31 (eg, the 1996 rate from September 1, 1995–August 31, 1996) so that annual rates reflect patterns of use during only 1 respiratory illness season. Only actual months of enrollment were considered in the denominator. Because subjects were observed for varying periods of time, the number of antibiotics dispensed per child is not meaningful; rather, the denominators of all rates are person-years of observation within specified age groups (the number of observed individuals multiplied by the number of years, or fraction thereof, they were included in the study). Cohort members could have contributed data during more than 1 study year and/or age group. A small fraction of enrolled children at each health plan had no visit or pharmacy claims over the study period. To reduce overestimation of our denominator size, we calculated the fraction of children <2 years of age, enrolled continuously for an entire year, who had no claims (pharmacy or visit). Because children <2 years are likely to have a least 1 visit, we assumed that the absence of any service was evidence that claims for the child’s care were not being submitted to the participating health plan (eg, because of dual insurance). We decreased the population denominator at each plan by this fraction (range: 1%–13%; median: 2%).
For each individual, claims for all outpatient encounters and all antimicrobial agents dispensed were analyzed. Pharmacy claims were generated at retail or health plan-based pharmacies, so they represent actual dispensings of medications rather than whether they were simply prescribed. Medications given to hospital inpatients are not included. Pharmacy claims routinely included a patient identifier, date dispensed, and National Drug Code (NDC) to identify the preparation. We used a commercially available list of NDC codes for all antimicrobial agents, and an iterative process using pharmacy database extracts from the participating health plans to ensure capture of all antimicrobials. We also analyzed all ambulatory visit claims for the same individuals during their periods of health plan enrollment. Data extracts from visit claims included a study identifier, date of encounter, 1 or more International Classification of Diseases, Ninth Revision (ICD-9)19 diagnosis codes, and 1 or more procedure codes. Approximately 22% of visit claims included >1 diagnosis. Based on previous work,14 we used an algorithm to identify a primary diagnosis for each visit, giving priority to any diagnosis of a bacterial infection, and using a hierarchy of bacterial diagnoses if >1 of these was recorded. Final diagnosis categories with ICD-9 codes19 include: acute otitis media (382, 384.0–384.2) otitis media with effusion (OME) (381.0–381.4), pharyngitis (034.0, 034.1, 041.0, 462, 463), bronchitis (466, 490), sinusitis (461, 473), pneumonia (481-483, 484.8, 485–487, 033.0, 033.9) cold/URI (460, 464 [excluding 464.3], 465, 487 [excluding 487.0], 786.2), bacterial-other (signifying a nonrespiratory tract bacterial infection), and other (including well-child care, nonspecific viral diagnoses, and those unrelated to infection). Each antibiotic dispensing was linked to the most recent visit occurring within the 3 days preceding the dispensing. Dispensings that could not be linked to a visit (including refills of previously prescribed medications or those prescribed after a telephone encounter) were included in rates of antibiotics received, but excluded from calculation of prescriptions attributed to each condition. Antimicrobial agents were divided into the following categories for reporting: first-line penicillins (penicillin, ampicillin, amoxicillin, dicloxicillin), second-line penicillins (amoxicillin/clavulanate), first-line macrolides (erythromycin), second-line macrolides (azithromycin, clarithromycin) cephalosporins, trimethoprim/sulfamethoxizole, tetracyclines, and others.
Trends in prescribing and differences among health plans were analyzed for statistical significance using generalized linear mixed models20 for Poisson regression on the rates of use. This was done to account for the clustering of observations across time within each health plan. The results are presented in terms of the actual measured rates in each health plan and the sample overall. The percentage decline over the 5-year period is reported, accounting for clustering and assuming a linear trend. All analyses were conducted using SAS statistical software, version 8 (SAS Institute, Cary, NC). The study was approved by Institutional Review Boards of each of the 9 participating sites.
Attributes of the 9 geographically dispersed health plans are shown in Table 1. Because included children were enrolled for varying periods, the number of person-years of observation varied among the health plans from 55 000 to 73 000. Over the 5-year observation period, children aged 3 months to <3 years received, on average, 2.2 antibiotics per year (range among health plans: 1.6–3.1). Those 3 years to <6 years of age received 1.3 antibiotics per year (range among health plans: 0.93–1.7); and those 6 years to <18 years received 0.77 antibiotics per year (range among health plans: 0.61–0.92). Analysis controlling for clustering of observations showed statistically significant variation among health plans in each age group (P < .05).
We present population-based rates of each condition for the 5-year study period overall (Fig 1), excluding dispensings not linked to a visit (20%) or with no diagnosis provided (3%). Among children 3 months to <3 years, otitis media (including acute otitis media and OME) was responsible for the majority 1.22 antibiotics/p-y (antibiotic dispensings per person-year) (68%) of antibiotic dispensings. Other respiratory tract conditions were, together, responsible for an additional 0.33 antibiotics/p-y (18)%: sinusitis, 0.08 antibiotics/p-y (4%); pharyngitis, 0.07 antibiotics/p-y (4%); pneumonia, 0.05 antibiotics/p-y (3%); bronchitis, 0.04 antibiotics/p-y (2%); and cold/URI 0.09, antibiotics/p-y (5%). An additional 0.19 antibiotics/p-y (9%) of prescriptions were associated with nonspecific viral or noninfectious diagnoses or with well-child care visits. Soft tissue infections, urinary tract infections, and rare, but more serious, bacterial infections were all represented in these data, but account for a small minority (0.05 antibiotics/p-y [3%]) of antibiotic prescriptions among young children, and are grouped under “other-bacterial.” The distributions of diagnoses associated with antibiotic dispensings for children aged 3 to <6 years and those aged 6 to <18 years are also provided (Fig 1). The rate of antibiotic use was dramatically lower in older age groups with 0.47 antibiotics/p-y (46%) and 0.12 antibiotics/p-y (22%) accounted for by otitis media in children 3 to <6 years and children 6 to <18 years, respectively. Not surprisingly, pharyngitis and sinusitis accounted for increasing proportions of antibiotics in older children. Combined prescribing for cold/URI and bronchitis accounted for 7%, 8%, and 9% of antibiotics in the 3 age groups, respectively.
The trend in rates of antibiotic dispensing for the 5 study years, 1996–2000, is shown for each health plan in Fig 2, and the rates for all patients combined is highlighted for each age group (thick line). In each of the 3 age groups we found a substantial decline in the rate of antibiotic use. The largest absolute drop was in young children, who had the highest rates of antibiotic use overall, from 2.46 antibiotics/p-y (range by health plan: 1.89–3.95) to 1.89 antibiotics/p-y (range by health plan: 1.38–2.41), an adjusted decline of 24% (P < .001; range by health plan: 6%–39%). Data from children aged 3 to <6 years showed a similar adjusted relative drop of 25% (P < .001; range by health plan: 8%–41%) from 1.47 antibiotics/p-y (range by health plan: 1.20–2.13) to 1.09 antibiotics/p-y (range by health plan: 0.72–1.40), and older children and adolescents (6 years to <18 years) had more modest declines of 16% (P < .001; range by health plan: −1% to 32%) from .85 antibiotics/p-y (range by health plan: 0.70 to 1.06) to .69 (range by health plan: 0.51–0.90) antibiotics/p-y.
We examined changes in absolute prescribing rates associated with specific conditions among children 3 months to <3 years of age, as well as the proportions of children diagnosed with particular conditions who received antibiotics. Together this information can help assess whether changes in the incidence or management of particular conditions were responsible for the overall decrease in antibiotic prescribing. Two health plans were excluded from this analysis because their proportions of visits with missing diagnoses was higher in the early study years than later ones, which could have biased this analysis. Table 2 shows the change in population-based rates of antibiotic dispensing for each condition over the 5 study years and the percent of the overall decrease in antibiotic use accounted for by each. For children <3 years, the most substantial drop in prescribing was for otitis media, 0.35 antibiotics/p-y from 1.3 to 0.96. This was accompanied by a decrease of 0.53 visits with a diagnosis of acute otitis media, from 1.8 to 1.3, but no change in the rate of diagnosis of OME. Lower absolute decreases were found for antibiotic treatment of other diagnoses of interest: pharyngitis, 0.04 antibiotics/p-y; bronchitis, 0.02 antibiotics/p-y; sinusitis, 0.02 antibiotics/p-y; and cold/URI, 0.05 antibiotics/p-y. Prescribing for pnuemonia increased slightly by 0.01 antibiotics/p-y.
One possible explanation for this decrease in antibiotic use would be a decline in the population-based rates of medical visits overall. The rate of visits among children 3 months to <18 years for any reason, including preventive care, decreased 9% from 3.3 visits per person-year in 1996 to 3.0 visits per person-year in 2000. The decreases by age group were 7% among children 3 months to 3 years, 12% among those 3 to <6 years, and 10% among those 6 to <18 years. Among diagnoses of respiratory tract diagnoses, several relevant trends emerge (Table 2). Among children 3 months to 3 years, in addition to the 30% decrease in the diagnosis of acute otitis media, substantial decreases were seen in the diagnosis of pharyngitis and bronchitis in this age group, with the latter decreasing by 58%. There was no compensatory increase in the diagnosis of cold/URI or of OME. Among children 3 years to 6 years old, the diagnosis of acute otitis media also decreased by .24 visits per person-year (35%), with pharyngitis, sinusitis, and bronchitis accounting for smaller absolute decreases. Finally, in older children and adolescents the decrease in diagnosis of pharyngitis exceeded that for acute otitis media.
There were few notable trends in the proportion of cases treated within each diagnosed condition. For example, the proportions of otitis media (76%), sinusitis (84%), and pneumonia (62%) treated with an antibiotic remained constant. (These estimates include follow-up visits for these conditions, so they are lower than would be expected for new episodes only.) The proportion of those diagnosed with bronchitis, considered an inappropriate indication for antibiotics, who were treated with antibiotics decreased from 72% to 67% among all age groups combined and from 61% to 56% among children 3 months to 3 years. The proportion of visits with a cold/URI diagnosis associated with a prescription also decreased from 15% to 10% overall, and from 12% to 8% in the youngest age group.
The decrease in treatment of otitis media accounted for 59% of the total decrease in antibiotics prescribed among children 3 months to 3 years (Table 2). The proportions of the decrease in prescribing attributable to other respiratory tract infections were smaller in this age group. Eight percent of the decrease was attributable to prescribing for cold/URI, 6% for pharyngitis, and 3% and 4% for sinusitis and bronchitis, respectively. Also in this age group, 14% of the decrease was accounted for by a decrease in antibiotics not associated with an ambulatory encounter in the prior 3 days. Among children 3 to <6 years and 6 to <18 years, otitis media also accounted for substantial proportions of the decrease in antibiotics (47% and 31%, respectively), although the absolute rates of prescribing are lower than in younger children. With increasing age, a larger proportion of the decrease in antibiotic use was attributable to less treatment of pharyngitis, accounting for 13% and 30% of the decrease in the 3- to <6- and 6- to <18-year-old age groups. The decrease in antibiotic use overall in the 2 older age groups was partially attributable to bronchitis (6% and 11%) and sinusitis (10% and 16%). Decreased prescribing for cold/URI accounted for only 5% of the decrease in overall antibiotic use in children 3 to <6 and 6 to <18 years.
Despite increasing antibiotic resistance, most authorities continue to recommend amoxicillin as the agent of choice for community-acquired upper respiratory tract infections in children. Among children 3 to <3 years, first-line penicillin use, as a proportion of antibiotics dispensed, rose slightly from 48% (range among health plans: 35%–58%) of all antibiotic dispensings in 1996% to 53% (range among health plans: 43%–60%) in 2000. Prescribing of trimethoprim-sulfamethoxizole decreased most dramatically from 19% to 10% of dispensings over the same period. The largest percentage increase was for broad-spectrum macrolides, which increased from 2% to 8% concurrent with the introduction of clarithromycin and azithromycin. Patterns of use according to drug class were similar in the older age groups. By 2000, first-line penicillins accounted for 59% of antibiotics among children 3 to <6 years and those 6 to <12 years. Among children 12 to <18 years, first-line penicillins accounted for only 34%, primarily because of increased proportions of tetracyclines (23%) and macrolides (15%).
This study analyzed data on ∼225 000 children from 9 health plans across the United States. The results show that antibiotic use in children has decreased markedly from 1996–2000. Although decreases were found in all 9 health plans and across all age groups, there was substantial variation among the health plans in rates of antibiotic use and in the magnitude of decrease. Linking each dispensing to a primary diagnosis allowed us to analyze the changes in antibiotics prescribed for a range of common conditions. Although otitis media continues to account for the majority of antibiotic use in children, a decrease in the number of cases accounts for most of the decrease in antibiotic use overall. The percent of diagnosed cases of otitis media, bronchitis, and sinusitis treated with antibiotics remained relatively constant over this same period. Finally, there is evidence for some increase in the use of first-line penicillins, as recommended by national guidelines.21
A recent report using the NAMCS reported a downward trend of 40% in antibiotic use for children and adolescents over the decade of the 1990s.16 The NAMCS collects data on ambulatory visits for a 1-week period in selected practices. Statistical techniques are then used to estimate population-based rates. In contrast, our study uses direct estimates of the denominator of enrolled children who both did and did not have medical care contact during a period of interest. Health plan data also capture antibiotics dispensed without an office visit, such as patient-initiated refills and prescriptions made by clinicians after evaluation of a patient by telephone. That these 2 studies, using different methods, show similar results is strong evidence that these trends are real. Confidence in the results is further bolstered by the consistency observed across the 9 distinct populations studied here.
The increase in antibiotic use in the 1980s9 occurred in concert with a rapid increase in the rate of diagnosis of otitis media.13,22 Whether this represented the use of lower diagnostic thresholds for this potential bacterial infection or an actual increase in its incidence is difficult to assess. Increased use of group child care may have led to more infectious illness in young children23–25 but probably does not explain the entire increase. Pressure on clinicians to see patients more quickly has also been raised as a contributing factor.26 Finally, parental pressure or antibiotic demand is cited by clinicians as an important driver of antibiotic overprescribing.26 Gaps in parental knowledge regarding appropriate indications for antibiotic use have been documented,27 although such parental pressure to prescribe may be overestimated.28,29
Competing explanations likewise exist for the observed decrease in antibiotic use. A decrease in population-based rates of use may reflect: 1) lower rates of care-seeking for acute illness by the population; 2) lower rates of diagnosis of potential bacterial infections when care is sought; or 3) less frequent treatment of infections presumed to be bacterial (eg, using a watchful waiting approach for some cases of acute otitis media). These data show a small decrease in visit rates overall, which may account for a portion of the observed reduction in antibiotic use. However, reduced prescribing is primarily accounted for by lower rates of diagnosis of bacterial infections, most notably otitis media in younger children (accounting alone for 59% of the decrease in prescribing) and a combination of otitis media, pharyngitis, and sinusitis in older children (accounting for 78% of the decrease in those 6 to <18 years). It is unlikely that changes in the epidemiology of disease account for the lower proportion of acute illness visits given a bacterial diagnosis. Haemophilus influenzae vaccine was introduced well before this study period, and pneumococcal conjugate vaccine was introduced only at the very end, so neither can account for the trends observed. Therefore, we conclude that the change in these patterns likely reflects changes either in likelihood of parents seeking care for these conditions, or in the criteria or thresholds clinicians use for assigning bacterial diagnoses.
The proportion of cases treated within each diagnosis was relatively stable. Attention has recently focused on the modest benefit of antibiotic treatment on a number of common conditions30–32; adoption of a watchful waiting approach for otitis media has been advocated by some.33,34 The data reported here show little change in clinicians’ treatment under each diagnostic label. This suggests that it may be more difficult to persuade clinicians to withhold treatment for minor infections than to adjust their diagnostic thresholds toward more judicious antibiotic use. Given this, one might expect to see the rates of OME increase as an alternative diagnosis to acute otitis media.35 We observed no such increase, suggesting that clinicians may be currently considering tympanic membranes normal that they previously would have diagnosed as having acute infection.
The diagnoses of cold/URI and bronchitis deserve special mention. Previous studies using NAMCS data have reported very high rates of antibiotic prescribing for cold/URI, which are considered inappropriate indications for antibiotic treatment.15,17 In previous health plan-based studies and in these data as well, the proportion of children with these viral diagnoses receiving antibiotics is much smaller.14 This does not prove that clinicians in these health plans do not prescribe for colds; rather, it shows that their assignment of diagnoses is more consistent with the treatment chosen. Cough illness, even characterized by deep cough, is generally viral.36 However, physicians in practice (and patients) use the diagnosis of bronchitis as an indication for antibiotics. The proportion of children with this diagnosis who received antibiotic treatment remains >60%—against current expert recommendations. If physicians become convinced that bronchitis is not an appropriate indication for antibiotic treatment, they may use competing diagnoses with overlapping symptoms, such as sinusitis, for which antibiotics are indicated according to recent guidelines.37
The strengths of health plan data for epidemiologic analyses have been described elsewhere,38 but these results must be interpreted with a number of caveats. First is the concern that the patients and physicians represented in data from these health plans are not representative of general medical care in the United States. The consistency of the findings across geographic regions and delivery system type, however, argues for generalizability across these features. Many of the clinicians represented in the data of 5 of the included health plans provide care to patients of many insurers within their geographic areas. We drew random samples of 25 000 members from each plan. This number allowed us to determine very precise estimates of rates of prescribing, while simplifying data management processes and giving equal weight in our analysis to each of the represented health plans. Though Medicaid members are represented in the data of 6 of these plans, they represent a small minority of enrollees. Therefore, we cannot make reliable inferences about care for low-income children or for those without health insurance. Finally, we lack information on any drugs that were not billed to the health plans. These account for a small proportion of antibiotics overall, but would include drug samples given in the office or those paid for completely out-of-pocket or billed to a second insurer. Furthermore, we have no reason to believe that the fraction of antibiotics dispensed in these ways changed over the period of study.
The substantial decrease in antibiotic prescribing is rightfully hailed as a victory for public health campaigns, but is also likely a reflection of increased patient and clinician awareness of antibiotic overuse and resistance from other sources. Our data highlight how changing diagnostic patterns are intimately connected with current and presumably future antibiotic use rates among children. For example, it may be more fruitful to ask clinicians to use stricter criteria for the somewhat subjective diagnosis of acute sinusitis than to ask them to recommend symptomatic care alone for diagnosed cases, as recommended in guidelines for adults.39,40 As we achieve the goal of reducing unnecessary antibiotic exposure for children, it will be critical to continue to monitor changes in patterns of resistance of common pathogens such as S pneumoniae in the community to gauge the benefit of decreased prescribing. Conversely, it will be critical to carefully monitor rates of mastoiditis and other rare complications of common bacterial infections as clinicians and parents raise their thresholds for using antibiotics. Such ongoing assessment will allow a more fully informed consideration of the risks and benefits of antibiotic use by children in an era of increasing resistance.
This study was conducted by the HMO Research Network Center for Education and Research on Therapeutics and supported by grant U8 HS10391 from the Agency for Healthcare Research and Quality.
We are indebted to Courtney Adams for her tireless work in coordinating the project; Parker Pettus for his meticulous analysis of the data; and the data analysts at each of the participating health plans, without whom this work could not have gone forward.
- Received January 30, 2003.
- Accepted May 27, 2003.
- Address correspondence to Jonathan A. Finkelstein, MD, MPH; Department of Ambulatory Care and Prevention, Harvard Medical School and Harvard Pilgrim Health Care, 133 Brookline Ave, 6th Floor, Boston, MA 02215. e-mail:
- ↵US Interagency Task Force. A Public Health Action Plan to Combat Antimicrobial Resistance. Bethesda, MD: US Interagency Task Force; 2001
- ↵Dowell SF, Marcy SM, Phillips WR, Gerber MA, Schwartz B. Principles of judicious use of antimicrobial agents for pediatric upper respiratory tract infections. Pediatrics.1998;101(suppl) :163– 165
- ↵Smith K. Who’s Minding the Kids? Child Care Arrangements: Spring 1997. Current Population Reports. Washington, DC: US Department of Commerce; 2002
- ↵Schappert MA. Office visits for otitis media: United States, 1975–90. Adv Data Vital Health Stat.1992;214 :1– 20
- ↵Platt R, Davis R, Finkelstein JA, et al. Multicenter epidemiologic and health services research on therapeutics of the HMO Research Network Center for Education and Research Therapeutics (CERTs). Pharmacoepidemiol Drug Safety.2001;10 :377
- ↵International Classification of Diseases, Ninth Revision. Reston, VA: Medicode; 2001
- ↵Breslow NE, Clayton DG. Approximate inference in generalized linear mixed models. J Am Stat Assoc.1993;88 :25– 41
- ↵Dowell SF, Marcy SM, Phillips WR, Gerber MA, Schwartz B. Otitis media—principles of judicious use of antimicrobial agents. Pediatrics.1998;101(suppl) :165– 171
- ↵Lanphear BP, Byrd RS, Auinger P, Hall CB. Increasing prevalence of recurrent otitis media among children in the United States. Pediatrics.1997;99(3) . Available at: http://www.pediatrics.org/cgi/content/full/99/3/e1
- ↵Koopman LP, Smit HA, Heijnen M-LA, et al. Respiratory infections in infants: interaction of parental allergy, child care, and siblings—the PIAMA Study. Pediatrics.2001;108 :943– 948
- Schwartz B, Giebink GS, Henderson FW, Reichler MR, Jereb J, Collet J. Respiratory infections in day care. Pediatrics.1994;94(suppl) :1018– 1020
- ↵Haskins R, Kotch J. Day care and illness: evidence, costs, and public policy. Pediatrics.1986;77(suppl) :951– 982
- ↵Trepka MJ, Belongia EA, Chyou P-H, Davis JP, Schwartz B. The effect of a community intervention trial on parental knowledge and awareness of antibiotic resistance and appropriate antibiotic use in children. Pediatrics.2001;107(1) . Available at: http://www.pediatrics.org/cgi/content/full/107/1/e6
- ↵Mangione-Smith R, McGlynn EA, Elliott MN, Krogstad P, Brook RH. The relationship between perceived parental expectations and pediatrician antimicrobial prescribing behavior. Pediatrics.1999;103 :711– 718
- ↵Takata GS, Chan LS, Shekelle P, Morton SC, Mason W, Marcy SM. Evidence assessment of management of acute otitis media: I. The role of antibiotics in treatment of uncomplicated acute otitis media. Pediatrics.2001;108 :239– 247
- Garbutt JM, Goldstein M, Gellman E, Shannon W, Littenberg B. A randomized, placebo-controlled trial of antimicrobial treatment for children with clinically diagnosed acute sinusitis. Pediatrics.2001;107 :619– 625
- ↵Ioannidis JPA, Lau J Technical report: evidence for the diagnosis and treatment of acute uncomplicated sinusitis in children: a systematic overview. Pediatrics.2001;108(3) . Available at: http://www.pediatrics.org/cgi/content/full/108/3/e57
- ↵van Buchem FL, Peeters MF, van’t Hof MA. Acute otitis media: a new treatment strategy. BMJ.1985;290 :1033– 1037
- ↵Damoiseaux RAMJ, van Balen FAM, Hoes AW, Verheij TJM, de Melker RA. Primary care based randomised, double blind trial of amoxicillin versus placebo for acute otitis media in children aged under 2 years. Med J.2000;320 :350– 354
- ↵O’Brien KL, Dowell SF, Schwartz B, Marcy SM, Phillips WR, Gerber MA. Cough illness/bronchitis—principles of judicious use of antimicrobial agents. Pediatrics.1998;101(suppl) :178– 181
- ↵American Academy of Pediatrics, Subcommittee on Management of Sinusitis and Committee on Quality Improvement. Clinical practice guideline: management of sinusitis. Pediatrics.2001;108 :798– 808
- ↵Selby JV. Linking automated databases for research in managed care setting. Ann Intern Med.2001;127 :719– 724
- Copyright © 2003 by the American Academy of Pediatrics