Antibiotic Prescribing for Canadian Preschool Children: Evidence ofOverprescribing for Viral Respiratory Infections

Elaine E. L. Wang, Thomas R. Einarson,James D. Kellner, and John M. Conly

From the Departments of Pediatrics and Medicine and the Faculty ofPharmacy, University of Toronto, and the Department of Pediatrics and

Department of Microbiology and Infectious Diseases, University ofCalgary, Canada

Antibiotic resistance is associated with prior receipt of antibiotics. An analysis of linked comput-erized databases for physician visits and antibiotic prescriptions was used to examine antibioticprescribing for different respiratory infections in preschool children in Canada. In 1995, 64% of61,165 children aged <5 years made 140,892 visits (mean, 3.6 visits per child) for respiratoryinfections; 74% of children who made visits received antibiotic prescriptions. Antibiotics wereprescribed to 49% of children with upper respiratory tract infection, 18% with nasopharyngitis, 78%with pharyngitis or tonsillitis, 32% with serous otitis media, 80% with acute otitis media, 61% withsinusitis, 44% with acute laryngitis or tracheitis, and 24% with influenza. Acute otitis mediaaccounted for 33% of all visits and 39% of all antibiotic prescriptions. The estimated Canadian-dollar cost of overprescribing was $423,693, or 49% of the total cost of antibiotics ($859,893) used inthis group. This population-based study confirms antibiotic overprescribing in Canada.

In the 50 years since the introduction of penicillin, there hasbeen the perception of control of many microbial diseases.However, in the 1990s, the rise of antimicrobial resistanceamong common organisms—such as the nonsusceptibility topenicillin of 25% of Streptococcus pneumoniae strains isolatedin the United States [1] and of .50% of those isolated in Spain[2], Italy [3], and Hungary [4]—has caused alarm about thepublic health threat of emerging infectious diseases and anti-microbial resistance.

There is a strong association between prior receipt of anti-biotics and development of antibiotic resistance [5–10]. Respi-ratory tract infections account for three-quarters of all antibi-otic prescriptions [11]. Data collected from a pediatric grouppractice [12], a Kentucky Medicaid-claims database [13], andan analysis of pediatric antibiotic prescribing in the NationalAmbulatory Medical Care Survey [14] suggest that 50% ofchildren who see a physician for a viral respiratory tract infec-tion receive antibiotics.

The amount of increase in antibiotic prescription in Europevaries, with a lower rate in the United Kingdom than inGermany or France [15]. In Canada, the number of antibiotic

prescriptions filled in 1996 was 26.3 million [16], in a totalpopulation of nearly 30 million [17]. To determine the magni-tude of antibiotic prescribing for specific respiratory tract in-fections, data from the Saskatchewan Health Databases focus-ing on young children ,5 years of age were obtained for theyear 1995. This age group was targeted because children,particularly those ,5 years of age, receive antibiotics morefrequently than those in older age groups [18, 19]. We alsoexamined patterns of use, costs of antibiotics, rates of inappro-priate antibiotic prescribing, and associated hospitalizations.

Methods

Saskatchewan Health Databases

Saskatchewan, a province in western Canada, had a totalpopulation of 1,015,200 in 1995 [17]. As in all Canadianprovinces, Saskatchewan residents have universal health careaccess through provincially funded health insurance. Physicianclaims, outpatient prescription-drug claims, hospital dis-charges, and health insurance registration can be linked acrossadministrative databases and over time by means of the uniquenumber assigned to each Saskatchewan Health beneficiary[20]. The administrative databases included prescribing infor-mation for the total population of Saskatchewan, with thesingle exception of registered American Indians. As of 30 June1995, 61,165 registrants eligible for prescription drug benefitswere ,5 years of age.

Saskatchewan has a relatively restricted formulary for anti-biotics for treatment of respiratory tract infections; it includesamoxicillin, ampicillin, penicillin V and G benzathine, clox-acillin, erythromycin, cefixime, cephalexin, and tetracycline.Physicians wishing to prescribe agents such as amoxicillin/clavulanate, cefaclor, cefuroxime, clarithromycin, and the

Received 3 November 1998; revised 24 February 1999.The interpretation and conclusions contained herein represent those of the

authors and do not necessarily represent those of the Government of Saskatche-wan or the Saskatchewan Department of Health.

Financial support: The study was funded by grants-in-aid from Abbott Lab-oratories Ltd. Canada, Bristol-Myers Squibb Canada, and Hoechst MarionRoussel Canada.

Reprints or correspondence: Dr. Elaine E. L. Wang, Clinical and MedicalAffairs, Pasteur Merieux Connaught, 1755 Steeles Avenue West, Toronto,Ontario, Canada M2R 3T4 (ewang@ca.pmc_vacc.com).

Clinical Infectious Diseases 1999;29:155–60© 1999 by the Infectious Diseases Society of America. All rights reserved.1058–4838/99/2901–0024$03.00

155

quinolones must fill in forms for “Exception Drug Status”coverage before their patients may receive these agents throughthe drug plan.

The above administrative databases were examined to ex-tract a subgroup of patients whose diagnoses could be classifiedas respiratory tract infections with use of the InternationalClassification of Diseases (ICD-9-CM) diagnostic codes [21].Codes corresponded to the following diagnoses: serous (381)and acute suppurative (382) otitis media, common cold (460),acute sinusitis (461), acute pharyngitis (462), acute tonsillitis(463), acute laryngitis (464), acute upper respiratory tract in-fection (URI; 465), acute bronchitis or bronchiolitis (466),pneumonia (including viral, pneumococcal, and other bacterialpneumonia, pneumonia due to another specific organism, pneu-monia without a causative organism specified, and broncho-pneumonia) (480–486), and influenza (487).

Only one diagnostic code was allowed for each visit. If therewere multiple diagnoses at the visit and an antibiotic wasprescribed, it was assumed that the diagnostic code accuratelyreflected the infection most likely to be of bacterial origin andtherefore likely to respond to antibiotics. An antibiotic pre-scription was attributed to a specific respiratory diagnosis whenit was filled within 7 days of the physician-patient encounter.When there were several visits during a 7-day period, theprescription was attributed to the diagnosis made at the mostrecent visit. The cost of the prescription (which, for purposes ofthe study, consists of the acquisition cost, the markup, and thedispensing fee) was summarized in 1995 Canadian dollars.

Appropriateness of Antibiotic Prescriptions

Wherever possible, antibiotic appropriateness was deter-mined on the basis of evidence-based guidelines. These in-cluded recent publications of a consensus group consisting ofmembers of the Centers for Disease Control and Prevention(Atlanta), the American Academy of Pediatrics, and the Amer-ican Academy of Family Practice, dealing with otitis media,nasopharyngitis, bronchitis, pharyngitis, and sinusitis [22–27].In the guideline for pharyngitis, it was noted that 85% of acutepharyngitis is of viral etiology, for which antibiotics are notindicated [26].

For conditions not reviewed by the above group, furtherinformation was gathered from other consensus guidelines orreference textbooks. For many of the conditions, there have notbeen clinical trials of antibiotic management, which wouldoffer evidence of the highest quality to indicate efficacy of anintervention [28]. In such situations, data on etiologic agentsand appropriateness of antibiotics for treatment against suchagents were used. However, these data constitute a lower levelof evidence than that from randomized trials.

Acute bronchiolitis is a viral infection in which secondarybacterial infections are rare, and there is no benefit fromantibiotic treatment, according to the single randomized clini-

cal trial examining this question [29]. Thus, an expert Canadianconsensus panel recommended against the use of antibiotics ininitial management of bronchiolitis [30]. Another Canadianconsensus panel has recommended against antibiotic treatmentwhen viral pneumonia is suspected on the basis of epidemio-logical factors, presence of wheezing, or identification of viraletiology through antigen detection [31]. Otherwise, antibioticsare recommended.

A single ICD-9-CM code applies for acute laryngitis, croup,and tracheitis. Acute laryngitis is a viral infection, for whichantibiotics are not indicated. Croup is managed with supportivetherapy, racemic epinephrine, or systemic or inhaled steroids asnecessary, but not antibiotics [32]. Both epiglottitis and bacte-rial tracheitis are bacterial infections for which antibiotics areindicated, but they are extremely rare and present as acuteserious illness.

Therefore, we considered the following as inappropriateindications for antibiotics: nonspecific URI, the common cold,serous otitis media, acute bronchitis or bronchiolitis, laryngitis,and influenza; in addition, we considered 85% of antibioticsprescribed for pharyngitis or tonsillitis to be inappropriate.Given difficulties in differentiating etiologic agents in pneu-monia, the heterogeneous nature of other respiratory infections,and the benefit of antibiotic treatment of acute otitis media andsinusitis, these conditions were considered appropriate indica-tions for antibiotic treatment.

Results

A total of 140,892 visits were made during the 1995 calendaryear by 38,848 children in Saskatchewan (3.62 visits per child),of whom 28,929 (74%) received at least one course of antibi-otic. The number of visits and the frequency of antibioticprescribing are listed in table 1. Because a subject may havehad more than one physician visit for a diagnosis, the numberof subjects and the frequency of prescribing per subject are alsoincluded.

Figure 1 shows the distribution of subjects prescribed anti-biotics, according to illness syndrome and costs of those anti-biotics. The most frequent reason for a visit and for an antibi-otic prescription was treatment of acute otitis media,accounting for 39% of all antibiotics prescribed.

The penicillin class of antibiotics, including ampicillin, re-mains the most frequently prescribed. There was variation inthe next most frequently used class of antibiotics, depending onthe clinical diagnosis (figure 2). Specifically, trimethoprim-sulfamethoxazole (co-trimoxazole) was the next most fre-quently used agent for acute otitis media, and erythromycinwas next most frequently used antibiotic for acute URI, pneu-monia, and pharyngitis or tonsillitis.

The mean (6SD) antibiotic cost per patient, as shown intable 2, varied from $13.04 (67.04) for a patient with acommon cold to $23.84 (621.08) for a patient with acute otitis

156 Wang et al. CID 1999;29 (July)

media. Patients with pneumonia were excluded from this com-parison because the use of antibiotics in the hospital was notcaptured.

Associated hospitalizations for the same diagnosis within 30days were generally uncommon. They occurred with 172(0.4%) of the visits for acute URI; 53 (0.3%) of the visits forpharyngitis or tonsillitis; 371 (4.4%) of the visits for bronchitisor bronchiolitis; no visits for a common cold; 25 (0.5%) of thevisits for serous otitis media; 223 (0.5%) of the visits for acuteotitis media; 356 (9.4%) of the visits for acute laryngitis/tracheitis; 272 (6.4%) of the visits for pneumonia; 9 (0.4%) ofthe visits for influenza; no sinusitis visits; and 86 (2.2%) of thevisits for other respiratory conditions. With the exception ofacute bronchitis or bronchiolitis, acute laryngitis or tracheitis,and pneumonia, morbidity for the other syndromes is low, asindicated by low associated hospitalization rates.

If the evidence-based management guidelines were fol-

lowed, antibiotics should have been withheld from all patientswith acute URI, acute bronchitis and bronchiolitis, the commoncold, influenza, serous otitis media, and acute laryngitis. Onecould also withhold antibiotics from 85% of those with acutetonsillitis or pharyngitis. Under these assumptions, 33,680 of atotal of 66,419 courses of antibiotics (51%) could have beeneliminated in 1995. The cost of overprescribing is in the orderof $423,693, 49% of the $859,893 total expenditure for outpa-tient antibiotics for respiratory tract infections in this agegroup. Acute otitis media was the most common diagnosiscontributing to antibiotic prescription costs, accounting for$369,296.

Discussion

Provincial health insurance allows Canadians to have returnvisits to their physicians without the burden of out-of-pocket

Table 1. Frequency of diagnoses and antibiotic prescriptions for Canadian preschool children.

Appropriateness of antibioticprescription; diagnosis No. of visits

Percentage of visitsresulting in prescription

No. ofsubjects

Percentage of subjectsreceiving antibiotics

Usually inappropriateAcute URI 43,160 36 23,255 49Acute bronchitis or bronchiolitis 8,504 50 5,473 65Common cold 5,963 16 4,743 18Serous otitis media 5,039 21 2,515 32Acute laryngitis, croup 3,795 34 2,647 44Influenza 2,340 22 1,971 24

Possibly inappropriateAcute pharyngitis or tonsillitis 16,445 69 12,322 76

Usually appropriateOther respiratory condition 3,887 32 3,157 35Acute otitis media 46,789 57 19,359 80Pneumonia 4,251 33 2,488 51Acute sinusitis 693 56 588 61

NOTE. URI 5 upper respiratory tract infection.

Figure 1. Contribution to totalantibiotic prescription costs for re-spiratory infections, by diagnosis(URI 5 upper respiratory tract in-fection).

157Antibiotic Overprescribing for Preschool ChildrenCID 1999;29 (July)

costs, a practice that may lead to differences in the frequencyof antibiotic prescribing as compared with that in the UnitedStates. The strength of these observations lies in the ability ofthe Saskatchewan Health Databases to capture the vast major-ity of all physician visits and outpatient antibiotic prescriptions.Our finding that ;50% of antibiotics prescribed are not indi-cated on the basis of evidence-based guidelines is remarkablyconsistent with rates reported from the United States [11–13,33].

In the United Kingdom, antibiotic prescribing increased by46% between 1980 and 1991 [15], and overprescribing ingeneral practice, especially for antibiotics, is estimated to cost275 million pounds annually [34]. Half of all antibiotic pre-scriptions in the United Kindgom were for respiratory tractinfections [35]. Thus, our findings likely have wide generaliz-ability.

There were limitations to our study. The disease to which theantibiotic is attributed is dependent on the diagnostic codeindicated on the physician claim form. For the purpose ofdetermining antibiotic appropriateness, the listed diagnosis wasassumed to be the indication for the antibiotic. Because claimforms allow only one disease code to be entered, it is possiblethat an uncoded illness was in fact the one for which antibioticswere indicated. We have observed in another study of prescrib-ing by pediatricians for respiratory tract infection that thesecodes may be incorrect in up to 41% of cases when comparedwith chart diagnoses [36].

It was assumed that physicians’ diagnoses were correct.Physicians may diagnose certain conditions as bronchitis, si-nusitis, or otitis media to justify antibiotic prescriptions [37,38]. This would lead to an underestimate of inappropriateantimicrobial prescribing.

Another limitation of this study is the inability to capture

antibiotic treatment initiated during a hospitalization. The lowrates of antibiotic prescription for pneumonia, for example, arelikely due to a selection bias to include milder cases or are dueto capture of follow-up visits during recovery. One wouldpredict that most patients with pneumonia had been hospital-ized, and antibiotics initiated in that setting are not captured inthe databases included in this study. It is assumed that many ofthe visits were follow-up visits, when no further antibioticcourses were being prescribed.

It is not possible to differentiate multiple visits for the sameillness vs. single visits for illnesses recurring in the same year.Because physicians often have patients with acute otitis mediaor sinusitis return for follow-up visits, this may explain the lowrate of prescribing for acute otitis media and sinusitis, as thesecond visit usually does not prompt a subsequent antibioticprescription. Similarly, it is not possible to determine whethersome of the antibiotic courses were prescribed only at thesecond visit, because of lack of improvement.

The estimate in the guidelines that 85% of pharyngitis casesare of viral etiology was based on studies that included olderchildren [26]. In this population of young children, the propor-tion of pharyngitis that is viral in origin may be closer to 100%.Thus, the estimate of antibiotic overuse for viral infections maybe low.

The relatively restricted formulary in Saskatchewan likelylowers the cost of agents and limits the variation in types ofantibiotics prescribed through the drug plan. Where there is alarger number of drugs included in a formulary, one couldpredict greater use of newer, more expensive agents. Further-more, actual prescribing may be underestimated in that thedatabases capture only antibiotic prescriptions that are actuallyfilled; if prescriptions were not filled by families, they wouldnot be noted in the database.

Only a single year was sampled, which fell during a periodbefore media reports of increasing antibiotic resistance and of

Figure 2. Number of prescriptions, per antibiotic class, used intreatment of common respiratory infections (Erythro-sulfa 5 eryth-romycin-sulfisoxazole; URI 5 upper respiratory tract infection).

Table 2. Antibiotic treatment costs (in 1995 Canadian $), by diag-nosis, for preschool children in Canada.

DiagnosisMean cost/patient

(SD)Total antibiotic cost

for 1995

Acute URI 17.08 (11.97) 192,813Acute bronchitis or

bronchiolitis 16.37 (10.42) 58,017Common cold 13.04 (7.04) 10,977Serous otitis media 20.63 (15.51) 16,587Acute laryngitis 14.25 (7.63) 16,744Influenza 13.82 (6.75) 6,606Acute pharyngitis or tonsillitis 16.83 (11.77) 143,475Other respiratory condition 14.92 (9.11) 16,488Acute otitis media 23.84 (21.08) 369,296Pneumonia 18.46 (10.81) 23,133Acute sinusitis 16.09 (13.19) 5,759

Total 859,893

NOTE. URI 5 upper respiratory tract infection.

158 Wang et al. CID 1999;29 (July)

educational efforts aimed at reducing inappropriate antibioticprescribing. A follow-up study is warranted to determinewhether some of these educational and media efforts directedtoward parents have substantially affected antibiotic prescrip-tion rates.

Comparisons of prescribing by type of physician (pediatri-cian vs. family practitioner) were not made. A higher rate ofinappropriate antibiotic prescribing has been described withregard to family practitioners than to pediatricians in the man-agement of acute laryngotracheobronchitis [39] and purulentnasopharyngitis [37]. However, in Saskatchewan, the majorityof primary care of children is delivered by family practitioners,while pediatricians are mainly consultants.

Despite study limitations, the major contribution to antibi-otic exposure is obvious for inappropriate antibiotic therapy forfrequently seen nonbacterial conditions such as acute URI.With use of evidence-based guidelines, almost half of the costsincurred for antibiotic treatment of respiratory infections couldhave been saved. There was also inappropriate selection ofantibiotics, which suggests a lack of differentiation betweenillness syndromes by some prescribers.

For example, ;6% of children with pharyngitis receivedtrimethoprim-sulfamethoxazole, which would not be appropri-ate for treatment against group A streptococci. However, afterpenicillins there is a difference in the next most frequentantibiotic used for different syndromes, suggesting that mostphysicians do consider the syndromes different. Such substan-tial savings would be supplemented by the savings associatedwith the occurrence of fewer infections due to antibiotic-resistant organisms, which cause twice as much morbidity andmortality than susceptible organisms causing the same infec-tions [40].

A number of factors other than the clinical condition influ-ence antibiotic prescribing. Such factors include lack of cer-tainty of diagnosis [41], heavy workload and inadequate timefor each patient [42], fear of litigation for not treating a bac-terial infection, the belief that antibiotics are innocuous, socialand psychological characteristics of the patient [43], and per-ceived pressure from parents to prescribe [44–46]. Guidelinesmust be adapted for use in the office-based setting, with rec-ognition of the existence of these potential obstacles. Theacceptance of these practice guidelines by both the generalpublic and physicians may be enhanced if the guidelines stressthe harm of inappropriate antibiotic prescription [22].

Acknowledgments

Data abstraction was performed under the supervision of Mary-Rose Stang, Ph.D. (consultant, Research Services, SaskatchewanHealth). This study was based in part on data provided by theSaskatchewan Department of Health.

References

1. Butler JC, Hofmann J, Cetron MS, et al. The continued emergence ofdrug-resistant Streptococcus pneumoniae in the United States: an update

from the Centers for Disease Control and Prevention’s PneumococcalSentinel Surveillance System. J Infect Dis 1996;174:986–93.

2. Nava JM, Bella F, Garau J, et al. Predictive factors for invasive disease dueto penicillin-resistant Streptococcus pneumoniae: a population-basedstudy. Clin Infect Dis 1994;19:884–90.

3. Marchese A, Debbia EA, Arvigo A, Pesce A, Schito GC. Susceptibility ofStreptococcus pneumoniae strains isolated in Italy to penicillin and tenother antibiotics. J Antimicrob Chemother 1995;36:833–7.

4. Marton A. Pneumococcal antimicrobial resistance: the problem in Hun-gary. Clin Infect Dis 1992;15:106–11.

5. Arason VA, Kristinsson KG, Sigurdsson JA, Stefansdottir G, Molstad S,Gudmundsson S. Do antimicrobials increase the carriage rate of peni-cillin resistant pneumococci in children? Cross sectional prevalencestudy. BMJ 1996;313:387–91.

6. Block SL, Harrison CJ, Hedrick JA, et al. Penicillin-resistant Streptococ-cus pneumoniae in acute otitis media: risk factors, susceptibility patternsand antimicrobial management. Pediatr Infect Dis J 1995;14:751–9.

7. Boken DJ, Chartrand SA, Goering RV, Kruger R, Harrison CJ. Coloniza-tion with penicillin-resistant Streptococcus pneumoniae in a child-carecenter. Pediatr Infect Dis J 1995;14:879–84.

8. Duchin JS, Breiman RF, Diamond A, et al. High prevalence of multidrug-resistant Streptococcus pneumoniae among children in a rural Kentuckycommunity. Pediatr Infect Dis J 1995;14:745–50.

9. Leibovitz E, Raiz S, Piglansky L, et al. Resistance pattern of middle earfluid isolates in acute otitis media recently treated with antibiotics.Pediatr Infect Dis J 1998;17:463–9.

10. Radetsky MS, Istre GR, Johansen TL, et al. Multiply resistant pneumo-coccus causing meningitis: its epidemiology within a day-care centre.Lancet 1981;2:771–3.

11. McCaig LF, Hughes JM. Trends in antimicrobial drug prescribing amongoffice-based physicians in the United States. JAMA 1995;273:214–9.

12. Arnold KE, Leggiadro RJ, Breiman RF, et al. Risk factors for carriage ofdrug-resistant Streptococcus pneumoniae among children in Memphis,Tennessee. J Pediatr 1996;128:757–64.

13. Mainous AG, Hueston WJ, Clark JR. Antibiotics and upper respiratoryinfection: do some folks think there is a cure for the common cold? JFam Pract 1996;42:357-61.

14. Nyquist A-C, Gonzales R, Steiner JF, Sande MA. Antibiotic prescribingfor children with colds, upper respiratory tract infections, and bronchitis.JAMA 1998;279:875–7.

15. Davey PG, Bax RP, Newey J, et al. Growth in the use of antibiotics in thecommunity in England and Scotland in 1980–93. BMJ 1996;312:613.

16. Consensus conference: controlling antimicrobial resistance. An integratedaction plan for Canadians. Can Commun Dis Rep 1997;23(suppl 7):S1–32.

17. Statistics Canada. Canadian statistics: population. CANSIM (StatisticsCanada’s online statistical database), Matrices 6367–79. Available at:http://www.statcan.ca/english/pgdb/people/population/demo02.htm.

18. Reichler MR, Allphin AA, Breiman RF, et al. The spread of multiplyresistant Streptococcus pneumoniae at a day care center in Ohio. J InfectDis 1992;166:1346–53.

19. Bergus GR, Levy AT, Levy SM, Slager SL, Kiritsy MC. Antibiotic useduring the first 200 days of life. Arch Fam Med 1996;5:523–6.

20. Miller E, Blatman B, Einarson TR. A survey of population-based drugdatabases in Canada. Can Med Assoc J 1996;154:1855–64.

21. International classification of diseases, 9th revision. Clinical modification.Washington, DC: Public Health Service, U.S. Department of Health andHuman Services, 1988.

22. Dowell SF, Marcy SM, Phillips WR, Gerber MA, Schwartz B. Principlesof judicious use of antimicrobial agents for pediatric upper respiratorytract infections. Pediatrics 1998;101:163–5.

23. Dowell SF, Marcy SM, Phillips WR, Gerber MA, Schwartz B. Otitismedia—principles of judicious use of antimicrobial agents. Pediatrics1998;101:165–71.

159Antibiotic Overprescribing for Preschool ChildrenCID 1999;29 (July)

24. O’Brien KL, Dowell SF, Schwartz B, Marcy SM, Phillips WR, GerberMA. Cough illness/bronchitis—principles of judicious use of antimi-crobial agents. Pediatrics 1998;101:178–81.

25. Rosenstein N, Phillips WR, Gerber MA, Marcy SM, Schwartz B, DowellSF. The common cold—principles of judicious use of antimicrobialagents. Pediatrics 1998;101:181–4.

26. Schwartz B, Marcy SM, Phillips WR, Gerber MA, Dowell SF.Pharyngitis—principles of juidicious use of antimicrobial agents. Pedi-atrics 1998;101:171–4.

27. O’Brien KL, Dowell SF, Schwartz B, Marcy SM, Phillips WR, GerberMA. Acute sinusitis—principles of judicious use of antimicrobialagents. Pediatrics 1998;101:174–7.

28. Woolf SH, Battista RN, Anderson G, Logan A, Wang E, other members ofthe Canadian Task Force on the Periodic Health Examination. Assessingthe clinical effectiveness of preventive maneuvers: analytic principlesand systematic methods in reviewing evidence and developing clinicalpractice recommendations. J Clin Epidemiol 1990;43:891–905.

29. Fris B, Andersen P, Brenoe E, et al. Antibiotic treatment of pneumonia andbronchiolitis: a prospective randomised study. Arch Dis Child 1984;59:1038–45.

30. Law BJ. Hospital management of respiratory syncytial virus infection. CanJ Infect Dis 1993;4:307–8.

31. Jadavji T, Law B, Lebel MH, Kennedy WA, Gold R, Wang EEL. Apractical guide for the diagnosis and treatment of pediatric pneumonia.Can Med Assoc J 1997;156:S703–11.

32. Kenna MA. Laryngotracheobronchitis. In: Johnson JT, Yu VL, eds. In-fectious diseases and antimicrobial therapy of the ears, nose, and throat.1st ed. Philadelphia: WB Saunders, 1997:389–95.

33. Gonzales R, Steiner JF, Sande MA. Antibiotic prescribing for adults withcolds, upper respiratory tract infections, and bronchitis by ambulatorycare physicians. JAMA 1997;278:901–4.

34. Working Party of the British Society for Antimicrobial Chemotherapy.Hospital antibiotic control measures in the UK. J Antimicrob Che-mother 1994;34:21–42.

35. MacDonald TM, Collins D, McGilchrist MM, et al. Antimicrobial prac-tice: the utilisation and economic evaluation of antibiotics prescribed inprimary care. J Antimicrob Chemotherapy 1995;35:191–204.

36. Arnold SR, Allen UD, al Zahrani M, Tan DHS, Wang EEL. Antibioticprescribing by pediatricians for respiratory tract infection in children.Clin Infect Dis 1999 (in press).

37. Schwartz RH, Freij BJ, Ziai M, Sheridan MJ. Antimicrobial prescribingfor acute purulent rhinitis in children: a survey of pediatricians andfamily practitioners. Pediatr Infect Dis J 1997;16:185–90.

38. Vinson DC, Lutz LJ. The effect of parental expectations on treatmentof children with a cough: a report from ASPN. J Fam Pract 1993;37:23–7.

39. Pianosi P, Feldman W, Robson MG, McGillivray D. Inappropriate use ofantibiotics in croup at three types of hospitals. Can Med Assoc J1986;134:357–9.

40. Holmberg SD, Solomon SL, Blake PA. Health and economic impacts ofantimicrobial resistance. Rev Infect Dis 1987;9:1065–78.

41. MacFarlane J, Lewis SA, MacFarlane R, Holmes W. Contemporary use ofantibiotics in 1089 adults presenting with acute lower respiratory tractillness in general practice in the U. K.: implications for developingmanagement guidelines. Resp Med 1997;91:427–34.

42. Bain J. Justification for antibiotic use in general practice. BMJ 1990;300:1006–7.

43. Howie JR. Clinical judgment and antibiotic use in general practice. BMJ1976;2:1061–4.

44. Mayefsky JH, el-Shinaway Y, Kelleher P. Families who seek care for thecommon cold in a pediatric emergency department. J Pediatr 1991;119:933–4.

45. Cowan PF. Patient satisfaction with an office visit for the common cold. JFam Pract 1987;24:412–3.

46. Hamm RM, Hicks RJ, Bemben DA. Antibiotics and respiratory infections:are patients more satisfied when expectations are met? J Fam Pract1996;43:56–62.

160 Wang et al. CID 1999;29 (July)