Acute respiratory tract infection is the most common illness to visit to pediatric health facilities1). Although it has been reported that upper respiratory tract infections (URTIs) are mostly viral in origin and are typically self-limiting in children, unnecessary antibiotics are frequently prescribed23). In Korea, the overall antibiotic prescription rate in pediatric outpatients with URTIs was 55.4% in 20114). Despite the recent decline in antibiotic prescription rates, this rate is still much higher than that of Western countries56).

The proper use of antibiotics has been challenging due to difficulties distinguishing between bacterial and viral infections in clinical practice78). These limitations have inspired a great deal of research into methods for verifying possible viral pathogens. Prior retrospective studies have suggested that a rapid antigen test could reduce the use of antibiotics and the duration of hospitalization910). However, other studies have reported contradictory results11). Recently, reverse transcriptase-polymerase chain reaction (RT-PCR) analyzes have become more sensitive than traditional methods, such as viral culture, direct fluorescent assays, and rapid antigen detection tests12).

The development of a RT-PCR test was expected to result in further increases in diagnostic yield and a reduction in antibiotic prescription13). However, several recent studies in pediatric populations have shown that the introduction of RT-PCR test has not reduced antibiotic use in clinical practice814).

This study was conducted to evaluate the factors associated with antibiotic use in children hospitalized with URTI and to identify the clinical characteristics associated with continued use of prescription antibiotics notwithstanding the detection of a respiratory virus by RT-PCR.

During the study period, RT-PCR assays were performed on a total of 1,277 patients upon admission. Of these, 393 inpatients met the inclusion criteria. The median (interquartile range [IQR]) age of inpatients meeting the inclusion criteria was 23 months (IQR, 13 to 41 months) (Table 1). Acute nasopharyngitis (67.4% , 265/393), acute pharyngotonsillitis (27.2%, 107/393), and acute laryngitis (5.4%. 21/393) were diagnosed in these patients. AOM was observed in 45 children (11.5%) and acute rhinosinusitis in 14 (3.6%). Four patients (1%) were diagnosed with concurrent AOM and acute rhinosinusitis.

Antibiotics were prescribed for 79 patients (20.1%) with URTI upon admission. The demographic and clinical characteristics of the patients who received antibiotics were compared with those of the 314 patients (79.9%) who did not receive antibiotics treatment (Table 2). Among those who received antibiotics, diagnoses of AOM and acute rhinosinusitis were more common (48.1% vs. 2.2%, 16.5% vs. 0.3%, respectively). The median WBC count (12,190/mm³ vs. 10,840/mm³, P=0.006), the median serum level of hS-CRP (2.1 mg/dL vs. 0.7 mg/dL, P<0.001) and the median ESR (23 mm/hr vs. 14 mm/hr, P<0.001) were higher in patients who received antibiotics than in those who did not. Risk factors for initial antibiotics use were AOM, acute rhinosinusitis, and HS-CRP. Patients having AOM, as compared with those who did not, had a relative risk of receiving antibiotics of 91.1 (95% confidence interval, 30.5 to 271.7; P<0.001). The odd ratio for initial antibiotics use, when level of WBC count and ESR increased, did not significantly increased on multivariate analysis.

Antibiotics were used in 18.5% (49/265) children diagnosed with acute nasopharyngitis and in 24.3% (26/107) with acute pharyngotonsilitis (Table 3). A single antibiotic agent was prescribed to 77 patients (97.5%), and combination therapy was administered to two patients. Among single-agent therapies, the penicillin family (68.3%, 54/79) was the most common agent used for each URTI, followed by cephalosporins (29.1%, 23/79).

Respiratory viruses were detected using the RT-PCR test in 223 (56.7%) of 393 patients. Of these, 43 patients (19.3%) were positive for two viruses, two patients (0.9%) were positive for three viruses, and one patient (0.5%) was positive for four concurrent viruses . Adenovirus (28.3%, 52/184) and rhinovirus (27.7%, 51/184) were commonly identified in acute nasopharyngitis. Adenovirus (44.6%, 29/65) was the most frequently found pathgen related to acute pharyngotonsilitis. In patient with acute laryngitis, parainfluenza was more commonly detected (50%, 12/24) (Fig. 1). In one-thirds of patients (17/52) diagnosed with AOM, viruses did not detected. Among the rest, adenovirus (21%) and rhinovirus (15.4%) was identified.

Of the 79 patients who received antibiotics, 35 patient s were discharged or had discontinued antibiotics earlier. The remaining patients (n=44) have treated antibiotics when RT-PCR results had been reported. Antibiotics were continued in 28 patients (63.6%) after clinicians were informed that viruses had been identified using the RT-PCR method. A comparison of the characteristics of children who either continued or discontinued antibiotic therapy after the identification of a viral pathogen is presented in Table 4. AOM was a statistically significant diagnosis in patients who maintained antibiotic usage (P=0.002). There were no significant differences in age, duration of fever, laboratory findings, or length of hospitalization between the two groups.

In this study, we analyzed the factors related to use antibiotics in patients with URTI. Children who were administered antibiotics at admission had a higher prevalence of AOM compared with those who did not. Similarly, the proportion of AOM was also greater in patients who continued antibiotics therapy, even when informed of respiratory virus presence, than in patients who stopped antibiotics. Therefore, diagnosis of AOM might affect antibiotic prescription behavior in children with URTI.

AOM is a major reason for antibiotic use in children1516). The rate of antibiotic prescriptions for AOM in our trial (84.4%) was consistent with rates reported in other Korean studies (86.1%)17). This rate varies from 56% in the Netherlands to 95% in the USA, Canada, and Australia18). In response to concerns about the emergence of antibiotic-resistant microorganisms and the economic burden of disease caused by antibiotic resistance, the need for accurate diagnosis of AOM to minimize unnecessary treatment has been globally agreed upon192021).

However, accurate diagnosis of AOM could be difcfiult in infants and young children because of incooperation with the examination and cerumen barrier of tympanic membrane. Recently published American Academy of Pediatrics guideline of AOM endorses strict otoscopic diagnostic criteria to differentiate AOM from otitis media with effusion19). Therefore, a cautious examination of the tympanic membrane and understanding the wide clinical spectrum of otitis media are required to achieve a high degree of diagnostic accuracy22). In addition to this effort to accurate diagnosis of AOM, “watchful waiting” in selected children without progression of signs or symptoms within 48 to 72 hours before prescribing antibiotics for AOM has been recommended to clinicians1923).

Twenty percent (79/393) of inpatients with URTIs were treated with antimicrobial agents during this study period. This proportion was lower than that of tertiary hospitals analyzed in a recent study evaluating pediatric outpatients with URTI from 2009 to 2011 in Korea (49.5%)4).

Measuring serum CRP as a means of differentiating between viral and bacterial infections is known to be helpful in making the decision whether or not to prescribe antimicrobial agents to children with acute respiratory illness24). A previous study of 176 children determined that serum CRP level exceeding 4 mg/dL probably implied bacterial respiratory infection25). Of this study's subjects, children who received antibiotics upon admission had higher level of hS-CRP than those who did not. However, the difference in the level of mean hS-CRP was not likely to impact on initial antibiotics use in the clinical setting. The lower odd ratio for hS-CRP associated with initial antibiotics prescription could explain its discrepancy.

In the present study, although a virus was detected using an RT-PCR assay, the antibiotic administration was continued in 28 patients (62.2%, 28/45). This finding corresponded with previous studies suggesting that detecting a viral respiratory pathogen using RT-PCR analysis result did not reduce antibiotic use81426). Other randomized trials evaluating the effect of viral RT-PCR results have presumed that similar findings were explained by the clinician's concerns regarding bacterial superinfection in patients with acute respiratory illness8). Although it is accepted that respiratory virus infection predisposes patients to subsequent bacterial respiratory superinfection, the virtual incidence of these serious bacterial infections remains ambiguous1127).

Our study has several limitations. First, the diagnosis of AOM was not made precisely. The results were not deduced based on an exact diagnosis protocol but electronic medical records. Second, the RT-PCR analyses in our study were performed only twice a week, resulting in discharged patients before PCR results was reported. This fact could have decreased the advantage provided by RT-PCR as a rapid diagnostic tool, which could have an effect on antibiotic prescription rates. However, other studies have revealed that rapid communication of RT-PCR results did not reduce antibiotic use28). Third, it is unlikely that positive RT-PCR results always imply a true viral respiratory tract infection. In a recent study that compared viral PCR results from children with respiratory infection to asymptomatic matched controls, respiratory viruses, such as bocavirus, rhinovirus, adenovirus, coronavirus, and enterovirus, were frequently detected in asymptomatic matched controls29). Fourth, this trial was not performed for all inpatients who were diagnosed with an upper respiratory virus but was limited to subjects who received viral RT-PCR analysis. In addition, the performance of our study in a tertiary care center, where antibiotic use is less likely than in a primary care setting, limits the ability to extend our results to a general pediatric clinic. Despite these limitations, the strength of the present study is that it was based on an actual clinical practice.

In this study of children with URTI, antibiotic prescriptions were more frequently found among patients with AOM than among those without AOM. After respiratory viral pathogen was identified using RT-PCR, maintenance of antibiotic treatment were more common in the children with AOM than those without AOM. Therefore, accurate diagnosis of AOM could be a clue in reducing the unnecessary antibiotics use in children with viral URTIs. Well-designed education programs and studies to select patients who might benefit from antimicrobial agents are also important to improve management of AOM. Future prospective studies will be required to develop strategies that can change the prescribing behavior of physicians to minimize unjustified antibiotic use in viral infections.