BACKGROUND AND OBJECTIVES: The Infectious Diseases Society of America recommends that clinicians forego testing for group A Streptococcal (GAS) pharyngitis in patients with clinical features of viral illness. The prevalence of viral features in patients tested for GAS pharyngitis is not known. The objectives of this study were as follows: to describe the prevalence of viral features in pediatric patients for whom rapid antigen detection tests (RADTs) for GAS pharyngitis are performed; and to compare the prevalence of GAS and the sensitivity of the RADT in patients with and without viral features.
METHODS: This secondary analysis of data from a prospective cohort study included children aged 3 to 21 years for whom RADTs were performed for sore throat in an urban tertiary care emergency department. The primary outcome was the prevalence of viral features, defined as cough, rhinorrhea, oral ulcers/vesicles, and/or conjunctival injection. Secondary outcomes were the prevalence of GAS and sensitivity of the RADT; these outcomes were compared between patients with and without viral features.
RESULTS: Overall, 63% (95% confidence interval [CI]: 57%–68%) of patients had at least 1 viral feature. The prevalence of GAS pharyngitis was higher in patients without viral features (42% [95% CI: 33%–51%]) than in patients with viral features (29% [95% CI: 23%–35%]) (P = .01). The sensitivity of the RADT was 84% (95% CI: 77%–91%) and was not significantly different in patients with and without viral features.
CONCLUSIONS: Because many asymptomatic children are carriers of GAS, judicious use of laboratory testing for GAS pharyngitis remains an important target for antimicrobial stewardship.
- CI —
- confidence interval
- ED —
- emergency department
- GAS —
- group A Streptococcal
- IDSA —
- Infectious Diseases Society of America
- RADT —
- rapid antigen detection test
What’s Known on This Subject:
The Infectious Diseases Society of America recommends that clinicians forego laboratory testing for group A Streptococcal (GAS) pharyngitis in patients with clinical features of viral illness, as positive test results in these patients suggest GAS carriage rather than true infection.
What This Study Adds:
This study shows that nearly two-thirds of patients tested for GAS pharyngitis have overt clinical features of viral illness. Judicious use of laboratory testing for GAS pharyngitis remains the most important way to avoid overtreating carriers of GAS.
Sore throat is one of the most common reasons for pediatric ambulatory care visits and is the presenting symptom in ∼12 million pediatric visits per year in the United States.1,2 Because there is overlap between the symptoms and signs of viral and group A Streptococcal (GAS) pharyngitis, clinicians cannot reliably differentiate these 2 entities on the basis of clinical grounds alone.3 Accordingly, the Infectious Diseases Society of America (IDSA) recommends laboratory testing with a rapid antigen detection test (RADT) and/or throat culture to distinguish viral pharyngitis from GAS pharyngitis except when “overt viral features,” such as rhinorrhea, cough, oral ulcers, conjunctivitis, stomatitis, or hoarseness, are present.4
Although RADTs and throat cultures are highly sensitive and specific for the detection of GAS on the pharynx, these tests cannot distinguish patients with GAS pharyngitis from pharyngeal carriers of GAS.5,6 Because 5% to 21% of asymptomatic children are GAS carriers,7 performing RADTs on patients with overt viral features may result in the identification of a large number of GAS carriers who are unlikely to benefit from antimicrobial therapy. To reduce the duration of symptoms and prevent suppurative and nonsuppurative complications of GAS pharyngitis, clinicians may feel compelled to treat symptomatic patients for whom laboratory test results for GAS are positive. As a result, identifying and foregoing laboratory testing in patients with overt viral features is an important component of antimicrobial stewardship because it may prevent overtreatment of GAS carriers. The prevalence of overt viral features in ambulatory pediatric patients for whom laboratory testing for GAS pharyngitis is performed according to usual practice is unknown.
The present study had 3 objectives. The primary objective was to determine the prevalence of overt viral features in children with sore throat for whom laboratory testing for GAS pharyngitis was performed. This prevalence reflects the extent to which IDSA recommendations for testing are followed in practice. A secondary objective was to compare the prevalence of GAS in patients with and without viral features. Significant differences in the prevalence of GAS between patients with and without viral features would provide support for the IDSA’s recommendation to forego testing in patients with overt viral features. Finally, a tertiary objective was to compare the sensitivity of the RADT in patients with and without viral features. Previously, significant differences in the sensitivity of the RADT were shown in patients with different severities of disease as measured by using the Centor score (a property known as the spectrum bias)8; however, the sensitivity of the test in patients with high scores was still low enough to justify performing cultures when RADT results were negative. We sought to determine whether the sensitivity of the test in patients with fewer viral features was higher than previously measured sensitivities, a finding that might support foregoing cultures in a subset of patients with negative RADT results and a high pretest probability of GAS disease.
Study Design and Population
This secondary analysis of data from a prospective cohort study included children between 3 and 21 years of age who presented to an urban tertiary care emergency department (ED) between October 1, 2013, and January 31, 2015.9 Patients were included in the study if they had a sore throat and a RADT performed during the study period. Patients were excluded if they were unable to communicate symptoms, had a high-acuity triage level (<3 on a 5-point scale), had a comorbid medical condition (significant cardiac, renal, or pulmonary disease; global developmental delay; or were immunocompromised), had been evaluated by a medical professional for sore throat during the preceding 48 hours, reported symptoms for >7 days, or had received antibiotics within 7 days of presentation. In our ED, nurses and physicians share responsibility for performing RADTs. Theoretically, RADTs should be performed in accordance with the ED’s clinical practice guideline for sore throat, which reflects current IDSA guidelines. A complete description of data collection procedures is available elsewhere.9
As part of the study, attending-level physicians completed a history and physical examination of all patients. Thereafter, these physicians documented the presence of specific clinical symptoms and signs in a standardized Web-based survey. Symptoms and signs included all elements of the modified Centor score,10 as well as features of viral illness. In accordance with IDSA guidelines, cough, rhinorrhea, conjunctival injection, and oral ulcers/vesicles were considered to be features of viral illness.4
Per the protocol at our institution, RADTs were performed on all pharyngotonsillar swab samples at the point of care. In accordance with guidelines from the IDSA and the American Academy of Pediatrics, throat cultures were performed whenever the RADT result was negative.4,11
The primary outcome in the present study was the proportion of patients who had features of viral illness, defined as cough, rhinorrhea, conjunctival injection, and/or oral vesicles/ulcers. We calculated the proportion of patients with any viral feature, the proportion with each individual feature, and the proportion with multiple features.
Secondary outcomes included the prevalence of GAS and the sensitivity of the RADT in patients with and without viral features. The prevalence of GAS was calculated as the percentage of patients who tested positive for GAS according to results of either RADT or culture. The sensitivity of the RADT was calculated as the number of patients with a positive RADT result divided by the number of patients with a positive culture finding. Because cultures are not routinely performed after a positive RADT result, and because the specificity of the RADT approaches 100%, we assumed culture positivity in all patients who had a positive RADT result.
Proportions were compared by using the χ2 test or, where appropriate, Fisher’s exact test. Tests for trends in proportions were assessed by using the Cochrane-Armitage test. Normal approximation methods were used to calculate 95% confidence intervals (CIs). Significance was considered at 2-sided P values <.05. All analyses were performed by using Stata 14 (Stata Corp, College Station, TX). This study was approved by the institutional review board at Boston Children’s Hospital.
Of the 1069 patients screened for possible inclusion in the study, 320 were enrolled. Figure 1 contains a flow diagram of included patients. Characteristics of patients included in the study are presented in Table 1.
Prevalence of Viral Features
Among all patients, 63% (95% CI: 57%–68%) had at least 1 viral feature, and 30% (95% CI: 25%–35%) had >1 viral feature (Table 2). The most prevalent viral features were cough (49% [95% CI: 43%–54%]) and rhinorrhea (40% [95% CI: 35%–45%]). Rhinorrhea was significantly more prevalent in patients who tested negative for GAS (44% [95% CI: 38%–51%]) than in those who tested positive for GAS (30% [95% CI: 22%–29%]) (P = .02).
Prevalence of GAS
The prevalence of GAS among all patients in the study was 34% (95% CI: 28%–39%). The prevalence of GAS was higher in patients with no viral features (42% [95% CI: 33%–51%]) than in patients with at least 1 viral feature (29% [95% CI: 22%–35%]) (P = .01). There was a statistically significant trend toward decreased GAS prevalence as the number of viral features increased (Fig 2).
Sensitivity of the RADT
The sensitivity of the RADT was 84% (95% CI: 77%–91%). There were no statistically significant differences in the sensitivity of the RADT in patients with varying numbers of viral features (Fig 3).
Despite the IDSA’s recommendation to forego laboratory testing for sore throat in children with overt viral features, approximately two-thirds of patients for whom RADTs were performed for sore throat in our ED had at least 1 viral feature. In addition, we found that GAS was less prevalent in patients with viral features than in patients without viral features and that the number of viral features was associated with GAS status. Finally, we detected no statistically significant variation in the sensitivity of the RADT according to number of viral features.
To our knowledge, this study is the first to assess the prevalence of viral features in a cohort of pediatric patients for whom RADTs were performed via usual care. Although some patients with true GAS pharyngitis may have viral features, the finding that almost two thirds of patients in our study had these features is an indication that we may be testing a large number of patients who are unlikely to have GAS pharyngitis. Because typical practice is to treat symptomatic patients with positive laboratory test results for GAS, indiscriminate testing of all patients with sore throat without consideration of the presence of viral features may lead to excessive antibiotic prescribing for GAS carriers. Furthermore, these findings may provide a conservative estimate of overtesting because many patients with symptoms other than sore throat (eg, fever, abdominal pain) may also be tested inappropriately for GAS.
There are several possible reasons why patients with sore throat were tested for GAS despite having overt viral features. First, inappropriate testing may be a negative consequence of the workflow in the ED. To improve efficiency, nurses often perform the RADT in triage before a physician examines the patient. It is possible that some nurses are less familiar than their physician counterparts with IDSA guidelines and/or the clinical features that suggest viral illness. Second, physicians may face similar gaps in understanding and may not appropriately consider the patient- and population-level costs of overtesting and overtreating GAS carriers. Third, parents’ preferences may play a role in determining testing practices. Numerous studies have described the relationship between parents’ expectations about receiving antibiotics and clinicians’ prescribing practices,12–15 and it is conceivable that similar expectations about testing for GAS influence clinicians’ decisions to perform testing.
Another possible explanation for overtesting of patients with viral features is the inappropriate use of clinical scoring systems in the management of patients with sore throat. Although scores devised by Centor, Wald, and Breese have been validated in clinical practice for the identification of patients who are unlikely to have GAS pharyngitis, even patients satisfying all criteria in the scoring systems have GAS pharyngitis in only 35% to 75% of cases.10,16–19 Accordingly, clinicians must understand that these scoring systems, together with a consideration of the viral features which do not appear in many of the scoring systems, are most useful for determining who should not have laboratory testing performed. Reframing clinicians’ decision-making process toward identifying patients without GAS pharyngitis, rather than trying to “rule in” GAS pharyngitis, may result in more appropriate testing and fewer unnecessary courses of antibiotic treatment.
Our study confirms findings from previous studies that patients with viral features have a lower prevalence of GAS and thus supports IDSA recommendations to avoid laboratory testing in this low-risk population. The absolute prevalence of GAS in our study (34%) was similar to estimates of prevalence from 2 meta-analyses.7,20 In addition, 1 of these meta-analyses found that the absence of cough (positive likelihood ratio: 1.5) and the absence of rhinorrhea (positive likelihood ratio: 1.3) were predictors of GAS pharyngitis.20 Our data revealed an inverse relationship between the number of viral features and the prevalence of GAS. Compared with patients with no viral features, patients with any viral feature were ∼30% less likely to have GAS, and patients with ≥2 features were >40% less likely to have GAS.
Assuming that patients with positive laboratory test results for GAS receive antibiotics (a practice that is customary in our ED and is supported by IDSA recommendations), interventions to promote judicious use of testing for GAS could significantly reduce the number of antibiotic prescriptions for sore throat in the substantial proportion of children with sore throat and viral illness who are carriers of GAS. Judicious use of antibiotics in these patients may help to prevent antibiotic resistance21 and reduce antibiotic-associated adverse events such as allergic reactions and Clostridium difficile infections.22 In addition, there is evidence that prescribing antibiotics for sore throat may produce a “medicalizing” effect, whereby patients who are prescribed antibiotics are more likely to return to care for uncomplicated and self-limited upper respiratory tract infections.23 Given that ∼12 million children per year seek care for sore throat in ambulatory settings, and that ∼60% of these children receive antibiotics, antibiotic stewardship in children with sore throat remains a public health priority.1
These harms of antibiotics are particularly important given the potentially evolving role of antibiotics in the treatment of GAS pharyngitis. Although antibiotics reduce the duration of symptoms in patients with GAS pharyngitis, the absolute benefit of antibiotics in reducing the risk of rheumatic fever has likely decreased in developed countries, where rheumatic fever is now extremely rare.24–26 Similarly, although early randomized controlled trials found a modest reduction in rate of suppurative complications (eg, peritonsillar abscess) with antibiotic treatment,27 more recent large observational studies have shown that these complications are often discovered at the first presentation of sore throat and are only rarely a result of withholding antibiotics for GAS pharyngitis.28,29 Limiting laboratory testing in patients with viral features might cause providers to overlook, or delay definitive diagnosis in, a small number of patients with true GAS pharyngitis. However, this growing body of data suggests that the costs of missing cases of GAS pharyngitis have decreased while the dangers of overprescribing antibiotics have grown.
Although the present study found that the prevalence of GAS decreased with increasing number of viral features, we found no statistically significant differences in the sensitivity of the RADT in patients with varying numbers of viral features. In addition, the sensitivity of the RADT (88%) in patients with no viral features was still too low to support foregoing backup cultures for negative RADT results in these patients. Previous studies have documented a spectrum bias in the sensitivity of the RADT among patients with GAS pharyngitis, whereby the sensitivity was improved in patients with higher modified Centor scores.8,30 The objective in those studies was to determine whether the sensitivity of the RADT was high enough to forego culture confirmation of negative RADT results in certain patients with a high pretest probability of disease. Although the authors of previous studies concluded that sensitivity was too low in all patients to forego cultures after a negative RADT result, they hypothesized that the high degree of discordance between RADT and culture results (ie, the low sensitivity of the RADT) in patients with low pretest probability of disease could reflect a higher bacterial burden of GAS in patients with more severe disease than in those with mild disease or GAS carriage. Additional studies and/or testing methods are needed to confirm this hypothesis. The point estimates for the sensitivity of the RADT in our study decreased with the number of viral features; however, the differences in sensitivity were not statistically significant. Given that only 107 patients in our study had RADT or culture results positive for GAS, the study may have lacked the statistical power to detect potential variation in the sensitivity of the test among patients with varying numbers of viral features. Alternatively, patients with no viral features may need additional clinical or epidemiologic characteristics for the RADT to perform well enough to obviate the need for cultures after negative RADT results in these patients. Additional studies should assess for a spectrum bias in patients with other combinations of clinical and epidemiologic features.
The present study has several limitations. First, our data, which were collected during usual care in 1 large academic pediatric ED, may not reflect testing practices in all ambulatory care settings. For example, testing practices may differ in centers where physicians (rather than nurses) perform the initial evaluation of all patients with sore throat, in EDs that follow different diagnostic algorithms, or in nonacademic settings such as private urgent care or walk-in clinics. Second, although we determined the prevalence of viral features among patients for whom RADTs were performed, our data did not include the symptoms and signs of patients with sore throat for whom laboratory testing was not performed. These data suggest that a large proportion of RADTs were not performed in accordance with IDSA recommendations; however, an analysis of the symptoms and signs of all patients with sore throat would help to determine the true extent to which IDSA guidelines are implemented. Third, the 320 patients analyzed in this study were only a fraction of the 1069 patients who were screened for inclusion. Although we acknowledge that excluding patients from the analysis may have influenced our results, we believe that excluding patients with comorbid medical conditions was important to control for potential differences in case mix at our tertiary care ED. Fourth, the relatively small sample size may have prevented us from detecting statistically significant differences in the prevalence of GAS or the sensitivity of the RADT in specific subsets of patients. Furthermore, because this study was a secondary analysis of data collected for a different purpose, we do not have information about all viral features. If other viral features such as hoarseness or exanthemata were present in our study population, our findings may then have underestimated the true prevalence of viral features. Finally, because it is not standard practice in our ED to perform throat cultures in patients with positive RADT results, we assumed (for the purpose of calculating the sensitivity of the RADT) that any patient with a positive RADT result would also have a positive culture result. To the extent that there were false-positive RADT findings, which are rare given the high specificity of this test, we then would have overestimated the test’s sensitivity.
Even with highly accurate RADTs, distinguishing viral pharyngitis from GAS pharyngitis remains a challenging and important component of antimicrobial stewardship. Our study suggests that a large proportion of patients tested for GAS pharyngitis have symptoms that are more consistent with viral illness than with true GAS infection. Until we have laboratory tests that can accurately distinguish between GAS infection and GAS carriage, judicious use of RADTs will remain the most important method to avoid unnecessary treatment of GAS carriers. Future studies are warranted to validate these findings in larger and different populations and to assess the extent to which IDSA recommendations for testing are implemented in all patients with sore throat (not just in those for whom laboratory testing is performed). In addition, quality improvement initiatives should promote more consistent implementation of current recommendations for GAS testing.
- Accepted February 22, 2017.
- Address correspondence to Daniel J. Shapiro, MD, Boston Combined Residency Program, Boston Children’s Hospital and Boston Medical Center, Menino Building, Dowling 3 South Resident Mailboxes, 1 Boston Medical Center Place, Boston, MA 02118. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Shulman ST,
- Bisno AL,
- Clegg HW, et al; Infectious Diseases Society of America
- Hersh AL,
- Jackson MA,
- Hicks LA; American Academy of Pediatrics Committee on Infectious Diseases
- Hall MC,
- Kieke B,
- Gonzales R,
- Belongia EA
- Lindgren C,
- Neuman MI,
- Monuteaux MC,
- Mandl KD,
- Fine AM
- McIsaac WJ,
- White D,
- Tannenbaum D,
- Low DE
- Watson RL,
- Dowell SF,
- Jayaraman M,
- Keyserling H,
- Kolczak M,
- Schwartz B
- Mangione-Smith R,
- McGlynn EA,
- Elliott MN,
- Krogstad P,
- Brook RH
- Bauchner H,
- Pelton SI,
- Klein JO
- McIsaac WJ,
- Goel V,
- To T,
- Low DE
- Le Marechal F,
- Martinot A,
- Duhamel A,
- Pruvost I,
- Dubos F
- Centers for Disease Control and Prevention
- Linder JA
- Little P,
- Gould C,
- Williamson I,
- Warner G,
- Gantley M,
- Kinmonth AL
- Beaudoin A,
- Edison L,
- Introcaso CE, et al; Centers for Disease Control and Prevention
- Lee GM,
- Wessels MR
- Copyright © 2017 by the American Academy of Pediatrics