European Journal of Clinical Microbiology & Infectious Diseases

, Volume 32, Issue 6, pp 787–793

Effect of clinical spectrum, inoculum size and physician characteristics on sensitivity of a rapid antigen detection test for group A streptococcal pharyngitis

Authors

    • INSERM, UMR S953Epidemiological Research Unit on Perinatal Health and Women’s and Children’s Health
    • Department of Pediatrics, Necker-Enfants-Malades Hospital, Assistance Publique-Hôpitaux de ParisUniv Paris Descartes, Sorbonne Paris Cité
  • M. Chalumeau
    • INSERM, UMR S953Epidemiological Research Unit on Perinatal Health and Women’s and Children’s Health
    • Department of Pediatrics, Necker-Enfants-Malades Hospital, Assistance Publique-Hôpitaux de ParisUniv Paris Descartes, Sorbonne Paris Cité
  • C. Levy
    • Association Clinique et Thérapeutique du Val-de-Marne (ACTIV)
    • Centre de Recherche Clinique (CRC)Centre Hospitalier Intercommunal de Créteil
  • P. Bidet
    • Department of Microbiology, Robert Debré Hospital, Assistance Publique-Hôpitaux de ParisUniv Paris Diderot, Sorbonne Paris Cité
  • M. Benani
    • Association Clinique et Thérapeutique du Val-de-Marne (ACTIV)
  • M. Koskas
    • Association Clinique et Thérapeutique du Val-de-Marne (ACTIV)
  • E. Bingen
    • Department of Microbiology, Robert Debré Hospital, Assistance Publique-Hôpitaux de ParisUniv Paris Diderot, Sorbonne Paris Cité
  • R. Cohen
    • Association Clinique et Thérapeutique du Val-de-Marne (ACTIV)
    • Department of MicrobiologyCentre Hospitalier Intercommunal de Créteil
Article

DOI: 10.1007/s10096-012-1809-1

Cite this article as:
Cohen, J.F., Chalumeau, M., Levy, C. et al. Eur J Clin Microbiol Infect Dis (2013) 32: 787. doi:10.1007/s10096-012-1809-1

Abstract

We aimed to assess the independent effect of clinical spectrum, bacterial inoculum size and physician characteristics on the sensitivity of a rapid antigen detection test (RADT) for group A streptococcus (GAS) in children. Double throat swabs were collected from 1,482 children with pharyngitis and 294 asymptomatic children in a French prospective, office-based, multicenter (n = 17) study, from October 2009 to May 2011. Patient- and physician-level factors potentially affecting RADT sensitivity were studied by univariate and multivariate multilevel analysis, with laboratory throat culture as the reference test. In children with pharyngitis and asymptomatic children, the prevalence of GAS was 38 % (95 % confidence interval 36–41 %) and 11 % (7–14 %), respectively. Overall, RADT sensitivity was 87 % (84–90 %). On stratified and multivariate multilevel analysis, RADT sensitivity was higher for children with pharyngitis than asymptomatic children (89 % vs. 41 %), children <9 than ≥9 years old (88 % vs. 79 %) and those with heavy than light inoculum (94 % vs. 53 %). RADT sensitivity was influenced by the physician performing the test (range 56–96 %, p = 0.01) and was higher for physicians with hospital-based clinical activity in addition to office-based practice (adjusted odds ratio 3.4 [95 % confidence interval 1.9–6.3], p < 0.001); inter-physician variations in RADT sensitivity were largely explained by this variable (proportional change in variance >99 %). The sensitivity of the RADT is independently affected by patient- and physician-level factors. Physicians who base their diagnosis of GAS pharyngitis on the results of a RADT alone should consider diagnostic accuracy monitoring and adequate training when needed.

Introduction

Group A streptococcus (GAS) is found in 20–40 % of childhood pharyngitis cases; the remaining cases are considered mainly of viral origin [1]. The reference test for the diagnosis of GAS pharyngitis is throat culture on a blood agar plate (BAP) in a microbiology laboratory for 48 h. Rapid antigen detection tests (RADTs) can provide immediate results and are generally considered highly specific as compared to laboratory culture (about 95 %) [2]. However, the American Academy of Pediatrics and the Infectious Diseases Society of America recommend the confirmation of negative RADT results by a throat culture because the sensitivity of RADTs is estimated to be insufficient, at about 85 % [3, 4]. Furthermore, RADT sensitivity is subject to variations by clinical spectrum of the disease (spectrum effect) [510], bacterial inoculum size (inoculum size effect) [1013] and the physician who performed the test (physician effect) [14, 15].

About 10 % of school-age children are asymptomatic pharyngeal GAS carriers [1], for which antibiotic treatment is usually not required [16]. Studying the sensitivity of RADTs in asymptomatic carriers might be important because the sensitivity might approach that in children with GAS carriage and intercurrent viral pharyngitis. Previous studies that evaluated a RADT in asymptomatic children [1720] reported heterogeneous sensitivities (from 20 % to 100 %) and were limited by their use of an obsolete RADT technique (latex agglutination) [17], a non-consensual reference test (reading of cultures after 24 h) [18, 19], and a low GAS carriage rate (1.6 %), which suggested selection bias [20].

To date, several studies have evaluated the effects of clinical spectrum, inoculum size and person performing the test on the sensitivity of the RADT, but no study analyzed these factors concomitantly. Furthermore, no study included asymptomatic carriers, nor analyzed physician-level factors that might explain the physician effect. We assessed the independent effect of patient- and physician-level factors on the sensitivity of a RADT.

Materials and methods

The study protocol was approved by the Ile-de-France XI institutional review board (n°09016). Parents and children gave their informed approval for participation before inclusion. The STARD statement was followed for reporting [21].

Patients

Eligible patients were 3–14 years old who were evaluated by their pediatrician and who did not receive antibiotics for 7 days before inclusion. Recruitment of a first case-cohort of symptomatic children was based on a clinical diagnosis of acute pharyngitis at the time of consultation (i.e., clinical signs of inflammation of the pharynx and/or tonsils). We recruited a second group of asymptomatic children with no symptoms at the time of consultation (i.e., well-child visits for evaluation of physical and cognitive development and/or vaccination), nor pharyngitis in the previous month.

Sample size

In children with pharyngitis, the sample size was estimated so that the 95 % confidence interval (95CI) for RADT sensitivity would be a ±5% estimation. Assuming sensitivity to vary between 75 % and 95 % and a GAS prevalence of 35 %, a sample of 714 children with pharyngitis was needed. In asymptomatic children, sample size was estimated so that the 95% CI for the frequency of GAS carriage would be a ±4% estimation. Assuming a GAS carriage frequency of 10 %, a sample of 250 asymptomatic children was needed. Enrollment was initially planned from October 1, 2009 to May 31, 2010 (year 1), but the study was extended for both groups from October 1, 2010 to May 31, 2011 (year 2) because of difficulties in sampling asymptomatic children.

Throat swabs

Throat samples were obtained by use of a contiguous double–swab collection–transportation system (Venturi Transystem Amies agar, COPAN Diagnostics, Corona, CA, USA). Throat samples were taken as recommended by the manufacturer of the RADT, that is, by use of a tongue depressor to avoid contact with any area of the mouth and by vigorously rubbing the surface of the tonsils, the posterior pharynx and any area of inflammation of the pharynx. The RADT (StreptAtest, Dectrapharm, Strasbourg, France) was performed immediately with swab #1 collected in the pediatrician’s office. Swab #2 was held at ambient temperature and sent to arrive within 72 h to the hospital laboratory by an express messenger service. On receipt, swab #2 was used to perform a throat culture. During year 1, the swab was directly inoculated onto a trypticase soy agar plate with 5 % sheep blood [10], and during year 2, the swab was moistened with 1 mL sterile distilled water, and one drop of the fluid expressed from the swab was streaked on a BAP containing colistin and nalidixic acid (Columbia CNA + 5 % sheep blood, Biomérieux, Lyon, France). The plates were incubated anaerobically at 37 °C and read at 24 and 48 h. ß-hemolytic colonies were investigated by latex agglutination (Prolex, Pro-Lab Diagnostics, Richmond Hill, ON, Canada). Inoculum size was estimated as follows: light, ≤50 GAS colonies per plate; heavy, >50 GAS colonies. Microbiologists were blinded to individual clinical data and RADT results.

Office-based pediatricians and survey tool

The 17 participating pediatricians were part of a French research-and-teaching network (Association Clinique et Thérapeutique du Val-de-Marne [ACTIV]) [22]. After completion of the clinical study, all participating pediatricians were invited to answer an online survey. The questionnaire was developed after review of previous publications aiming to explore the influence of contextual and organizational factors on physician practices [2325]. The questionnaire was designed as a four-domain survey tool including questions about physician general characteristics (age [continuous], gender, year of medical graduation [continuous]), practice type and setting (number of physicians in the office [solo/2-person practice vs. >2 clinicians], hours of weekly clinical activity in the office [continuous], estimated proportion of patients covered by the French universal health insurance [continuous], whether the physician had a hospital-based clinical activity in addition to office-based practice [yes/no]), physician habits in performing the RADT (person collecting the throat sample and reading the result of the RADT [physician vs. office staff], site of throat sampling [1 tonsil vs. both tonsils], following manufacturer’s instructions in terms of number of drops for each chemical reagent [4 drops vs. other] and use of a timer to control chemical reactions [yes/no]), and knowledge and beliefs of pharyngitis and GAS disease (estimated GAS prevalence in childhood pharyngitis [<20 % vs. 20–40 % vs. >40 %] and estimated incidence of acute rheumatic fever in France in children 3–14 years old [continuous]). The final version of the survey tool consisted of 14 questions organized in a random fashion. Physician data without identifier variables were entered into the database and linked to patient data.

Statistical analysis

Hospital laboratory throat culture was considered the reference test. First, the general characteristics of the study population, the frequency of GAS recovery in children with pharyngitis and asymptomatic children and the overall diagnostic accuracy of the RADT (sensitivity and specificity) were described. Second, the characteristics of the pediatricians who participated in the study were described. Third, we used univariate analysis with chi-square test to examine patient- and physician-level characteristics that may affect RADT sensitivity. Fourth, we used a multilevel logistic model to account for the hierarchical structure of data, with patients (level 1) nested within physician offices (level 2) [26, 27]. Restricting the analysis to patients with a positive throat culture for GAS and using the result of the RADT as the outcome in the model allowed us to study the independent effect of patient- and physician-level variables on the sensitivity of the RADT. The model estimated the odds ratio (OR) of having a positive RADT result with a positive throat culture result for GAS. Patient-level variables were clinical status (pharyngitis vs. asymptomatic), age dichotomized into 2 categories of equal range (<9 year vs. ≥9 year) and inoculum size (light or heavy GAS growth). Physician-level variables were those from the questionnaire described above, with continuous variables dichotomized at their median.

First, an empty model that included only the random effect was adjusted (model 0) to estimate the baseline physician-level variance (var(0)), i.e., the physician effect. Next, we adjusted a multivariate multilevel model that included patient-level covariates (model 1). It allowed us to assess the association between RADT sensitivity and patient-level variables and to calculate the residual physician-level variance after adjustment for patient-level variables (var(1)). We used the proportional change in the variance (PCV = [var(0)var(1)]/var(0)) to assess the extent to which the physician effect may be explained by differences in the distribution of patient-level characteristics. Then, we adjusted multivariate multilevel models that included physician-level covariates (models 2–5). Physician-level characteristics that were associated with RADT sensitivity with p ≤ 0.2 on univariate analysis were entered stepwise in multilevel models. We also calculated the PCV for each physician-level variable. Data were double entered into 4D software v6.4, and statistical analysis involved use of Stata/SE 12 (StataCorp, College Station, TX, USA).

Results

Patient characteristics, GAS prevalence and overall RADT accuracy

In total, 1,781 children met the inclusion criteria. Two patients with uninterpretable RADT results and three with a lost throat swab were secondarily excluded. Therefore, analysis involved 1,776 children (790 girls, 44 %), including 1,482 with pharyngitis and 294 asymptomatic children. The mean (SD) age was 6.1 (2.5) and 7.0 (2.7) years for children with pharyngitis and asymptomatic children, respectively (p < 0.001). The prevalence of GAS in children with pharyngitis was 38 % (95 CI 36–41 %), and the frequency of GAS carriage was 11 % (7–14 %) in asymptomatic children (p < 0.001). Overall, the sensitivity and specificity of the RADT was 87 % (84–90 %) and 93 % (91–94 %), respectively.

Physician characteristics

The mean (SD) age of physicians was 53.4 (7.1) years and 8 (47 %) were women (Table 1). In all, ten physicians (59 %) had a hospital-based clinical activity in addition to their office-based practice. All but one physician declared collecting throat samples themselves and all physicians declared reading the result of the RADT themselves.
Table 1

Characteristics of physicians who participated in the study (n = 17)

Characteristics

Value

Age, years, mean (SD)

53.4 (7.1)

Women

8 (47)

Year of medical graduation, median (range)

1987 (1979–2002)

Practice type and setting

 Physicians in the office

   Solo/2-person practice

6 (35)

   Group practice (>2 clinicians)

11 (65)

 Weekly clinical activity in the office, hours (median, range)

37 (20–55)

 Estimated proportion (%) of patients covered by the French universal health insurance, median (range)

5 (1–15)

 Hospital-based clinical activity

10 (59)

Physician’s habits for performing the RADT

 Site of throat sampling

   One tonsil

5 (29)

   Both tonsils

12 (71)

 Use of the correct quantity of each reagent for the RADTa

13 (76)

 Use of a timer to control time of reactions for the RADTa

6 (35)

Knowledge and beliefs of GAS disease

 Estimated GAS prevalence in childhood pharyngitis

   <20 %

2 (12)

   20–40 %

12 (71)

   >40 %

3 (18)

 Estimated incidence of ARF in France in children aged 3–14 years, cases/year, mean (SD)

2.6 (3.7)

Data are no. (%) unless indicated

RADT rapid antigen detection test, GAS group A streptococcus, ARF acute rheumatic fever

aManufacturer’s instructions are as follows: (1) place four drops of reagent A and four drops of reagent B in the tube; (2) add the throat swab in the tube and turn the swab ten times; (3) leave the swab in the tube for 1 min; (4) press the swab against the side of the tube; (5) discard the swab; (6) place the test strip into the tube; (7) leave the strip in the tube and read the result at 5 min

Effect of clinical spectrum and inoculum size

On univariate analysis, the sensitivity of the RADT was higher in children with pharyngitis than asymptomatic children (89 % [95% CI 87–92 %] vs. 41 % [24–59 %], p < 0.001; Table 2). RADT sensitivity was higher but not significantly in children <9 than ≥9 years (88 % [85–91 %] vs. 79 % [70–89 %], p = 0.053). Among 602 GAS-positive throat cultures, 501 (83 % [80–86 %]) showed heavy inoculum. Heavy inoculum was less frequent in asymptomatic carriers than in children with pharyngitis (34 % [17–52 %] vs. 86 % [83–89 %], p < 0.001). On univariate analysis, RADT sensitivity increased with inoculum size (53 % [44–63 %] vs. 94 % [91–96 %], p < 0.001; Table 2). Sensitivity was higher during year 2 than year 1 (92 % [89–95 %] vs. 82 % [78–86 %], p < 0.001). On multivariate analysis, RADT sensitivity was higher for children with pharyngitis than asymptomatic children (adjusted odds ratio [aOR] 5.0 [95% CI 1.9–13.0], p = 0.001), children <9 than ≥9 years (aOR 2.3 [1.1–5.1], p = 0.04) and children with heavy than light inocula (aOR 10.2 [5.7–18.5], p < 0.001; model 1; Table 3). These associations were stable after adjusting for physician-level variables (Table 3).
Table 2

Sensitivity of the RADT by patient- and physician-level characteristics: univariate analysis

Characteristics

No. a (%)

Sensitivity, % (95CI)

p

Patient-level characteristics

Clinical status

 Asymptomatic

32 (5)

41 (24–59)

<0.001

 Pharyngitis

570 (95)

89 (87–92)

Age, years

 3–8

534 (89)

88 (85–91)

0.053

 9–14

68 (11)

79 (70–89)

Inoculumb

 Light

101 (17)

53 (44–63)

<0.001

 Heavy

501 (83)

94 (91–96)

Physician-level characteristics

Estimated proportion of patients covered by the French universal health insurance, %

 ≤5

10 (59)

89 (86–92)

0.03

 >5

7 (41)

83 (78–88)

Hospital-based clinical activity

 No

7 (41)

78 (72–84)

<0.001

 Yes

10 (59)

91 (88–94)

Year of medical graduation

 ≤1987

10 (59)

89 (85–92)

0.10

 >1987

7 (41)

84 (79–89)

Use of the correct quantity of each reagent for the RADTc

 No

4 (24)

81 (70–91)

0.15

 Yes

13 (76)

88 (85–90)

95CI 95 % confidence interval

a Number of patients (n = 602) or physicians (n = 17)

b Light, ≤50 GAS colonies per plate; heavy, >50 GAS colonies

c Manufacturer’s instructions are to use four drops of reagent A and four drops of reagent B

Table 3

Sensitivity of the RADT by patient- and physician-level characteristics: multivariate multilevel analysis

Characteristics

No. a (%)

Model 1

Model 2

Model 3

Model 4

Model 5

aOR (95CI)

p

aOR (95CI)

p

aOR (95CI)

p

aOR (95CI)

p

aOR (95CI)

p

Patient-level characteristics

Clinical status

 Asymptomatic

32 (5)

1

0.001

1

0.001

1

0.001

1

0.001

1

0.001

 Pharyngitis

570 (95)

5.0 (1.9–13.0)

 

4.8 (1.8–12.4)

 

4.5 (1.8–11.4)

 

5.0 (1.9–12.9)

 

5.1 (2.0–13.1)

 

Age, years

 3–8

534 (89)

2.3 (1.1–5.1)

0.04

2.4 (1.1–5.4)

0.03

2.5 (1.1–5.5)

0.02

2.3 (1.1–5.2)

0.04

2.3 (1.1–5.1)

0.04

 9–14

68 (11)

1

 

1

 

1

 

1

 

1

 

Inoculumb

 Light

101 (17)

1

<0.001

1

<0.001

1

<0.001

1

<0.001

1

<0.001

 Heavy

501 (83)

10.2 (5.7–18.5)

 

10.3 (5.7–18.6)

 

11.0 (6.1–20.0)

 

10.1 (5.6–18.3)

 

10.1 (5.6–18.3)

 

Physician-level characteristics

Estimated proportion of patients covered by the French universal health insurance, %

 ≤5

10 (59)

1.5 (0.7–3.3)

0.35

 >5

7 (41)

1

 

Hospital-based clinical activity

 No

7 (41)

1

<0.001

 Yes

10 (59)

3.4 (1.9–6.3)

 

Year of medical graduation

 ≤1987

10 (59)

1.2 (0.5–2.6)

0.67

 >1987

7 (41)

1

 

Use of the correct quantity of each reagent for the RADTc

 No

4 (24)

1

0.75

 Yes

13 (76)

1.2 (0.4–3.2)

 

Residual physician-level variance

0.21

 

0.21

 

<0.001

 

0.18

 

0.21

 

PCV (%)*

n/a**

 

n/a**

 

>99 %

 

14 %

 

n/a**

 

All models adjusted for year of inclusion (year 1 vs. year 2)

aOR adjusted odds ratio, 95CI 95 % confidence interval, PCV proportional change in variance, n/a not applicable

a Number of patients (n = 602) or physicians (n = 17)

b Light, ≤50 GAS colonies per plate; heavy, >50 GAS colonies

c Manufacturer’s instructions are to use four drops of reagent A and four drops of reagent B

* PCV is calculated on the basis of the physician-level variance of model 0 (var(0) = 0.19, p = 0.03) for model 1 and physician-level variance of model 1 (var(1) = 0.21) for models 2–5

** Not applicable because of equal or increasing physician-level variance between the two models compared

Physician effect

RADT sensitivity was significantly affected by the physician performing the test (range 56–96 %, p = 0.01). On univariate analysis, RADT sensitivity was higher for physicians with ≤5 % of their patients covered by the French universal health insurance system (89 % [86–92 %] vs. 83 % [78–88 %], p = 0.03; Table 2) and those with hospital-based clinical activity in addition to their office-based practice (91 % [88–94 %] vs. 78 % [72–84 %], p < 0.001). RADT sensitivity was higher but not significantly for physicians who had more clinical years in practice (89 % [85–92 %] vs. 84 % [79–89 %], p = 0.10) and those who declared they used the correct number of drops of each chemical reagent needed for the RADT (88 % [85–90 %] vs. 81 % [70–91 %], p = 0.15). On multilevel analysis, the empty model confirmed a significant physician effect (Model 0, var(0) = 0.19, p = 0.03). This physician effect was not explained by differences in distribution of patient-level variables (Model 1, var(1) = 0.21; Table 3). Hospital-based clinical activity was the only physician-level variable significantly associated with RADT sensitivity (aOR 3.4 [1.9–6.3], p < 0.001), and it accounted for 99 % of the physician effect (Model 3, var(3) < 0.001, PCV >99 %).

Discussion

The sensitivity of a RADT is a key determinant defining its role in diagnostic strategies for pharyngitis. Several authors have suggested that the RADT sensitivity can vary by patient-level characteristics, as well as by the person performing the test [715]. To our knowledge, this is the first study to examine the independent effect of patient- and physician-level factors on the sensitivity of a RADT. Our results suggest that physician-level factors may affect the sensitivity of the RADT independently of patient clinical spectrum and bacterial inoculum size.

The results from this study confirm the effect of patient clinical spectrum on the sensitivity of the RADT, with increased sensitivity for children with pharyngitis than asymptomatic carriers. The variation in sensitivity of the RADT by clinical status may be explained by locoregional changes due to infection and inflammation, such as biochemical changes, bacterial biofilm formation, increased antigen exposure, internalization phenomenon, changes in antigenic structures and differences in the distribution of GAS genotype (emm gene) in children with pharyngitis and asymptomatic carriers [28].

Our results also confirm the importance of inoculum size and physician effect on the sensitivity of the RADT. In general, inoculum size effect and the physician effect are thought to be linked through sampling variation, with the most experienced operators more frequently obtaining heavy inocula, and heavy inocula being associated with higher RADT sensitivity. However, our results show that RADT sensitivity may vary significantly depending on the physician who performed the test, independent of inoculum size. We found wide variation in RADT sensitivity by physician (range 56–96 %), which did not decrease when taking inoculum size into account on multivariate multilevel analysis (model 1, Table 3).

Among several physician-level factors associated with variations in RADT sensitivity on univariate analysis, having a hospital-based clinical activity in addition to office-based practice was the only variable remaining significant on multivariate multilevel analysis and explained most of the physician effect (PCV >99 %). However, we do not believe that having a hospital-based clinical activity is likely to increase the sensitivity of the RADT per se, but rather, this variable might indirectly capture the effect of other physician characteristics that we did not assess. As well, previously, the appropriate management of childhood pharyngitis was found to depend on physician characteristics such as specialty (family physicians vs. pediatricians) and type of practice (solo/2-person vs. group practice and rural vs. urban location) [25].

Our study has several limitations. First, the study is based on a small sample of French pediatricians (n = 17), all of whom are involved in a research-and-teaching network (ACTIV) and practice in urban areas. Therefore, our data might not be representative of the practices of office-based physicians, and our results need to be confirmed in larger samples of physicians and in different practice types and settings.

Second, although our results confirm that heavy inocula are less frequent in asymptomatic carriers than in children with pharyngitis (34 % vs. 86 %) [29], we were not able to investigate whether RADT false-negative results would be associated with GAS carriage in children with pharyngitis [11]. We included asymptomatic children in an attempt to approach the sensitivity of the RADT in GAS carriers who have an intercurrent viral pharyngitis. However, because the sensitivity of the RADT is affected by clinical status (pharyngitis vs. asymptomatic children) independent of inoculum size, this approximation does seem reliable. Most experts consider that the only way to differentiate true GAS infection and GAS carriage is by use of GAS-specific antibodies (i.e., anti-streptolysin O and anti-DNase B antibodies) [11]. However, the assessment of streptococcal antibody response would have required repeated blood samples that seemed untenable because of the cross-sectional design of the study. Moreover, the interpretation of streptococcal antibody titers is debated with regard to most recent available data [30].

Third, we cannot exclude that sample discordance between the two swabs might explain some discrepancies between results from the RADT and from throat culture, but this hypothesis could not be further investigated. Indeed, sample discordance was previously reported to be 5–10 % in studies involving duplicate throat cultures [2]. Although we used a contiguous double-swab collection system to minimize sample discordance, RADT sensitivity may have been low for some physicians because of asymmetrical throat sampling.

In conclusion, the results of this large prospective study expand our understanding of the factors affecting RADT sensitivity for detecting GAS. Diagnosis, like any decision-making process, is affected by various factors [23, 24]. Our results confirm that the sensitivity of a RADT should not be considered a fixed value, but rather, several patient- and physician-level factors may independently affect the sensitivity of the test. That physician characteristics affect the results of the RADT independent of patient clinical features and inoculum size should be emphasized in medical educational programs to improve physician practices and reduce inter-physician variations in RADT sensitivity. Our results support the American Academy of Pediatrics recommendation for physicians who have chosen to base their diagnosis of GAS pharyngitis solely on the result of a RADT to first evaluate the performance of this RADT in their office [3] and consider taking adequate educational and training programs when needed.

Acknowledgments

The authors thank A. Liboz (Department of Microbiology, Robert Debré Hospital); M. Boucherat, MD (database design); F. de La Rocque, MD (study conception); I. Ramay, D. Menguy, S. Tortorelli and M. de Pereira (ACTIV); B. Khoshnood, MD, PhD (multilevel analysis methodological support); and all the physician investigators who participated in the RADT GAS study: F. Corrard, MD, P. Deberdt, MD, A. Elbez, MD, M. Goldrey, MD, J. Gosselin, MD, P. Martin, MD, A.S. Michot, MD, N. Panis, MD, D. Qutob, MD, C. Romain, MD, O. Romain, MD, C. Schlemmer, MD, F. Thollot, MD, A. Wollner, MD.

Funding sources

This study was internally funded by ACTIV and externally by Dectrapharm, manufacturer of the RADT (Rapid Antigen Detection Test). JFC was supported by educational grants from Agence Régionale de Santé d’Ile-de-France, Laboratoires Guigoz - Société Française de Pédiatrie - Groupe de Pédiatrie Générale - Groupe de Recherches Epidémiologiques en Pédiatrie, and a research grant from the French Ministry of Health (PHRC Régional 2012, AOR 12089). The external funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Conflicts of interest

The authors declare that they have no conflict of interest.

Previous presentations

This work was presented in part at the 30th Annual Meeting of the European Society for Pediatric Infectious Diseases, Thessaloniki, Greece, May 2012 (Abstract A-435-0007-01213).

Copyright information

© Springer-Verlag Berlin Heidelberg 2013