This cross-sectional study was undertaken in the department of pediatrics of a tertiary-care teaching hospital in Chennai, India, from July 2015 to April 2016. Children aged between 1 to 18 years presenting to the outpatient department with symptoms of pharyngitis and clinically suspected as streptococcal pharyngitis were included in the study. Children who presented　with obvious signs of viral respiratory infection (having rhinorrhea, coryza, conjunctivitis, coughing and/or sneezing) were excluded.　Children who were already on antibiotics were also excluded.

All screened children were recruited by purposive sampling sequentially, till the sample size was reached. Approval of the Institute Human Ethics Committee was obtained. Informed written consent was obtained from the parents or guardians of all the participants.

All the children were evaluated by a pediatrician. Demographic and clinical data were collected from the children. Clinical examination of the pharynx and tonsils was conducted under flash light illumination. All the variables related to modified Centor score were documented and score was calculated for each child. The clinician　collected the sample from pharyngeal wall (after depressing tongue with tongue depressor) by a sterile cotton swab applied on the tip of 6-7 inches long swab stick. Cotton swab was rubbed gently against pharyngeal mucosa. All precautions to avoid contamination were taken and the samples were immediately transported to the microbiology department in a sterile test tube at room temperature. Smear was examined by Gram staining, and Blood Agar was the culture media used for culture. Any　growth　observed on culture medium was assessed for morphology (smear examination), colony characteristics and biochemical tests. Isolates were subjected for antibiotic sensitivity as per CLSI guidelines.

Throat swab culture positivity was considered as the primary outcome variable. Various sociodemographic, lifestyle related parameters were considered as exploratory variables. The children with and without culture positive infection were compared with respect to demographic, clinical variables and modified Centor score. The association between the symptoms, signs and modified Centor score with culture positivity was assessed by cross tabulation and chi-square test. IBM SPSS version 21.0 was used for statistical analysis. P <0.05 was considered statistically significant.

A total of 120 participants were included in the analysis. About 35% in the study population had streptococcal sore throat. Only one child had throat swab culture positive for Staphylococcus aureus. The streptococcal positivity was 28% among 1 to 5 year, 40% in both 6 to 10 year and more than 11-year-old age groups. The proportion of streptococcal pharyngitis increased with increasing Centor score category. Among the symptoms, cough, throat pain, painful swallowing and difficulty in swallowing showed statistically significant difference among streptococcal and nonstreptococcal pharyngitis (Table I).

TABLE I Factors Associated with Streptococcal Pharyngitis in the Study Population (N=120)

The area under the curve value for modified Centor score was 0.589 with a P value 0.11 and 95% CI was 0.481 to 0.697, respectively in the study population. Among symptoms, cough had the highest sensitivity (63.4%), but poor specificity (36.7%). Palatine petechiae had 100% specificity, but very poor sensitivity (4.9%). Tender anterior cervical nodes and palatine exudates also had very high specificity (88.6% and 97.5%, respectively) but poor sensitivity (Table II).

TABLE II  Sensitivity and Specificity of various Clinical findings in Diagnosis of Streptococcal Pharyngitis

The proportion of antibiotic resistance was highest for Cotrimoxazole (16.7%), followed by tetracycline (9.5%). Ciprofloxacin and Ofloxacin resistance was seen in 7.1% of the study population. Erythromycin resistance was seen in 4.8% of the strains. Intermediate resistance was found for Erythromycin (4.8%), Azithromycin and Ofloxacin (2.4%).

Modified Centor score of 4 had a poor sensitivity (12.2%, 95% CI 4.8% to 26.2%), but very high specificity (96.2%, 95% CI 89.3% to 99.2%) in predicting streptococcal pharyngitis. The overall diagnostic accuracy was 67.5% (95% CI 58.4% to 75.8%). Even though the positive likelihood ratio was 3.21, the 95% confidence interval ranged from 0.81 to 12.77. The negative likelihood ratio was 0.91 (95% CI 0.81 to 1.03).

In this study, the overall predictive validity of modified Centor score, as assessed by Area under ROC curve was poor. Palatine petechiae, tonsillar exudates and tender anterior cervical lymph nodes had high specificity and positive predictive value in diagnosis of streptococcal pharyngitis. Streptococcus pyogenes strains were sensitive to penicillin. However, resistance to Cotrimoxazole (16.7%), tetracycline (9.5%), fluoroquinolones and macrolides was high.

Fig. 1 ROC analysis of predictive validity of modified Centor score in predicting streptococcal pharyngitis.

The AUC reported in the original 1981 study by Centor [3] was 0.78, which is considerably higher compared to our study. McIssac, et al. [4] modified the score, and reported high (83.1%) sensitivity. Few other studies from the West have reported AUC value higher than 0.7, indicating good predictive validity [6]. Another study had reported an area under the ROC of 0.633 [7]. A study had reported a strong positive association between tonsillar exudates and culture positive Group A streptococcal (GAS) sore throat [7]. But there are other studies, which have documented sensitivity as low as 3.9% for a modified Centor score of 4, with a high specificity of 96%, as in current study [8]. Similar pattern of low sensitivity and higher specificity has been demonstrated by few other studies [9,10]. Lowering the cut-off level of modified Centor score, has shown to increase the sensitivity, but at the cost of specificity, which may result in increase in unnecessary antibiotic prescription [6].

The reasons for such a wide variation in reported predictive validity needs to be evaluated. The key differences in the demographic characteristics, differences in the clinical presentation, and quality of clinical assessment may be responsible for these differences. The same factors may determine the clinical utility of modified Centor score.

The limited sample size in the study was a major limitation that did not allow for controlling for the effect of confounding or detailed analysis of factors associated with culture positive infection. Exclusion of possible viral etiology on clinical grounds alone was another limitation that can affect generalizability of these results.

We conclude that the predictive validity of Centor score for diagnosis of streptococcal pharyngitis is poor, and its utility as screening tool is questionable. Considering very low sensitivity, high proportion of streptococcal sore throat may be missed, if modified centor score is used a screening tool.

Contributors: JV: had conceptualized the idea, prepared the study proposal, engaged in data compilation, analysis and preparation of all the drafts of the manuscripts; AM,GG: have fine-tuned the study proposal, engaged in data collection, reviewed all the drafts and provided their inputs. They have compiled, verified and approved the final draft. All authors have contributed, designed and approved the study.

Funding: None; Competing interest: None stated.