• 
• 

Three Part Question

In [patients presenting with tonsillitis] can [measurements of inflammatory markers] [differentiate between streptococcal and viral tonsillitis]?

Clinical Scenario

In the primary care setting, sore throat is a very common cause for consultation. However, the cause of tonsillitis is not known in many cases. Relying on clinical presentation and history alone is unreliable in differentiating streptococcal from viral tonsillitis, creating the problem of unnecessary and ineffective use of antibiotic therapy in many situations. Microbiological culture is the best way to establish an aetiological diagnosis, but is time consuming. The question therefore arises whether inflammatory markers such as C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), white cell count (WCC) and procalcitonin (PCT) can differentiate streptococcal from viral tonsillitis.

Search Strategy

Medline (PubMed), EMBASE and the Cochrane Controlled Trials Register (CCTR) were searched separately for articles published in English before 30 April 2010.
Pubmed:[“tonsillitis” [MESH] OR “pharyngitis” [MESH] ]AND [“C-reactive protein” OR “leukocyte count” OR “erythrocyte sedimentation rate” OR “ procalcitonin”]. The references of all relevant articles were examined to identify any publications potentially not retrieved by the electronic searches.

Search Outcome

8 articles were identified from a total of 202 abstracts as fulfilling our selection criteria and provided information on the various inflammatory markers used in differentiating streptococcal from viral tonsillitis. A total of 1031 patients with sore throat were examined, of which 377 had positive culture results for streptococcal tonsillitis.

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Putto A et al. 1986 Finland	Case reports of 62 children mean age 4.9 years presenting to private paediatric practice and hospital with tonsillitis – defined by presence of tonsillar exudates and axillary or rectal temperature ≥38°C.	Case-control study (level 3b)	CRP, WCC, ESR	Difference not significant	P-values were not provided for all mean values. Retrospective analysis, which is prone to selection bias.
			Age	Median (range) ages (years): Adenoviral: 1.8 (0.5-10.5); EBV: 5.4 (1.5-16.0); Streptococcal: 7.5 (2.8-17.0), p<0.001
Ylikoski et al. 1989 Finland	257 army conscripts age 18-27 years presenting to military hospital with sore throat, axillary temperature >38°C and local inflammatory changes in palatine tonsils. Excluded were patients who received antibiotics in preceding 2 weeks.	Case control study (level 3b)	CRP	Mean (SD): 70 (35) mg/L (Group A strep), 59 (7) mg/L (virus)	P-values, confidence intervals and sensitivities and specificities were not calculated. Population group not representative of general population.
			WCC	Mean (SD): 13.3 (4.3) x 109/L (Group A strep), 7.9 (2.4) x109/L (virus)
			WCC	Suggested cut-off level of >12 x109/L for group A strep
			ESR	No significance
Tandeter et al 1993 Israel	39 patients mean age 17.3 years presenting to a collection of 6 family practices with a sore throat and a clinical picture of streptococcal tonsillitis.	Case-control study (level 3b)	WCC	Median(range) streptococcal: 13,400/mm3 (4100 – 22,100); Sensitivity: 0.80; Specificity: 0.714; Positive predictive value: 0.833; Negative predictive value: 0.667, (p<0.01)	No guidelines or standardisation regarding diagnostic criteria for streptococcal tonsillitis.
Hjortdhal et al 1994 Norway	174 patients age >18 years old presenting to emergency department between 4pm to 9pm, with a chief complaint of sore throat.	Case-control study (level 3b)	CRP	Mean: 50.4mg/L streptococcal, 19.5mg/L viral. (p<0.01)	Confidence interval of sensitivity, specificity, likelihood ratio and positive predictive values were not given.
			CRP	CRP levels >40mg/L had discriminatory properties if the fever lasted less than a week, and >20mg/L if a week or longer.
			WCC	Mean: 12.1x109/L streptococcal, 8.8x109/L viral. (p<0.01)
			WCC	WCC cut-off value of 10 x109/L clinically useful.
			ESR	Difference not significant
Gulich et al 1999 Germany	161 consecutive patients age ≥16 years presenting to general practice with sore throat.	Prospective cohort study (level 2b)	CRP	CRP cut-off value of 35mg/L for bacterial tonsillitis: Sensitivity 0.78 (95%CI 0.61–0.90) Specificity 0.82 (0.73–0.88) Positive predictive value 0.57 (0.42-0.70) Negative predictive value 0.92 (0.85–0.96)	Inclusion and exclusion criteria not specified. Definition of original presentation of sore throat unclear.
			CRP	ROC: 0.85
			WCC	ROC: 0.68
Elsammak et al. 2006 Egypt	45 patients mean age 4.91 years seen in paediatric emergency department, divided equally into 3 groups of healthy children, bacterial tonsillitis, and nonbacterial tonsillitis. Controls were healthy, non-anaemic children. Excluded were children who received any antibiotic therapy or with any renal, liver or systemic disease.	Prospective cohort study (level 2b)	PCT	Best marker of streptococcal tonsillitis.	All patients were evaluated after having fever ≥12 hours, which coincides with peak of PCT. CRP peaks at 48-72 hours. This could be a possible reason for the better specificity of PCT reported.
			PCT	Median (range): 0.374ng/mL (0.11–6.5) bacterial, 0.105ng/mL (0.01–0.53) non-bacterial, 0.03ng/mL (0.01–0.08) control. (p<0.01)
			PCT	ROC: 0.862
			PCT	Cut-off 0.2275ng/mL: Sensitivity: 0.73; Specificity 0.87
			CRP	Median (range): 50mg/L (22.4-71.1) bacterial, 23.6mg/L (5.9-61.2) non-bacterial, 2.6mg/L (0.9-6) control. (p<0.01)
			CRP	ROC: 0.809
			CRP	Cut-off 39.2 mg/L: Sensitivity 0.80; Specificity 0.73
			WCC	ROC: 0.636
Wolf et al. 2007 United Kingdom	120 patients mean age 25.04 years presented to ENT hospital department with EBV tonsillitis and 100 with bacterial tonsillitis – defined as symptoms of sore throat, pyrexia, dysphagia, odynophagia, redness of throat and tonsils, and white plaques on tonsils. Excluded were patients with quinsy, parapharyngeal or retropharyngeal abscesses, peritonsillitis or compromised immune systems.	Case control study (level 3b)	Lymphocyte/WCC ratio	Higher in EBV than bacterial. Mean (SD) L/WCC ratio: 0.54 (0.14) vs 0.10 (0.08), p<0.01.	Statistically significant difference in age between the patients with EBV and bacterial tonsillitis. Retrospective analysis.
			L/WCC	>0.35: Sensitivity 0.90 and specificity of 1.00 of EBV detection.
			WCC	Higher in bacterial than EBV. Mean (SD) WCC: 16560/µL (54100/µL) vs 11400/µL (4670/µL), p<0.01.
			Lymphocyte count	Higher in EBV than bacterial. Mean (SD): 6490/µL (3590/µL) vs 1590/µL (680/µL), p<0.01.
			Neutrophil count	Higher in bacterial than EBV. Mean (SD): 13770/µL (5230/µL) vs 3830/µL (1920/µL), p<0.01.
Peltola et al. 2007 Finland	Children aged from 1 month to 16 years who presented to paediatric hospital outpatient clinics - identified according to CRP and WCC measurements. Excluded were temperature <38°C, ongoing treatment for malignancy, and missing records.	Case control study (level 4).	WCC and CRP	Not sensitive to distinguish bacterial from viral infection. However, both WCC and CRP should be used together to improve accuracy.	The comparison between streptococcal and viral tonsillitis was only a subset of the entire investigation. Retrospective analysis with arbitrarily determined ranges for inflammatory markers may both have introduced bias.

Comment(s)

Where reported, WCC, CRP and PCT were significantly higher in patients with streptococcal tonsillitis. ESR was not different between the groups. Where a receiver operating characteristics (ROC) analysis was performed, PCT had a larger area under the curve at 0.86 compared to CRP at 0.80 and 0.85 in different studies. The eight studies included in this review had different approaches to comparing inflammatory markers between patients with and without streptococcal tonsillitis. Some studies compared the mean or median values of the parameters, while others explored a variety of arbitrary cut-offs to determine a value with the highest sensitivity and specificity. A minority performed a ROC analysis. This heterogeneity prevented a meta-analysis of these observational studies. The disparate results may also be due to the different timing of blood sampling in relation to duration of illness. As one study demonstrated, timing of blood sampling may influence which cut-off level has the best performance characteristics. The majority of studies did not comment on the timing of the determination of inflammatory markers in relation to duration of disease. Putto et al demonstrated that adenoviral tonsillitis is associated with high CRP and WBC levels, similar to streptococcal tonsillitis. Future studies comparing inflammatory markers in children with and without streptococcal tonsillitis need to investigate viral aetiology in the comparison group to enable a comparison of results of studies in different settings. Such a comparison may have to correct results for percentage of adenovirus tonsillitis. Only Ylikoski et al examined the differences in inflammatory markers in group A streptococci against non-group A streptococci and other bacterial aetiologies. CRP and WCC levels are also consistently higher in group A streptococcal tonsillitis when compared with non-group A streptococcal tonsillitis or other types of bacterial tonsillitis such as those caused by Streptococcus pneumoniae, Haemophilus influenzae or Moraxella catarrhalis. If studies only looked for bacteria in the form of positive cultures for group A streptococcus, patients with other undetected bacterial tonsillitis in the comparison group may have been missed, reducing what may have been a significant difference between groups if the comparison group had only included patients with definite viral tonsillitis. Future studies need to explore ROC analysis for all parameters to enable a comparison of studies. A marker with intrinsically high specificity for the detection of a bacterial infection like PCT should be explored in future studies regarding a cut-off with optimum sensitivity. PCT levels can be determined rapidly and could form part of an algorithm to aid decision making concerning antibiotic use in patients with tonsillitis.

Editor Comment

CRP, C reactive protein; ESR, erythrocyte sedimentation rate; PCT, procalcitonin; ROC, receiver operating characteristics; WCC, white cell count.

Clinical Bottom Line

WCC, CRP and PCT levels are higher in patients with streptococcal tonsillitis compared to patients with tonsillitis or pharyngitis without group A streptococcus isolated from a throat swab. Which of these markers has the best test performance characteristics requires further study.

References

1. Putto A, Meurman O, Ruuskanen O. C-reactive protein in the differentiation of adenoviral, Epstein-Barr viral and streptococcal tonsillitis in children. Pediatrics 1987;80:6–12.
2. Ylikoski J, Karjalainen J. Acute tonsillitis in young men: etiological agents and their differentiation. Scand J Infect Dis 1989;21:169–74.
3. Tandeter H, Namir S, Herman J. Pride and prejudice: the case of streptococcal pharyngitis and the WBC. J Clin Epidemiol 1993;46:1071–3.
4. Hjortdahl P, Melbye H. Does near-to-patient testing contribute to the diagnosis of streptococcal pharyngitis in adults? Scand J Prim Health Care 1994;12:70–6.
5. Gulich MS, Matschiner A, Glück R, Zeitler HP. Improving diagnostic accuracy of bacterial pharyngitis by near patient measurement of C-reactive protein (CRP) Br J Gen Pract 1999;49:119–21.
6. Elsammak M, Hanna H, Ghazal A, Edeen FB, Kandil M. Diagnostic value of serum procalcitonin and C-reactive protein in Egyptian children with streptococcal tonsillopharyngitis. Pediatr Infect Dis J 2006;25:174–6.
7. Wolf DM, Friedrichs I, Toma AG. Lymphocyte-white blood cell count ratio: a quickly available screening tool to differentiate acute purulent tonsillitis from glandular fever. Arch Otolaryngol Head Neck Surg 2007;133:61–4.
8. Peltola V, Toikka P, Irjala K, Mertsola J, Ruuskanen O. Discrepancy between total white blood cell counts and serum C-reactive protein levels in febrile children. Scand J Infect Dis 2007;39:560–5.