Humair J, Revaz SA, Bovier P, Stalder H. Management of Acute Pharyngitis in AdultsReliability of Rapid Streptococcal Tests and Clinical Findings. Arch Intern Med. 2006;166(6):640-644. doi:10.1001/archinte.166.6.640
How to use clinical score, the rapid streptococcal antigen test (RSAT), and culture results is uncertain for efficient management of acute pharyngitis in adults.
This prospective cohort study included 372 adult patients with pharyngitis treated at a Swiss university-based primary care clinic. In eligible patients with 2 to 4 clinical symptoms and signs (temperature ≥38°C, tonsillar exudate, tender cervical adenopathy, and no cough or rhinitis), we performed an RSAT and obtained a throat culture. We measured sensitivity and specificity of RSAT with culture as a gold standard and compared appropriate antibiotic use with cost per patient appropriately treated for the following 5 strategies: symptomatic treatment, systematic RSAT, selective RSAT, empirical antibiotic treatment, and systematic culture.
RSAT had high sensitivity (91%) and specificity (95%) for the diagnosis of streptococcal pharyngitis. Systematic throat culture resulted in the highest antibiotic use, in 38% of patients with streptococcal pharyngitis. Systematic RSAT led to nearly optimal treatment (94%) and antibiotic prescription (37%), with minimal antibiotic overuse (3%) and underuse (3%). Empirical antibiotic treatment in patients with 3 or 4 clinical symptoms or signs resulted in a lower rate of appropriate therapy (59%) but higher rates of antibiotic use (60%), overuse (32%), and underuse (9%). Systematic RSAT was more cost-effective than strategies based on empirical treatment or culture: $15.00, $26.00, and $32.00, respectively, per patient appropriately treated.
The RSAT we used is a valid test for diagnosis of pharyngitis in adults. A clinical approach combining this RSAT and clinical findings efficiently reduces inappropriate antibiotic prescription in adult patients with acute pharyngitis. Empirical therapy in patients with 3 or 4 clinical symptoms or signs results in antibiotic overuse.
Pharyngitis is a common symptom, accounting for 1% to 2% of the visits to primary care physicians.1 Acute pharyngitis in adults is mainly caused by a virus; only about 10% of incidences are bacterial, mainly caused by group A β-hemolytic streptococci, which is the only indication for antimicrobial therapy.2 In group A streptococcal pharyngitis (GASP), penicillin V effectively reduces symptom duration by 1 to 2 days, spread of the disease, and incidence of suppurative complications and rheumatic fever, which is exceptional in developed countries.3- 7 Antibiotic therapy provides moderate clinical benefits in a minority of patients with pharyngitis, which is a self-limited disease with low complication rates.3,4 Although antibiotic therapy should be prescribed in only the few patients with GASP, physicians prescribe antibiotics for 73% of patients with pharyngitis.8 Widespread antibiotic use frequently causes adverse effects (eg, allergy or diarrhea), increases use of medical services and costs, and contributes to bacterial resistance.3,9
Physician clinical diagnosis of GASP has low sensitivity and specificity because no element in the history or clinical examination is accurate enough.10- 12 Clinical scores of Centor et al11 and McIsaac et al13 based on 4 clinical findings are valid and reliable instruments that enable clinicians to estimate the probability of GASP in adults. Recent rapid streptococcal antigen tests (RSATs) using optical immunoassays give immediate results with higher average sensitivity (80%-90%) and specificity (≥95%).4,14- 17 Diagnosis based on RSAT results may reduce unnecessary antibiotic prescription and costs and does not increase complication rates even without culture-confirming negative RSAT results.13,15- 19 However, throat culture remains the gold-standard test for the diagnosis of GASP despite its suboptimal performance, cost, and delayed results in clinical practice.3,4,15
Recent studies in adults suggest that a clinical approach based on throat culture is better than RSAT or empirical treatment to optimize antibiotic prescription and cost-effectiveness.16,20 Inasmuch as no clinical trial has yet compared effectiveness, feasibility, and costs of various clinical approaches, there is no evidence-based consensus on the best clinical approach for management of acute pharyngitis in adults when the key issue is diagnosis and antimicrobial therapy of GASP. The US guidelines based on expert opinions recommend 4 alternative strategies, using clinical score, RSAT results, or throat culture findings to limit antibiotic prescription in patients most likely to have GASP and to benefit from antimicrobial therapy.3,4
This observational study conducted in a cohort of adult outpatients with pharyngitis had 2 goals—to measure the performance of RSATs across different clinical scores and to compare antibiotic prescription with direct costs for various recommended strategies using the clinical score, RSAT results, and culture findings to manage acute pharyngitis in adult outpatients.
We conducted a prospective cohort study in a walk-in clinic of a university-based primary care center in Geneva, Switzerland, from March 1, 1999, to September 30, 2001. Care was provided by 15 medical residents and 8 attending physicians, among whom many change yearly, some work part time, and replacements and rotations are common. We included all consecutive patients older than 15 years with pharyngitis with at least 2 of 4 of the clinical findings of Centor et al11: temperature of 38°C or higher, tonsillar exudate, tender cervical adenopathy, and no cough or rhinitis. Patients with no or 1 symptom or sign were excluded because most experts agree that they do not require further testing or antibiotic therapy because GASP prevalence is lower than 5%.3,4 No patient refused to participate in the study. The study complied with ethical regulations of the University Hospital of Geneva, Geneva, Switzerland.
For each included patient, a physician obtained a history, performed a clinical examination, and determined the Centor score corresponding to the sum of the 4 clinical findings. For all patients with a score of 2 to 4, physicians obtained 2 throat swabs, 1 for RSAT and 1 for culture. Physicians were trained to perform RSAT, an optical enzyme-linked immunosorbent assay (TestPack Plus Strep A w/OBC [On-Board Controls] II; Abbott Laboratories, Abbott Park, Ill), according to the manufacturer's instructions. A comparative study showed that this RSAT had the highest sensitivity and specificity in ambulatory medical practice.21 Throat cultures were incubated on 2 blood agar plates in anaerobia without inhibitor and were read after 48 hours in the Laboratory of Bacteriology, University Hospital of Geneva. Physicians prescribed antibiotic treatment immediately in all patients with positive RSAT results or within 48 hours in patients with a positive throat culture. Our local guidelines recommend as a first choice penicillin V, 1 million units 3 times a day for 10 days, or a macrolide in patients who are allergic to penicillin.
We compared performance of the RSAT with that of throat culture, considered the gold standard. For the entire sample and each clinical score, we determined sensitivity, specificity, positive and negative predictive values using the Newcombe-Wilson method without continuity correction,22 and positive and negative likelihood ratios with their 95% confidence intervals.23
Using clinical data and test results, we constructed a decision analysis model (Data 3.5; Treeage Software, Williamstown, Mass) comparing antibiotic prescription for 5 management strategies for acute pharyngitis in adults:
Symptomatic treatment without test or treatment.
Systematic RSAT with antibiotic therapy in patients with positive results.
Selective RSAT in patients with 2 or 3 clinical criteria and empirical antibiotic therapy in patients with a score of 4.
Empirical antibiotic therapy without testing in patients with 3 or 4 clinical criteria.
Systematic culture with antibiotic treatment in patients with positive results.
To assess the effectiveness of each clinical approach, we used the decision model to calculate the following rates: antibiotic use in the entire cohort, appropriate antibiotic use in patients with GASP, antibiotic overuse in patients without GASP, antibiotic underuse in patients with GASP, and appropriate treatment with antibiotics in patients with GASP and without antibiotics in patients without GASP. We did not assess quality of life, complications, and adverse drug effects.
For each clinical approach, we computed the medical costs per patient appropriately treated. In the base-case analysis using 2002 prices in Swiss francs converted to US dollars, the costs were $25.00 for a 10-day course of penicillin treatment, $5.00 for the RSAT, and $18.00 for the throat culture. Costs of the consultation and symptomatic treatment were excluded because they were assumed to be identical for each clinical approach. To test the robustness of our findings, we performed a sensitivity analysis to assess whether prevalence of GASP, sensitivity and specificity of RSAT, cost of RSAT, costs of antibiotics, and sensitivity of culture would influence the cost-effectiveness of each clinical approach.
We included 372 consecutive ambulatory patients with pharyngitis, most of whom were young, with a slight female predominance (Table 1). Frequency of clinical findings of Centor et al11 were fairly similar (60%-75%). More than 80% of the patients had moderate illness, with clinical scores of 2 and 3. The overall prevalence of GASP based on throat culture was 37.6% and steadily increased with the clinical score (23.6%, 41.0%, and 60.3%, respectively, for 2, 3, and 4 criteria). Prevalence of group G streptococci was 5.1% and for group C streptococci was 3.8%, with 1 patient in each group having positive RSAT results. Eighty-eight physicians treated, on average, 4.2 patients during 3 successive academic years.
Compared with the throat culture, RSAT achieved high global sensitivity (91.4%; 95% confidence interval, 85.6%-95.0%) and specificity (95.3%; 95% confidence interval, 91.7%-97.3%). Performance of RSAT was good, with high positive predictive value (92%), negative predictive value (95%), and positive likelihood ratio (19.3) (Table 2). Sensitivity and positive predictive values increased progressively with the clinical score, suggesting a possible spectrum bias, which was not significant because confidence intervals overlap. Specificity and negative predictive values remained fairly constant regardless of clinical score.
Systematic throat culture yielded the highest rates of appropriate treatment and antibiotic use but was also the most expensive clinical approach (Table 3). With systematic RSAT use, rates of appropriate treatment and antibiotic prescription were nearly optimal, with minimal antibiotic overuse and underuse. Moreover, this clinical approach cost half as much and enabled immediate decisions about antibiotic prescription. Selective RSAT in patients with clinical scores of 2 or 3 and empirical treatment in patients with clinical scores of 4 increased antibiotic overuse at a higher cost. Empirical antibiotic treatment in patients with clinical scores of 3 or 4 resulted in a high rate of antibiotic prescription, with more overuse in patients without GASP than appropriate use for patients with GASP. With this clinical approach, fewer than 60% of patients were appropriately treated, at a higher cost. Symptomatic treatment without testing or antibiotic therapy was the least expensive clinical approach but yielded only 62% of patients appropriately treated because of antibiotic underuse in all patients with GASP.
In the cost-effectiveness analysis, only 3 strategies can be considered (Table 3). Symptomatic treatment was the least costly and less effective option, systematic RSAT had the lowest cost-effectiveness ratio, and systematic culture was the most effective but most expensive option.
These results were not sensitive to variations in prevalence of GASP. When prevalence was less than 41%, systematic RSAT was the least costly clinical approach compared with systematic culture. With higher prevalence, empirical treatment in patients with clinical scores of 3 or 4 was less costly but also less effective than systematic RSAT (extended dominance).
For cost of antibiotics between $5.00 and $50.00, systematic RSAT was again the best option. Other strategies were either less effective for similar costs (direct dominance) or had higher marginal cost-effectiveness ratios (extended dominance). As long as the cost of the antibiotics was less than $22.00, empirical treatment in patients with clinical scores of 3 or 4 was the least costly option but was less effective than empirical treatment (direct dominance).
Varying costs of RSAT from $1.00 to $20.00 did not affect these results. Systematic RSAT remained the least expensive option as long as its cost was less than $5.70. Thereafter, empirical treatment in patients with clinical scores of 3 or 4 and selective RSAT became less costly but was less effective than systematic RSAT (extended dominance).
RSAT characteristics did not change our findings, because systematic RSAT was the best option when RSAT sensitivity was greater than 37% and its specificity was greater than 77%. Finally, varying the sensitivity of culture to 90% did not affect these results.
Our study shows that the RSAT can be a valid diagnostic test for the diagnosis of GASP in adults with pharyngitis, particularly when combined with the Centor clinical score and used in patients with a high probability of GASP. According to our results, the best clinical approach for diagnosis and treatment of pharyngitis in adults is systematic RSAT in patients with at least 2 clinical findings suggestive of GASP. Among the various options recommended in guidelines, this clinical approach optimizes antibiotic prescription by limiting both overuse and underuse and is cost-effective. The strengths of this study include a design based on a fairly large and real cohort of adult primary care patients, comparison of relevant strategies recommended in current guidelines,3,4 and a cost-effectiveness analysis with robust findings.
Our study has several potential limitations. We assessed theoretical effects of the various strategies on antibiotic use and costs from a decision analysis model but not from a clinical trial evaluating prospectively real practices and their effect on symptoms, quality of life, complications, health care use, and long-term costs. Because patients were recruited from a walk-in clinic of a university hospital, our study may have included a selected sample with more severe illness than patients seen in community primary care practices. Diagnostic tests could not differentiate patients with GASP from streptococcal carriers, of which the prevalence is minimal among adults. The eventual presence of a selected population and streptococcal carriers may overestimate the prevalence of GASP. However, it is unlikely to change the choice of the best clinical approach because the sensitivity analysis showed that a lower prevalence of GASP does not affect antibiotic use and cost-effectiveness.
Systematic RSAT misses most cases of groups C and G streptococcal pharyngitis. However, it probably has little effect on the most efficient clinical approach because prevalence of these types of pharyngitis is less than 10%, they do not cause serious complications, and no convincing evidence shows that antibiotic therapy is beneficial.2
A single throat culture is not an optimal standard because its sensitivity is 90% to 95%. Use of 2 different cultures as the standard may have changed the performance of RSAT; however, this effect is probably small because the sampling technique for both tests is identical and was performed simultaneously by the same person. The sensitivity analysis showed that appropriate treatment and cost-effectiveness were not influenced by culture sensitivity. Use of 2 cultures as the standard is controversial and does not reflect actual conditions of clinical practice or similar research in primary care.15- 17
Comparisons of performance of RSAT are limited by differences in tests, populations, and disease profile between studies. In earlier research, performance of the RSAT was lower and more variable; however, recent studies testing RSAT based on high-sensitivity optical immunoassays reported similar specificity but lower sensitivity.14- 16 Most studies tested RSAT in children and included all patients with pharyngitis regardless of clinical score. Many false-negative results of RSAT may be due to smaller bacterial load in patients with a lower clinical score. This increasing sensitivity with higher clinical scores suggest a spectrum bias reported by others, with similar sensitivity.24 This effect may have boosted the sensitivity in our study, which included only patients with clinical scores of 2 to 4 and a higher prevalence of GASP. Therefore, the clinical approach based on RSAT in patients with higher clinical scores increased the rate of diagnosis of GASP and appropriate prescription of antibiotics. False-positive results of RSAT are unrelated to clinical score, individual physicians, or group of residents, and only a few are due to cross-reactions with non–group A streptococci. They may be caused by problems in transport or storage of some cultures. Our sensitivity analysis confirms that a systematic RSAT probably is the most efficient clinical approach because it is not influenced by plausible variations of RSAT sensitivity and specificity.
In a similar study, strategies based on culture yielded slightly better results than those using RSAT for the diagnosis of GASP with appropriate antibiotic therapy.16 These conflicting results may be explained by the different prevalence of GASP between the 2 studies (21.9% vs 37.6%). This discrepancy may result from different scoring methods, settings, patient characteristics, and the spectrum bias suggested by our findings. However, both studies found that empirical treatment in patients with higher clinical scores results in massive antibiotic overuse.16 Neuner et al20 recently reported that culture was slightly more cost-effective than RSAT. However, their study was not based on clinical observations but on a decision model using assumptions drawn from the literature. Furthermore, some of their assumptions differed from our data, such as strategies tested in all patients regardless of clinical score, lower prevalence of GASP, and cost of RSAT 2.7 times higher.
Our results indicate that combining systematic RSAT with a clinical score is the most efficient clinical approach for the diagnosis and treatment of acute pharyngitis in adults. They do not support approaches based on results of throat culture or empirical treatment, which are recommended in a recent guideline.3 The clinical approach based on a clinical score and RSAT limits antibiotic prescription essentially to patients with GASP, who benefit from this treatment.3,4 Targeting these patients makes antibiotic therapy more effective and reduces the duration of symptoms and the incidence of suppurative complications.5,6 The clinical benefit may be small in individual patients but substantial in a population in which the disease is common. This clinical approach is also convenient in clinical practice because it enables immediate therapeutic decisions and limits overprescription of antibiotic therapy.13,17 It prevents treatment delays, which reduce the effect on symptom relief, and increase workload for follow-up telephone calls. The clinical approach based on RSAT may also reduce adverse effects and costs of unnecessary antibiotic therapy.3 Because physicians prescribe antimicrobial therapy, usually nonrecommended broad-spectrum antibiotics, for most patients with pharyngitis, systematic RSAT can contribute to limiting the development of antibiotic resistance in communities.3
Our cost-effectiveness analysis is applicable to other countries with different economic status because our sensitivity analysis shows that RSAT is the preferred clinical approach for a wide range of costs. For example, results remain unchanged for the US retail price of a 10-day course of penicillin treatment ($5.00-$20.00), which is recommended at lower doses than in Switzerland.3,20
Although our results suggest that one clinical approach is more efficient, recent studies, including ours, did not use an experimental design and, therefore, reached different conclusions. To define the optimal approach to treat pharyngitis in adults, randomized controlled trials could test and compare various treatment strategies in primary care and measure their effects in real practice on antibiotic prescription, symptoms, quality of life, complications, adverse effects of therapy, health care use, and costs.
We concluded that RSAT is a valid test for the diagnosis of pharyngitis in adults. Combining an RSAT with a clinical score seems to be an effective and cost-effective clinical approach to limit antibiotic prescription and to appropriately treat acute pharyngitis in adults.
Correspondence: Jean-Paul Humair, MD, MPH, Medical Outpatient Clinic, Department of Community Medicine, University Hospital of Geneva, 24 rue Micheli-du-Crest, CH-1211, Geneva 14, Switzerland (Jean-Paul.Humair@hcuge.ch).
Accepted for Publication: August 17, 2005.
Financial Disclosure: None.
Previous Presentations: This study was presented in part at the 16th World Conference of Family Doctors; May 14, 2001; Durban, South Africa; and at the 25th Annual Meeting of the Society of General Internal Medicine; May 4, 2002; Atlanta, Ga.