Clinical score and rapid antigen detection test to guide antibiotic use for sore throats: randomised controlled trial of PRISM (primary care streptococcal management)
- “Clinical score and rapid antigen detection test to guide antibiotic use for sore throats: randomised controlled trial of PRISM (primary care streptococcal management)” by Paul Little and colleagues (BMJ 2013;347:f5806, doi:10.1136/bmj.f5806).
Objective To determine the effect of clinical scores that predict streptococcal infection or rapid streptococcal antigen detection tests compared with delayed antibiotic prescribing.
Design Open adaptive pragmatic parallel group randomised controlled trial.
Setting Primary care in United Kingdom.
Patients Patients aged ≥3 with acute sore throat.
Intervention An internet programme randomised patients to targeted antibiotic use according to: delayed antibiotics (the comparator group for analyses), clinical score, or antigen test used according to clinical score. During the trial a preliminary streptococcal score (score 1, n=1129) was replaced by a more consistent score (score 2, n=631; features: fever during previous 24 hours; purulence; attends rapidly (within three days after onset of symptoms); inflamed tonsils; no cough/coryza (acronym FeverPAIN).
Outcomes Symptom severity reported by patients on a 7 point Likert scale (mean severity of sore throat/difficulty swallowing for days two to four after the consultation (primary outcome)), duration of symptoms, use of antibiotics.
Results For score 1 there were no significant differences between groups. For score 2, symptom severity was documented in 80% (168/207 (81%) in delayed antibiotics group; 168/211 (80%) in clinical score group; 166/213 (78%) in antigen test group). Reported severity of symptoms was lower in the clinical score group (−0.33, 95% confidence interval −0.64 to −0.02; P=0.04), equivalent to one in three rating sore throat a slight versus moderate problem, with a similar reduction for the antigen test group (−0.30, −0.61 to −0.00; P=0.05). Symptoms rated moderately bad or worse resolved significantly faster in the clinical score group (hazard ratio 1.30, 95% confidence interval 1.03 to 1.63) but not the antigen test group (1.11, 0.88 to 1.40). In the delayed antibiotics group, 75/164 (46%) used antibiotics. Use of antibiotics in the clinical score group (60/161) was 29% lower (adjusted risk ratio 0.71, 95% confidence interval 0.50 to 0.95; P=0.02) and in the antigen test group (58/164) was 27% lower (0.73, 0.52 to 0.98; P=0.03). There were no significant differences in complications or reconsultations.
Conclusion Targeted use of antibiotics for acute sore throat with a clinical score improves reported symptoms and reduces antibiotic use. Antigen tests used according to a clinical score provide similar benefits but with no clear advantages over a clinical score alone.
Why do the study?
The aim of this study was to identify patients presenting in a primary care setting with acute sore throat who should be prescribed antibiotics.
Acute sore throat in children and adults is one of the most common reasons for consultation with general practitioners and is an important and difficult problem. Many episodes of acute sore throat are caused by infection. It is particularly important that general practitioners and other clinicians identify and treat bacterial related acute sore throat, particularly streptococcal infection, since untreated the consequences can be serious, such as rheumatic fever and suppurative (pus forming) complications (for example, peritonsillar abscess, otitis media (the most common), sinusitis, mastoiditis, pneumonia). But because history and clinical examination on their own are not enough to confirm or exclude bacterial infection, and because signs and symptoms of bacterial infection can often overlap and be difficult to distinguish from those of viral infection (which are the most common), patients with acute sore throat are commonly prescribed antibiotics. Yet many episodes of acute sore throat are a consequence of self limited viral rather than bacterial respiratory tract infections. Patients with non-bacterial infections who are prescribed antibiotics are subject to the adverse effects of antibiotic drugs (for example, diarrhoea and cutaneous eruptions caused by the aminopenicillin group of antibiotics) and contribute towards the growing problem of antimicrobial resistance worldwide.
However, several diagnostic strategies are available to help doctors to target better the prescription of antibiotics, ultimately reducing their overall use, while at the same time preventing progression of the illness and complications and improving symptom control. These include the use of rapid streptococcal antigen detection tests and clinical scores such as the Centor criteria, which were designed to gauge the presence of Lancefield group A β haemolytic streptococcus. Group A β haemolytic streptococcus causes about 15-30% of sore throats in young people and 10% in adults.
The authors of this study had previously developed a clinical score to predict the presence of Lancefield group A, C, and G streptococci (all of which share similar clinical presentations). More empirical management strategies (relying merely on the observation and experience of the clinician) include delayed antibiotic prescribing (the patient receives a prescription but is advised to take the drug only if symptoms worsen) or no offer of antibiotics.
Nevertheless, the evidence base supporting the use of both antigen tests and clinical scores is scant. In this study, the authors wanted to determine the effect of rapid streptococcal antigen tests and clinical scores that predict the presence of Lancefield group A, C, and G streptococcal infection compared with delayed antibiotic prescribing.
What did the authors do?
The authors carried out a randomised controlled trial to compare delayed antibiotic prescribing, use of a clinical score, and rapid antigen tests to identify the extent to which these strategies limit or target antibiotic use in patients with acute sore throat. This trial had three arms, with one including the clinical score alone and another including both the clinical score and the rapid antigen detection test. The patients in the delayed prescribing arm comprised the control group.
The authors developed two different clinical scores in diagnostic studies (score 1 and score 2) carried out before the trial, and the trial thus had two separate parts. In the first part of the study, 1129 patients were randomised to the three arms of the study (with two of them including the clinical score 1). In the second part of the study, 631 patients were randomised to the same three arms (but with two of them now including clinical score 2 rather than clinical score 1). The results that are reported as the main findings of the study pertain to the use of score 2 (the acronym of which is FeverPAIN) because it performed better than score 1—that is, there were beneficial effects when using score 2 but little or no benefit when using score 1. The results of the first part of the study are available in appendix 1 on bmj.com (www.bmj.com/content/suppl/2013/10/10/bmj.f5806.DC1/litp010913.ww1_default.pdf).
The FeverPAIN score is a five item score, and includes fever, purulence, rapid attendance to the general practice (within three days after the onset of symptoms), severely inflamed tonsils, and absence of cough or coryza (“common cold”).
Healthcare professionals, mostly general practitioners, recruited patients over three years of age presenting to several general practices in south and central England with acute sore throat (≤2 weeks) and with an abnormal appearance to the throat (erythema or pus, or both). Exclusion criteria were non-infectious causes of sore throat (for example, aphtous ulceration, candida, drugs) and an inability to obtain consent (patient or guardian), such as in cases of dementia or uncontrolled psychosis.
At baseline, the recruiting health professionals assessed the patient’s temperature as well as the presence and severity of baseline symptoms (sore throat, difficulty swallowing, fever during the illness, runny nose, cough, feeling unwell, diarrhoea, vomiting, headache, muscles ache, abdominal pain, sleep disturbance) through four point Likert scales (none, a slight problem, a moderately bad problem, a bad problem) and the presence of clinical signs (pus, nodes, tender nodes, raised temperature).
Patients were then randomised to the three arms of the study through a web based computer randomisation program. A total of 207 patients was randomised to the control group (delayed antibiotic prescription), whereas 211 and 213 patients were randomised to the clinical score and rapid test groups, respectively.
Patients in the delayed prescription group were told to pick up a prescription at the reception of the general practice if their symptoms did not improve or worsened after three to five days. In the clinical score group, patients with an overall score of 0 or 1 did not receive antibiotics, those with a score of 2 or 3 received a delayed antibiotic prescription, and those with a score of 4 or higher received antibiotics.
In the rapid antigen detection test group, the FeverPAIN score was also used in all patients. Patients with scores of 0 or 1 did not receive antibiotics or a rapid antigen test, those with a score of 2 received a delayed prescription, and those with a score of 3 or higher received the rapid antigen test. Patients with a negative test result did not receive antibiotics.
Patients were also requested to fill in a symptom diary every night up to 14 nights or until they believed that their symptoms had settled. They scored each symptom on a seven point scale from 0 to 6 (with 0 meaning “no problem” and 6 meaning “as bad as it could be”). The symptoms eligible to be scored were sore throat, difficulty swallowing, feeling unwell, fever, and sleep disturbance.
This randomised controlled trial used an open design, which meant that even though the patients were blinded to the specific details of the different arms of the study, both patients and researchers were aware of which intervention was being deployed.
This study was also a pragmatic trial since even though clinicians assigned to patients in each group were supposed to abide by predefined prescribing strategies for that group (provided the patient did not refuse it), they also had the freedom to discuss, if necessary, other strategies with the patient. A pragmatic trial reflects the real life conditions of everyday primary care practice.
What did the study find?
The study essentially found that it is possible to target antibiotic use with both clinical score and antigen test combined and clinical score alone, with no real difference between the two arms.
The study had both primary and secondary outcomes (described in table 2 of the paper on bmj.com). The primary outcome was symptom severity—namely, the mean score of sore throat and difficulty swallowing in the two to four day period after the consultation—and this was assessed through the diary. Both the clinical score group and the rapid antigen test group had lower mean scores than the delayed prescribing group (−0.33, 95% confidence interval −0.64 to −0.02, and −0.30, −0.61 to 0.004, respectively.)
The secondary outcomes were the median duration of symptoms rated moderately bad or worse, antibiotic use, side effects (reported in the diary and in the notes), the “medicalising belief in the need to see a doctor in future episodes” (patients reported through Likert scales that they would either be “slightly likely” or “less likely” to see the doctor for future episodes), and the return of the patient to the practice.
The median duration of symptoms rated moderately bad or worse in the delayed prescribing group was five days, and four in both the clinical and the rapid antigen test group (hazard ratio 1.30, 95% confidence interval 1.03 to 1.63, and 1.11, 0.88 to 1.40, respectively), although the results for the rapid antigen test group were not statistically significant. Hazard ratios are calculated in a type of statistical analysis called Cox regression, which is used in time to event outcome variables such as time to recurrence after treatment, and was used for the variable duration of symptoms rated moderately bad or worse.
Overall, 46% (75/164) of patients in the delayed prescribing group reported using antibiotics, but only 37% (60/161) and 35% (58/164) reported using them in the clinical score group and rapid antigen test group, respectively. This translates into a relative risk reduction of 29% and 27% in the clinical score group and rapid antigen test group (risk ratio 0.71, 95% confidence interval 0.50 to 0.95, and 0.73, 0.52.to 0.98, respectively).
The patient’s belief in the need to see the doctor in the future or the patient’s return to the surgery did not differ significantly between groups. These two variables were statistically analysed by logistic regression, which is used in binary outcome variables, which was the case here (“slightly likely” and “less likely” for belief to see the doctor and “return” or “no return” to the surgery).
What are the strengths and limitations of this study?
Firstly, it is important to note that this study was a randomised controlled trial and that these types of trials are difficult to do, even for common conditions such as acute sore throat. The authors say this is the first randomised trial to look at whether both rapid antigen detection tests and clinical scoring methods influence symptom control and antibiotic use in patients with acute sore throat.
The sample size was underpowered to ensure significant results for dichotomous outcomes (outcomes for which there are only two possibilities, such as the outcome of return to the surgery), but had enough power for the symptomatic outcomes.
Because changes in the expected direction for symptomatic outcomes in both intervention groups were significant, the authors considered it is less likely that the results arose by chance.
The association between a certain exposure and a certain outcome can often be partly or completely caused by a different exposure, called a confounder or confounding factor. Table 3 in the paper on bmj.com shows the additional adjustment the authors made to see whether variation between general practices acted as a confounding factor (“adjustment for practice as a covariate”). The authors also made an “adjustment for clustering by practice,” which meant that they took into consideration that the intervention effects may be more similar between people within the same practice than would be expected by people being completely independent. No differences were found with the adjustment for practice as a covariate and only minor differences for the adjustment for clustering by practice, but the confidence intervals tended to become wider.
Compliance with the intended strategy was good across all groups (79% in the delayed prescribing group, 85% in the clinical score group, and 84% in the antigen test group), and was probably not greater because this was a pragmatic trial. The authors carried out what is called an intention to treat analysis, which reflects what happens in clinical practice. This means that all participants randomised to their respective trial arms were analysed regardless of whether, for instance, they dropped out of the study or ended up receiving a different strategy.
Information about non-recruitment of patients was limited (it referred to those with acute sore throat who were not approached because of lack of time to consent and recruit, and also to those who were asked but then declined, but particularly the latter), which the authors attribute to the health professionals being busy attending to patients with acute illness.
Clinicians participating in this study used management prompts when carrying out the clinical assessment of patients (history and examination), which are some sort of written information to assist with history taking, physical examination, and also to guide the management, such as a protocol or guidelines, and is either available on paper or by computer). As such, we do not know whether clinicians in real life practice obtain similar results as they are less likely to use management prompts.
What does the study mean?
The take home message of this study is that clinical scores can lead to better symptomatic outcomes and can also be useful in helping clinicians decide which patients should be prescribed antibiotics, ultimately reducing antibiotic use. Moreover, the use of rapid antigen detection tests in addition to clinical scores seems equivalent for impact on symptom control and antibiotic use to the use of clinical scores alone, with no apparent advantages.Tiago Villanueva, general practitioner and editorial registrar, BMJ
1BMA House, London
Correspondence to: firstname.lastname@example.org
I thank Paul Little and Domnhall Macauley for their input into this article.
Competing interests: None declared.
Provenance and peer review: Commissioned; not externally peer reviewed.
- Choby BA. Diagnosis and treatment of streptococcal pharyngitis. Am Fam Physician 2009;79:383-90.
- National Institute for Health and Care Excellence. Sore throat—acute. NICE clinical knowledge summaries, 2012: http://cks.nice.org.uk/sore-throat-acute#!background.
- Patient.co.uk. Upper respiratory infections—coryza. 2013. www.patient.co.uk/doctor/upper-respiratory-infections-coryza.
- Peacock J, Peacock P. Oxford handbook of medical statistics. Oxford Medical Publications, 2013.
- Sedgwick P. Confounding in clinical trials. BMJ 2012;345:e7951.
- Sedgwick P. Analysis by intention to treat. BMJ 2011;342:d2212.
Cite this as: Student BMJ 2013;21:f7026