Best management of patients with an acute sore throat is still best management for these patients

To the Editor,

We appreciate that Jan Matthys from Belgium has submitted a Letter to the Editor [1] with five interesting comments to our original publication ‘Best management of patients with an acute sore throat – a critical analysis of current evidence and a consensus of experts from different countries and traditions’ [2]. Below are our reactions to these comments:

1. Matthys states that it is unlikely that our proposed management will be implemented in any country for two reasons: (a) ‘it is time consuming’ and (b) ‘it incurs the cost of a rapid test’.

   Our reply: Actually, we propose to first sort out patients in apparently uncomplicated, potentially complicated and potentially critically ill. Doing this initial triage is the most important message in our publication. Most patients will have an apparently uncomplicated illness and for those we state that they only need reassurance of the self-limiting nature of the illness, if they have a low probability for presence of group A Streptococci (GAS) such as Centor score 0–2 (McIsaac 0–2 or FeverPain 0–2).

   For patients with a higher probability for presence of GAS, such as Centor score 3–4 (McIsaac 3–4 or FeverPain 3–5), we propose two alternative strategies. One of these options is to reassure patients with no testing or prescribing of antibiotics, as is currently recommended in Belgium and the Netherlands. The other options is to test 10–15% of patients and prescribe antibiotics to 3.5–6.6% of patients which would be much less than is currently been done in any country. We are perfectly fine with Belgium and the Netherlands choosing the restrictive no test no antibiotics policy to patients with an apparently uncomplicated acute sore throat. Other countries may not like to adopt to this policy and our other suggested alternative policy of prescribing antibiotics to 3.5–6.6% of patients would be a significant reduction compared to their current prescribing habits. Furthermore, in countries like Sweden and the US, their current guidelines already recommend the use of rapid testing for patients with a higher probability for presence of GAS and our recommendation only adds rapid acting analgesics to their current decision trees to further reduce testing and antibiotic prescribing.

2. Matthys states that delayed prescription strategy is in many cases done to meet the patient’s expectations, the patient’s pressure for antibiotics or to end the consultation quickly and limit the time needed to allay the real concerns about the complaint, but is usually unnecessary.

   Our reply: This anecdotal statement is not really justified and does not seem to be based on the evidence regarding why clinicians use delayed prescriptions. The evidence suggests that clinicians’ factors determine what clinicians decide to do and length of consultation is unrelated to this decision [3].

   Antibiotics, immediate or delayed, may be given to meet patient expectations – and we agree that this is a poor reason to provide antibiotics. However, this is unrelated to the question about delayed versus immediate antibiotics. The rationale for delayed antibiotics is that clinicians have no good predictors of subsequent adverse outcomes [4] and where the clinician is uncertain, it is a reasonable strategy: delayed antibiotics are not necessary but might be a reasonable approach since they do result in more efficient use of health care resources (significantly lowering costs for revisits), and by providing rapid access to antibiotics if the illness deteriorates, they are probably associated with lower complications when compared to no offer of antibiotics.

   Our paper was focussed on testing and when to treat. Of course, general principles of clinical practice remain pivotal in all encounters with patients. Asking about fears and expectations of patients is of course important, as is adequate patient information in word and if available, leaflets/online patient information. We agree with Matthys that we could have mentioned these general issues, but they were not the focus of our paper.

3. Matthys perceive we talk too much about rheumatic fever and refer to a Cochrane review that has not seen a single case of rheumatic fever since 1975.

   Our reply: Matthys refers to analysis 4.3 (page 66) in the Cochrane review about antibiotics for the sore throat [5]. The studies post 1975 included 1389 + 978 = 2367 patients from predominantly high income countries and found no cases of rheumatic fever among them. This finding supports our statement, ‘Although the risk for rheumatic heart disease remains high in most low-income countries it has declined steeply in high-income countries where it has almost vanished’ [2]. Rheumatic fever is of no concern in Belgium and many other high-income countries. However, rheumatic fever remains to be of high concern in parts of Australia and New Zealand, as well as in many low income countries. Many readers of infectious diseases are from countries where rheumatic fever remains a concern and we aimed to provide some guidance to them as well.

4. In the first comment by Matthys, we clarified in our response that we actually propose two alternative pathways for patients with an apparently uncomplicated acute sore throat and a high scoring for possible presence of GAS. In his fourth comment, Matthys criticises that we only mention the first alternative pathway, no testing and no antibiotics, in an asterisk in Figure 2 in our original publication.

   Our reply: Apart from the asterisk in Figure 2, we also wrote in the text under the heading, ‘Apparently uncomplicated acute sore throat’: ‘There are two principally different approaches to antibiotic prescribing to patients attending with apparently uncomplicated acute sore throat. One wait-and-see approach recommends no antibiotic use irrespective of the presence of GAS while the other recommends the use of antibiotics to a few selected patients harbouring GAS to reduce acute symptoms where analgesics are insufficient. There is no definite evidence to recommend one approach over the other.’ Further down in the same section, we write: ‘An important message in Figures 1 and 2 is that patients with apparently uncomplicated acute sore throat and low clinical scoring should neither be tested nor prescribed antibiotics. Those with a higher clinical score can either be tested, and treated if the POCT is positive, or a wait-and-see strategy could be used’. Hence, apart from the asterisk in Figure 2, we also talked about the wait and see policy twice in the text and we are happy with Belgium and the Netherlands adhering to it.

5. Matthys states that ‘Even with Centor scores 3–4 or McIsaac 3–4 or FeverPain 3–5 scores, given the high rate of carriage, the probability that the infection is actually caused by streptococci is no more than 50%, equal to tossing a coin.’

   Our reply: As stated before, the most important message we want to emphasise is to first sort out patients who are apparently uncomplicated, potentially complicated and potentially critically ill. The debate here is what to do with patients with an apparently uncomplicated acute sore throat. For these patients, clinical scoring becomes redundant if you want to adopt a strict wait and see policy with no antibiotics (as in Belgium and the Netherlands). That means it also becomes irrelevant if the aetiologic agent is GAS. However, those countries that want to allow a minimal use of antibiotics for patients with an apparently uncomplicated acute sore throat, such as Sweden and the US, focus on GAS and this cannot be done in a meaningful way without the use of tests to detect presence of GAS.

   It is true that the pre-test probability for GAS being the aetiologic agent is around 50% in patients with Centor scores 3–4 or McIsaac 3–4 or FeverPain 3–5 scores. However, the post-test probability in patients with a Centor score of 0–4 for GAS being the aetiologic agent if the test is positive is given in Table 1 in our original publication, which estimates this probability giving a point estimate, with 95% confidence intervals (CIs) [2]. For adults, it is 92% (CI 87–95%) and for children, it is 65% (CI 42–80%). However, if we only select patients with Centor scores 3–4 and a positive test for presence of GAS, the post-test probability for GAS to actually cause the infection goes up and is 95% (CI 90–97%) for adults and 83% (CI 63–93%) for children [6]. A previous publication by Malmberg et al. [7] analysing the probability for GAS causing the sore throat in adult patients with Centor 3–4 found a post-test probability of 97% (CI 91–100%) if the test was positive. Another publication by Gunnarsson and Manchal [8] found the probability for GAS being the causative agent in adult patients with 3–4 Centor scores and a positive test was 94% (CI 90–96%).

   Hence, there is ample evidence that the post-test probability for GAS being a causative agent in patients with 3–4 Centor scores and a positive test for presence of GAS is much higher than 50%, especially in adult patients. We agree that the probability for GAS being the causative agent is lower for children than for adults due to the higher carrier rate in children, but it remains well above 50% also for children with 3–4 Centor scores and a positive test for GAS.

Disclosure statement

No potential conflict of interest was reported by the author(s).