Skip navigation

Reducing unnecessary consultation - a case of NNNT?

Tom Marshall
Specialist Registrar in Public Health Medicine
Northamptonshire Health Authority


Readers of Bandolier will be familiar with the concept of NNT as a convenient summary of the beneficial impact of a treatment. But what if the treatment has an effect - such as encouraging unnecessary consultation - which we want to avoid? We could think of this in terms of NNNT (numbers needed to not treat).

Consultation for throat infection

Throat infection is a common reason for a GP consultation. There is not much support for using antibiotics. The Cochrane Collaboration's most recent review confirmed the extremely modest effects of antibiotics in reducing the duration of throat infection [1]. Because simple analgesics are all that is needed, dealing with it can also be a frustrating exercise. Yet patients still consult. Could GPs have an influence on this pattern of behaviour? A recent randomised controlled trial explored this question in a novel way [2].

Patients were randomly allocated to immediate antibiotics, an offer of delayed antibiotics (three days later, if symptoms persisted) and no antibiotics. However, patients were not blind to their treatment. The clinical outcomes and patient satisfaction were the same for each strategy. However, compared to either offering a delayed prescription or not prescribing, issuing a prescription for an antibiotic tended to persuade patients that they were effective. It also affected their intention to consult in the future.

Numbers-needed-to-not-treat (NNNT)

From this we can work out a NNNT (numbers needed to not treat - Table 1). For every 3 patients we "not treat" (avoid or delay prescribing antibiotics) we persuade an additional one that they are not effective. Similarly for every 4 patients whom we "not treat" we persuade one that they need not consult with the problem in the future.

Table 1: Effect of prescribing an antibiotic for throat infection on patients' beliefs and intentions
Effect on patient Penicillin V prescribed No Penicillin V or offer delayed for 3 days NNNT
Believe "antibiotics are effective" 87% 57% 3
Say that they "will consult next time" 79% 55% 4
"satisfied" with consultation 96% (ns) 92% (ns)

Rather than simply relying on patients' stated intentions, a follow up to the initial study looked at the factors which led patients who had suffered one throat infection to consult again in the next year [3]. Again the message was fairly clear. An immediate prescription for antibiotics increases the number of patients who consult again (Table 2). It was also clear that those patients who did consult again were much more likely to have been prescribed antibiotics for throat infection at some time in the past.
Table 2: How prescribing antibiotics affects patients' re-attendance with throat infection
Effect on patient Penicillin V prescribed No Penicillin V or offer delayed for 3 days NNNT
Consult again with throat infection 38% 27% 10
Penicillin V prescribed in the past No Penicillin V prescribed in the past NNNT
Consult again with throat infection 50% 27% 4


We can sum this up by illustrating what would happen if a GP prescribed antibiotics to 100 fewer patients with throat infection in a year. Thirty three fewer would believe antibiotics were effective, 25 fewer would intend to consult with the problem in the future and 10 fewer would come back within the next year. Game, set and match to masterly inactivity.


  1. CB Del Mar, P Glasziou. Antibiotics for the symptoms and complications of sore throat. In: The Cochrane Database of Systematic Reviews. Available in The Cochrane Library [database on disk and CDROM]. The Cochrane Collaboration; Issue 1. Oxford: Update Software; 1997. Updated quarterly.
  2. P Little et al. Open randomised controlled trial of prescribing strategies in managing sore throat. British Medical Journal 1997 314: 722-7.
  3. P Little et al. Reattendance and complications in a randomised trial of prescribing strategies for sore throat: the medicalising effect of antibiotics. British Medical Journal 1997 315: 350-2.

previous or next story in this issue