Need to know
Professor Paul Little tackles questions on antibiotic prescribing dilemmas from GP Dr Stephen Gardiner
1. Studies seem to be confirming the dangers of macrolide antibiotics, which are linked to sudden cardiac death. I use them mostly for those with 'penicillin allergy'. Do you think these patients, whom studies suggest are rarely truly penicillin-allergic, risk more from taking erythromycin or penicillin?
There are 14 membered macrolides (particularly erythromycin and clarithromycin) that cause prolongation of the QT interval, and this has been associated with rare instances of cardiac arrhythmia (especially torsade de pointes) and sudden cardiac death.
A retrospective evaluation documented an 80 per cent increase in risk of sudden cardiac death for patients taking erythromycin alone compared with patients not taking any anti-biotic. But since patients taking amoxicillin had a 50 per cent increased risk, the risk of sudden death with erythromycin alone is likely to be mainly the result of the risk from the infection and/or severity of the infection.
However, there is another major danger: drugs that inhibit the hepatic metabolism of erythromycin via the cytochrome P450 3A (CYP3A) system may increase the serum
levels of erythromycin and thereby increase the risk of cardiac arrhythmia.
Studies have demonstrated interactions between either erythromycin or clarith-romycin and ciclosporin, pimozide, disopyramide, carbamazepine, midazolam, digoxin, statins, calcium channel blockers and warfarin.
In a retrospective evaluation patients taking these drugs (predominantly diltiazem or verapamil in this study) had five times the risk of sudden death.
Whether this effect is due to increased serum levels of erythromycin, or the direct and/or combined effect of such calcium channel blockers, is unclear.
But whatever the reason some caution should be used in combining these drugs.
So although the absolute risk of death among patients taking erythromycin is small (three deaths in 194 person years), it is probably wise to use azithromycin rather than erythromycin or clarithromycin in patients receiving concurrent therapy with CYP3A inhibitors, where there is a good history of penicillin allergy.
2. There has been a move to make some antibiotics available over the counter, such as chloramphenicol for conjunctivitis. Given that evidence is appearing that it has little or no benefit, is this a good idea?
Chloramphenicol eye drops went OTC in June 2005, and trimethoprim for cystitis is also being considered for an OTC move.
For conjunctivitis, a systematic review found modest effectiveness with antibiotics, and a recent primary care trial in children suggested even less benefit. Some argue that an OTC increase in chloramphenicol use is not a major issue, since it has rarely been implicated in the development of resistance to date.
However, there is still the issue of the denuding of local naso-pharyngeal flora (since the lachrymal duct drains into the nasopharynx), the inappropriate use of resources and the medicalisation of self-limiting illness.
My view is therefore that the first-line use of an immediate course of chloramphenicol drops is probably misplaced; arguably symptomatic treatment in the first instance (bathing with boiled cooled water) and a 'delayed' prescription would be reasonable, and is supported by recent evidence.
Chloramphenicol ointment has advantages in that it stays around for much longer, but for very young children drops are easier to use, and may suit a wide age range.
Regarding cystitis, the evidence from the few placebo-controlled trials suggests that for women who attend primary care, antibiotics are likely to provide benefit of one to two days in an illness lasting in total seven to eight days.
Given the unpleasantness of cystitis, many women would opt for antibiotics on this basis. But only a small minority of women with cystitis-like symptoms come to the doctor. Most self-medicate and use OTC remedies. The women we don't see probably have a milder form, so it is debatable whether a significant increase in antibiotic prescribing for this group is appropriate.
3. What can be done to help us resist the pressure from parents for treatment for conjunctivitis when schools and nurseries adopt exclusion policies for affected children?
I explain that conjunctivitis is no more dangerous or contagious than the common cold, and that national guidance suggests that children should not be excluded from schools and nurseries any more than children with colds are excluded.
I also tell parents I am very happy to speak to any nurseries or schools about this issue (and have so far rarely been taken up on the offer).
4. I give deferred scripts for some conditions, such as otitis media, on a Friday. But the evidence they are used appropriately seems sketchy, so I sometimes postdate them to prevent early use. Is there any evidence for the benefit of this?
Where a patient is not unwell at all and not demanding antibiotics it is probably reasonable not to offer any antibiotics for any respiratory infection. Delayed prescribing is useful, especially where there are high patient expectations.
In addition, given our lack of evidence about who is at risk of complications or prolonged symptoms, it is also a useful technique where the prescriber feels uncertain about whether the patient is at risk.
When properly used it is effective in reducing antibiotic use and is acceptable to patients. For otitis media in particular, there is also reasonable evidence that using a well-defined 'wait and see' policy is likely to be safe.
A large cohort study of 5,000 children that used the Dutch 'wait and see' guidance (to use antibiotics only in the small minority where there is still significant otalgia or fever after 72 hours, or a discharge that continues for more than 10 days) resulted in only one child getting mastoiditis, and the parents in that case waited far longer than the 72 hours.
There is now evidence from trials in otitis media, sore throat and chest infections that a clearly defined delayed prescription strategy is acceptable, and results in low use of antibiotics.
The key issue is making it clear you don't believe the prescription is necessary and that antibiotics are not likely to help in their particular case.
It's also important to provide clear guidance about the natural history and about good symptomatic self-treatment (analgesic antipyretics). It should be made clear that the prescription is to be used only if the illness is getting significantly worse or not starting to improve after a clear time period related to its natural history.
The time periods for each illness are: otitis media, three days after seeing the doctor (or four days in total); sore throat, five days after seeing the doctor (eight days in total); rhinosinusitis, seven to 10 days after seeing the doctor (two weeks in total); chest infections, 10 days after seeing the doctor (or three weeks in total).
For patients who are more unwell, or where you are concerned that complications might develop, it is reasonable to halve these delaying times.
5. I think there has been an increase in antibiotic prescribing for conditions such as otitis media since GPs gave up their out-of-hours responsibilities. Is there any evidence to back this up? What is the latest advice about antibiotics for otitis media and are we seeing mastoiditis or other complications re-emerging?
It will be some years before trends due to changes in out-of-hours care become apparent. There is some evidence that since the first paper on delayed prescribing for sore throat in 1997, there has been a small increase in the admission rate for mastoiditis, but the evidence is a little patchy since not all the data sets show this.
It is clear that an increase in prescribing to stop mastoiditis would require thousands of scripts to prevent one admission, which is difficult to justify and would result in an increasing spiral of antibiotic resistance. As with the other respiratory infections, the key bit of missing information is knowing which patients are particularly at risk.
6. Is there any place for parenteral antibiotics in primary care other than in patients suspected to have meningococcal infections?
I think there is little case for parenteral antibiotics unless it is clear a systemically unwell patient will have difficulty getting to hospital, although there have been few trials of community-based parenteral antibiotic management, so the question is as yet unanswered. Even with meningitis there is debate about how safe it is.
7. I tend to prescribe antibiotics more often and earlier in patients with chronic chest problems. It is likely that most of the time they have viral LRTIs. Do you think it is still worthwhile doing this to reduce the complication and hospital admission rates?
This is one of the areas where there is little research to inform practice. Many exacerbations will be of viral illness, but on the other hand those with chronic bronchitis do harbour significant pathogens in the lower tract, and are also at high risk of admission.
So in the absence of more evidence, I think it is reasonable to treat exacerbations with antibiotics, particularly those with very green sputum – where the research suggests bacterial super infection is more likely.
8. Recently it was reported in a study that children with proven flu improve more quickly if given antibiotics than if not. It was suggested we should move away from the idea of bacterial or viral infections and think of conditions which do or don't respond to antibiotics. Do you think there is any merit in this?
I think there is some merit in this concept. If some infections which apparently start as viral infections show a marked deterioration, then it is certainly reasonable to treat. However, I think some caution is needed before advocating antibiotic treatment in patients with flu and other viral illnesses, which would clearly have a major effect on the volume of antibiotic prescription.
The paper concerned showed the effect in children with flu and was based on a secondary analysis. So until more definitive evidence is available, and we understand which patients do and which do not benefit from antibiotics, it cannot really be used to justify prescribing of antibiotics.
9. Despite all our efforts to reduce antibiotic prescribing, it is impossible to avoid giving
them to some patients, particularly for chest or sinus problems. Which would you consider the least harmful antibiotic (in terms of the wider community) in these instances?
Interestingly, chest and sinus infections are two of the infections where there is little evidence that antibiotics work very well – the Cochrane reviews suggest that, at best, prescribing for either condition results in modest benefits. The issue is more about patient expectation and traditions of prescribing.
In the case of antibiotic prescribing for sore throat and otitis media, patients nearly always used to get antibiotics, but now there is less pressure. Similarly the tide is turning for chest infections and it is getting easier to avoid prescribing, but sinusitis remains a major problem.
Key techniques are to explain the natural history (see question on delayed prescribing) and provide good advice on analgesics/
antipyretics for the symptoms. The least harmful antibiotic for chest and sinus infections is probably amoxicillin, 500mg three times a day for five days.
We should be aiming for higher doses of antibiotics for a shorter time.
Regarding the dose and current resistance, there is less evidence of resistance in representative community settings in the UK, but 500mg is probably sensible now and if anything should help minimise the development of resistance.
What i will do now
Dr Stephen Gardiner responds to the answers to his questions
• The absolute risk to patients from macrolides is low, and in those who need one, azithromycin seems to be safe for those taking CYP3A inhibitors
• I will continue to use delayed and postdated scripts and it is reassuring to see evidence accumulating for the benefits of this
• I am reassured that there is no pressure to prescribe to prevent mastoiditis; the Dutch guidelines on prescribing in otitis media seem sensible and should be promoted
• My lower threshold for antibiotics for those with exacerbations of chronic chest problems seems to be justified
• I will continue to give amoxicillin to those who 'need' their antibiotics
Stephen Gardiner is a GP in Bridgwater, Somerset
Paul Little is professor of primary care research at the University of Southampton and a GP in Romsey, Hampshire – his research interests include the management of acute infections and enabling patients to self-manage common conditions
Competing interests Abbott Laboratories has paid Paul Little for two consultancy sessions in the last five years on the issue of complications of acute respiratory infections