Antibiotic treatment failure in four common infections in UK primary care 1991-2012: longitudinal analysis
BMJ 2014; 349 doi: https://doi.org/10.1136/bmj.g5493 (Published 23 September 2014) Cite this as: BMJ 2014;349:g5493
All rapid responses
What's labelled Cochrane contains not always Cochrane
The publication of Craig J Currie et al. from the Cochrane Institute of Primary Care and Public Health, Cardiff University, Cardiff, UK, and Global Epidemiology, Pharmatelligence, Cardiff, UK, is not so intelligent as it wants to be.
1. Conclusions like "from 1991 to 2012, more than one in 10 first line antibiotic monotherapies for the selected infections were associated with treatment failure" mean in the publications reality that in 22 years of observation only 13.9 to 15.4 percent of antibiotic treatment failed. In the meantime 86.1 to 84.6 percent of the antibiotic treatment obviously succeeded instead.
2. When "overall failure rates increased by 12% over this period..." of 22 years this is an annual increase of 0.545 percent only.
In my opinion most of the British GP and Family Doctors did right with their targeting antibiotic treatment from 1991 until 2012. Nearly constant failure rates from 13.9 to 15.4 percent over such a long period instead of augmenting antibiotic resistance, increasing allergies e. g. against penicillin and amoxicillin, ß-lactamase active germs, less effectiveness of older antibiotics e.g. trimethoprim and tetracycline, non indicated ciprofloxacin therapy, and new multi-resistant germs like MRSA or others should be taken into consideration.
This seems to be a very undifferentiated BMJ-publication caricaturing the former idea of critical Cochrane investigation.
Competing interests: No competing interests
Though impressive in scope, there is a critical problem in the interpretation of this analysis due to the majority of the illnesses studied. The authors include “upper respiratory tract infections (for example, pharyngitis, laryngitis, tonsillitis, sinusitis), lower respiratory tract infections (for example, pneumonia, bronchitis, whooping cough), skin and soft tissue infections (for example, cellulitis, impetigo, abscesses), and acute otitis media”.
Of these, sore throat, laryngitis, sinusitis, bronchitis, and otitis media are usually self-limiting, and antibiotics have been found to have either no effect, or minimal effect.[1–5] Antibiotics do not affect the clinical outcome in whooping cough, but are effective in eliminating pertussis, and therefore may reduce transmission (which would not show in this analysis).[6]
Antibiotics are regarded as effective for pneumonia and skin infection by consensus, since placebo-controlled trials would be unethical. [7, 8]
94.4% of treatment failures were identified by switching of antibiotic treatment, rather than any evidence of deterioration, or development of a complication. Given the illnesses studied, one obvious explanation for this might be unrealistic expectations of treatment. Are patients with viral illnesses returning to their doctors when the first course of antibiotics has not worked, and receiving a second?
The authors state that “signals of actual bacterial resistance might therefore not become apparent against a background of unnecessary antibiotic use.” This, surely, is a serious understatement. It is difficult to see how the effects of antibiotics on what are primarily self-limiting viral infections provides any useful information about bacterial resistance.
1. Spinks A, Glasziou PP, Del Mar CB (2013) Antibiotics for sore throat. The Cochrane database of systematic reviews 11: CD000023. Available: http://www.ncbi.nlm.nih.gov/pubmed/24190439.
2. Reveiz L, Cardona AF (2013) Antibiotics for acute laryngitis in adults. The Cochrane database of systematic reviews 3: CD004783. Available: http://www.ncbi.nlm.nih.gov/pubmed/23543536.
3. Ahovuo-Saloranta A, Rautakorpi U-M, Borisenko OV, Liira H, Williams JW, et al. (2014) Antibiotics for acute maxillary sinusitis in adults. The Cochrane database of systematic reviews 2: CD000243. Available: http://www.ncbi.nlm.nih.gov/pubmed/24515610.
4. Smith SM, Fahey T, Smucny J, Becker LA (2014) Antibiotics for acute bronchitis. The Cochrane database of systematic reviews 3: CD000245. Available: http://www.ncbi.nlm.nih.gov/pubmed/24585130.
5. Venekamp RP, Sanders S, Glasziou PP, Del Mar CB, Rovers MM (2013) Antibiotics for acute otitis media in children. The Cochrane database of systematic reviews 1: CD000219. Available: http://www.ncbi.nlm.nih.gov/pubmed/23440776.
6. Altunaiji S, Kukuruzovic R, Curtis N, Massie J (2007) Antibiotics for whooping cough (pertussis). The Cochrane database of systematic reviews: CD004404. Available: http://www.ncbi.nlm.nih.gov/pubmed/17636756.
7. CREST Management of Cellulitis in Adults (n.d.). Available: http://www.gain-ni.org/images/Uploads/Guidelines/cellulitis-chart.pdf. Accessed 26AD–9AD.
8. Lim WS, Baudouin SV, George RC, Hill AT, Jamieson C, et al. (2009) BTS guidelines for the management of community acquired pneumonia in adults: update 2009. Thorax 64 Suppl 3: iii1–i55. Available: http://www.ncbi.nlm.nih.gov/pubmed/19783532.
Competing interests: No competing interests
This phenomenal analysis must be a first, and deserves congratulations.
Analysing GP records is hugely difficult, and GPRD goes to great lengths to secure accurate reliable data, with mitigation of bias.
The authors expected to see a general rise in ‘treatment failure’, no doubt in anticipation of rising antibiotic resistance. But the study revealed unexpected results, not easy to explain. Standard trusty amoxicillin appears to have remained reliable throughout the decades ! Whilst Trimethoprim, Ciprofloxacin and Cefalexin are increasingly failing. The authors suggest this may be because of inappropriate off-guideline usage, which would explain their initial failure-rate back in the 1990s. But how does it explain the steady deterioration vis-à-vis amoxicillin ?
Maybe there is an unrecognised bias resulting from the definition of ‘monotherapy’ and ‘treatment failure’? Only episodes of ‘monotherapy’ are examined, whereby the initial antibiotic is the first in a 30-day period, and no other antibiotic should have been given within 30 days. But then the definition of ‘failure’ includes patients given another antibiotic within 30 days, lumped together with other criteria, which may represent very different things. Surely there should be no one with a second antibiotic within 30 days of the first, since that was an exclusion criterion ? To my mind, a switch of antibiotic within the period of illness under observation (2-15 days), should be the better indicator of a GP thinking that treatment has failed.
Given that the really significant ‘failures’ relate to hospital admissions for lower respiratory infections, it would be useful if the authors followed up with a detailed analysis of this subset, with a more refined definition of failure. I suggest something akin to that used by Hess et al., in their reference 14:
“ defined as ≥1 of the following events ≤30 days after index date: a refill for the index antibiotic after completed days of therapy, a different antibiotic dispensed >1 day after the index prescription, or hospitalization with a pneumonia diagnosis or emergency department visit >3 days postindex. “
There is a wealth of information here ready to answer many more questions. An especially useful tool would be an ‘antibiotic failure’ monitoring system, telling us when our ‘first-line’ schedules need updating !
Well done, Currie et al, and GPRD !
Reference
Hess G, Hill JW, Raut MK, Fisher AC, Mody S, Schein JR, et al. Comparative antibiotic failure rates in the treatment of community-acquired pneumonia: results from a claims analysis. Adv Ther2010;27:743-55.
Competing interests: No competing interests
As touched on briefly by the authors in the discussion and conclusions, the main reasons for antibiotic treatment failure are often (in not so many words): 1) wrong diagnosis (viral aetiology) and 2) inadequate communication (to explain cause and management, side effects, to elicit and manage true patient agenda) - all often due to lack of time in consultation.
It would be interesting to know if there was a demonstrable seasonal difference in failure rate, with the majority of viral illness (and antibiotic prescribing for urtis) happening in the autumn/winter months.
The overall message of the need to prescribe judiciously and to follow local guidance is true as ever
Competing interests: No competing interests
Re: Antibiotic treatment failure in four common infections in UK primary care 1991-2012: longitudinal analysis
Another Curiosity
Normally I do not check simple statistical analysis expecting failure-free percent calculations in scientific papers. But the authors' say:
"In 1991, the overall failure rate was 13.9% (12.0% for upper respiratory tract infections; 16.9% for lower respiratory tract infections; 12.8% for skin and soft tissue infections; and 13.9% for acute otitis media). By 2012, the overall failure rate was 15.4%, representing an increase of 12% compared with 1991"
and this neglects totally that the difference of 1.5% points between 13.9% and 15.4% overall failure rates is not at all "representing an increase of 12%"! It is an increase of 10.8% over a period of 22 years. That means an annual increase of 0.49% only.
This publication must be revised.
Competing interests: No competing interests