Skip navigation

Reducing antibiotic prescribing


Brownie points are seldom available for doing the simple things that make a difference. Bandolier was taken, therefore, by a study that investigated a method of reducing unnecessary antibiotic prescribing [1] published in a journal with one of the highest impact factors, namely JAMA. It seemed to be just the sort of thing that the new primary care groups in England could be doing to find ways to ensure that best practice becomes the norm.


This was a non-randomised study of four medical office practices in Colorado. Two practices (47,000 people) acted as controls, one (36,000) had a limited intervention, and one (35,000) had a full intervention to try to limit unnecessary antibiotic prescribing. Baseline information as collected over November 96 to February 97, and during the same period the following year when the interventions were applied.

Interventions were based on preliminary studies that identified key factors that had to be addressed in making a change. As usual, this needed a multifactorial approach. The full intervention site received household and office-based patient education materials and a clinician educational intervention. Information was mailed to 25,000 households. It included:

In the physician's office there were posters attached to the wall of each room, with information sheets on the limited role of antibiotics in acute bronchitis or chest colds for patients, plus graphics describing the whole problem of antibiotic resistance in Colorado.

There was also a full physician and nurse education programme, including education on evidence-based management of acute bronchitis and how to say no to patient demands for antibiotics. Meetings were led by the medical director who became the programme champion.

The limited intervention site received office-based educational material only. The control sites received no material.

Eligible patients included all adults who made an office visit for acute bronchitis, sinusitis or upper respiratory tract infection during baseline or study period. Information about diagnosis, prescriptions and return visits was obtained from databases.


All four sites had similar rates of visits and prescription of antibiotics during the baseline collection of data. In the following year in the full intervention site the percentage of patients with uncomplicated bronchitis fell from 78% to 44%. There was no change for the limited intervention site or the control sites (Figure). At the full intervention site the number of return visits did not increase, nor was there any increase in the proportion of patients (about 1%) returning with pneumonia. There was no change in prescribing of antibiotics for uncomplicated upper respiratory tract infections (low) or sinusitis (high). The numbers of visits was the same between baseline and intervention periods.

Figure 1: Reduction in antibiotic prescribing, by intervention


This wasn't a randomised trial, nor could it be blinded. It was very good, though, because it set out to study a problem, found the levers of change, and pulled them. It involved patients, their professional carers, it informed, and it was based on evidence. And, of course, it worked. No surprise there for anyone who has ever been involved with quality improvement in industry. It uses classic techniques.

It is interesting to think about how expensive this or similar schemes would be to implement in a primary care group. Some organisational costs, some printing costs, some time to obtain and present the evidence well. This paper is worth a read for anyone who wants to make things better.


  1. R Gonzales et al. Decreasing antibiotic use in ambulatory practice. Impact of a multidimensional intervention on the treatment of uncomplicated acute bronchitis in adults. JAMA 1999 281: 1512-1519. Materials on line at ( )

previous or next story in this issue