Skip navigation

Better prescribing of NSAIDs


Providing information to doctors about better prescribing, also called academic detailing, has been shown to be a good thing. A Cochrane review [1] of 18 randomised or quasi randomised trials found positive effects on practice in all trials, but only one measured a patient outcome. A new study from Adelaide [2] demonstrates an apparent remarkable decrease in admissions for perforations, ulcers and gastrointestinal bleeding following detailing about appropriate use of NSAIDs.


The study was conducted in a particular area of metropolitan Adelaide with a population of 154,000. Two surgery visits were made in 1992. These visits focussed on better use of prescribed NSAIDs. The visits were preceded by a review of the literature, with a written summary of useful information prepared, and externally reviewed by experts and opinion leaders.

Printed materials providing a source of 'balanced, unbiased information' were then left at each doctor visit. The programme highlighted the extensive use of NSAIDs and large number of adverse reactions, specifically in high use and high risk groups.

Just under 90% of the 210 doctors practising in the area (80% of whom were GPs) received the service conducted by pharmacists with teaching hospital clinical experience (and most continued to receive advisory visits from the pharmacists about other areas of prescribing). A neighbouring comparison area of 72,000 people did not have the intervention.

Outcomes measured for NSAID use were defined daily doses of NSAIDs prescribed per person per day, and units of NSAIDs delivered to pharmacies from manufacturers and wholesale suppliers. Hospital admissions were monitored for persons with ICD codes indicating ulceration or bleeding events in the upper GI tract with or without bleeding in the intervention and comparison area.


NSAID prescribing in the intervention area fell (Table). Two to four years after the visits prescribing in the intervention area was 9% lower in the intervention area compared to the control area. Two years after the intervention supplies of NSAIDs to pharmacies were 25% lower.

Table: Outcomes of NSAID prescribing - use of NSAIDS and hospital admissions for perforation and ulcers

Outcome Period Intervention area Comparison area
NSAID prescribed 94-96 compared with 91 -16% -7%
NSAID delivered to pharmacies 94 compared with 91 -23% +5%
Hospital admissions for perforation or ulceration 1992 20/100,000 14/100,000
Hospital admissions for perforation or ulceration 1998 6/100,000 14/100,000

Over the period 1986 to 1998 hospital admission rates for upper GI tract ulceration or bleed in the comparison area was unchanged at about 14 per 100,000 people (Table). In the intervention area rates were rising before the intervention to a peak of about 20 per 100,000 before the intervention. Thereafter rates began to fall, and by 1998 were about 6 per 100,000, a fall of 70% from the peak.


There is much to say about this. The study could be criticised in a number of ways, but studies of every day interventions are hard to do. The results are remarkable because we are given patient outcomes, starkly unchanged in the comparison area but with big changes in the intervention area. The size and duration of the effect astonish.

But wait. If one thing shines out from this study it is the way in which the intervention was made. Evidence was gathered, it was weighed, it was honed by experts and opinion leaders, and it was 'balanced, unbiased information'. The doctors who received it were treated as the responsible people they were, and not like mushrooms.

Perhaps this is the secret. If so, this is a clear lead for how PCGs, of about the same average size as the intervention area in this study, could make a difference.


  1. T O'Brien et al. Educational outreach visits: effects on professional practices and health care outcomes. Cochrane Library 1999 issue 4.
  2. FW May et al. Outcomes of an educational-outreach service for community medical practitioners: non-steroidal anti-inflammatory drugs. Medical Journal of Australia 1999 170: 471-474.
previous or next story in this issue