Intended for healthcare professionals

General Practice

A randomised controlled trial of feedback to general practitioners of their prophylactic aspirin prescribing

BMJ 1997; 315 doi: (Published 05 July 1997) Cite this as: BMJ 1997;315:35
  1. Peter McCartney (phpspmcc{at}, general practitionera,
  2. Wendy Macdowall, research assistanta,
  3. Margaret Thorogood, senior lecturera
  1. a Health Promotion Sciences Unit, Department of Public Health and Policy, London School of Hygiene and Tropical Medicine, London WC1E 7HT
  1. Correspondence to: Dr McCartney
  • Accepted 17 February 1997


Although low dose aspirin reduces risk in patients with heart disease, many such patients do not receive daily prophylactic aspirin.1 We report a trial of feedback of general practitioners' data on aspirin prescribing aimed at increasing coded aspirin prescribing in patients with heart disease.

Subjects, methods, and results

Computerised practices were randomised to receive feedback on their prescribing, either of aspirin for patients with ischaemic heart disease or of hormone replacement therapy for women who had had hysterectomies. We approached 48 practices in north London; nine refused, and 11 were excluded. Of the 28 (58%) practices in the study, seven were single handed and six had five or more partners. All participating practices used the emis computer system except two that used Paradoc; both systems yielded sufficiently reliable and comparable data. Eligible practices had to have computerised information on hysterectomies and ischaemic heart disease and use their systems for repeat prescribing. Practices were then randomised by using sealed envelopes to an intervention or a control group.

Patients were considered to have ischaemic heart disease if they had such a diagnosis coded; had had a myocardial infarction or angina; had had a revascularisation procedure; or were taking a nitrate by repeat prescription. We collected data for prescribing of both aspirin and hormone replacement therapy in all practices at baseline and follow up (at least three months later).

Feedback on the baseline data—together with appropriate educational input—was given at a practice meeting.2 3 The one hour session had approval for the postgraduate education allowance. We encouraged practices to audit certain patients—for example, patients with ischaemic heart disease not apparently taking aspirin—and facilitated support through medical audit advisory groups. We calculated the change in the proportion of patients with ischaemic heart disease taking daily aspirin over four months (the average time to follow up). The standard error of this change was calculated:


where Pa1 represents proportion of patients taking aspirin in the aspirin group at baseline, Pc1 represents these patients at follow up, Pc2 represents the control group, and n1, n2, n3, and n4 are the appropriate denominators.

There were 14 practices in each arm and a total of 182 200 patients. We identified from computer searches a diagnosis of coronary heart disease or repeat prescriptions for nitrates, or both, in 2813 patients, of whom 1354 took aspirin. In three randomly selected practices we validated the computerised data. We examined a random sample of 20% of the written records of patients aged 40-69 years for missed computer diagnoses of ischaemic heart disease and a random 50% of the written records of patients identified by computer searches as having ischaemic heart disease. There were no important discrepancies.

Figure 1 shows the proportion of patients with ischaemic heart disease prescribed daily aspirin before and after the intervention. Such prescribing rose from 787/1646 (47.8%) to 1004/1725 (58.2%) in the intervention group, compared with 567/1167 (48.6%) to 610/1220 (50.0%) in the control practices (P<0.001). As practices in the aspirin arm were reminded of the code for “buying aspirin over the counter” but those in the hormone replacement arm were not, we ignored these codings when measuring the outcome. In the written records that we examined, three (4.6%) patients with ischaemic heart disease bought their aspirin over the counter.

Fig 1
Fig 1

Average percentage of patients with ischaemic heart disease taking aspirin at entry to study and at follow up (on average four months after feedback) for treatment and control groups (14 practices in each group). Vertical bars represent 95% confidence intervals around the overall mean


Feedback of prescribing practice can increase the proportion of patients with ischaemic heart disease receiving prescribed daily aspirin by 9%. Some of the apparent increase may be due to improved coding, but only 2% of patients with ischaemic heart disease had only a written record (no computer record) of aspirin use, and less than 5% were buying their aspirin over the counter.

About a million patients have ischaemic heart disease in the United Kingdom.4 If aspirin use in 86 high risk patients prevents one death in two years3 then a 9% increase in prescribing daily aspirin would reduce mortality from ischaemic heart disease by 1134 deaths every two years. In practices similar to those studied, feedback to 20 general practitioners (number of doctors needed to treat) or 6.4 practices (number of practices needed to treat) would be needed to prevent one death from ischaemic heart disease in two years.


We thank the 28 practices that took part, especially the practice managers and general practitioners. The medical audit advisory groups provided a support network that facilitated this project.

Funding: Department of Health, Camden and Islington Health Authority, North Thames Research and Development, London Implementation Zone Educational Initiative.

Conflict of interest: None.


  1. 1.
  2. 2.
  3. 3.
  4. 4.