Biostatistics in Research Practice: The ESPRIT trial

This website is for students following the M.Sc. module Biostatistics in Research Practice at the University of York.

The ESPRIT Trial was a comparison of aspirin and dipyridamole combined versus aspirin alone for the secondary prevention of vascular events after ischaemic stroke of presumed arterial origin.

Patients were recruited within 6 months of a transient ischaemic attack or minor stroke of presumed arterial origin. T hey were randomised to aspirin (30–325 mg daily) with dipyridamole (200 mg twice daily) (n=1363) or to aspirin alone (n=1376). The primary outcome event was the composite of death from all vascular causes, non-fatal stroke, non-fatal myocardial infarction, or major bleeding complication, whichever happened first. Treatment was open, but auditing of outcome events was blinded. Primary analysis was by intention to treat.

Mean follow-up was 3·5 years (SD 2·0). Median aspirin dose was 75 mg in both treatment groups (range 30–325); extended-release dipyridamole was used by 83% (n=1131) of patients on the combination regimen.

The following graph was produced:

Two graphs side by side, entitled 'Primary outcome event' and 'Ischaemic events'. Each has 'Time from randomisation (years)' on the horizontal axis and 'Cumulative event rate (%)' on the vertical axis. Each graph shows two rising stepped lines labeled 'Aspirin alone' and 'Aspirin + dipyridamole'. The aspirin alone line rises more steeply in each graph.

Primary outcome events arose in 173 (13%) patients on aspirin and dipyridamole and in 216 (16%) on aspirin alone (hazard ratio 0·80, 95% CI 0·66–0·98). Patients on aspirin and dipyridamole discontinued trial medication more often than those on aspirin alone (470 vs. 184), mainly because of headache.

Questions

1. What is meant by “primary outcome event” and “primary analysis”?

Check suggested answer 1

2. Mean follow-up was 3·5 years (SD 2·0). One last time for an old favourite: what can we deduce about the distribution of follow-up time? Why should the follow-up times vary so much?

Check suggested answer 2

3. What type of graph is this? Why is the vertical axis labelled “Cumulative event rate”?

Check suggested answer 3

4. What is meant by hazard and hazard ratio and how can we interpret a hazard ratio = 0·80, 95% CI 0·66–0·98? What assumptions are made about the data for this calculation?

Check suggested answer 4

5. The hazard ratio = 0·80 with 95% CI 0·66–0·98. The centre of this confidence interval would be 0.82. Why is the hazard ratio point estimate not in the centre of its confidence interval?

Check suggested answer 5

6. Patients on aspirin and dipyridamole discontinued trial medication more often than those on aspirin alone, 470 out of 1363 compared to 184 out of 1376. What method would we use to test the null hypothesis that discontinuation is unrelated to medication?

Check suggested answer 6

Reference

The ESPRIT Study Group. Aspirin plus dipyridamole versus aspirin alone after cerebral ischaemia of arterial origin (ESPRIT): randomised controlled trial. Lancet 2006; 367: 1665–73.

Back to Biostatistics in Research Practice index.

To Martin Bland's home page.

This page maintained by Martin Bland.
Last updated: 5 April, 2007.