UCL Institute of Health Informatics


Farr Institute researchers report novel method of assessing the true relevance of trial findings in real world patients

19 July 2016

Clinical trials of new drug treatments are traditionally conducted in patient populations that are very different from the “real world” populations doctors treat in day-to-day practice. The relevance of these trials to the treatment of individual patients in the consulting room care is therefore a constant concern to the practicing clinician. This week in the BMJ investigators from the Farr Institute in London report a novel method of assessing the true relevance of trial findings in real world patients. 

The trial in question was the recent PEGASUS-TIMI-54 trial. It was the first study to examine the benefits and harms of combined aspirin and ticagrelor treatment in patients who had already survived one year after a heart attack. Ticagrelor which, like aspirin, acts by thinning the blood is usually stopped after the first year but the PEGASUS-TIMI-54 trial showed that in patients who received continued treatment beyond 1 year the risk of further heart attack, stroke or death was reduced (benefit) but the risk of major bleeding was increased (harm) compared with patients who received only aspirin.  

To be selected for the PEGASUS-TIMI-54 trial patients needed to fulfil some very specific criteria. Consequently they were quite unlike most patients who have survived a year after heart attack that doctors meet in the consulting room.  Can doctors be sure that the trial findings apply to these real world patients? To answer this important question the investigators from the Farr Institute turned to the CALIBER dataset of anonymised GP records of real patients linked to their hospital records and to the national heart attack and death records. They identified a real-world population of 7328 patients who had survived at least 1 year after a heart attack. Key findings were as follows: 

• CALIBER’s real world patients were very different from the trial patients.  Only 1676 (23%) met the selection criteria for the PEGASUS-TIMI-54 trial.

• Even this “target population” of 1676 patients was different from the trial population, being an average of 12 years older with considerably higher proportions of women. 

• The risks of death, stroke or further heart attack (18%) and bleeding events (3%) in CALIBER’s target population during three years follow-up were double those reported in the PEGASUS-TIMI-54 trial. The high risk of patients who have survived a year after heart attack was an important finding that has not hitherto been widely recognized.

• The main finding was that in CALIBER’s target population treatment of just 99 patients with aspirin plus ticagrelor would prevent one death, stroke or further heart attack. Treatment of 133 patients would be needed to cause one additional bleeding event. Moreover when the calculation was applied to all of CALIBER’s 7328 real world patients who had survived a year or more after a heart attack, the benefits and the bleeding risks appeared very similar.

This study from the Farr Institute has applied novel methodology using linked electronic health records to show that patients who have survived at least a year after a heart attack remain at high risk of death, stroke or further heart attack. This risk can be reduced by continued treatment with a combination of aspirin and ticagrelor which has the potential to save many lives. In future, the methodology reported in this BMJ paper should be applied in other settings to provide doctors with real-world evidence of how trial findings might be applied in clinical practice.

Authors: Timmis A, Rapsomaniki E, Chung SC, Pujades-Rodriguez M, Moayyeri A, Stogiannis D, Shah AD, Pasea L, Denaxas S, Emmas C, Hemingway H2. Prolonged dual antiplatelet therapy in stable coronary disease: comparative observational study of benefits and harms in unselected versus trial populations. BMJ. 2016 Jun 22;353:i3163. doi: 10.1136/bmj.i3163.