The derivation and validation of an electrocardiograph prediction model for the diagnosis of non-ST-elevation myocardial infarction in the Emergency Department.

  • Niall Morris

Student thesis: Phd


Background: The electrocardiograph (ECG) has been integral to the diagnosis of acute coronary syndromes challenge since the mid-20th Century and is the first investigation people undergo on presentation to the Emergency Department. It is now used in non-ST-elevation myocardial infarction (NSTEMI) clinical prediction models and clinicians are asked to make a subjective assessment of whether there is evidence of ischaemia. Aims: I aimed to derive and validate an ECG prediction model that would produce an objective measure of ECG ischaemia in the diagnosis of NSTEMI. Methods: I derived the model using logistic regression in a single centre retrospective cohort. The model was then validated in a multi-centre prospective cohort. The primary outcome was NSTEMI. Results: The model had a sensitivity of 25.6% (19.0-33.2%), specificity of 96.3% (95.0-97.4%), a positive predictive value of 50.0% (40.0-60.0%) and a negative predictive value of 90.1% (89.2-90.9%). Validated, the sensitivity was 23.5% (17.4-30.6%), specificity was 95.2% (93.6-96.4%), positive predictive value of 46.0% (36.6-55.7%) and a negative predictive value of 87.6 % (86.7-88.5%). It was not statistically inferior to clinician assessment on direct comparison. Conclusion: I have derived and validated an electrocardiograph prediction tool that is comparable to an Emergency Doctor in the diagnosis of NSTEMI.
Date of Award1 Aug 2019
Original languageEnglish
Awarding Institution
  • The University of Manchester
SupervisorRick Body (Supervisor), Garry Mcdowell (Supervisor) & Kevin Mackway-Jones (Supervisor)

Cite this