Intended for healthcare professionals

Practice Diagnosis in General Practice

Chest pain

BMJ 2009; 339 doi: (Published 03 November 2009) Cite this as: BMJ 2009;339:b4117
  1. Michael Jelinek, cardiologist and associate professor of medicine1,
  2. Kevin Barraclough, general practitioner2
  1. 1St Vincent’s Hospital and University of Melbourne, Melbourne, Australia
  2. 2Painswick GL6 6TY
  1. Correspondence to: M Jelinek michael.jelinek{at}

    Having a sense of the accuracy of diagnostic tests will help general practitioners to interpret and use the tests appropriately and, as in the example of chest pain, avoid unnecessary testing (doi:10.1136/bmj.b3823)

    A 45 year old man attends an emergency department having had 10-15 minutes of severe retrosternal chest pain at a party earlier that evening. The pain occurred at rest and resolved spontaneously when he sat quietly. His electrocardiogram and troponin concentration are normal on arrival at hospital and again eight hours later. He is discharged that night and advised to see his general practitioner the next day for follow-up.

    The diagnostic dilemma

    The likeliest cause of retrosternal pain in this patient is gastrointestinal. However, this presentation could be due to lethal coronary artery disease. The clinician needs to know whether the risk of coronary disease is low or whether the patient requires further testing to clarify the probability of coronary artery disease and the risk of cardiovascular death. An exercise electrocardiogram might refine the probability of disease and its prognosis.

    The diagnostic approach: probabilistic reasoning

    Probabilistic (or bayesian) reasoning provides the framework with which a clinician can track how the probability of disease changes at each step in the diagnostic reasoning process. Sensitivity and specificity are reformulated as positive and negative likelihood ratios. These multiplying factors convert pre test odds to post-test odds.1

    In this case probabilistic reasoning can help in three steps:

    • How common is important coronary artery disease in the population (the pretest probability when the only information we have is age and sex)?

    • If a patient has symptoms, how does this affect the probability of coronary artery disease?

    • How useful is an exercise electrocardiogram in refining this diagnostic probability upwards or downwards and past some decision threshold for angiography? In addition, how useful is it for prognosis?

    For the general …

    View Full Text

    Log in

    Log in through your institution


    * For online subscription