Article Text

Download PDFPDF
Competing risk analysis of factors related to long-term incidence of CHD
  1. Gilbert MacKenzie1,2,3,
  2. Mary Greig3,
  3. Iris Hay3,
  4. John Pemberton3
  1. 1Centre of Biostatistics, University of Limerick, Limerick, Ireland
  2. 2Department of Statistics (CREST), ENSAI, Bruz, France
  3. 3The Department of Community Medicine, The Queen's University of Belfast, Belfast, UK
  1. Correspondence to Professor Gilbert MacKenzie, The Chalet, 85, Maryville Park, Belfast BT9 6LQ, UK; gilbert.mackenzie{at}


Background The 5-year follow-up results for the 1202 middle-aged men prospective study of coronary heart disease (CHD) incidence were published in 1980. This paper extends the follow-up, relating the development of CHD to 10 risk factors.

Methods The population studied comprised all men born in the 10-year period 1909–1918 aged 45–64 years at baseline who were registered in 6 group practices in Belfast. Some 1202 CHD-free men entered the study. Study end points included: (a) the development of CHD, (b) the development of myocardial infarction, (c) the development of angina pectoris, (d) death from myocardial infarction and (e) death from other causes. The men were followed for an average of 6.9 years, and the influence of 10 risk factors was assessed by Cox's proportional hazards model in a competing risk framework.

Results The analysis of first major CHD event identified four risk factors—diastolic blood pressure, serum cholesterol, T wave abnormality and tobacco index. First myocardial infarction depended on diastolic blood pressure, ST abnormality and tobacco index, while time to first angina pectoris depended on serum cholesterol, T abnormality, tobacco index and age at entry. These findings do not support the hypothesis of a common risk factor profile in the myocardial infarction and angina pectoris groups. The sensitivity of all models was poor.

Discussion The study confirms the role of known risk factors in the development of first CHD event. It also suggests that the risk factors involved in developing myocardial infarction and angina pectoris differ. The poor sensitivity of models suggests the presence of unmeasured risk factors in the aetiology of CHD.

  • Epidemiology of cardiovascular disease
  • Epidemiological methods

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • This paper was drafted by the first author in September 1986, 10 years after JP had retired from the department which he had led from his appointment in 1958

  • MG and JP deceased.

  • Contributors GM drafted the paper and analysed the data. MG conducted the baseline examinations and assisted with the compilation of the Methods section. IH prepared the original data on punch cards and computerised the data for analysis. JP contributed to drafting and polishing the paper.

  • Funding The Nuffield Provincial Hospitals Trust, the Northern Ireland Hospitals Authority and the British Heart Foundation supported this study with generous grants over a number of years.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The data are computerised and held in a file compatible with SPSS, but are not in a state for general circulation as it is many years since they were analysed.