Intended for healthcare professionals


Comparison of case fatality in smokers and non-smokers after acute cardiac event

BMJ 1997; 315 doi: (Published 18 October 1997) Cite this as: BMJ 1997;315:992
  1. Gabe S Sonke, postgraduate studenta,
  2. Alistair W Stewart, biostatisticiana,
  3. Robert Beaglehole (r.beaglehole{at}, professora,
  4. Rod Jackson, associate professora,
  5. Harvey D White, cardiologistb
  1. a Department of Community Health, Faculty of Medicine and Health Science, University of Auckland, Private Bag 92019, Auckland, New Zealand
  2. b Department of Cardiology, Green Lane Hospital, Auckland,
  1. Correspondence to: Professor Beaglehole
  • Accepted 25 April 1997


Although smoking is a major modifiable risk factor for acute myocardial infarction, it has also been associated with an up to twofold lower risk of dying in hospital after an acute myocardial infarction.1 2 We analysed data from a community based register of coronary heart disease to determine whether differences in case fatality (the proportion of those dying) between smokers and non-smokers are restricted to patients who have been admitted to hospital and to evaluate possible explanations for this smoker's paradox.

Subjects, methods, and results

All deaths related to coronary causes and all admitted patients aged 25–64 who met predefined criteria for myocardial infarction or coronary death were identified in Auckland, New Zealand, between 1986 and 1992 as part of the World Health Organisation MONICA (monitoring trends and determinants in cardiovascular disease) project. Study criteria, and methods of case finding and data collection procedures have been published.3 4 Postmortem examinations were performed on 63% of those who died from cardiac causes. Deaths before admission to hospital, deaths within 28 days after admission, and the total number of deaths were measured. Smoking was determined by direct questioning of surviving patients and of relatives of those who died. Patients were classed as current smokers (those who smoked at least one cigarette a week at the onset of symptoms or gave up smoking less than one month before the index event), ex-smokers (those who had abstained from smoking for at least one month before the onset of symptoms), or non-smokers (those who had never smoked). Logistic regression models were used to assess the effects of smoking on case fatality after adjusting for age, sex, history of myocardial infarction, and history of angina. For those admitted to hospital, adjustments were based on whether they received thrombolytic treatment. An adjustment for the year of infarction was included to account for time trends in event rates.

Between January 1986 and December 1992, 5106 patients with a definite myocardial infarction or who died from coronary causes were identified. Of these, 2166 were current smokers, 1477 were ex-smokers, and 1088 were non-smokers; information on smoking was missing for 375 patients, 231 of whom died before admission to hospital. Smokers were younger, more likely to be men, and fewer of them had a history of coronary heart disease when compared with non-smokers (1). The ex-smokers were older, more likely to be men, and more of them had previously had a myocardial infarction when compared with non-smokers.

Demographic information, case fatality, crude odds ratio, and adjusted odds ratio by smoking status for acute cardiac events in patients aged 25–64 years, 1986-92

View this table:

Compared with non—smokers, smokers had a higher risk of dying before hospital admission but this was not significant. The risk of dying after hospital admission was significantly lower in smokers. Overall, there was no significant effect of smoking on total case fatality because smokers who die before admission have a bigger effect on total case fatality than smokers who survive to be admitted. Ex-smokers had lower risks of dying both before and after hospital admission, resulting in an overall reduction in case fatality when compared with non-smokers.


We found a lower case fatality within 28 days after an acute cardiac event for smokers who had been admitted to hospital when compared with non-smokers; there was a non-significant rise in case fatality before admission to hospital in smokers. The lower case fatality after hospital admission among smokers is balanced by an excess in the number of smokers who died before hospital admission. There was no overall effect of smoking on case fatality from an acute cardiac event.

Adjusting for confounding reduced the apparent beneficial effect of smoking shown in the crude analysis of deaths after admission and increased the magnitude of the detrimental effect of smoking in the analysis of deaths before admission. The apparent decrease in case fatality in smokers after an acute cardiac event is restricted to patients who have been admitted, and the smoker's paradox is largely explained by a greater case fatality before admission to hospital in smokers.


Funding: Health Research Council of New Zealand and the National Heart Foundation of New Zealand.

Conflict of interest: None.


  1. 1.
  2. 2.
  3. 3.
  4. 4.