Of the 6676 patients enrolled in the TRACE registry, 5983 were discharged alive from hospital and evaluated in this study. During follow up until June 1994, 1659 patients died. Among the deaths, 1261 were classified to be due to cardiovascular causes, 256 were classified to be due to non-cardiovascular causes, and 142 were unclassifiable because of inadequate information. Of the cardiovascular deaths 536 were classified as SCD and 725 as non-SCD.
The baseline characteristics differed between the four age groups (table 1). Table 2 shows the number of different modes of death in each age group.
Baseline characteristics of 5983 myocardial infarction survivors according to age group
Frequency of different modes of death according to age group
The effect of increasing age on all modes of mortality was demonstrated to be log linear. There was no interaction between age and sex. Hazards were proportional for both age and sex.
The cumulative incidence of SCD in the entire population was 5.3%, 7.4%, and 9.4% at one, two, and three years, respectively. The risk of both SCD and non-SCD increased significantly with increasing age (fig 1). The risk ratio associated with a 10 years increase in age (independently of sex) was 1.56 and 2.13 for SCD and non-SCD, respectively (p < 0.0001) (table 3). The increase in non-SCD associated with increasing age was 40% higher than the increase in SCD risk (p < 0.0001).
Mortality probability curves for sudden cardiovascular death and non-sudden cardiovascular death, according to age group.
Multivariate Cox analysis of age, sex, and mode of death
Table 4 gives the three year mortality from SCD and non-SCD in each age group. The ratio between risk of SCD and risk of non-SCD decreased significantly with increasing age (table 4), from 1.44 in the youngest to 0.55 in the oldest (p < 0.001).
SCD and non-SCD three year cumulative mortality probability
In a multivariate analysis, male sex was associated with an increased risk of SCD and all cause mortality (table 3), but male sex had no predictive value concerning non-SCD independently of age (table 3). The increase in SCD risk associated with male sex was 28% higher than the increase in non-SCD risk (p = 0.05).
Stratification of the analyses by sex showed that the age effect was similar in men and women, since no interaction was present. Figure 2 illustrates this, which shows the cause specific cumulative three year mortalities. Furthermore, fig 2 shows that the SCD risk of women lagged less than 10 years behind that of men.
Cause specific three year mortality probability per 1000 patients for sudden cardiovascular death (SCD) and non-SCD according to age and sex.
Since the registry included patients both with and without randomly assigned treatment, the analyses were repeated with inclusion of information on randomisation and assigned treatment (trandolapril or placebo). Adjustment for these covariates did not affect the results regarding the impact of age and sex. The analysis of trandolapril versus placebo showed a similar effect of trandolapril treatment on SCD and non-SCD. Information about other medical treatment issued at discharge was not recorded and could not be analysed.
Adjustment for other risk factors likely to be associated with SCD or non-SCD risk and their relation did not change the results. Additional risk factors included in the model were wall motion index, congestive heart failure, New York Heart Association functional class, ventricular fibrillation, atrial fibrillation, smoking, body mass index, history of MI, hypertension, angina, or diabetes, and thrombolysis. The independent risk ratio associated with male sex was 1.25 (p < 0.05) for SCD and 1.12 (p = 0.2) for non-SCD. For a 10 year increase in age the risk ratios were 1.25 and 1.73, respectively (p < 0.0001 for both SCD and non-SCD). Furthermore, decreased wall motion index, congestive heart failure, New York Heart Association functional class, atrial fibrillation, and history of hypertension, angina, or diabetes were all independently associated with increased risk of both SCD and non-SCD.