AMI data for the Basque Country are part of the IBERICA (Investigación, Búsqueda Específica y Registro de Isquemia Coronaria Aguda) population-based register, a project started in 1997, in which eight Spanish regions participate. The identification of hospitalized AMI patients and description of variables and quality controls have previously been reported in detail.20

The subjects of our study consisted of all the patients resident in the Basque Country who had a first AMI between 1999 and 2000 and were registered in the IBERICA database. This population register included individuals aged 25 years or more and covered around 80% of the population in the Basque Country, corresponding to five of the seven Basque health areas. The estimated mean population resident in the Basque Country covered by IBERICA for 2001 (aged ≥ 25 years) was 636,920 men and 679,509 women.

All first definite “fatal and non-fatal cases”, “fatal possible cases” and “fatal insufficient data events” classified according to the MONICA study criteria were included.21 Non-fatal possible cases (n=62) (most of which clinically corresponded to angina) and cases without socioeconomic information were excluded from the incidence analysis (n=355), while all patients who died before being admitted to any private or public hospital in the area of study were excluded from the survival analysis (n=1,616).

Monitoring of cases from symptom onset to day 28 was undertaken via an active search of hospital clinical records. Survival at 5 years was cross-checked using the Spanish National Death Index (including manual checking), individual health records and hospital clinical records.

The following variables were taken into account: sex, age at disease onset, the presence of classical coronary risk factors (diabetes, hypercholesterolemia, hypertension and smoking), history of prior angina, AMI characteristics [typical symptoms, location, Killip class on admission, left ventricular ejection fraction (LVEF) and exercise test results], time from symptom onset to first monitoring and time from symptom onset to thrombolytic treatment, hospital complications during the acute phase (severe arrhythmia, postinfarction angina, reinfarction, and maximum Killip class during hospitalization), the therapeutic approach adopted during hospitalization and at discharge (antiplatelet therapy, beta blockers, angiotensin converting enzyme inhibitors [ACEI], early revascularization [thrombolysis within 6h of symptom onset and primary angioplasty on the day of infarction] and cardiac surgery), and place of treatment (admission or not to an intensive care unit [ICU]).

SES was studied using a deprivation index ecologically assigned to each patient according to the census tract (average population of 1,226) of the residence at diagnosis of AMI. The deprivation index for each census tract was calculated using four socioeconomic indicators derived from the 2001 census:22 proportion of active people unemployed, proportion of manual workers with no qualification, proportion of people with education only to the primary level or below and proportion of houses with a low level of comfort.23 A principal-components analysis was used to obtain a summary score, in such a way as to explain as much of the total variation in the data as possible. Despite its ecological nature, this index has been proven to perform well as a measure socioeconomic differences.24 Patients were categorised into quintiles (Q1 being the least deprived quintile and Q5 the most deprived) according to the deprivation index of their census tract of residence. For clarity, individuals were classified into three groups as a function of the deprivation index: (Q1 [highest], Q2-Q3-Q4 [medium] and Q5 [lowest]) for the figures presented in this study.

Continuous variables were described as means (with standard deviations), and discrete variables as absolute and relative frequencies. The Chi-squared test was used to check the equality of distributions by socioeconomic status, and an ANOVA was employed to test the equality of means. The Levene and Kolmogorov-Smirnov tests were used to check for homocedasticity and the normality of continuous variables.

The reference population of each socioeconomic level stratum consisted of all persons aged 25 years or more living in the census tracts to which that level was assigned. The populations were categorised by sex and 5-year age group (from 25-29 years to 85 or more years). Age-standardized incidence rates were calculated by direct methods using the Basque Country population in 2001 as the standard population. Relative risks were calculated for every quintile of the deprivation index with respect to the highest socioeconomic group, using Poisson models. Ninety-five percent confidence intervals were estimated for these incidence rates and relative risks.

Observed survival was estimated for the acute phase (first 28 days) and relative survival for the late phase (5 years). Deaths occurring in the acute phase were assumed to be directly related to AMI, with the expected mortality in the general population virtually null (thus observed survival and relative survival are very similar).

The Kaplan-Meier estimator was used to calculate observed survival and the log-rank test to assess univariate intra-group differences. Relative survival (with 95% confidence intervals) was calculated using the Hakulinen method,25 employing age-, sex- and deprivation index (quintile)-specific life tables for the general population in the area. Specifically, relative survival was the ratio of the observed survival to the expected survival of the corresponding general population and expresses the probability of AMI survival after adjustment for competing causes of death, a figure which is known to have a good performance for long follow-up periods.14

Mortality and population tables for the two periods 1999-2001 and 2002-2005 for the Basque Country, stratified by age (single year), sex and socioeconomic group were obtained from the Basque Statistics Institute.22

Data analysis showed a very different pattern with respect to the cardiovascular event in the first 28 days compared with the remainder of the follow-up period (from 29 days to 5 years). Since the usual proportional hazard assumption was violated for the whole follow-up, we decided to study the two periods separately. All variables were considered for multivariate analysis in long-term survival whereas treatments prescribed at hospital discharge were excluded from analysis of the first 28 days.

For the acute phase, a Cox proportional hazards model was used to study the effects on risk of the different covariates considered and the Akaike information criterion was used as a covariate exclusion criterion. The models satisfied (p>0.150) the proportional hazards hypothesis; this assumption was verified using the test based on Schoenfeld residuals.26

For the period from 29 days to 5 years, a multivariate model was constructed from the generalization for relative survival of the proportional hazards model proposed by Andersen.27 The test proposed by Stare was used to verify the hazards proportionality hypothesis28 (p>0.250) and again the Akaike information criterion was used as the covariate exclusion criterion.

All the analysis was performed using STATA 10, R 2.6.2 (relsurv package29) and SAS 9.2 software. Ninety-five per cent confidence intervals were also calculated for hazard ratios and the level of significance was set at p<0.05.

A total of 3,974 incident AMI events were included from the period 1999 to 2000. Socioeconomic data were missing for 355 (8.9%) of these events. Comparison of patients with socioeconomic information with those without this information revealed some differences. The patients with no socioeconomic data were younger (mean age 67.8 vs 71.7 years, p<0.001) and showed lower mortality (17.5% vs 55.1%, p<0.001) and there was a smaller proportion of AMI cases with insufficient data (7.3% vs 31.3%, p<0.001); furthermore, most of these patients were from the Gipuzkoa health area (92.3% vs 40.2%, p<0.001).

This study included 3,619 patients: 62.7% men and 37.3% women. Age-standardized incidence rates by socioeconomic group and sex are listed in table 1. These rates were higher in men than in women across the five socioeconomic quintiles considered, increased with increasing deprivation in men and showed a non-clear trend in women; the highest rates were found in the most privileged group compared with the remaining groups. In men, the risk of AMI was higher in the lowest socioeconomic group (RR=1.17; 95%CI: 1.02-1.34) than in the most privileged group, but in women no significant differences in any of the quintiles compared with the reference group (Q1) were observed (data not shown).

The data from 2,003 patients was used for survival analysis, after excluding 1,616 out of hospital deaths. The data for this analysis included 1,391 (69.4%) men and 612 (30.6%) women. The mean age at disease onset was higher in women (75.0 [12.5] years) than in men (63.8 [13.4] years). Excluded patients had a higher mean age (77.9 [12.3] years), 54.0% were men, and distribution differed by socioeconomic status in both men and women, with a gradient from the highest socioeconomic group (Q1: 25.1% in men and 29.1% in women) to the lowest (Q5: 13.5% and 11.2%, respectively).

Tables 2 and 3 show the distribution of baseline characteristics by socioeconomic group in men and women, respectively, after their first AMI. In men, the proportion who never underwent exercise tests or left ventricular ejection fraction tests significantly increased from the most privileged socioeconomic groups to the most deprived groups, while among women, diabetes showed a socioeconomic gradient with a significantly higher proportion of women with diabetes in the most deprived groups (42.9% Q5 vs 23.6% Q1).

Of the 2,003 patients with a first AMI included in this study, 384 patients (200 men and 184 women) died within the first 28 days following their AMI, and a further 667 patients (381 men and 286 women) died within 5 years.

Figure 1 shows the observed survival curves in the first 28 days following a first AMI (left) and the relative survival curves from day 29 to 5 years (right), by socioeconomic level in men and women. During the acute phase, men in the most privileged quintile and women in the most deprived quintile had better rates of survival than persons in the remaining quintiles (although these differences were not significant). In the late phase, relative survival was higher for quintile 1 in men (non-significant) and for quintiles 2, 3 and 4 in women (p<0.001) compared with the remaining socioeconomic groups.

Figure 1.

Observed survival curves over the first 28 days (left) and relative survival curves from day 29 to five years (right) after a first myocardial infarction by socioeconomic levela and sex, IBERICA-Basque Country 1999-2000.

Q: quintile

aSocioeconomic level assigned at census tract of residence.

(0,25MB).

Tables 4 and 5 show the results of the hazard ratio for short- and long-term survival, respectively. In men, the age-adjusted model in the first 28 days following a first AMI identified a significantly higher relative risk of death in quintiles 3 (HR=1.60; 95%CI: 1.02-2.51) and 5 (HR=1.65; 95%CI: 1.02-2.69) compared with the highest socioeconomic group. After adjustment for age and for the most statistically relevant variables plus health area, this effect was reduced (the differences became non-significant). In women, there were no significant differences or clear gradients in either of the models. In the late phase, no significant differences were observed in either men or women.