Mortality from cardiovascular diseases has been declining in most developed countries during the past 30–40 years.1^–4 In the 1970s and 1980s in Finland, most of the decrease in mortality from ischaemic heart disease5 and stroke2 was explained by decreases in the levels of traditional risk factors (smoking, hypertension, total cholesterol). Despite the positive trend in cardiovascular mortality, the socioeconomic differences have persisted and even grown wider in Western countries.6^–13 During the period 1972–87 the changes in risk factors among different socioeconomic groups did not explain the increase in socioeconomic differences in ischaemic heart disease mortality in Finland.14 Risk factors accounted for 47% of the observed decline in mortality from ischaemic heart disease among male white-collar workers but 75% among blue-collar workers.14 So, even though there was a favourable development in the absolute level of risk factors in all socioeconomic groups, the decrease in mortality from ischaemic heart disease in the higher economic groups was clearly steeper than predicted on the basis of the changes in classic risk factors.

This study aims to update the information on the relation between the trends in traditional risk factors and socioeconomic differences in cardiovascular mortality. After the start of the North Karelia Project in the early 1970s there was a continuous improvement in all main risk factors in Finland from the early 1970s to the late 1990s. Thereafter, however, smoking has actually increased in women and stayed at the same level in men and cholesterol levels have stopped decreasing.15 16 These developments make it interesting to analyse the possibly changing role of the traditional risk factors in explaining the changes in socioeconomic differences in cardiovascular mortality.

METHODS

FINRISK cohorts

The study population comprised people aged 35–64 years at baseline who were selected from the population-based FINRISK surveys in 1987, 1992, 1997 or 2002 in three areas of Finland (North Karelia, Kuopio and Turku/Loimaa). FINRISK surveys are independent random samples drawn from the population register and stratified according to gender, area and 10-year age group. The total sample size for all surveys in these areas was 19 452 out of whom 15 235 participated (78%). Participants who had missing information on risk factors (n = 582) or socioeconomic status (n = 11) were excluded from the study. Thus, 6964 male and 7678 female participants aged 35–64 years at baseline were included in the analyses.

The surveys were approved by the ethics committee, conducted according to the ethical rules of the National Public Health Institute and carried out in accordance with the Declaration of Helsinki.

Risk factor measurements

The study protocol for the FINRISK surveys included a questionnaire on health behaviour (smoking, use of alcohol, eating habits and physical exercise) and a personal health examination including measurements of blood pressure, height, weight and total serum cholesterol. Recipients were asked to fill in the questionnaire and bring it to the survey site where the health examination was performed. If a person did not appear at the survey site, he/she was contacted by phone to provide a new survey date.

Methods for measuring data on serum cholesterol and blood pressure in FINRISK surveys have been explained elsewhere in detail.2 16 Participants were defined as hypertensive if their systolic blood pressure exceeded or equalled 140 mm Hg or diastolic blood pressure exceeded or equalled 90 mm Hg or they had used antihypertensive drugs during the past week. Those subjects who were currently smoking (cigarettes, cigars or pipes) or had quit less than six months ago were classified as smokers. The rest were classified as non-smokers.

Socioeconomic indicators

Occupational class and household income were used as indicators of a person’s position in society. Occupational class was divided into six groups: upper non-manual workers, lower non-manual workers, manual workers, farmers, other employers and others. The group “others” was highly heterogeneous and included students, people with a long unemployment history or whose occupational class was unknown. Pensioners were classified according to their past occupational group. Women’s occupational class was primarily determined by their own present or past status. Housewives and others without present or past occupational class were classified according to the occupation of the spouse.

Income was defined as total household income per year adjusted for family size using the OECD (Organisation for Economic Co-operation and Development) equivalence scale, where the first adult in the household was weighted as 1.0, other adults as 0.7 and children under 18 years old as 0.5.17

Statistical analysis

Data on mortality were taken from the National Causes of Death Register and cardiovascular deaths (the sum of ischaemic heart disease and stroke deaths) included the following International Classification of Diseases (ICD) codes as the underlying causes of death: ICD-8: 410-414, 430-434, 795, ICD-9: 410-414, 430, 431, 433, 434, 436, 437, 798, ICD-10: I21-I25, I46, I60, I61, I63, I64, R96, R98, R99. With the help of this country-wide register, the follow-up was 100% complete in practice. These data were further linked to the individual level records of the censuses carried out in Finland in 1985, 1990, 1995 and 2000 in order to obtain data on occupational class and household income. The linkage was performed by Statistics Finland using personal identification numbers unique to every resident of Finland.

Two FINRISK cohorts (1982 and 1987, pooled together and followed until the end of 2001) were used to determine a logistic regression model for the probability of cardiovascular death based on risk factor values at the baseline. Age and baseline value of total cholesterol were used as continuous variables and hypertension and smoking as dichotomised variables. Interactions between variables were not statistically significant and were not included in the analyses. Based on these models the probability of cardiovascular death for men was: 1/(1 + exp (9.08 − 0.09 × age − 0.69 × smoking − 0.56 × hypertension − 0.23 × cholesterol)). The probability of cardiovascular death for women was: 1/(1 + exp (10.66 − 0.10 × age − 0.88 × smoking − 0.89 × hypertension − 0.14 × cholesterol)). All terms were statistically significant at a 0.0001 risk level except cholesterol for women (p = 0.02). Average age for men and women, instead of actual age, was used in calculating the risk of cardiovascular death in order to avoid the influence of age difference in different study years between socioeconomic groups.

Using these formulas the predicted probability of cardiovascular death was calculated for each study year by entering risk factor values into the formulas and calculating the mean of these individual risk scores. The outcome was the predicted changes in cardiovascular mortality rates since 1987 based on the changes in risk factor levels in subsequent FINRISK surveys. These predicted rates were then compared to observed mortality rates which were calculated using actual cardiovascular deaths occurred in the study population. Owing to the relatively large annual variation in observed cardiovascular deaths we used the mean rates of five-year intervals (1985–9, 1990–4, 1995–9, 2000–4) in order to obtain sufficient stability in the analyses.

RESULTS

Table 1 presents the participation rates for the FINRISK surveys in 1987–2002 in the three different areas (North Karelia, Kuopio and Turku/Loimaa) for 35–64-year-old men and women. The participation rates varied among men from 83% to 68% and among women from 88% to 74%, depending on the study year and area. In general, there was a declining trend over time. The average total participation rate for all the surveys was 75% for men and 82% for women.

Age standardised risk factor levels (using European standard population) have improved over the FINRISK survey years 1987–2002. The improvement has mainly occurred during the period 1987–97 while little change is seen from 1997 to 2002. Among men, mean total cholesterol has decreased from 6.27 mmol/l in 1987 to 5.76 mmol/l in 2002. The proportion of hypertensive participants has diminished during the same period from 65% to 52% and the proportion of smokers from 38% to 34%. Among women there is a similar decreasing trend except in smoking, where the proportion of smokers has actually increased from 16% to 21%.

Table 2 shows the corresponding changes in risk factor levels by occupational class and income. A positive trend is clear in all occupational classes in 1987–97 among men. Cholesterol levels declined in all groups, and the differences between occupational classes remained similar: upper non-manual workers had the lowest cholesterol levels in each study year. The prevalence of hypertension and smoking also decreased in all groups (except in “others” group in hypertension). The latest period 1997–2002 shows, however, a halt in the positive trend in risk factor levels. During this period the mean cholesterol level increased in all occupational classes (except in “other employers”) and the prevalence of smoking grew among upper non-manual workers, workers and “other employers.”

Among women, the decreasing trend in risk factor levels was sustained throughout the whole period of 1987–2002, except in smoking where an increasing trend is seen in all occupational classes (except in upper non-manual workers). Upper non-manual workers had the lowest risk factor levels during the whole period (except in smoking where farmers were the best-off group).

When using household income as the indicator of socioeconomic position, the picture is similar to that observed in the analyses by occupational class. In the lowest income tertile, time trends in the risk factors follow largely similar patterns as were observed for manual workers and, correspondingly, the time trends in the highest income tertile resemble those among upper non-manual employees. Cholesterol levels decreased during 1987–97 among men, and the prevalence of hypertension decreased for the whole period but the prevalence of smoking increased in the two lowest income tertiles during 1997–2002. Among women, mean cholesterol and the prevalence of hypertension decreased during 1987–97 whereas the proportion of smokers increased in all income tertiles during 1997–2002 (table 2).

Age standardised mortality (using European standard population) from cardiovascular diseases decreased substantially during 1987–2002 (table 3). For example, male upper non-manual workers had 186 cardiovascular deaths per 100 000 in 1987 but only 52 in 2002. This development occurred in all socioeconomic groups and among both sexes, but the decrease was fastest in the highest groups. Among women, the death rates are quite low, which makes it difficult to estimate the true nature of the trend, although it looks similar to men.

Table 4 shows the trend in observed cardiovascular mortality rates in 1987–2002 compared to the predicted change that was based on changes in risk factors during the same period. In 2002 among men cardiovascular mortality had decreased 57–72% compared to 1987, depending on the occupational class. The decrease was greatest among upper non-manual workers and smallest among manual workers. The decline in mortality rates was steady during the whole period 1987–2002. The predicted mortality decline was much smaller than the observed decline. For example, in male upper non-manual workers, only 29% of the decline in mortality rates from 1987 to 1997 was explained by improving levels of risk factors. In the other occupational classes the corresponding contribution of risk factors was higher, ranging from 34% among lower non-manual workers to 44% among farmers. During the last period, 1997–2002, observed mortality continued to decline whereas predicted mortality increased. As a result, during the whole period 1987–2002 only 16% of the mortality decline among male upper non-manual workers and 28% to 34% in the other occupational classes (if the heterogeneous “others” class is excluded) was attributable to changes in risk factor levels.

The corresponding results concerning income groups are similar. In men the changes in risk factor levels explained 34–41% of the decrease in mortality rates from 1987 to 1997 but during the whole study period 1987–2002 the contribution of the risk factor trends was markedly smaller, only 22–34%. The risk factors explained a larger proportion of the mortality decline in the lowest income tertile than in the upper two thirds. Among women, the contribution of risk factors to the mortality decline appears to have been smaller than among men, at least in the higher groups. In the lowest income tertile improving risk factor levels contributed 45% to the decline in mortality during 1987–97, whereas in the two highest tertiles the contribution was only 14–19%. When analysing the whole period 1987–2002, the contribution of risk factor trends to the mortality decline varied from 15% to 25%, the lowest income tertile having the highest proportion.

DISCUSSION

Mortality rates from cardiovascular diseases have declined considerably in Finland since the early 1970s. This decline has taken place in all socioeconomic groups for both men and women.2 18 In the 1970s this development was mainly caused by the improvement in traditional risk factor levels (smoking, total serum cholesterol, hypertension), but since the 1980s onwards the role of traditional risk factors in explaining cardiovascular mortality has diminished.5 18 Despite the positive trend in mortality the differences between socioeconomic groups have remained large and even grown wider.

This study aimed at investigating to what extent the changes in traditional risk factors explain the changes in socioeconomic differences in cardiovascular mortality in Finland during the past two decades. Our study showed that cardiovascular mortality has decreased remarkably in all occupational classes and income tertiles in both genders. However, mortality rates have decreased more in higher socioeconomic groups than in lower groups. For example, in male upper non-manual workers the mortality decline was 72% during 1987–2002 and more than 65% among the two highest income tertiles, whereas in male manual workers the decline was 57% and in the lowest income tertile 62%.

Comparison of the observed mortality decline with the predicted mortality trends that were calculated using the coefficients obtained from the two FINRISK cohorts (1982 and 1987) provided valuable results. Based on improving risk factor levels during 1987–97 the model predicted the observed mortality decline rather well: 29–44% of cardiovascular mortality decline was explained by the improvement in risk factor levels depending on the socioeconomic group in men. There was a clear socioeconomic gradient, though, because in male upper non-manual workers the risk factors contributed only 29% to mortality decline whereas in manual workers it was 41%.

During the latest period, 1997–2002, the positive effect of the risk factors came to a standstill or even reversed. This can be seen very clearly in predicted mortality rates which tend to increase during this latest period. The observed mortality rates, however, continued to decline and did not correspond to the predicted values.

According to a previous study by Vartiainen et al, time trends in traditional risk factors accounted for more than half of the observed decline in mortality from ischaemic heart disease during 1972–87.14 This finding and our results concerning mortality from all cardiovascular diseases suggest that the contribution of traditional risk factors to the decline in cardiovascular mortality has gradually diminished around the turn of the century. The declining role of traditional risk factors in determining time trends in cardiovascular mortality has emerged first in the upper socioeconomic groups.

A possible explanation for this diminishing power of traditional risk factors to account for socioeconomic differences in cardiovascular mortality is the introduction of invasive cardiac procedures from the 1980s onwards, other powerful treatments and improving counselling on risk factors.19^–24 It has been suggested that patients belonging to the higher socioeconomic groups have had better access to these new services than patients in lower socioeconomic groups.19^–24 Laatikainen et al estimate that 23% of the reduction in mortality from coronary heart disease during 1982–97 was explained by improved treatments and the contribution of risk factors was 53–72%.23 It is also possible that socioeconomic differences in other factors not measured in our study such as fibrinogen concentration and other markers of inflammation or other modifiable risk factors (for example, diabetes or body mass index) could explain part of the mortality differences that our model could not account for.25 From the point of view of health policy, it has been suggested that preventive treatment should be directed more towards the socially deprived.26 27 Our study confirmed these findings.

The main strengths of this study are its population-based design, the large number of cardiovascular deaths and the long and complete follow-up. The possibility of linking these data at an individual level with socioeconomic indicators provided by Statistics Finland made it possible to receive uniquely accurate information on cardiovascular mortality in different socioeconomic groups. The cardiovascular diagnoses in the National Causes-of-Deaths Register have recently been validated.28 The main limitation of the study was the declining response rate in the FINRISK studies over the years. It has been established that mortality and health behaviour differ substantially between non-participants and participants.29 For example, analysis of the FINRISK samples from the years 1972, 1977, 1982, 1987 and 1992 showed that non-participants had twice as high overall mortality as participants. We have recently shown that lower socioeconomic groups were over-represented among non-participants in FINRISK surveys.30 This resulted in underestimation of the absolute risk of mortality because of the negative socioeconomic mortality gradient. However, non-participation did not essentially distort the relative socioeconomic differences in mortality.30

Another limitation in the study was the small number of deaths among women which is the result of a remarkable improvement in women’s cardiovascular mortality in Finland. Mortality rates among women are quite low at present, which makes it difficult to obtain sufficient statistical power in the analyses. A further limitation is related to the latency period of the risk factor effects on mortality. The period of declining risk factor levels is fairly long and the improved risk factor situation has clearly manifested itself as lower cardiovascular disease mortality. However, the more recent levelling off and partial worsening of the risk factor situation may not yet have fully manifested its effect on mortality owing to the relatively short time period.

Our study shows that socioeconomic differences in cardiovascular diseases are a more complex phenomenon than just differences in traditional risk factor levels. Even though classic risk factors still account for 25–50% for the excess cardiovascular mortality among lower socioeconomic groups compared to higher socioeconomic groups, there are other explanatory factors such as improved treatments that are becoming more important in developed countries. Both primary prevention and equal use of healthcare resources are needed for narrowing the socioeconomic disparities in cardiovascular mortality.