Introduction

Cardiovascular (CVD) disease is the leading cause of death and a major cause of morbidity in most parts of the world. However, large differences in mortality and morbidity rates and their trends have been described among different countries and populations [13]. Also, the clinical manifestations of atheromatous disease differ among these populations: the ratio of myocardial infarctions/cerebrovascular accidents in a number of countries was found to range between 4.63 and 0.08 during the years 1985–1989 [4]. Greece, in the landmark Seven Countries Study [5] in the middle 1960s, was found to be a low-risk country for CVD. This is especially true for coronary heart disease (CHD) [6], but not to the same degree for stroke [7], for which Greece is considered to be an intermediate risk country. Although the Seven Countries Study included only two rural populations from Corfu and Crete, it probably represented the whole country, since Greece was mostly rural at the time. On the other hand, during the period from 1970 to 1990 Greece was the only European Union country to show an increase in CHD mortality [2], although the all-cause and cerebrovascular disease mortality was markedly reduced and life expectancy was increased [8]. Since the early 1990s a decline in CHD mortality was evident in Greece as well [9, 10]. Therefore, it is worthwhile to study the epidemiology of the Greek population during the last 50 years and investigate the possible reasons for the disparities in the time course of cause-specific mortality and morbidity of the various clinical manifestations of CVD.

The scope of this paper is to analyze all available data on CVD mortality and morbidity in Greece, spanning five decades from 1956 to 2007, highlighting differences in the trends of various clinical manifestations of CVD, as well as gender differences in mortality trends.

Methods

The number of deaths, the cause of death and the estimated mid-year Greek population were derived from data published by the National Statistical Service of Greece for the years 1956 through 2007 (the last year for which there was available data at the time of manuscript preparation). Data from the Health and Social Welfare Statistical Yearbook were also used. Hospitalization rates for acute myocardial infarction (AMI) and stroke were calculated for the period 1979–2003 based on the number of patients discharged with the corresponding diagnoses.

During the period from 1956 to 2007 four different revisions of the International Classification of Diseases (ICD) have been used in Greece (ICD-6 until 1959, ICD-7 until 1967, ICD-8 until 1978 and ICD-9 since then). The most notable difference has been caused by the shift from ICD-7 to ICD-8, in which the term “Ischemic Heart Disease” was first used. Previously, these deaths were classified under the term “Arteriosclerotic heart disease, including coronary disease”. This necessitated the calculation of separate regression models for mortality from CHD for the periods before and after this time. The following ICD-9 codes were used: All causes ICD: 001-E999. CVD ICD codes: 390–459. CHD: ICD codes 410–414. AMI: ICD code 410, Cerebrovascular disease (stroke): 430–438.

The age range for all rates presented has been truncated to 45–74 years. Mortality rates were calculated per 100,000. Age-standardization was performed by the direct method, using the European standard population. Trends over time were determined using log-linear regression models. Specifically, the Joinpoint Regression Program, created by the US National Cancer Institute, was used to determine “joinpoints”, i.e., one or more points in time at which a significant change in the slope of the trend occurs. A Monte Carlo permutation method is used to determine the simplest model, with the smallest number of joinpoints, that best fits the data. The addition of one or more joinpoints does not offer a statistically significant improvement of the resulting model [11]. The minimum number of joinpoints was defined as zero (a straight line) and the maximum as three (to limit the complexity of the model). This resulted in the calculation of 1–4 trends, named accordingly. For each trend, the estimated annual percentage change (EAPC) was calculated. This is done by using the previously computed log-linear regression model y = bx + a, where y is the natural logarithm of the mortality rate and x is the calendar year. EAPC is equal to 100 × (eb − 1), where e is the base of the natural logarithm. Note that the EAPC is calculated based on the values of the regression line, and therefore only approximately predicts the actual values observed.

Results

Tables 1 and 2 present the results of the joinpoint analysis of CHD, stroke, CVD and all cause mortality trends over time in Greek men and women, respectively, including the estimated annual percentage change for each separate trend, its corresponding 95% confidence interval and the cumulative percentage difference for each cause of death from 1956 to 2007.

Table 1 Joinpoint analysis of the mortality rate for CHD, stroke, CVD and all causes in Greek men, 1956–2007
Table 2 Joinpoint analysis of the mortality rate for CHD, stroke, CVD and all causes in Greek women, 1956–2007

In 1956, age-adjusted CHD mortality rate per 100,000 was 142 in men and 80 in women aged 45–74 years old and remained practically stable until 1960. By 1968 it had increased by 57 and 42%, respectively. After 1969, it increased by 45% in men until 1976 and by 32% until 1972 in women. Later, mortality remained essentially stable until 1989 in men and until 1990 in women. Since then, it has started to decline by 1.2 and 2.6% per year in men and women, respectively. However, only in women has it returned below 1956 levels. Of note, mortality from CHD has been the one most affected by the adoption of newer revisions of ICD in 1960, 1968 and 1979. Otherwise, its initial increase would have been even more pronounced (Fig. 1).

Fig. 1
figure 1

Age-adjusted mortality from coronary heart disease

In 1979 the AMI morbidity rate (hospitalizations and out of hospital deaths per 100,000, all ages) was 308 in men and 105 in women. By 2003, it had increased by 39% in men, to 428, and by 70% in women, to 178.

Furthermore, in 1979 the age-adjusted hospitalization rate for AMI per 100,000 amongst the population aged 45–74 years old was 309 cases in men and 70 in women. This rate increased by 36% in men and 37% in women by 2003 (Fig. 2).

Fig. 2
figure 2

Age-adjusted hospitalization and mortality rate for acute myocardial infarction

In contrast, in-hospital case-fatality declined steadily during that same period, from 16.2% in men and 26.7% in women in 1979 to 9.2% and 13.4%, respectively, by 2003 (Fig. 3).

Fig. 3
figure 3

In-hospital case-fatalities from acute myocardial infarction

The total mortality rate from AMI, however, did not decrease accordingly. In this 25-year period it decreased only by 26% in both sexes.

Age-adjusted stroke mortality rate in 1956 exceeded that of CHD, accounting for 159 deaths per 100,000 population in men and 155 in women. In men, it increased initially by 0.4% per year until 1978, eight times less rapidly than mortality from CHD, whereas in women, it increased by 1.4% per year until 1965. This was followed by a gradually accelerating decline by 2–4.6% per year in men (52% cumulative decrease) and by 1.2–6.1% in women (74% cumulative decrease) (Fig. 4). As a result, the ratio of stroke mortality/CHD mortality, which was 1.1 in men and 1.9 in women in 1956, was reversed. By 2007, it had become 0.4 in men and 0.9 in women.

Fig. 4
figure 4

Age-adjusted mortality from stroke

On the other hand, the age-adjusted hospitalization rate for stroke per 100,000 amongst the population aged 45–74 years old increased significantly from 1979 to 2003, from 408 cases in men and 318 in women to 636 and 363, an increase of 56% and 13%, respectively (Fig. 5).

Fig. 5
figure 5

Age-adjusted hospitalization rate for stroke

The combined effect of the aforementioned trends on age-adjusted CVD mortality is shown in Fig. 6. After an initial increase in men, by a maximum of 28% until 1987, it has decreased by 2% per year. In women, CVD mortality was stable until 1963, declining slowly at first and by 3–5% from 1988 onwards.

Fig. 6
figure 6

Age-adjusted mortality from cardiovascular disease (CVD) and all cause mortality

Finally, the age-adjusted all-cause mortality rate per 100,000 is also presented in Fig. 6. After remaining almost stable for 8 years, it then started declining steadily in both sexes by 1–3% per year, reaching a 34% cumulative decrease in men and 56% decrease in women by 2007.

Sex ratios (men/women) in 1956 were 1.8 for CHD, 1.0 for stroke, 1.3 for CVD and 1.5 for all-cause mortality. Note the practically equal rate of stroke mortality in both sexes. Mortality in women decreased by 2007 to a significantly greater degree, resulting in sex ratios of 4.1 for CHD, 1.8 for stroke, 2.8 for CVD and 2.3 for all-cause mortality, an increase of 129, 75, 118 and 48%, respectively. Meanwhile, sex ratios for the AMI mortality and hospitalization rates remained essentially unchanged between 1979 and 2003, at 3.8 and 4.4, respectively. The sex ratio for hospitalization for stroke was 1.3 in 1979 and increased to 1.8 by 2003.

Discussion

In the late 1950s, all-cause mortality in the Greek population was one of the lowest in the world, resulting in a longer life expectancy than that of many other nations [2, 3, 12, 13]. This was mostly due to the low incidence of CHD [5]. Since then, major changes have occurred in Greece. The burden of infectious disease has greatly diminished. The gross national product and the per capita income have increased more than 30-fold. Internal immigration has taken place and a large proportion of the population has moved from the countryside to the large cities. At the same time the healthcare system has been constantly improving. All these changes have had a great impact on all-cause mortality, resulting in an early decline that is continuing up to this day.

As expected, overall mortality is higher in males, a finding that has been observed in all countries [2, 12]. The same is true for mortality from CVD, CHD and, more recently, stroke, with the gap between men and women steadily widening.

The improvement in mortality is not uniform across all the major causes of death. There has been a steady decline in CVD mortality (after an initial increase) over the last 3 decades, attributed mostly to a parallel decrease in mortality from stroke. On the contrary, mortality from CHD increased for several decades and has only recently started to decline. As a result, although CHD and stroke contributed equally to CVD mortality in men until 1960, by 1991 CHD mortality was twice as high as that of stroke. Overall, CVD is now responsible for more than 50% of all deaths each year.

In comparison to central and north European populations, the Greek population still has a lower CHD and higher cerebrovascular mortality [2, 10], despite the aforementioned changes and despite blood cholesterol levels being as high as in other populations [14]. A similar pattern (higher cerebrovascular and lower CHD mortality) is present in Portugal, southern Spain and southern Italy [2]. This similarity could be attributed to the local lifestyle (diet and smoking habits). The Mediterranean diet, thanks to the antioxidants and monounsaturated fatty acids it contains [15], may protect the population from CHD but not from stroke. The argument for a relative protection of the Mediterranean population from high cholesterol levels is supported by epidemiological data. The Seven Countries Study has shown that equal blood cholesterol levels do not always signify equal risk [16]. Also, in a study from Italy it is concluded that for the same level of risk factors the estimated total coronary risk is lower for the Italians than for the population of the Framingham Study [17].

The main risk factor for stroke is hypertension [18, 19]. The decline in cerebrovascular disease mortality is probably due to better control of arterial hypertension in the last 4 decades. The age-adjusted prevalence of hypertension in the Greek population was relatively low in 1997 in comparison to other developed countries and the awareness, treatment and control of hypertension in Greece, although far for being ideal, is better than in other countries [20]. This, in turn, is thought to be the result of the introduction of more effective anti-hypertensive drugs, an increased public awareness of the dangers of hypertension, as well as the abandonment of the use of salt as a food preservative. It has been estimated that a 5 mmHg reduction in systolic blood pressure in a population results in a 14% overall reduction in mortality due to stroke [21]. In Finland, reduction of salt intake to one-third during the past 30 years has been accompanied by a more than 10 mmHg fall in the population average blood pressure and a 75–80% decrease in stroke mortality [22]. Interestingly, the prevalence of hypertension is essentially equal in Greek men and women (45 and 44% respectively in one recent study) [23], in contrast with most other Western countries, where men show a significantly higher rate [24]. Possibly, this could, at least partially, account for the lower male/female sex ratio observed for stroke mortality, in comparison to other forms of CVD.

The increase in the incidence of CHD and AMI in the Greek population during the 1960s and 1970s, when a decline in CHD mortality was already observed in most industrialized countries [25], is probably due to changes in several aspects of the lifestyle of Greeks during the past decades, with a resulting increase in the prevalence of most CVD risk factors.

First, the Greek population is slowly drifting away from the traditional Mediterranean diet described in the Seven Countries Study [5], which is associated with a low cardiac mortality rate [26, 27]. Although olive oil continues to provide the major part of the fat consumed, total fat intake has increased during the past decades. Per capita fat consumption increased from 125 g in 1980 to 148 g in 2002 [28]. The saturated fatty acid content rose from the 5–8% values of the 1960s in Crete and Corfu [5] to 9.5 and 10.8% in 2000 [29]. Total calorie intake has increased by 24% and the consumption of sugar, milk and meat has increased 2-, 3- and 5-fold, respectively [27, 28]. Over the past 40 years, the consumption of fish, vegetables and cereals has been declining [28]. Nevertheless, in comparison to other countries, animal fat intake is still low in Greece; it occupied the fifth place from last in a multi-country study [26].

Obesity is also on the rise, beginning even in adolescence [30]. Recent studies have found the prevalence of obesity to be 20–26% in Greek men and 15–18% in women [31, 32], with a cumulative prevalence of overweight and obesity reaching 67–73% in men and 46–48% in women, one of the highest in Europe [33]. A previous regional study in the 1980s reported 24% of adult men and 23% of adult women as having a body mass index (ΒΜΙ) >27 kg/m2 [14].

The aforementioned changes have led to an increase in cholesterol blood levels [34]. In a study in Crete [35] in the early 1990s, the average total cholesterol was 6.2 mmol/1 (240 mg/dl). Twenty-five years earlier in the same island the total cholesterol was 5.5 mmol/1 (212 mg/dl) [36], whereas in 1960, during the Seven Countries Study, it was 4.7 mmol/1 (182 mg/dl) [5]. In recent years (2001–2002), the prevalence of hypercholesterolemia (total serum cholesterol >5.1 mmol/l) was found to be 46% in men and 40% in women (aged >18 years old) in a population sample from the Athens area [37]. This is in contrast to the declining serum total cholesterol levels in US adults [38].

Greeks have been and still are heavy smokers [5, 14, 34, 35]. The low CHD mortality during the 1960s, despite the high prevalence of smoking, may be due to low cholesterol blood levels. It is possible that in a low cholesterol environment, smoking is not as harmful to the coronaries as it is in hypercholesterolemic patients. The Japanese, also heavy smokers, still have low CHD mortality [16], probably thanks to their low cholesterol levels. Despite the increasing prevalence of smoking among women, men continue to be more avid smokers, by about 23%. [37].

The prevalence of type 2 diabetes mellitus has been increasing in the Greek population. In a sample of an urban Greek population, it increased from 2.4 to 3.1% (an increase of 29%) during the period from 1974 to 1990, with even more prominent changes in the 50–79 age group [39]. A recent study exploring the metabolic syndrome in Greece reported a prevalence of diabetes mellitus of 10.6% among the adult population [40].

The increased urbanization of the Greek population over the last 50 years may be an important factor in the overall increase of CHD mortality. Stressful city occupations have been associated with high overall and CHD mortality [41]. Α decrease in physical exercise, a characteristic of urban life, may also have played an important role [42]. The ATTICA study [43] reported more than half of all men and women in the greater Athens area as adhering to a sedentary lifestyle.

The recent decline in CHD mortality is probably due to newer medical and invasive interventions that prevent death and not to a reduction in the incidence of the disease, since the CVD risk factors have not been adequately reduced in the population [37, 38, 44]. The frequency of smoking in the Greek cohort of EUROASPIRE II [45] was found to be 25.1%, of diabetes 30%, of hypercholesterolemia 64.5%, of obesity 28% and of overweight 79.2%. This is also supported by the observation that although the in-hospital fatality of AMI has decreased by half between 1979 and 2003, deaths from AMI have decreased only slightly, as hospitalization and morbidity rates have increased greatly during the same period. A similar pattern was seen for stroke, although the reduction in mortality rates in this case was substantial.

The accuracy of mortality rates derived from vital statistics depends on each deceased’s physician correctly diagnosing the cause of death. This results in some error, which however has been shown by data from the Framingham registry and the Monica Project to be rather small, below 5–8% [46, 47]. The trends observed in hospitalization rates may have also been influenced by gradual changes in hospitalization policies. However, Greece has had a national health system with universal health coverage for over two decades now, so any such changes should be minor.

Conclusions

Although the various types of CVD share common risk factors, the trends of their respective mortality rates over the past 5 decades have differed significantly in the Greek population. This could partly be explained by the fact that not all risk factors equally contribute to CHD and stroke and they have not all been equally well controlled.