OUP user menu

★ editor's choice ★

Economic burden of cardiovascular diseases in the enlarged European Union

José Leal, Ramón Luengo-Fernández, Alastair Gray, Sophie Petersen, Mike Rayner
DOI: http://dx.doi.org/10.1093/eurheartj/ehi733 1610-1619 First published online: 22 February 2006


Aims Cardiovascular disease (CVD), together with its main components, coronary heart disease (CHD), and cerebrovascular diseases, is the main source of morbidity and mortality in the European Union (EU), but to date, there has not been any systematic cost-of-illness study to assess the economic impact of CVD in the EU.

Methods and results CVD-related expenditure was estimated using aggregate data on morbidity, mortality, and healthcare resource use. Healthcare costs were estimated from expenditure on primary, outpatient, emergency, and inpatient care, as well as medications. Costs of unpaid care and lost earnings due to morbidity and premature death were included in the study. CVD was estimated to cost the EU €169 billion annually, with healthcare accounting for 62% of costs. Productivity losses and informal care represented 21% and 17% of costs, respectively. CHD represented 27% and cerebrovascular diseases 20% of overall CVD costs.

Conclusion CVD is a leading public health problem. Our study is the first to assess the economic burden of CVD across the EU, and our results should help policy makers evaluate policy impact and prioritize research expenditures. However, because of data unavailability, our study has important limitations, which highlight the need for more accurate and comparable CVD-specific information.

  • Cardiovascular disease
  • Coronary heart disease
  • Cerebrovascular diseases
  • Cost-of-illness study
  • Europe
See page 1521 for the editorial comment on this article (doi:10.1093/eurheartj/ehl075)


The Council of Ministers of the European Union (EU) has emphasized that cardiovascular disease (CVD) is the largest cause of sickness and morbidity and a major cause of death and premature death and of reduced quality of life for the citizens of the EU.1 CVD [defined as International Classification of Diseases (ICD)-10 category I00–I99] causes over 1.5 million deaths in the EU and is the main cause of years of life lost from early death.2 The most common CVDs are coronary heart disease (CHD; ICD-10 category I20–I25) and cerebrovascular disease (ICD-10 category I60–I69), accounting for ∼40 and 25% of CVD deaths, respectively.2

However, despite CVD being the main cause of morbidity and mortality in the EU, there has been no systematic cost-of-illness study to assess its economic impact. Cost-of-illness studies not only estimate the resources consumed in disease prevention, detection, and treatment but also estimate the opportunity costs of relatives providing care for patients and the foregone economic productivity associated with inability to work due to disability or premature death.

The objectives of this study were to provide an estimate of the economic costs of CVD for the EU, including healthcare costs, informal care costs, and productivity loss, and to estimate the proportion of total CVD cost attributable to CHD and cerebrovascular diseases.


Methodological background

Cost-of-illness analyses involve the identification, measurement, and valuing of resources related to an illness. A societal perspective was adopted with all costs being considered, not only healthcare costs but also those falling outside the healthcare sector, including opportunity costs associated with unpaid care and productivity losses associated with premature death or morbidity.

An annual time frame was adopted for our analysis, whereby all costs within the most recent year for which data were available were measured. Costs were converted to 2003 prices using the health component of the consumer price index for each country,3 with earnings being adjusted using wage inflation indices.47 National currencies were converted to Euros (€) using 2003 exchange rates. However, as comparisons using currency exchange rates do not necessarily reflect real price differences between countries, we also employed the purchasing power parity (PPP) method.9 The PPP method measures the price of the same bundle of goods in different countries using Euros as a common currency, thus allowing the comparison of costs, adjusted for cost of living, between countries.3

International sources,811 national ministries, and statistical institutes were consulted for epidemiological and healthcare utilization data. When data were not obtained from these sources, the published literature was consulted. If no data were found, extrapolations were performed from similar countries. A country was judged to be similar if it shared comparable gross national income, healthcare expenditure per capita, physician density, life expectancy, and location.

We employed a ‘top-down’ approach to calculate total expenditure using aggregate data on morbidity, mortality, hospital admissions, and other related indicators. A summary of the methods used is given here; for a more detailed overview of methods used, see Supplementary material.

Healthcare expenditure

The following categories of CVD healthcare service were included: primary care, accident and emergency (A&E) care, hospital inpatient care, outpatient care, and medications. Other types of activities relating to the prevention of CVD such as health education in community-based settings were not included because of the difficulties in identifying activity levels.

For Germany, France, and the Netherlands, CVD-related healthcare expenditure was derived from cost-of-illness studies.1214 In the Netherlands, pharmaceutical expenditure was updated using the latest data.15

To account for private spending on healthcare, in countries reporting public resource use only, estimates were inflated using the proportion of private spending on healthcare.9

Health service utilization

The methods used to estimate CVD-related healthcare utilization are described subsequently; these methods were also used to determine CHD- and cerebrovascular disease-related utilization.

Primary care

Primary care activities consisted of CVD-related visits to general practitioners (GPs), together with GP visits to patients' homes, and, where available, visits to or by a primary care nurse.

National sources1619 were used to derive CVD-related consultations in the UK, Finland, Malta, and Belgium. For the remaining countries, the total number of consultations was obtained, and those related to CVD were derived by applying the proportion of hospital discharges due to CVD, assuming that a proportion of patients leaving inpatient care due to CVD would be followed up in GP consultations.

Hospital outpatient care

Outpatient care comprised specialist consultations taking place in outpatient wards, clinics, or patients' homes. National data2022 were used to obtain the number of CVD-related outpatient visits in the UK, Portugal, and Lithuania. For the remaining countries, the total number of consultations was obtained, and the proportion related to CVD was estimated by applying the proportion of hospital discharges, as before, assuming that CVD-patients being discharged from hospital were followed in an outpatient setting.

Accident and emergency

A&E care consisted of all CVD-related hospital emergency visits. For Denmark and the UK, national sources23,24 provided the total number of CVD-related A&E attendances. For the remaining countries, all-cause attendances were obtained, and the proportion of CVD-related hospital discharges was applied. This was assumed to be a good approximation of people with CVD attending A&E.

For the earlier categories, the assumptions used to estimate resource use in countries with no CVD-related data were also applied to countries where these data were available. The estimated and actual resource use were evaluated together with the resulting variation on total healthcare cost estimates.

Hospital inpatient care

Inpatient care was estimated from the number of CVD-related days in hospital, including day case admissions and rehabilitation sessions. Disease-specific average length of stay and hospital discharges were obtained,8,9 and multiplied together to obtain the total number of CVD-related inpatient days.

Healthcare unit costs

Unit costs of an inpatient day were obtained from national sources or, when unavailable, by dividing the total expenditure on inpatient care by the total number of hospital inpatient days for all causes. No such information was available for Malta and Cyprus, and unit costs were then derived using the coefficient from a regression of hospital cost onto health expenditure per capita for all countries.

Unit costs for primary, outpatient, and emergency care were obtained in a similar fashion. However, to cost these activities, we made extensive use of an economic evaluation undertaken in 12 countries.25 If no information was available, regression techniques were used to estimate unit costs using data from other countries. Unit costs are reported in Table 1.

View this table:
Table 1

Average unit costs in the EU, by country, 2003

Expenditure on medication

In the UK, Spain, Belgium, Finland, and Italy, CVD-related pharmaceutical expenditure was obtained from national sources.2630 For the remaining OECD countries, this figure was obtained from OECD databases.9 For those countries where CVD-related pharmaceutical expenditure could not be obtained, we assumed that the proportion of total pharmaceutical expenditure attributable to CVD was similar to the average in those countries where expenditure had been established (variation around this average was found to be low in countries with available data).

Only Germany,12 France,13 and the Netherlands14 provided information on the proportion of CVD pharmaceutical expenditure attributable to CHD and cerebrovascular diseases. As the proportion of pharmaceutical expenditure due to CVD in these three countries was relatively similar to the proportion across remaining EU countries, the averaged proportions from Germany, France, and the Netherlands (i.e. 22% for CHD and 9% for stroke) were applied to the remaining countries.

Non-health service costs

Non-health service costs comprise informal care costs, productivity costs attributable to mortality and morbidity, patient travel costs, and out-of-pocket expenses (e.g. child care, home aids, and over-the-counter medication). However, as little information was found on patient travel and out-of-pocket expenses across countries, only informal care and productivity costs were estimated.

Estimation of informal care costs

Informal care costs were equivalent to the opportunity cost of unpaid care, i.e. the time (work and/or leisure) that carers forgo, valued in monetary terms, to provide unpaid care for relatives suffering from CVD. It was hypothesized that only people with CVD who were severely hampered in daily activities would receive informal care.

Hours of informal care were estimated by calculating the number of people with CVD who were hampered in daily activities because of health problems.3 Using data from a European study,31 we determined the probability of receiving care for those people with limiting conditions, which was then applied to the number of people with limiting CVD. Finally, the proportion of care given by carers of working age was determined,31 the number of hours spent by each informal carer was obtained from a UK survey,32 and these were multiplied by the number of informal carers in each age group.

In order to value the amount of unpaid care time, the hourly wage rate33 was applied to informal care provided by those carers of working age and employed. For those carers in retirement or not working, the hourly minimum wage34 was applied. For those countries with no minimum wage, the wage of the worst paid sector was used.

Estimation of productivity costs

Productivity costs included the foregone earnings related to CVD-attributable mortality and morbidity.

The productivity loss from CVD-mortality was estimated by calculating the sum of the age- and sex-specific products of the following:

  1. number of CVD-related deaths;10

  2. number of remaining work years at the time of death (in order to estimate the likely earnings that an individual who died would otherwise have received from paid employment);

  3. annual earnings;33

  4. economic activity and unemployment rates.11

As these costs will be incurred in future years, future earnings were discounted to present values using a 3.5% annual rate.35

Morbidity costs were those costs associated with CVD-attributable absence from work, estimated by multiplying the number of certified days off work due to CVD by the daily earnings.33

The number of CVD-related working days lost was obtained for Austria36 and the UK.37 For the remaining countries, the total number of working days lost due to all diseases was obtained and the proportion related to CVD was estimated by applying the proportion of CVD-hospital bed days in the working age population. We hypothesized that there was a positive correlation between the number of days in hospital and the working-days lost.

However, absent workers are likely to be replaced, and so the morbidity loss as computed earlier will be an overestimate of the actual loss. Hence, we estimated the ‘friction period’, i.e. the period of employee's absence from work due to illness before the employee is replaced, to be 90 days.38 The friction period adjusted morbidity loss was then estimated by multiplying the unadjusted productivity loss by the friction period and then dividing this product by the average duration of each spell of work incapacity, estimated in this study to be 232 days.39

Sensitivity analysis

The effects of 20% changes in healthcare costs, informal care costs, and earnings for males and females were examined. In addition, as GP, outpatient, and A&E visits were estimated using subjective assumptions, the effect of 50% changes in these categories was tested. We also assessed the effects of discounting on productivity costs by using rates of 0 and 10%. Furthermore, we compared the actual total CVD-related healthcare expenditure in France, Germany, the Netherlands, and the UK—the countries with extensive CVD-specific healthcare information—with the estimated total for these countries had it been necessary to estimate every resource category in these countries.


Costs of CVDs

Healthcare costs

CVD accounted for over 126 million hospital bed days in the EU, representing 277 hospital bed days per 1000 population (Table 2). The number of inpatient days varied significantly between countries, from 39 per 1000 population in Malta to 573 in Lithuania. The majority of CVD-related doctor consultations were with GPs, representing 687 visits per 1000 population, compared with 315 outpatient visits per 1000 population (Table 2).

View this table:
Table 2

Resource units per 1000 population in the EU, by country, 2003

CVD cost the EU healthcare systems approximately €105 billion in 2003 (Table 3), with this expenditure accounting for 12% of total healthcare expenditure in the EU (Table 4). The percentage of CVD-related healthcare expenditure varied significantly between countries, from 2% in Malta to 17% in the UK.

View this table:
Table 3

Costs of CVD (€ million) in the EU, by country, 2003

View this table:
Table 4

Healthcare costs of CVD-related diseases in the EU, by country, 2003

The major component of CVD-related healthcare expenditure was inpatient care, which accounted for €60 billion, representing 57% of total healthcare costs. However, in Slovakia, inpatient care accounted for 34% of CVD-related costs, whereas in the UK, this proportion accounted for 76% of healthcare costs. CVD-related pharmaceutical expenditure was also a major cost component, representing 27% (€28.4 billion) of total healthcare costs. Again, this proportion varied between countries, from 11% of total healthcare costs in Poland to >50% in Greece and Portugal. The other three cost components (i.e. primary, outpatient, and emergency care) accounted for 16% of costs, with A&E representing the smallest component.

CVD represented an annual healthcare cost of €230 per EU citizen (Table 4). The amount spent on healthcare for people with CVD varied widely across the EU, mainly reflecting the wealth of each individual country. For example, costs per person varied 19-fold between the lowest spender, Malta (€22 per person) and Germany, which had the highest cost per capita (€423). However, when price differentials were accounted for using PPP, cost differences between countries narrowed substantially (Table 4).

Informal care costs

Approximately 4.4 million people with CVD were severely hampered in daily activities, representing 1% of the EU population. However, out of these, 2.85 million received informal care, with 2.95 billion hours used to care for them. Informal care of CVD sufferers was estimated to cost the EU €29 billion (Table 3).

Productivity costs

CVD accounted for two million deaths in the EU, representing 2.18 million working-years lost. This was estimated to cost about €24.4 billion, after adjustments for working status, and discounting.

There were 268.5 million working-days lost because of CVD morbidity (i.e. 591 days per 1000 population). This represented a cost of €28 billion. However, when adjusted using the friction period, this estimate fell to €10.8 billion (Table 3).

Total costs

Overall, CVD is estimated to have cost the EU economy €169 billion in 2003 (Table 3). Germany and the UK represented over half (54%) of all CVD costs, whereas Malta and Cyprus represented <0.1%. Of the total cost of CVD, ∼62% of total costs were due to healthcare, 21% due to productivity losses, and 17% due to informal care.

Costs of CHD

Healthcare costs

CHD cost the healthcare systems of the EU just under €23 billion in 2003 (Table 5). The major component of health expenditure was inpatient care, which accounted for €14 billion (62%) of healthcare costs, followed by pharmaceutical expenditure, which represented 23% (€5.4 billion) of total healthcare costs. Primary, outpatient, and emergency care, accounted for ∼16% of healthcare costs.

View this table:
Table 5

Costs of CHD (€ million) in the EU, by country, 2003

Informal care costs

Over 678 000 people provided care to CHD patients, representing 702 million hours of care, which was estimated to cost the EU €6.8 billion (Table 5).

Productivity costs

Approximately one million working years were lost because of CHD mortality, accounting for 44% of all working years lost because of CVD-related deaths, with a cost of €11.7 billion (Table 5). Additionally, 90 million working days were lost because of CHD morbidity, representing a cost of €3.5 billion after adjusting costs using the friction period (€9.1 billion without adjustment).

Total costs

Overall, CHD is estimated to have cost the EU €45 billion in 2003: one-quarter of the overall cost of CVD. Over half of these costs (51%) were incurred in healthcare, 34% in productivity losses and 15% in informal care.

Costs of cerebrovascular disease

Healthcare costs

The cost of cerebrovascular diseases to the EU healthcare systems was €21 billion in 2003 (Table 6), with 82% (€17 billion) of healthcare costs being inpatient care. When compared with CVD and CHD, pharmaceuticals represented a small proportion, 5% (€1.1 billion), of total healthcare costs.

View this table:
Table 6

Costs of cerebrovascular diseases (€ million) in the EU, by country, 2003

Informal care costs

Approximately 697 000 people provided care to those with cerebrovascular disease in the EU, representing 723 million hours of care, which was estimated to cost €7.2 billion (Table 6).

Productivity costs

Over 430 million working years were lost from deaths due to cerebrovascular disease, which accounted for 20% of all working years lost because of CVD, with a cost of €4.4 billion (Table 6). An additional 44 million working days were lost in the EU because of morbidity from cerebrovascular disease, representing a cost of approximately €1.7 billion after adjusting costs using the friction period (€4.3 billion without adjustment).

Total costs

Overall, cerebrovascular disease was estimated to have cost the EU €34 billion in 2003: around one-fifth of the overall cost of CVD. Of the total cost of cerebrovascular disease, 61% of costs were due to healthcare costs, 18% to productivity losses and 21% to informal care.

Sensitivity analysis

The methods used to estimate primary, outpatient, and emergency care visits in countries where such data were not available were evaluated by applying them to countries where such data were in fact available. Hence, the estimated numbers of GP visits per 1000 population were 422 in Belgium (an underestimate of 32% from actual visits), 396 (−6%) in Finland, 67 (−35%) in Malta, and 433 (−2%) in the UK, which accounted for a variation in the baseline healthcare costs of these countries of −2.1, −0.1, −0.2, and −0.1%, respectively.

The estimated number of emergency attendances per 1000 population was nine in the UK (an underestimate of 18% from actual attendances) and 20 (+1%) in Denmark, representing a variation in the total baseline CVD-related healthcare costs of −0.06 and +0.02%, respectively. The estimated outpatient visits resulted in a variation in baseline total healthcare costs of −0.6, −1.1, and +2% in Lithuania, Portugal, and the UK, respectively.

In addition, we compared the actual total CVD-related healthcare expenditure in France, Germany, the Netherlands, and the UK—the four countries with most data—with the estimated total for these countries had it been necessary to estimate every resource category in these countries. For France and Germany, the result was an underestimate of 16 and 24%, respectively, whereas for the Netherlands and the UK, the result was an overestimate of 16 and 11%, respectively.

Varying inpatient care, medication, A&E, and outpatient unit costs upwards and downwards by 20% produced a variation in the baseline total EU CVD-related costs of ±7, ±3.4, ±0.24, and ±0.61%, respectively. Our results did not vary significantly when the assumptions used to derive primary, outpatient, and A&E cost estimates were varied simultaneously by 50%, resulting in changes in total costs of ±4.8%. Without discounting, future foregone earnings costs increased by 4.7%, whereas a 10% discount rate was associated with a reduction of 4.5% in costs.

Factors associated with CVD-related healthcare costs

To explore potential reasons for variation in CVD-related health expenditure between countries, we undertook ordinary least-squares (OLS) regression analysis using national income, life expectancy, standardized CVD mortality rates, physician density, and hospital beds per 10 000 population as independent variables. OLS was chosen because it assumes an additive impact of the independent variables on CVD-related health expenditure. Diagnostic statistical tests were performed to evaluate for relevant omitted variables (RESET test), heteroskedasticity (Breusch–Pagan test), and significance of model parameters (t-test).

Our results showed that when income data were included in the OLS regression, factors such as CVD mortality rates or life expectancy that might be expected a priori to predict CVD healthcare costs were not statistically significant (P>0.05). Figure 1 shows this strong positive correlation between income and CVD-related health expenditure. However, this correlation does not necessarily indicate a causative relationship and could arise, for example, due to systematic differences in how expenditures are calculated.

Figure 1

Correlation between CVD-healthcare expenditure and national income (GDP). Lt, Lithuania; Pl, Poland; Sk, Slovakia.


This study is the first to estimate the costs of CVD in the enlarged EU and the proportion of these costs attributable to CHD and cerebrovascular disease. We estimate the total cost of CVD to be €169 billion a year. A recent study40 in the USA estimated the cost of CVD to be $394 billion (€296 billion). In line with our results, that study found that healthcare expenditure accounted for ∼61% of costs, which were estimated using comparable resource categories and methods. Productivity losses accounted for the remainder of total costs, although it is unclear how these were estimated. On a per capita basis, the USA devoted $839 (€ PPP 715) in CVD-related healthcare, approximately €485 more per citizen than the EU, after adjusting for price differences between the two economies.9

Even though CHD and cerebrovascular diseases account for 64% of all CVD deaths, these two diseases represent 47% of total CVD costs (27 and 20% for CHD and cerebrovascular diseases, respectively). Therefore, other CVDs such as hypertension (ICD-10: I10–I15) or other forms of heart disease (ICD-10: I30–I52) may account for a significant proportion of total costs. For example, in Germany, hypertension represented 23% of total CVD costs and other CVDs (i.e. excluding CHD, cerebrovascular diseases and hypertension) accounted for 35% of costs.12 Furthermore, for the 19 EU countries in the OECD, the majority (51%) of CVD-bed days was due to CVDs other than CHD and cerebrovascular diseases.9

Few empirical cost-of-illness, studies have been published to evaluate the impact of other diseases across Europe. As part of a study calculating the global economic burden of diabetes,41 the healthcare costs in the enlarged EU ranged between €32 and €61 billion, significantly lower than our estimates for CVD. The hospital cost of vertebral fractures across 15-member countries of the EU was estimated to be €377 million,42 significantly lower than the estimated CVD-related inpatient costs (€57 billion) for the same countries.

Establishing the cost of an illness does not permit us to say whether a country is spending too much or too little on a disease; rather, the main aim of a cost-of-illness study is to help inform decisions concerning allocation of research funding, by providing a measure of the economic burden of particular health problems. At the 2000 Lisbon summit, EU governments called for a better use of research efforts. For this purpose, the Framework Programme was created as the main instrument for European research funding with a budget of €2.3 billion for health research.43 Studies such as ours enable comparisons between the burden of different diseases, aiding decision makers to prioritize scarce research funds to areas with the highest burden.44 Furthermore, if such studies are performed at regular intervals, they can measure the impact of health policy decisions. In the European setting, they could potentially be used to monitor the effect of interventions aiming to reduce the burden of CVD, such as legislation to curb smoking on CVD.

In order to be in a better position to inform policy decisions aimed at reducing the burden of disease improved information on epidemiology, and accurate information on resource use and unit costs is imperative. Although the EU recognizes the importance of CVD-related comprehensive information in its Member States,1 very little specific information was available for this exercise for a majority of countries. We were therefore compelled to make many assumptions and extrapolations.

Comparability of epidemiological estimates between countries is also important. For example, MONICA data45 from France suggested that official mortality statistics underreported deaths from CVD compared with other countries and showed a 75% underestimate in CHD deaths. Reports2 have highlighted that French doctors have a much higher rate of reporting death from ‘other causes’ than that of other countries. Therefore, our mortality costs for France, and possibly for other countries, will be underestimates.

There were also problems of comparability in resource use information. For instance, some countries reported the total number of visits to doctors and emergency departments on the basis of information collected from statistical institutes; however, other countries based this information on surveys.

In a similar fashion, unit costs were derived from numerous sources and extrapolations were made using multivariate analysis. As in other international costing studies,46 we found wide differences in unit costs, especially in physician visits and hospitalizations. These differences may be due to inter-country variations in duration of consultations, but may also be related to differences in definitions and methodologies. Furthermore, some of the available unit costs were based on fees rather than costs, with fees being set centrally for the purposes of providing incentives or transferring funds within the healthcare system rather than to depict true costs. Hence, it is very difficult to make unit costs more comparable, as their exact nature is hard to identify. A more uniform costing methodology across the EU would have been difficult to perform but might have provided more accurate cost estimates, reflecting the true costs of CVD and easing the interpretation of our results.

Finally, our estimates are likely to be an underestimate. Some categories of healthcare costs, such as health education, were not included because of data limitations; however, other studies have shown these to represent a small proportion of total costs.39,47 Furthermore, patient travel and out-of-pocket expenses were not included but were found not to be of particular importance in other diseases.47,48 However, it is not clear whether the impact of these omissions will affect our results substantially and further research is required in this area. Additional research is also necessary to assess the costs incurred by working people with CVD returning to their post but whose productivity is diminished because of illness.

Despite these acknowledged and important data limitations, our study is the first to quantify the burden of CVD in the EU. We believe that our study will be of particular interest to European policy makers. It highlights, above everything else, the need for comparable and accurate information on the prevalence, mortality, and resource use associated with CVD in the countries of the EU.

Supplementary material

Supplementary material is available at European Heart Journal online.


This project was funded by a grant from the European Heart Network and the British Heart Foundation. The Health Economics Research Centre (HERC) obtains financial support from the National Health Service Research Capacity Development (NHSRCD) programme. We are grateful to Nicola Boulton, Judit Simon, Jorge Felix, and Annelie Niklasson for useful contributions to this project. The comments from three anonymous reviewers are also acknowledged.

Conflict of interest: none declared.


View Abstract