An Italian Chart for Cardiovascular Risk Estimate Including High-Density Lipoprotein-Cholesterol
Most current clinical guidelines on primary coronary heart disease prevention emphasize the importance of risk stratification. Tools for cardiovascular risk estimation have been produced in many countries, although their use has been limited. The availability of new tools that include additional risk factors might lead to their more widespread use. The objective of our study was to produce an updated version of an existing chart for the estimation of cardiovascular disease risk using Italian population data and including high-density lipoprotein-cholesterol (HDL-C) levels as a predictor.
Data were analyzed from nine population studies run in Italy, which included a total of 8054 men and 3206 women aged 45–74 years. The individuals included in these studies had no history of cardiovascular events or diabetes mellitus.
During a mean follow-up of 10 years (range 5–15), incidence data were collected for non-fatal and fatal cases of major cardiovascular diseases (coronary heart disease, cerebrovascular diseases, and peripheral artery diseases). Findings for major cardiovascular risk factors (i.e. sex, age, systolic blood pressure, serum total cholesterol levels, HDL-C levels, smoking habits) at study entry and their relationship with the occurrence of events during the follow-up were used to develop models for the prediction of cardiovascular events. These were multivariate models, based on a log-linear model incorporating the Weibull distribution, and separate models were developed for men and women.
In 10 years, 461 new cardiovascular events occurred among men and 147 among women. The models showed good predictive power, with around 30% of events located in the upper decile of the estimated risk, and around 50% in the upper quintile of estimated risk. The area under the receiver operating characteristic curve, calculated based on internal validation only, was 72%, indicating favorable diagnostic performance of the models.
The independent predictive power of HDL-C was strong, with 1% increase in HDL-C level being associated with a decrease in the incidence of cardiovascular diseases of almost 1% among men and almost 2% among women.
A chart accommodating sex, age, total cholesterol level, HDL-C level, systolic blood pressure, and cigarette consumption was subsequently produced. The inclusion of HDL-C levels in this chart was novel for a risk chart in Italy, as it had not been included in previous editions of the same tool. A special feature of this chart was a new section dealing with the estimate of the ‘relative risk,’ defined by the ratio of absolute risk to the risk expected on the basis of the age, sex, and average age-specific risk factor levels of the involved populations.
The cardiovascular risk assessment devised in the current study represents an improved means for physicians to determine cardiovascular risk and discuss the risk with patients. The chart could be used in countries where the background risk is similar to that of the Italian population; however, external validation of the model is required to adequately assess transferability, and until then the chart should be used with caution in non-Italian populations. Compared with earlier tools, it has the advantage of including HDL-C levels as a predictor of cardiovascular risk.
Estimations of cardiovascular risk have become increasingly popular during the past few decades, as data from large population studies have become available and multivariate functions have allowed estimates of risk to be produced based on the measurement of numerous risk factors. It is widely accepted that cardiovascular risk estimation is more sensitive and specific when more risk factors are considered at the same time, instead of one by one. A great number of tools intended to estimate cardiovascular risk in individuals without cardiovascular disease have been produced worldwide during the past few decades.[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17] The use of tools for the estimation of risk began in the US in the 1970s, then spread to other countries; initially, the tools developed were manuals, followed by charts and interactive software.
The aim is to identify high-risk individuals who merit individual and personalized preventive treatment, sometimes including the use of drugs.[3,6,7,18,19] Preventive action in these high-risk individuals, although they are still free from overt disease, is more urgent and more promising, in terms of expected benefit, than considering individuals who are not at high risk.
It has been demonstrated that cardiovascular risk can be grossly over- or under-estimated by using tools ‘imported’ from other countries.[20, 21, 22] In particular, the Seven Countries Study showed that risk functions produced in Northern European and North American populations grossly overestimated cardiovascular risk in Southern European populations.[20,21] In a dedicated analysis, it appeared that the Framingham risk function for coronary heart disease incidence overestimated the risk in Italian populations by at least 2-fold. This finding is of particular interest, as many predictive tools of this type are directly and indirectly based on the Framingham risk functions. There is contrasting evidence regarding the validity of the Framingham models for the prediction of coronary heart disease events, even in Northern European populations.[23, 24, 25]
More recently, the EuroScore chart has been designed to serve a greater number of countries in Europe; however, different charts were needed for high- and low-risk countries. Moreover, even after calibration for local background mortality, it had a substantial limitation as the considered endpoint was mortality, and not the incidence of cardiovascular events.[26,27]
For these reasons, in Italy there is a long tradition of producing risk functions and tools for local use, based on Italian population data. The first Italian tool for cardiovascular risk prediction was published in 1980 in the form of a manual.
In 2005, the Research Group for the Estimation of Cardiovascular Risk in Italy produced and distributed a chart of cardiovascular risk, called Chart Riskard 2005. This was the second chart produced by the group, the first, called Riscard 2002, was created a few years beforehand. In both cases, the tools were based on risk functions generated solely from national population studies. In Chart Riskard 2005, the endpoint was the first major cardiovascular event within 10 years and the risk factors included were sex, age (range 45–74 years), systolic blood pressure, total cholesterol, diabetes mellitus, and cigarette consumption.
At the same time, the group produced and distributed a much more complex and interactive software, also called Riskard 2005, that included high-density lipoprotein-cholesterol (HDL-C) levels as a possible predictor. Despite the statistical, conceptual, and clinical superiority of software programs, charts are still more popular, resulting from the (partly wrong) assumption that they are easier to use. This is the perception of the authors about the attitude of most physicians, based on the authors’ personal experiences in post-graduate teaching and activities within the so-called ‘permanent education of the medical profession.’
It has been widely documented,[31, 32, 33, 34] even within studies used for the development of the Riskard 2005 software, that HDL-C has a powerful protective role for cardiovascular events. In a recent post hoc analysis an HMG-CoA reductase (statin) trial, it was shown that HDL-C levels had a strong predictive power for cardiovascular events, even in patients receiving statins who were achieving very low levels of low-density lipoprotein-cholesterol (LDL-C).
Therefore, the Research Group has explored the possibility and the opportunity to create another paper tool (chart) for prediction of cardiovascular events that includes HDL-C levels, although this variable was already included in the above-mentioned software.
The production of the chart has been a complicated process, and it is described in this article along with the presentation of the ‘relative risk’ section of the new chart.
Material and Methods
Data from nine Italian population studies[36, 37, 38, 39, 40, 41, 42] were provided by the members of the Research group to the central staff responsible for the production of risk functions and their transformation into a chart.
Originally these studies included individuals aged 35–74 years, a total of 12 045 men and 5108 women, with a follow-up that ranged from 5 to 15 years. The overall size of the population was then reduced when individuals with known cardiovascular disease or diabetes were excluded together with those aged <45 years.
Therefore, for the purpose of this chart, subjects were of the same age range (aged 45–74 years) as in an earlier chart named Riskard 2005 without a history of cardiovascular events or diabetes, resulting in a total of 8054 men and 3206 women. Events occurring during a mean follow-up of 10 years (range 5–15) were considered.
The decision to exclude diabetic patients was based on two reasons: (i) a chart with seven risk factors (the six from the previous chart plus HDL-C levels) demands an enormous number of pages, and it becomes a kind of manual; and (ii) diabetic patients are currently considered at high cardiovascular risk and therefore a quantitative estimate of their cardiovascular risk seems redundant,[43, 44, 45, 46, 47, 48, 49, 50, 51] although some guidelines do not equate type 2 diabetes with pre-existing cardiovascular disease.[12,52,53]
Risk factors measured at baseline and considered for the development of risk functions were as follows: sex, age (years), systolic blood pressure (mmHg), total serum cholesterol level (mg/dL); HDL-C level (mg/dL), and cigarette smoking (average number of cigarettes per day).
The measurement techniques differed slightly across the various studies. In particular, in the Italian Rural Areas (IRA) cohorts, blood pressure was measured in a supine position, while in the other studies the subjects were sitting. For all the other details, the procedure was that suggested by the WHO Cardiovascular Survey Methods Manual. Measurements of total and HDL-C levels were made by laboratories that were under external quality control. More details are available in basic papers of the various studies.[36, 37, 38, 39, 40, 41, 42]
During the follow-up, mortality data and causes of death were systematically collected and coded by a single nosologist using the 9th revision of the WHO International Classification of Diseases, and recorded together with the date of death. In case of apparent multiple causes of death, preference was given, in rank order, to violent causes, advanced cancer, coronary heart disease, cerebrovascular accidents, and others.
Non-fatal cardiovascular events were coded only for ‘major’ cases, although in some cohorts ‘minor’ event data were also collected. Validation of events was made, for approximately 93% of cases, by the confirmation of a hospital discharge diagnosis.
For each subject, all events occurring in the 10-year follow-up period were coded in temporal sequence, each accompanied by a date, and the first event with its date of occurrence was considered for analysis.
The estimate of risk was based on major fatal and non-fatal cardiovascular events described in the following sub-categories:
major coronary events included sudden coronary death, non-sudden coronary death, definite fatal and non-fatal myocardial infarction, and surgery of the coronary arteries;
major cerebrovascular events included definite fatal and non-fatal cerebral thrombosis or hemorrhage, and surgery of the carotid arteries;
major peripheral artery events included fatal and non-fatal aortic aneurysm, fatal disease of the leg arteries, and surgery of the peripheral arteries or aorta.
This selection excluded the following:
minor forms of coronary, cerebral, or peripheral artery disease (such as angina pectoris, transient ischemic attack, and intermittent claudication);
cases of only ‘possible’ non-fatal myocardial infarction or cerebrovascular accident;
heart disease of uncertain etiology manifested only with heart failure, arrhythmia, or heart block;
chronic degenerative cerebral conditions where the vascular origin was doubtful.
Risk functions for prediction of cardiovascular events were produced using risk factors measured at entry examination, cardiovascular events within the 10-year period, and an accelerated failure time model (a log-linear model that incorporates hazard using a Weibull distribution).
There were three options for how HDL-C data could be incorporated into the predictive model. Option 1 was to use total cholesterol and HDL-C levels in the same model; option 2 was to use HDL-C and non-HDL-C (total cholesterol minus HDL-C) levels; and option 3 was to use the ratio of total cholesterol to HDL-C levels. Models produced along these lines showed that the ratio of total cholesterol to HDL-C level had the best discriminant power. This was suggested by the chi square of the likelihood ratio statistics of the models, the differences in which (for both men and women) were significantly in favor of option 3 (p < 0.05) when compared with options 1 and 2.
Despite this, the decision was taken to use the models derived from option 1 for reasons that will be given in the discussion.
Models Used for the Construction of the Chart
A reduction in HDL-C level of 10 mg/dL is associated with a relative risk for a major cardiovascular event of 1.21 for men and 1.34 for women.
This means that, with the use of the six risk factors selected for this analysis, more than 30% of all events are located in the upper decile (10%) of the risk distribution, and approximately 50% of the events were in the upper quintile (20%) of the distribution (i.e. of the events that occurred, approximately 50% were in individuals who were classified in the top 20% of the estimated risk distribution).
The evaluation of the models using the receiver operating characteristic (ROC) curves suggests that the area under the curve was 72% for both men and women, indicating favorable predictive performance. However, the area under the ROC curve was calculated based on internal validation only.
The structure of the chart derived from these risk functions is reported in the technical appendix.
The new tool developed in this study, Chart Riskard HDL 2007, was designed to predict the cardiovascular risk for individuals who do not have clinical manifestations of atherosclerotic damage to the coronary, cerebral, and peripheral arteries, and who are free from diabetes, using a function of six risk factors. As yet, external validation of the tool has not been performed.
A chart, because of physical constraints, can only incorporate a limited number of risk factors, while there are no such limitations for interactive software. This situation lead us to delete a dichotomic factor included in the previous chart, Riskard 2005, and to replace it with another risk factor subdivided into four classes. The loss of diabetes as a risk factor and the non-applicability of the system to diabetic patients is an acceptable limitation of the new chart, as there is a wide consensus on the existence of high cardiovascular risk in association with diabetes,[43, 44, 45, 46, 47, 48, 49, 50, 51] regardless of other risk factors, with the consequent need for individualized preventive treatment. Comparing the risk functions of the previous chart Riskard 2005 (without HDL-C levels, but including diabetes) with those of Riskard HDL 2007, the presence of diabetes roughly corresponds to a risk equivalent to a reduction in HDL-C level of 20 mg/dL.
It is necessary to note that the risk of cardiovascular events can be increased by the presence of factors not considered here, such as high levels of triglycerides, fibrinogen, and homocysteine, and low levels of vital capacity and forced expiratory volume, in addition to low levels of habitual physical activity, family history of cardiovascular disease, the presence of corneal arcus, and the coexistence of other cardiovascular diseases. Moreover, the estimates are likely to be lower than the true risk level, if the computation is made in individuals who already have cardiovascular disease of atherosclerotic origin.
In the Riskard HDL 2007 chart, as in the previous Riskard 2005 chart, we have also produced estimates in the form of relative risk (compared with the mean expected risk), in order to make the concept and the interpretation of the excess risk easier to comprehend. However, we suggest that the use of absolute risk should be used in preference to the relative risk for the identification of high-risk individuals of relatively advanced age (i.e. ≥60 years), whereas the estimation of relative risk (based on our operative definition) should be preferred for the identification of high-risk subjects of relatively young age (i.e. <60 years).
In fact, examining the section of the chart on relative risk reported here, large contrasts in the colors of cells (corresponding to large variation in risk) are found in boxes dealing with relatively younger age groups (45–59 years), whereas a substantial uniformity of colors of cells is typical for older ages (≥60 years). This is an empirical self-evident observation that has suggested the above statement. The reverse occurs when exploring the absolute risk of the chart (not reported).
The Riskard HDL 2007 tool is the first cardiovascular risk chart to be produced in Italy that has included HDL-C as a risk factor.[9,34, 35, 36, 37] Several tools have been produced and distributed in Italy during the past few years,[29,30,57, 58, 59, 60, 61, 62] and this may give the impression of a redundancy in this area. However, the subsequent editions of risk assessment tools tend to be more refined than predecessors, and are based on greater population samples and larger numbers of subjects, with different endpoints and longer periods of follow-up. Perfection and stability in this area are still far away. The choice of one tool instead of another, and of the various options within the same tool, should depend upon the preference of the users of these tools. Plans are being made to inquire, within a reasonable period of time after its introduction and in selected group of users, about the use, benefits from, and difficulties in using this chart.
Use of the Chart and its Limitations
The Riskard HDL 2007 chart is intended to be used in clinical settings where high-risk subjects have to be identified in view of a personalized preventive program. The chart has been produced using Italian population data and therefore the estimated cardiovascular risk reflects the relatively low risk of this country. Caution should therefore be exercised in the use of this tool in other settings, unless the background risk is similar to that in Italy. In fact, even within a single country, the use of a predictive tool might provide imprecise results if large differences in incidence exist that are dependent upon background risk levels not explained by risk factors. However, the availability of regional (within country) predictive models is far outside present capabilities.
Moreover, another readily apparent problem is that the Riskard 2005 and Riskard 2007 charts are yet to be externally validated. Although planned, this will only be possible when other data sources become available in Italy that are similar, from a methodological point of view, to those used to create this chart.
From a theoretical point of view, the presence in the same function of total cholesterol and HDL-C levels is not ideal, as HDL-C is a component of total cholesterol. However, the correlation between the two variables (HDL-C and total cholesterol levels) is not exceedingly high.
The use of total cholesterol and HDL-C levels has been a compromise. In fact, the use of HDL-C and non-HDL-C levels, and, to an even greater extent, the total cholesterol/HDL-C ratio, would have offered better discriminatory power. However, the pragmatic reason for using total cholesterol and HDL-C levels in the model was that it did not require the users of the chart to perform an arithmetic operation. This is considered an advantage over the two other options because of the perceived reluctance of physicians to be involved in trivial mathematics. This impression is also derived from the post-graduate teaching experience of the authors. Another reason why this option was chosen was that the measurement of non-HDL-C (an alternative to total cholesterol levels) is not common practice, although it has been used in important studies and represents a good approximation of LDL-C levels, and for this reason was used in the Riskard 2005 software.
On the other hand, it is clear that the presence of HDL-C level in the model, which is highly predictive, partly obscures the predictive power of the total cholesterol level, particularly in women. In other studies on the predictive power of the HDL-C level, a 1% difference in HDL-C was inversely associated with a 1% difference in coronary heart disease incidence, while a 2–3% difference in HDL-C level was associated with a 3–4% difference in coronary heart disease incidence. In our data, a 1% difference in HDL-C was inversely associated with differences in cardiovascular disease incidence of 0.9% among men and 1.6% among women. Apparently the role of HDL-C in cardiovascular and coronary disease risk is strong even when a different endpoint is considered, i.e. cardiovascular disease, instead of coronary heart disease.
The Communication of Risk
A special feature of the Riskard HDL 2007 chart (and of the two preceding charts from the same research group) is the presence of a section dealing with the so-called ‘relative risk’ (see supplementary material). This quantity is arbitrarily defined as the ratio of absolute risk over the risk expected on the basis of the age, sex, and average age-specific risk factor levels of the involved populations. This indicator has a higher probability of being understood, both by the physician and by patients, than the ‘absolute risk,’ which represents a number whose meaning is familiar only to a few specialists in the area. We believe that the so-called ‘relative risk’ provides an easier way to communicate the concept of risk as it represents how much the subject at risk of an event compared with the ‘average person’ of the same sex and age. The notion of carrying a risk that is two or more times greater than the average person may give a motivation for action that cannot be created by the notion of the ‘absolute risk,’ which is a number without a reference point.
Another open question in the process of decision making in the use of predictive tools is the absence of a consensus on the cut-off level used to define high risk, which cannot be found in existing guidelines.[3,6,7,18,19] If the definition of high-risk is based on absolute risk, the choice is largely dependent upon the overall risk of the population for which the estimate is made.
It is recognized that risk charts and similar tools, beyond their practical meaning, also have an educational role as they focus attention on risk factors, the predictive role of risk factors, and the need to start preventive action. In this particular case, the attention is drawn to the role of HDL-C, which is sometimes neglected or considered as a risk indicator rather than a target for preventive action. This attitude is partly justified by the difficulties in raising HDL-C levels through non-pharmacological means, the uncertainties about the real benefits of increasing HDL-C levels, and the modest influence of statins on HDL-C levels.[64,65]
However, there are some hints of possible future ways to increase HDL-C levels, thereby providing added benefit to the reduction of LDL-C levels with statins. This may partly close the gap in risk that still exists in the majority of people using statins. In particular, extended-release niacin, alone or in combination with a selective prostaglandin D2 receptor antagonist, is a potent agent that has been shown to increase HDL-C levels but is currently underused.
Other hopes come from new drugs whose actions include a substantial increase of HDL-C, such as rimonabant, a selective cannabinoid CB1 receptor blocker approved in Europe for the treatment of obesity.
The Chart Riskard HDL 2007 chart is the first chart for cardiovascular risk prediction in Italy that has included HDL-C in the estimation of cardiovascular risk. Previously in this country, only software-based risk assessment tools have considered HDL-C levels, despite HDL-C levels being strongly predictive of cardiovascular risk. In addition to enabling physicians to estimate cardiovascular risk and explaining this in terms of the ‘relative risk’ for individual patients, this tool may play an educational role in helping to reduce cardiovascular risk in the general population through the adoption of lifestyle modification and the use of appropriate treatments.
Sheridan Henness and Siobhan Ward of Wolters Kluwer Health Medical Communications provided English language assistance and advice on the preparation of this article for submission. Funding for this assistance was provided by Merck Sharp & Dohme, Italy.
The activity of the Research Group for the Estimate of Cardiovascular Risk in Italy and the production of the chart and of this report were sponsored by a scientific-educational grant from Merck Sharpe & Dohme, Italy, based in Rome.
At the time of the research, the Group included Enrico Agabiti-Rosei (University of Brescia, Italy), Gianfranco Botta (MSD Italy, Rome, Italy), Luigi Carratelli (MSD, Rome, Italy), Giuseppe Cavera (Villa Sofia Hospital, Palermo, Italy), Ada Dormi (University of Bologna, Italy), Antonio Gaddi (University of Bologna, Italy), Mariapaola Lanti (Association for Cardiac Research, Rome, Italy), Mario Mancini (University of Naples, Italy), Alessandro Menotti (Association for Cardiac Research, Rome, Italy) Mario Motolese (Centro per la Lotta contro l’Infarto, Rome, Italy), Maria Lorenza Muiesan (University of Brescia, Italy), Sandro Muntoni (University of Cagliari, Italy), Sergio Muntoni (Association MEDICO, Cagliari, Italy), Alberto Notarbartolo (University of Palermo, Italy), Pierluigi Prati (Centro per la Lotta contro l’Infarto, Rome, Italy), Stefano Remiddi (MSD Italy, Rome, Italy), Alberto Zanchetti (University of Milan, Italy)
The prototype of the chart was created by Medrisk srl, Rome, Italy (firstname.lastname@example.org).
Part of the data reported here were published, in Italian, in the proceedings of the Congress “Conoscere e Curare il Cuore 2007,” organized in Florence, Italy by the Centro per la Lotta contro l’Infarto, Rome, Italy.
- 1.American Heart Association. Coronary risk handbook: estimating the risk of coronary heart disease in daily practice. Dallas (TX): AHA, 1973Google Scholar
- 3.Prevention of coronary heart disease in clinical practice: recommendations of the Task Force of the European Society of Cardiology, European Atherosclerosis Society and European Society of Hypertension. Eur Heart J 1994; 15: 1300-31Google Scholar
- 6.Prevention of coronary heart disease in clinical practice: recommendations of the Second Joint Task Force of European and other Societies on coronary prevention. Eur Heart J 1998; 19: 1434-503Google Scholar
- 7.National Heart & Lung Institute. Joint British recommendations on prevention of coronary heart disease in clinical practice. British Cardiac Society, British Hyperlipidaemia Association, British Hypertension Society, endorsed by the British Diabetic Association. Heart 1998; 80(2): 1–29Google Scholar
- 8.Cardiac risk assessment V.98.02 [computer program]. PC version. Manchester: University of Manchester, 1998Google Scholar
- 11.Thomsen T. Prediction and prevention of cardiovascular diseases: Precard© [thesis]. Copenhagen: University of Copenhagen, 2000Google Scholar
- 13.American Heart Association. Health risk awareness [online]. Available from URL: http://www.americanheart.org [Accessed 2007 Sep 1]
- 14.Procam risk calculator [online]. Available from URL: http://www.CHD-Taskforce.com [Accessed 2007 Sep 1]
- 18.National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). Third report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. Circulation 2002; 106: 3143–421Google Scholar
- 19.Fourth Joint Task Force of the European Society of Cardiology and other societies on cardiovascular disease prevention in clinical practice (constituted by representatives of nine societies and by invited experts). European guidelines on cardiovascular disease prevention in clinical practice. Eur J Cardiovasc Prev Rehabil 2007 Sep; 14 Suppl. 2: S1-113Google Scholar
- 28.Manuale del Rischio Coronarico. Per stimare il rischio coronarico nella pratica medica. Udine: Associazione Nazionale Centri per le Malattie Cardiovascolari e CIBA-GEIGY, 1980Google Scholar
- 33.Assman G, Cullen P, Schulte H. The Munster Heart Study (PROCAM): results of follow-up at 8 years. Eur Heart J 1998; 19 Suppl. A: A2–A11Google Scholar
- 34.Sharrett AR, Ballantyne CM, Coady SA, et al. Atherosclerosis Risk in Communities Study Group. Coronary heart disease prediction from lipoprotein cholesterol esters, triglycerides, lipoprotein (a), apolipoproteins A-1 and B, HDL density subfractions: the Atherosclerosis Risk in Communities (ARIC) Study. Circulation 2001; 104: 1108–13PubMedCrossRefGoogle Scholar
- 36.Menotti A, Lanti M, Puddu PE. Epidemiologia delle malattie cardiovascolari. Insegnamenti dalle Aree Italiane del Seven Countries Study. Rome: Cardioricerca Ed., 1999: 1–532Google Scholar
- 41.Averna MR, Barbagallo CM, Montalto G, et al. Progetto di epidemiologia e prevenzione Ventimiglia di Sicilia: dati di popolazione. G Arterioscl 1992; 17: 83–7Google Scholar
- 48.Vongpatanasin W. Cardiovascular morbidity and mortality in high-risk populations: epidemiology and opportunities for risk reduction. J Clin Hypertens (Greenwich) 2007; 9 Suppl. 4: 11–5Google Scholar
- 49.Rydén L, Standl E, Bartnik M, et al. Guidelines on diabetes, pre-diabetes, and cardiovascular diseases: executive summary. The Task Force on Diabetes and Cardiovascular Diseases of the European Society of Cardiology (ESC) and of the European Association for the Study of Diabetes (EASD). Eur Heart J 2007; 28: 88–136PubMedCrossRefGoogle Scholar
- 51.Guidelines on diabetes, pre-diabetes, and cardiovascular diseases: executive summary. The Task Force on Diabetes and Cardiovascular Diseases of the European Society of Cardiology (ESC) and of the European Association for the Study of Diabetes (EASD). Eur Heart J 2007; 28: 88-136Google Scholar
- 52.New Zealand Guidelines. Evidence-based best practice guidelines: assessment and management of cardiovascular risk. Wellington (NZ): New Zealand Guidelines Group, 2003Google Scholar
- 53.The Royal College of General Practitioners. Effective Clinical Practice Unit. Sheffield/NICE: Clinical guidelines for type 2 diabetes: blood pressure man-agement/lipid management. London: National Institute for Clinical Excellence, 2002 OctGoogle Scholar
- 54.Rose G, Blackburn H. Cardiovascular survey methods. Geneva: World Health Organization, 1968Google Scholar
- 55.Classificazione delle Malattie, Traumatismi e Cause di Morte. IX Revisione 1975. Roma: Istituto Centrale di Statistica, Metodi e Norme Serie C, N.10, 1984Google Scholar
- 56.BMDP. Statistical software manual. Cork, Ireland: BMDP, 1992Google Scholar
- 57.Riscard 2002. Software per la stima del rischio cardiovascolare. Roma: Cardioricerca ed, 2002Google Scholar
- 59.CCM: il progretto cuore [online]. Available from URL: http://www.cuore.iss.it [Accessed 2007 Sep 1]
- 60.Carta del Rischio Cardiovascolare Globale. Progetto Cuore. Istituto Superiore di Sanità, Dicembre 2003 [online]. Available from URL: http://www.cuore.iss.it [Accessed 2007 Sep 1]
- 62.Il Progetto Cuore: Studi Longitudinali. Ital Heart J 2004; 5 Suppl. 3: 94S-101SGoogle Scholar
- 63.Ridker PM, Rifai N, Cook NR, et al. Non-HDL cholesterol, apoliproteins A-I and B100, standard lipid measures, lipid ratios, and CRP as risk factors for cardiovascular disease in women. JAMA 2005; 294: 325–33Google Scholar
- 64.Chapman J. Beyond LDL-cholesterol reduction: the way ahead in managing dyslipidaemia. Eur Heart J Suppl 2005; Suppl. F: F56–62Google Scholar
- 65.Barter PH, Rye KA. High density cholesterol: the new target. A handbook for clinicians. Birmingham (UK): Sherborne Gibbs Ltd, 2005Google Scholar