Cardiovascular diseases continue to be a major cause of morbidity, mortality, and disability.1

Primary prevention of these diseases aims to lower the risk of cardiovascular disease by combining different strategies. To maximize cost-effectiveness, these efforts mainly focus on persons considered to be at high risk, in order. Patients at higher cardiovascular risk are usually identified by cardiovascular risk (CVR) population calculations. Consequently, CVR calculators are an essential tool for primary prevention, as they are readily available and easy to use, making them an affordable approach to screening. Several clinical practice guidelines, such as the European Guidelines on Cardiovascular Prevention2 and its Spanish adaptation3 as well as the guidelines of the US Preventive Service Task Force4 and the NICE (National Institute for Clinical Excellence),5 already include CVR calculation among their cardiovascular disease prevention guidelines to implement various algorithms in clinical decision-making.

Numerous functions have been developed to estimate CVR based on a certain number of known CVR factors. These functions include the 10-year Framingham function6 for the US population, the SCORE function7 for European countries, and the updated QRISK3 algorithm8 based on primary care patient databases from England.

In the case of Spain, the researchers of the REGICOR study (Registre Gironí del Cor [Girona Heart Registry])9 performed a recalibration for the Spanish population, finding that the original Framingham function overestimates CVR. Likewise, the SCORE project provides specific tables for countries with a low incidence of cardiovascular disease and has also been adapted for Spain using data from the MONICA study (MONItoring Trends and Determinants in CArdiovascular Disease)-Catalonia.10 Recently, a study has been published on the validity of both equations, concluding that they significantly overestimate cardiovascular mortality among the Spanish population.11

The ICARIA (Ibermutuamur CArdiovascular RIsk Assessment) study, based on data obtained from occupational health assessments, showed high prevalences of CVR factors, metabolic syndrome, and patients with high CVR. A high incidence of nonfatal cardiovascular events has also been observed in patients with a moderate-to-high SCORE.12,13 Additionally, moderate-to-high CVR is associated with longer, more costly sick leave episodes due to cardiovascular and noncardiovascular illnesses, and an improvement in CVR has been seen to be accompanied by reduced sick leave during 1-year follow-up.14,15

Nevertheless, all these risk calculation tables, currently in use, estimate short-term risk, usually at 10 years, and can be used for middle-aged individuals aged 35 or 40 years or older. Most young adults, considered to have a low 10-year risk according to the current tables, would probably be at high risk if long-term or lifetime risk were analyzed. For this reason, the latest European guidelines for cardiovascular prevention recommend calculating lifetime risk, vascular age, or relative risk in young adults.2 Some cardiovascular primary prevention algorithms even recommend combining both tools and estimating the lifetime CVR of patients with low 10-year CVR.16 Furthermore, it has been observed that CVR factors are strongly implicated in subclinical arteriosclerosis prevalence and progression in individuals younger than 50 years.17 Along this line of research, a function has been developed to estimate the 30-year CVR of descendants of the original Framingham cohort,18 and even the lifetime risk,19–23 ie, the lifetime coronary, cerebrovascular, and CVR by using data obtained from numerous cohort studies conducted in the United States over the past 50 years. In England, a lifetime CVR function has also been developed from recently updated databases obtained by primary care physicians.24 These functions assign a long-term or lifetime probability of cardiovascular disease and are of particular interest among younger individuals, who usually do not have high short-term risk.

The aim of the present study was to prepare a predictive function of lifetime CVR (IBERLIFERISK) for fatal and nonfatal cardiovascular events in the Spanish working population.

This study was based on a retrospective cohort design with 7 to 10 years of follow-up. To conduct the study, we selected employees aged 18 to 65 years who were working for member companies of a large nationwide occupational insurance firm (Ibermutuamur), who had no history of cardiovascular disease, and who had passed an occupational health assessment at the Sociedad de Prevención de Ibermutuamur (now Cualtis) between 2004 and 2007. Informed consent was obtained from all participants prior to enrollment in the study, in accordance with the principles set forth in the Declaration of Helsinki. The protocol was reviewed and approved by the respective ethics committee.

The usual information for an ordinary health assessment was recorded at the baseline visit: sociodemographic variables, company industry (coded according to the National Business Activity Code, 1994 version), and clinical variables, such as other disease history, active drug therapy, and anthropometric and laboratory values.

The study collected information from Ibermutuamur's official records on all study participants’ sick leave episodes from the date of enrollment until 31 December 2014. For participants who withdrew from the study for other reasons (change in insurance firm, unemployment) before the data lock point, information was requested from the National Statistics Institute for all study participants and was available for mortality but not for nonfatal cardiovascular events.

The primary outcome variable, namely, incidence of fatal or nonfatal cardiovascular event, was coded using the ICD-9-CM (International Classification of Diseases, Ninth Revision, Clinical Modification), 2002: ischemic heart disease (codes 410-414), heart failure (code 428), cerebrovascular disease (codes 431-438, except 432.1, 437.2, 437.3, and 437.7), and peripheral arterial disease (codes 440-444, except 442, 443.0, and 443.1). Fatal cardiovascular events also included deaths where the cause was coded as hypertensive disease (codes 401-405) or arrhythmia (codes 426-427, except 427.5).

For purposes of analysis, the study cohort was randomly distributed into 2 groups at a ratio of 7:3. The larger cohort comprised 70% of all participants (referral cohort) and was used to develop the CVR lifetime equation in the Spanish working population, whereas the other (validation) cohort included the remaining 30% and was used to validate the CVR equation developed in the first cohort.

The present study obtained a model to calculate the lifetime CVR (IBERLIFERISK) from age 18 to 75 years in the Spanish working population.

The study found that use of the lifetime risk identifies more high-risk patients and reclassifies others from moderate risk to high risk. A study comparing the lifetime risk of the American College of Cardiology/American Heart Association model and of QRISK observed that 1.61% (95%CI, 1.55-1.66) patients were reclassified from moderate risk to high risk according to QRISK and 27.1% (95%CI, 27.11-27.70) according to the American College of Cardiology/American Heart Association score.29 Another study based on the 30-year Framingham model reclassified around 70% as high risk and also discriminated between participants with and without evidence of carotid plaques.30

Consistent with the QRISK lifetime risk model, our model is also flexible and allows risk to be calculated in different age brackets, not necessarily lifetime (in our case, up to age 75 years) but also allows CVR to be calculated at 5, 10, 15 years, or any whole number from the participant's age up to age 75 years.

The discrimination obtained from the model was satisfactory, with an area under the ROC curve of 0.84 (95%CI, 0.82-0.85) in men and 0.73 (95%CI, 0.66-0.80) in women. The QRISK model had an area under the ROC curve of 0.842 (95%CI, 0.840-0.844) in women and 0.828 (95%CI, 0.826-0.830) in men.

Calibration was better for men than for women. The validation was performed using data up to 5 years of follow-up, as this was the follow-up period of most participants in the sample. The calibration was repeated with 8-year data (median follow-up), which showed similar results to those obtained at 5 years (data not shown). One of the reasons for these calibration differences may be the low number of cardiovascular events, particularly in women. Mortality data were obtained from the National Statistics Institute; hence, loss of fatal events is unlikely although nonfatal cardiovascular events may have been lost. Although information is usually collected on this type of event and the analyses also considered any possible losses to follow-up, information could have been lost about nonfatal events in participants who switched their insurance or became unemployed. This study observed only 0.56% cardiovascular events and 1% competing events in the referral cohort, vs 5.19% and 6.34% reported by the QRISK lifetime risk calculation.24 The differences between our results and those obtained by QRISK could be explained, at least partially, by the fact that our population was much younger, with a mean age of 35 years, whereas the mean age was 48 years for QRISK. However, the follow-up time of this study (6.5 years on average) was very similar to that of QRISK (7 years). Some authors consider it necessary to observe 40 to 50 events per endpoint included in the predicted model for their estimate to be valid and reliable.31

The results of our study indicate that men's risk is overestimated, specifically in the high-risk deciles. In women, more erratic behavior was observed in the low-risk deciles and better adjustment in the high-risk deciles, except for the last decile, in which the model overestimates the risk. The QRISK function24 slightly underestimates the risk in the low-risk levels, but shows satisfactory calibration at higher risk.

The lifetime CVR model developed for the Spanish population, like those developed in the United Kingdom (QRISK) and in the United States (American College of Cardiology/American Heart Association), still raises questions that should be answered before the model can be recommended for clinical practice. There is no evidence regarding the actual threshold for considering a patient at high risk according to the lifetime risk, although some authors24 recommend using the 90th percentile value, which would correspond to a threshold > 50%.

There is also no evidence on the very long-term benefits (beyond those observed in clinical trials) of antihypertensive or lipid-lowering agents in young patients with a high lifetime CVR, although it is true that long-term benefits have been reported by post-trial follow-up studies, for instance, by the Heart Protection Study32 on statin use.

Like relative risk or vascular age, lifetime CVR may be a useful tool to improve communication with patients at low-to-moderate absolute 10-year risk, thus raising their awareness of the importance of CVR and encouraging them to make changes needed to acquire healthy habits. Although the lifetime risk model was derived from a different sample than the one used later to validate it, it is always desirable to perform a validation in a completely different sample from the study sample. This external validation is expected to be performed with 2 different samples: one of them would be a cohort of Ibermutuamur workers who pass a health assessment during 2008 and 2009 and another would be a sample of primary care patients. This score will also be compared with other lifetime risk scores (QRISK and American Heart Association) to assess its applicability. Likewise, a platform is being developed to calculate lifetime risk individually, as done by the QRISK investigators, as it is not viable to create a table listing all the combinations of the risk factors included in the model or to calculate risk from the individual's age up to any higher age.

This study has obtained a model to calculate lifetime CVR (IBERLIFERISK) in the Spanish working population aged 18 to 75 years. The model showed satisfactory discrimination and calibration, but was better in men than in women.

Project coordinated and funded by the Fondo de Investigaciones Sanitarias (Health Research Fund) (PI14/01177), Carlos III Health Institute, Ministry of Economy and Competitiveness.

None declared.