Methods

The Institute for Medical Screening at the Chaim Sheba Medical Center has annual screening program for executives, which performs about 10,000 screening examinations each year. All participants are self-referred asymptomatic men and women. Most participants have repeated annual visits for a number of years. This database is well suited for the characterization, follow-up, and clinical event identification of asymptomatic population.

A computerized database was established in the year 2000 and was used as the data source for this study. At baseline, all subjected fill a health questionnaire, undergo physical examination, routine blood tests, and undergo a treadmill exercise stress test (Bruce protocol). Participants are then followed annually and all clinical events occurring between visits are recorded systematically. If abnormal values were identified during an index visit, participants were referred with recommendations for management and possible further evaluation to their primary care physician, with the newly diagnosed conditions appropriately documented on the next follow-up visit.

The entire database included 25,515 subjects. Subjects were excluded if they had a single visit to the center (n = 6,952) or had a personal history of ischemic heart disease (IHD), cerebrovascular disease, diabetes mellitus, or hypertension at baseline (n = 3,367). We excluded 360 subjects (1.5%) due to insufficient data. Thus, the final study population comprised 14,966 subjects. The institutional review board approved this study.

The primary outcome of this study was the development of CVD defined as the composite incidence of hospitalization for acute coronary syndrome (ACS), symptom driven percutaneous coronary intervention (PCI), or ischemic cerebrovascular accident. The secondary end point was all-cause mortality. Incident events were based on medical summaries of annual visits, as assessed by the centers’ physicians and recorded in the computerized medical records.

In statistical analysis, continuous variables were compared using the Student t test, and categorical data were compared using the chi-square test. For the end point of cardiovascular free disease survival curves, we used Kaplan–Meier survival estimates with comparison of cumulative events across strata by the log-rank test. Multivariate Cox proportional hazards regression modeling was used to assess the independent risk for developing CVD. Models were adjusted for the following covariates: gender, age ≥50 years, systolic blood pressure ≥120 mm Hg, body mass index (BMI) >30, high low-density lipoprotein ≥130 mg/dl, low high-density lipoprotein (HDL) <50 mg/dl in women and <40 mg/dl in men, high triglycerides ≥150 mg/dl, impaired fasting glucose, and renal dysfunction that was defined as estimated glomerular filtration rate <60 ml/min/1.73 m ² and calculated using Modification of Diet in Renal Disease (MDRD) formula. The model was further adjusted for the following self-reported variables: smoking status, physically activity, and positive family history for IHD defined as history of IHD in a first-degree relative of 55 years old or younger. Estimated METS recorded during the exercise stress test were individually compared with the age and gender norms published by the American College of Sports Medicine. In our study, we defined high cardiorespiratory fitness as being in the upper median performance according to these norms within each gender group. Multivariate Cox proportional hazard regression model with interaction term analysis was used to compare risk for development of CVD disease in men versus women population in the various prespecified risk factor subgroups. The groups were dichotomized according to the following cut-off values: age ≥50 and <50 years, systolic blood pressure >120 mm Hg and 120 mm Hg or lower, BMI 30 or lower and BMI >30, high low-density lipoprotein ≥130 mg/dl, low HDL <50 mg/dl in women and <40 mg/dl in men, high triglycerides ≥150 mg/dl, impaired fasting glucose, renal dysfunction (estimated glomerular filtration rate <60 ml/min/1.73 m ² ), smoking status, physically active, and positive family history for IHD. Interaction term analysis was assessed by adding a gender-by-risk factor interaction term to each multivariate age-adjusted model. An association was considered statistically significant for a 2-sided p value <0.05 in all tests. All analyses were 2 tailed and performed with the SPSS version 22 statistical software (IBM Inc., Chicago, Illinois).