This retrospective study reviewed ambulatory electronic medical records of patients who presented to outpatient clinics at Tawam Hospital, Al-Ain, between April 1, 2008, and December 31, 2008. Al-Ain is the fourth largest city in the UAE, and it has a population of approximately 200,000 UAE nationals [12]. Tawam Hospital is a publicly funded facility, and, collectively, its outpatient medical clinics serve most of the residents of Al-Ain. Ethical approval for this study was obtained from Tawam Hospital and the United Arab Emirates University’s Research and Ethics Board (CRD 239/13). The requirement for informed consent was waived because the patients’ records and information were anonymized and de-identified before the analyses.

The study’s sample size was calculated using a formula for a study designed to estimate incidence in a population [13]. A sample size of 923 was determined based on an anticipated 10% incidence of CVD [6] and utilizing 80% power at a 2-sided level of significance of 0.05.

Sociodemographic and clinical data were collected through a systematic review of 1534 consecutive patients’ ambulatory electronic medical records. The study’s inclusion criteria were UAE nationals aged ≥18 years who had any of the following at baseline: a history of smoking, body mass index (BMI) ≥25 kg/m², systolic blood pressure (SBP) ≥120 mmHg, diastolic blood pressure (DBP) ≥80 mmHg, serum glycosylated hemoglobin A1c (HbA1c) level ≥ 5.7%, serum triglyceride (TG) level ≥ 1.69 mmol/L, serum total cholesterol (TC) level ≥ 5.17 mmol/L, serum low-density lipoprotein-cholesterol (LDL-C) level ≥ 3.36 mmol/L, and a serum high-density lipoprotein-cholesterol (HDL-C) level < 1.03 mmol/L, and lipid-lowering, antidiabetic, or antihypertensive medications.

Of the 1512 eligible subjects, we excluded 209 subjects who had histories of a CVD event, which was defined as a diagnosis of angina, a prior myocardial infarction (MI), angioplasty of the coronary arteries, coronary artery surgery, a stroke, a transient ischemic attack, peripheral arterial disease, or heart failure, at baseline. We also excluded 316 subjects whose data were missing at baseline. Follow-up data were collected annually between the baseline visits in 2008 and July 31, 2018. During the follow-up period, ten subjects, representing 0.77% of the initial cohort, had < 1-year follow-up after inclusion. These subjects were excluded from the final analyses. The final sample for the analyses comprised of 977 subjects (Fig. 1).

The data collected at baseline included sociodemographic parameters, clinical information, and treatment modalities. The sociodemographic data included age, sex, and history of smoking, which was defined as a current or past history of smoking tobacco. The clinical data included the BMI, which was calculated as the weight (kg) divided by the height (m²), SBP and DBP, the HbA1c, TC, HDL-C, LDL-C, and TG levels, the TC/HDL-C ratio, and the estimated glomerular filtration rate (eGFR). The eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation that is based on serum creatinine [14]. The treatment modalities included the use of antidiabetic medications, lipid-lowering agents, and antihypertensive drugs. Dyslipidemia was defined as documented treatment with lipid-lowering medications (e.g., bile acid sequestrant, fibrate, or 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitor) or the presence of ≥1 of the following: TG ≥1.69 mmol/L, TC ≥5.17 mmol/L, LDL-C ≥ 3.36 mmol/L, or HDL-C < 1.03 mmol/L [15]. Obesity was defined as a BMI of ≥30 kg/m². Blood pressure (BP) was measured by qualified registered nurses using an automated oscillometric BP measuring device as part of the daily routine clinical practice. HTN was defined as an SBP ≥140 mmHg, a DBP ≥90 mmHg, or the use of antihypertensive medications (e.g., diuretics, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, alpha blockers, beta blockers, or calcium channel blockers). According to the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure [16], we categorized BP into three groups: Those with SBP lower than 139 mmHg and DBP lower than 89 mmHg as normal BP group; Stage 1 HTN was defined as an SBP between 140 and 159 mmHg, or DBP between 90 and 99 mmHg; Stage 2 HTN was defined as an SBP higher than 160 mmHg or DBP higher than 100 mmHg. If systolic and diastolic pressures fell into different categories, subjects were assigned to the higher category. Patients with an HbA1c level of ≥6.5% or who were receiving antidiabetic medications (e.g., sulfonylurea, thiazolidinedione, dipeptidyl peptidase-4 inhibitor, biguanide, or insulin) were considered to have DM. Good glycemic control was considered if patients with DM had a HbA1c level of < 7% [17]. CKD stages 3–5 were defined as an eGFR < 60 mL/min/1.73 m² according to the Kidney Disease: Improving Global Outcomes clinical guidelines [18].

A major CVD event was defined as the first occurrence of a fatal or non-fatal MI, a fatal or non-fatal stroke, or acute peripheral arterial occlusion. A non-fatal MI was confirmed using the World Health Organization’s criteria in the context of elevated levels of cardiac biomarkers with either symptoms or diagnostic electrocardiograms suggestive of ischemia [19]. A non-fatal stroke was defined as typical neurological dysfunction lasting > 24 h and neuroimaging that excluded other diagnoses. The occurrence of fatal MIs or fatal strokes was confirmed by the hospital records or death certificates. Sudden death outside the hospital after the onset of symptoms with no evidence of a non-coronary cause of death was also considered a fatal MI. Acute peripheral arterial occlusion was defined as an arterial event that had a sudden onset, had a duration of < 2 weeks, and resulted in symptomatic limb ischemia [20]. If a subject had experienced > 1 endpoint, the first event to occur was used to define the onset of disease. The occurrence of the study outcomes was adjudicated on July 31, 2018, and it was based on a review of all of the electronic medical records and death certificates by an ad hoc committee of clinicians.

All of the laboratory assays undertaken at baseline and during follow-up were conducted within the medical laboratory department at Tawam Hospital. The circulating HbA1c levels were measured using an automated analyzer (Integra 400 Plus; Roche Diagnostics GmbH, Mannheim, Germany). Fasting lipid profiles and serum creatinine levels were measured using standard methods and a UniCel DxC-800 Synchron Clinical System (Beckman Coulter, Inc., Brea, CA, USA).

The data distributions and categories were examined, and the categories with small sample sizes and skewed distributions were documented. The continuous variables are presented as the means (standard deviations) or as the medians (1st quartile, 3rd quartile), and the categorical variables are presented as proportions. Differences between men and women at baseline were assessed using the independent samples t-test if the data for the continuous variables were normally distributed and the Mann-Whitney U test if the data for the continuous variables were non-normally distributed, and Fisher’s exact test (2-tailed) was used for the categorical variables.

The patient-years at risk for major CVD, MI, stroke, and acute peripheral arterial occlusion were calculated for each subject from the baseline visit to the time at which each outcome was diagnosed, the time of death, or the time of the last outpatient clinic visit, whichever occurred first. The incidence rates and 95% confidence intervals (CIs) were calculated by dividing the number of new events for each outcome by the respective patient-years at risk. Kaplan-Meier time-to-event analyses were conducted for men and women for major CVD, MI, stroke, and acute peripheral arterial occlusion, and the results were compared using log-rank tests.

Multivariable Cox proportional hazards regression analyses stratified according to sex were used to determine the predictors of incident CVD. Since previous studies’ findings have shown that the SBP, DBP, HbA1c level and BMI have curvilinear associations with the risk of incident CVD [21, 22], quadratic terms for these predictors were initially considered in the models and they were excluded if they were not statistically significant. The proportional hazards assumption was assessed using log-log plots. Multicollinearity was evaluated by examining the tolerance. The results are expressed as the hazard ratios (HRs) and 95% CIs.

Table 1 shows the cohort’s baseline characteristics and a comparison of men and women. Compared with men, women were older and took lipid-lowering medication more frequently, and they had a higher BMI, higher TC and HDL-C levels, a higher eGFR, a lower TG level, a lower TC/HDL-C ratio, lower SBP and DBP, and were less likely to have a history of smoking. Among patients with DM, women had better glycemic control than men.

During the median follow-up period of 8.92 years (interquartile range 7.85, 9.58 years), 97 subjects experienced incident major CVD, 66 subjects experienced incident MIs, 31 subjects experienced incident strokes, and 4 subjects experienced incident acute peripheral arterial occlusion. Overall, the incidence rate (per 1000 person-years) for major CVD was 12.7 (95% CI 10.4–15.4), and in men and women the incidence rates (per 1000 person-years) for major CVD were 16.8 (95% CI 12.9–21.4) and 9.0 (95% CI 6.4–12.4), respectively. The 9-year cumulative incidence of major CVD was 9.9% (95% CI 8.2–11.9). Major CVD and MI occurred significantly more frequently in men than in women, and the rates of stroke and acute peripheral arterial occlusion were similar in men and women (Table 2).

Table 3 presents the results from the multivariable Cox regression analyses of incident major CVD. The tolerance ranged from 0.45 to 0.98, indicating an absence of multicollinearity, and the log-log plot to test the proportional hazards assumption was not significant. In both sexes, a low eGFR was a significant risk factor for major CVD. The SBP was a strong predictor of major CVD, with risk increases of 20 and 58% per 10 mmHg SBP increase in men and women, respectively. The major CVD risk increased by 16% in men and 24% in women for each 1% rise in the HbA1c level. In women, the TC/HDL-C ratio was an additional significant independent risk factor, with an estimated 44% risk increase per ratio increment. Age and a history of smoking exacerbated the major CVD risk by 50% per decade and 80%, respectively, in men. In the model, the quadratic terms of the BMI, SBP, DBP, and HbA1c level were not statistically significant in either men or women.

One strength of this study is that it is the first longitudinal study to assess the epidemiology of CVD and sex-specific risk factors in high-risk UAE nationals with a median follow-up period of 9 years. A longer follow-up period is desirable as this would enable the long-term risk factors associated with CVD to be determined. Moreover, this study’s analyses relied on direct measurements of the laboratory variables and anthropometric parameters rather than self-reported information to classify the predictor variables.

This study has several limitations. First, this was a retrospective study. A prospective study design involving standardized measurements of the laboratory variables and anthropometric parameters may have furnished the investigation with data of better quality and reduced the potential for bias. Second, other parameters, for example, the waist circumferences, family histories of CVD, physical activity, diet, and the menopausal status were not included in this study, because the data were unavailable, and could have impacted the study outcomes. Third, our study consisted of UAE nationals at high cardiovascular risk from outpatient clinics of a single hospital, therefore, our results may not be applied to the general population in the UAE. Finally, changes in the status of medication use and the evolution of risk factors from baseline were not considered during the follow-up period. These and other confounding factors affecting the study outcomes cannot be ruled out. The changes in risk factors over time and their impact on CVD outcome may be worth assessing in future studies.