Prevalence of cardiovascular disease in type 2 diabetes: a systematic literature review of scientific evidence from across the world in 2007–2017

Criteria for eligibility were guided by the PICO reporting system (which describes the participants, interventions, comparisons, and outcome[s] of the systematic review), together with the specification of the type of study design (PICOS), from the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [24].

The search was undertaken between February 15 and March 6, 2017. Databases searched included Medline and Embase between January 2007 and March 2017. In addition, PubMed was searched from 2014 to identify articles that were “ahead of print” yet fully available. Evidence presented at selected conferences during the last 5 years were accessed, including the Annual Meetings of the International Society Pharmacoeconomic Outcomes and Research (ISPOR), American Diabetes Association (ADA), European Association for the Study of Diabetes (EASD) and American Association of Clinical Endocrinologists (AACE). Keywords linked to MeSH terms specific to each database were used in the search including prevalence, OR epidemiology, AND acute coronary syndrome, OR cardiovascular disease, OR cardiovascular death, OR non-fatal myocardial infarction, OR non-fatal stroke, OR obesity AND type 2 diabetes mellitus. Other keywords were cerebrovascular disease, cerebral arterial disease, intracerebral hemorrhage, cerebral infarction, coronary artery disease, ischemic heart disease, atherosclerotic heart disease, coronary heart disease, angina pectoris. Identified articles and previous reviews were hand searched for articles that may have included data useful to this search.

Two reviewers independently searched Medline, Embase and the proceedings of major scientific meetings for suitable papers. Results were compared and adjudicated through consensus discussion. A third reviewer checked all results for quality assurance.

Data extracted from articles included information concerning the publication, the patients involved, and outcomes of interest. Publication items included the first author, year of publication, the country in which the data were collected, and date of data collection. Patient data collected included the number of patients screened, percentages of males and females, average age, duration of T2DM, the proportion with obesity (or average body mass index (BMI) ± SD). Outcome data consisted of the numbers and percentages of patients having each cardiovascular outcome, overall and separately for males and females, where available. The same procedure (two independent reviewers plus a third judge) was followed for data collection as for article selection.

Data were analyzed descriptively, with sums, averages, and medians, and ranges reported. The primary outcome was the estimate of prevalence rates of CVD in patients with T2DM. No overall quantitative synthesis was undertaken. Weighted averages were calculated for individual countries and IDF regions. For patient characteristics, we calculated simple averages and medians across studies. Due to a single study with a sample size of more than three million people, which skewed the data, we calculated weighted averages for patient characteristics with and without that study. It should be noted that averages were based on the studies that reported the outcome, which may then represent a subgroup of the entire pool of studies.

The risk of bias was explored by applying the checklist from the STROBE initiative [25]. They have produced a validated checklist of items that should be addressed in reports of observational studies. There are 22 main items, each of which addresses an issue of research design and is presented in a list of recommendations. Items are scored as dichotomously as acceptable or not acceptable.

The flowchart in Fig. 1 depicts the article selection process. We initially identified 1539 papers that appeared to be suitable, but after examining them systematically, 57 studies were accepted. Three articles each presented two different sets of results [26‐28]; therefore, there are 60 sets of analyses within these 57 articles. Table 1 lists these studies along with their descriptive variables. There were 51 full articles and six abstracts presented at scientific meetings. Data collectively represent more than 4.5 million people with T2DM from around the globe.

In Table 2, results are presented geographically according to the classification system used by the IDF [1]. Studies from 25 countries were represented in this review: Australia, Belgium, Brazil, Cameroon, China, France, India, Indonesia, Iraq, Ireland, Italy, Korea, Mexico, Netherlands, Portugal, Qatar, Russian Federation, Saudi Arabia, Scotland, Spain, Sweden, Switzerland, Thailand, UK and USA. Details are provided in Table 3. Three areas were responsible for generating 80% of the studies. Europe produced most articles (46%), followed by the Western Pacific/China (21%), and North America (13%). The other 20% were from the rest of the world. There were no discernible patterns differentiating prevalence rates between countries, based on income status. Part of the problem is that there are few studies in low- and middle-income countries and none from those in the lowest income level.

In the 57 individual studies, data from 4.5 million people with T2DM were presented with nearly 3.1 million people coming from a single Russian study by Shestakova [29]. As presented in Table 1, using a simple average across studies, the average age was 63.6 ± 6.9 (median = 64.3 years; weighted average = 66.3 ± 6.9 years). The weighted average proportion of persons with obesity was 46.3 ± 15.0%, with a simple average of 47.0 ± 14.7% (median = 45.4%), defined as a BMI ≥ 30 kg/m². The mean percentage of males across the studies was 50.5 ± 10.3% (median = 52.0%); the weighted average of the proportion of males was 36.0 ± 10.0%, including the study by Shestakova [29], and 54.1 ± 9.9% excluding that study. The patients had T2DM for an average duration of 10.4 ± 3.7 years (median = 10.0 years; weighted average = 6.6 ± 3.7 years). Among the 23 studies that reported duration of follow-up, the average was 7.3% ± 4.5 years (median = 6.0 years; weighted average = 5.2 ± 4.3 years).

Prevalence rates of cardiovascular comorbidities are summarized in Table 4 for all patients as well as separately for males and females. In studies reporting gender-specific prevalence rates, males had higher prevalence rates than females for all outcomes except overall CVD, where both sexes had an overall prevalence rate of approximately 27%. Overall, in studies that presented prevalence rates for males and females combined, the prevalence of CVD among persons with T2DM was 32.2%. CAD and atherosclerosis were the most prevalent CVD comorbidities, with prevalence rates of 21.2 and 29.1%, respectively, whereas stroke was the least prevalent with a prevalence rate of 7.6%. It is unclear why people with T2DM have different susceptibilities to these diseases. An explanation for the high prevalence rate for atherosclerosis could be that it is an artifact of patient selection. In the studies that examined atherosclerosis, most patients were enrolled if they had high-risk scores for atherosclerosis, resulting in a very high rate of disease detection.

Table 5 presents the data regarding the rates of mortality associated with CVD in persons with T2DM. The weighted average of death rates from the eight studies with 3,208,557 patients with T2DM was 9.9% (95% CI 8.6–11.3%) [29‐36]. There were 6.3% who died due to CAD and another 1.5% from CBV. Comparing patients with both T2DM and CVD with patients having neither T2DM nor CVD, the odds ratio for death was 4.56 (95% CI 3.53–5.89) [32]. Using a weighted average from seven studies (N = 86,557) [29‐35], the calculated deaths due to CVD comprised 50.3% (95% CI 37.0–63.7%) of all deaths in patients with T2DM. The major contributors were CAD, which was responsible for 29.7% (95% CI 25.1–34.4%) and stroke/CBV for 11.0% (95% CI 8.8–13.3%).

About half of the patients included in this analysis had obesity. Three-quarters of the included studies reported on patients’ BMI or the percent of patients with obesity. While the definitions and BMI cut-off points of obesity varied across studies, the most commonly used definition of obesity was a BMI ≥ 30 kg/m², which was employed by 16 studies (43% of those providing a definition).

Five papers reported prevalence rates of CVD according to obesity status, and all of them found a positive relationship between obesity and increased prevalence rates of CVD [26, 37‐40]. Using logistic regression to control for multiple factors, Bhatti et al. [26] found a positive correlation between obesity and CAD (P = 0.021). Tamba et al. [37] reported positive correlations between obesity and both CAD (r = 0.3, P < 0.001) and stroke (r = 0.5, P < 0.001). Boonman-de Winter et al. [38] quantified the relationship between BMI and heart failure. The prevalence rate of heart failure was 38.7% (95% CI 31.2–46.1%) in patients with a BMI ≥ 30 kg/m² and 23.4% (95% CI 19.4–27.5%) in those with a BMI < 30 kg/m², which represents a 65% increase due to obesity.

Two studies explored the relationship between increasing BMI and risk of CVD [39, 40]. According to Wentworth et al. [39], for CAD in both males and females, the prevalence rate of CAD increased with each successive increase in BMI, with a five-fold increase between the lowest and highest categories [< 25 (normal), 25–30 kg/m² (overweight), 30–35 kg/m² (mild obesity), 35–40 kg/m² (moderate obesity) and > 40 kg/m² (severe obesity)]. The difference was that prevalence rates in males were about double those for females in every BMI category. For the outcome stroke/transient ischemic attack (TIA) in males, only the highest category (BMI > 40) had elevated prevalence rates, which were about double those for the lowest category (BMI < 25). For females, prevalence rates of stroke/TIA increased in those who were overweight and had mild or moderate obesity but decreased for those with severe obesity. Finally, Glogner et al. [40] had quite different results. They reported a steady increase in prevalence rates of MI from 6.86% in those with a BMI < 20–9.33% in patients who were overweight (BMI 25–30), a 36% increase. However, MI prevalence rates declined thereafter with each increasing category of obesity. The highest category (BMI ≥ 40) had a prevalence rate of 5.01%, which was 27% lower than those in the lowest category (BMI < 20). Thus, patterns vary quite widely, and studies often examined different outcomes.

In the assessment of risk for bias, the studies addressed 80% of the STROBE checklist items (i.e., research design and data presentation), on average. The mean was 80 ± 12%, and the median was 81%, with a range of 54–100%. The two items that were addressed by 100% of the articles were reporting of outcome data and reporting of outcomes. The two items addressed the least were the statement of funding (56%) and indicating the study design with a commonly used term in the title or abstract (60%).

Age is a well-known risk factor for CVD. Out of the 57 articles, thirteen (25%) reported on the relationship between age and CVD and the results were quite mixed. Nine studies identified a significant relationship between age and CVD [30, 38, 41‐47], but only two presented results across multiple age categories [38, 42]. Alonso-Moran [42] found that the odds ratio for IHD, stroke, heart failure and MI all increased sequentially with each increase in 5-year age category as compared with the age group 35–39 used as a reference. All of these individual outcomes achieved statistical significance (P < 0.001). Boonman-de Winter et al. [38] similarly reported a sequential increase in prevalence rates of heart failure for all patients in 5-year age categories from 60 to 64 to > 80 years of age. Other authors reported that older patients had higher prevalence rates than younger patients, but provided few details on age categories [30, 41, 43, 45, 46]. On the other hand, four studies reported no differences between age categories [26, 34, 48, 49]. Three other studies used age as a covariate in a logistic regression with no further details [28, 50, 51]. Therefore, few studies have quantified the effect of age on CVD prevalence rates among people with T2DM.

Obesity has long been established as an independent risk factor for CVD [7, 52], and is associated with CAD [53, 54], atherosclerosis [51], and cardiac death [55, 56]. Furthermore, it has been shown that overweight and obesity are highly prevalent in T2DM patients with high CV risk and that BMI and waist circumference are related to major cardiometabolic risk factors such as hypertension and elevated low-density lipoprotein cholesterol (LDL-C) [57].

Obesity is usually defined by body mass index (BMI, calculated as body weight in kg divided by the square of height in meters), with the World Health Organization (WHO) classifying adults with a BMI 30 kg/m² as obese [58]. However, BMI as a measure to stratify patients with obesity has limitations and does not account for the wide variation in body fat distribution nor the quality of fat, and may not account for associated health risk in different individuals and populations [58]. This has been shown to be true for South Asian populations [59]. In a study from Raji et al. [60] noted that compared with Caucasians, Asian Indians had significantly greater total abdominal and visceral fat matched with Caucasians of the same age, gender, and BMI, meaning that this population has an increased CVD risk. Besides, there is a weaker association between increasing BMI and T2DM in Asian populations compared with Caucasians due to the risk for T2DM begins increasing at comparatively normal BMI in Asian populations [61].

Seven of the included studies evaluated the relationship between obesity and/or BMI and CVD risk. Five of the studies included in this review identified a positive relationship between obesity and increased prevalence rates of CVD [26, 37‐40]. One of these studies [26] used lower BMI cut-off points to account for Asian populations in accordance with WHO recommendations on BMI for Asian populations [62] and evaluated abdominal adiposity with waist circumference measurements to determine the prevalence of obesity. Overall, the studies found a positive relationship between increasing BMI and CVD; except in one study [39], where women with severe obesity had a reduced prevalence of stroke. While the authors do not explain the reduced prevalence of stroke/TIA, it may be explained by differences in vascular risk markers in men, such as pre-existing ischemic heart disease, age, and smoking [63]. Furthermore, the presence of gonadal steroids, most notably estrogen, may lend a protective effect against stroke/TIA in women and it has been shown that adiposity is associated with increased levels of estrogen [64].

Although obesity is identified as a risk factor for CVD, it is associated with a paradox in that mortality is lower in patients who are overweight or obese than in those whose BMI is normal or underweight [65]. Lee et al. [66] reported that obesity provided a survival benefit to patients with heart failure who did not have comorbid diabetes, but not in patients who did have concomitant diabetes. In contrast, a group led by Abi Khalil [67] examined a cohort of 2492 T2DM patients in seven countries in the Middle East, Gulf region, with acute heart failure. They reported that BMI was inversely correlated with the risk of mortality, with severe obesity associated with less mortality risk.

It is clear that the relationship between obesity and the risk of CVD and CVD-related deaths requires further exploration to identify these mechanisms and relationships.

In persons with T2DM, CVD is responsible for at least half of the mortality, as previously mentioned. Among the specific diseases within that term, CAD was most lethal, followed by stroke. Similar results have been demonstrated with other models. In an incidence-based study, Straka et al. [68] followed 29,863 patients (5501 with T2DM and 24,362 without T2DM) over a 1-year period. Four of the incident cardiovascular outcomes they reported were significantly higher in those with T2DM. Patients with T2DM had 10% greater risk of CAD, 53% of MI, 58% of stroke, and 112% increased risk of heart failure. Therefore, T2DM is a substantial risk factor for CVD and its consequences.

As this was a global review, studies from across the world were included. Given the variation in which diabetes and its macrovascular complications are treated and managed across countries and income levels, it is relevant to look at prevalence rates across regions and countries. However, almost half (46.0%) of the research was produced in Europe, and very little information was obtained from the less developed regions of the world such as Africa, Latin America, and the Asian subcontinent.

As shown in Table 2, the regions with the highest prevalence of overall CVD were North America and Caribbean (46.0%; N = 4,327,503), Southeast Asia (42.5%, N = 537) and Western Pacific (including China) (33.6%; N = 44,062). Southeast Asia stands out with a higher prevalence of CAD (29.4%) compared with other regions. The prevalence of CAD in this region is driven by one study from India [69], which specifically investigated the pattern of CAD in 134 symptomatic T2DM patients in India. However, epidemiological studies on people of South Asian origin have shown an increased likelihood of developing CAD that is up to two times higher than in Caucasians [70]. The higher risk is due to both pathophysiological and life course-related risk factors [70].

The summaries across countries and regions provide an overview of the geographic spread of research but should be interpreted with caution given the limited number of studies for some of the regions and countries. Figure 2 illustrates the distribution of studies across regions and countries and clearly shows that few studies exist for several regions. For example, only one study from one country in the African region was identified and therefore should not be seen to represent findings for the region’s entire T2DM population.

Treatment of both T2DM and CVD vary greatly between and within countries, and although much of the CVD risk in T2DM can be associated with the long-term complications of T2DM, there has been growing interest to determine whether certain antidiabetic drugs influence this risk. For example, sulfonylureas which are the second most commonly used antidiabetic drug after metformin, have been shown to be associated with an increased risk of cardiovascular events and mortality [71]. Newer antidiabetic drugs have been shown to lower the risk of CVD in T2DM patients [17‐20]; however, these drugs are often intended to be used as second- to third-line treatments and many years may pass before patients can benefit.

Encouragingly, CVD mortality is declining in high-income countries among the general population due to reductions in cardiovascular risk factors as well as to recent advances in prevention, treatment, and management [72]. This trend has also been observed in people with T2DM in some countries. Jung et al. [73] estimated trends in CVD in people with and without T2DM in South Korea using data from the national health insurance system. The results show a significant reduction in CVD risk among people with T2DM brought on by improvements in the care and management of patients. However, in many developing countries where the burden of T2DM is rapidly rising and lifestyle patterns changing an increase in CV risk factors among people with T2DM can be expected [3]. A study from China analyzed the relationship between lifestyle behaviors and multiple CV risk factors in 25,454 people with T2DM [74]. The researchers found that unhealthy lifestyles were common, especially among those who are non-elderly, and above-college educated. Furthermore, it was found that an unhealthy lifestyle was associated with poor blood, blood lipid, and blood pressure control. Decreasing the impact of T2DM and CVD in developing countries will require interventions aimed at changing risky lifestyle behaviors.

Screening people with T2DM for CV risk is an important strategy for reducing mortality and CVD events. A study from Denmark [75] found that a single round of diabetes screening and cardiovascular risk assessment in middle-aged adults in general practice was associated with a significant reduction in risk of all-cause mortality and CVD events in people with T2DM. The same researchers found that population-based stepwise screening for T2DM and CVD among all middle-aged adults was not associated with a reduction in mortality or CV events. Therefore, the benefits of population-based screening are limited in this context [76]. Kesall et al. [77] found that targeting specific occupational and industry groups with health checks could help identify individuals at high risk of both T2DM and CVD. In a study of 500,000 members of the Australian working population, they found that high T2DM and CVD risk was increased significantly in many occupational groups and industries.

Recent research points to an increasingly better understanding of the markers for identifying high CVD risk in people with T2DM. Li et al. [78] found that the combined application of carotid and lower extremity ultrasonography may be helpful to identify patients with T2DM who have a higher CVD risk. In a study of 2830 hospitalized patients with T2DM, they found that the concomitant presence of carotid and lower extremity atherosclerosis further increases the risk of CVD in patients with T2DM, compared with those who had either carotid or lower limb atherosclerosis and those without atherosclerosis. A study by Mohammedi et al. [79] found that major peripheral arterial disease (PAD) presenting as lower-extremity ulceration or amputation and peripheral revascularization is associated with increased risk of death and CV events in people with T2DM. The researchers conclude that screening for PAD along with active management are crucial for prevention of CVD in people with T2DM. In addition, coronary artery calcium (CAC) assessments have been found to significantly improve the risk classification for CHD and atherosclerotic CVD events in people with T2DM—regardless of the duration of diabetes [80]. Thus, a CAC assessment can be a useful tool for classifying people with T2DM into lower- or higher-risk groups for long-term CVD risk.

Lipid profile has long been considered among the most important risk factors for CVD in T2DM, and several trials have confirmed that lowering low-density lipoprotein cholesterol (LDL-C) via statins in T2DM was effective in reducing the risk of CVD [81, 82]. It is also well known that statins also have a triglyceride-lowering effect [83]. In a cross-sectional study of 223,612 patients with T2DM in China, researchers found that although lower triglyceride was associated with reduced CVD risk in the short-term, it was associated with increased risk in the long-term [84]. This paradox could mean that low triglyceride is not necessarily associated with good clinical outcomes in all people with T2DM and that there are subgroup associations with CVD in patients with different durations of T2DM. Furthermore, Clua-Espuny et al. [85] suggest that the relative importance of risk factors wanes in complex chronic patients with T2DM with advancing age. In a cohort study of almost 3500 complex chronic patients above the age of 80 of whom 53% had diabetes and a high prevalence of associated classical risk factors, the researchers found that all-cause mortality was more affected by aging factors than by specific complications of diabetes. The authors make the recommendation that, for these patients, the care strategy may need to be redefined and adapted to comorbidities and functional autonomy rather than being focused on treatment outcomes.