Introduction

Cardiovascular disease (CVD) is the leading cause of death worldwide, being responsible for nearly 18.6 million global deaths in 2019.1 Moreover, current projections estimate that global deaths due to CVD will increase to 23 million by 2030.2 The high global burden of CVD can be partly attributed to the high prevalence of numerous CVD risk factors, such as tobacco use, high cholesterol, hypertension (HTN) and type 2 diabetes mellitus (T2DM).3 Furthermore, HTN and T2DM share common risk factors with CVD, often resulting in similar disease management plans.4

Previous epidemiological studies have found that trends in CVD,5 HTN6 and T2DM7 incidence are stabilising or declining in most high-income countries, while cardiometabolic diseases are on the rise in low-income and middle-income countries.8 However, few studies to this date have taken into account possible differences by population subgroups, despite evidence of sex differences in cardiometabolic disorders9 and evidence that individuals of a low socioeconomic status (SES) are at a greater risk of developing CVD,10 HTN11 and T2DM.12 Within Spain, higher prevalences of CVD,13 HTN14 and T2DM15 have been observed in lower SES population subgroups. However, detailed information on CVD, HTN and T2DM incidence rates is still scarce due in part to a lack of longitudinal data.

Longitudinal data allows for time trend analyses, which are important for monitoring chronic illness rates and planning interventions and prevention strategies directed specifically towards population subgroups whose trends are not improving or evolving at a desired rate. Understanding CVD, HTN and T2DM trends over time by population subgroup is necessary for developing more comprehensive care plans that take into account the nuanced needs of each specific group and is an important step towards achieving health equity. Our study aimed to estimate the time trends in the incidence of CVD, HTN and T2DM in Catalonia, Spain from 2009 to 2018 while considering possible differences by age, sex, and SES.

MethodsStudy design, setting and data source

We performed a cohort study with prospectively collected data from the Information System for Research in Primary Care (SIDIAP; www.sidiap.org).16 A description of the characteristics of SIDIAP can be found in the online supplemental material. Because the data in SIDIAP is pseudonymised, consent was waived per the International Ethical Guidelines for Epidemiological Studies. This study was approved by the Clinical Research Ethics Committee of the IDIAPJGol (project code: 20/237-P).

We included all adults at least 40 years of age at time of entry into the study who were registered in SIDIAP between 1 January 2009 and 31 December 2018, with at least 1 year of observed data, and free of cardiometabolic conditions. Though SIDIAP has data available starting in 2006, we chose for our study period to begin in 2009 to minimise the potential inclusion of prevalent cases of CVD, HTN and T2DM. The index date was the first recorded primary care visit during the study period. Individuals were followed from cohort entry until diagnosis with an outcome of interest, the end of the study (31 December 2018), death, or transferring out of SIDIAP.

Our conditions of interest were CVD, HTN and T2DM, as defined using ICD-10 codes (online supplemental table S1). We extracted data on sex (male/female) and age (40–54, 55–69 and 70+ years). Information on SES was available for urban areas at the census tract level through the Mortalidad en áreas pequeñas españolas y desigualdades socioeconómicas y ambientales (MEDEA) index of socioeconomic deprivation (online supplemental material).17 For our study, we categorised the MEDEA index into quintiles, with the first and fifth quintiles representing the least and most deprived urban areas, respectively. For descriptive purposes, we also extracted data on nationality, smoking status and alcohol use.

Patient and public involvement

Study participants were not actively involved in defining the research question, designing the study, or measuring the outcome, as the data were collected prior to the design of the study. Study participants were not involved in the write up or reporting of the results and there are currently no official plans for the results to be disseminated to the study population.

Statistical analyses

Individuals diagnosed with a CVD, HTN or T2DM were not considered as eligible incident cases for the same condition in future years after the year of their first diagnosis. We calculated the overall incidence rate of CVD, HTN and T2DM for each study year from 2009 to 2018, stratified by sex. Incidence was calculated as the number of new cases of each condition divided by 1000 person-years of follow-up, and person-years were calculated as the number of years each individual was at risk of developing one of the three conditions during the study period. Next, we calculated yearly incidence rates of all three conditions, stratified by sex and age group. Finally, we restricted these same analyses to individuals living in urban areas and stratified additionally by SES.

We calculated incidence trends over time in three equivalent subperiods, from 1 January 2009 to 31 December 2012, 1 January 2013 to 31 December 2015 and 1 January 2016 to 31 December 2018. We calculated incidence rate ratios (IRRs) and their corresponding 95% CIs for each age-sex subgroup to analyse the differences in incidence between two subperiods: 2009–2013 versus 2014–2018. This division through the midpoint of the study period allowed us to compare incidence and observe potential differences between the first and second halves of the entire period. To estimate differences in incidence trends by SES, we calculated IRRs and 95% CIs and per cent change between the years 2009 and 2018 for the least deprived and most deprived areas. We calculated per cent change of incidence by dividing the difference between incidence in 2018 and the incidence in 2009 by the incidence in 2009 and multiplying by 100. Analyses were conducted in Stata V.17.0 and in R Core Team 2016 (R Project for Statistical Computing).

Results

The study population included 3 247 244 adults over the age of 40 residing in urban areas, with a median age at study entry of 58 years (IQR: 47–71) (table 1, online supplemental figure S1 and online supplemental table S2).

Table 1

Descriptive characteristics of the total study population from SIDIAP database, 2009–2018 (n=3 247 244)

Figure 1

Annual incidence of cardiovascular disease, hypertension and type 2 diabetes mellitus stratified by age and sex, 2009–2018.

Cardiovascular disease incidence

Incidence rates of CVD increased from 2009 to 2018 in both sexes, except in the 70+ age group (figure 1). Incidence increased most sharply in the 40–54 age group in both sexes (eg, in men from 4.0/1000 person-years in 2009 to 6.6 in 2018) (online supplemental table S2). In women aged 40–54 years, CVD incidence was 61% (95% CI: 52% to 69%) higher between 2016 and 2018 compared with 2009 and 2012, but 22% (16% to 29%) higher between 2013 and 2015, suggesting a steeper increase in incidence in the most recent years (figure 2). Increase in CVD incidence were steeper in women than men in all age groups and nearly all compared study periods. Incidence rates of CVD increased in the 55–69 age group (eg, in women from 5.1 in 2009 to 7.4 in 2018), with a steeper increase in CVD incidence in the most recent years in both sexes (online supplemental table S2 and figure 2). In the 70+ age group, there was little or no change in CVD incidence in either sex, though we observed a slight decrease in CVD incidence in men between 2016 and 2018 (IRR=0.93, 0.90 to 0.95).

Figure 2

Incidence rate ratios (IRRs) of cardiovascular disease, hypertension and type 2 diabetes mellitus stratified by age and sex, between the periods 2013–2015 and 2016–2018 vs 2009–2012.

Trends in CVD incidence by SES mirrored the overall trends of CVD incidence (figure 3). We observed higher incidence levels in the most deprived areas, especially in the two youngest age groups. In women aged 40–54 years, the 2009 incidence of CVD was 81% (42% to 131%) higher in the most deprived areas compared with the least deprived areas (tables 2 and 3). The 2018 incidence of CVD was 68% (39% to 103%) higher in the most deprived areas compared with the least deprived. Incidence was also higher for the most deprived areas compared with the least deprived areas for the 55–69 age group in both sexes, remaining roughly the same between 2009 and 2018. Differences in incidence by SES were not evident in the 70+ age group. A potential interaction between sex and SES was observed, as SES differences were consistently higher in women than men for both comparison points (eg, 1.81 (1.42–2.31) vs 1.32 (1.14–1.53) for women vs men, respectively, in 2009).

Figure 3

Annual incidence of cardiovascular disease, hypertension and type 2 diabetes mellitus by age, sex and socioeconomic status (SES), 2009–2018.

Table 2

Incidence rate ratios for cardiovascular disease, hypertension and type 2 diabetes mellitus by sex, age and deprivation Index, 2009 to 2018

Table 3

Per cent change of incidence rates for cardiovascular disease, hypertension and type 2 diabetes mellitus by sex, age and deprivation Index, 2009 to 2018

Hypertension incidence

HTN incidence increased from 2009 to 2013, before decreasing from 2014 to 2018 for both sexes and in all age groups (eg, in men aged 55–69 from 36.83 in 2009 to 38.7 in 2013 to 20.6 in 2018) (figure 1 and online supplemental table S2). In both sexes, incidence decreased most sharply in the 70+ age group (eg, in men from 33.3 in 2013 to 15.9 in 2018). In women aged 70+, HTN incidence decreased by 39% (37%–40%) in 2016–2018 compared with 2009–2012 and by 10% (9%–12%) in 2013–2015 (figure 2). These trends differed in 40–54 years old, where HTN incidence was higher between 2013 and 2015, but then began to decrease between 2016 and 2018. For example, in women aged 40–54, HTN incidence was 13% (11%–15%) higher in 2013–2015 compared with 2009–2012, but then decreased by 4% (2%–6%) in 2016–2018. Decreases in HTN incidence tended to be steeper in men than women, especially in the 55–69 age group.

Higher incidence levels were observed in the most deprived areas in the two youngest age groups (figure 3). In women aged 40–54 years, the 2009 incidence was 52% (41%–65%) higher in the most deprived areas (tables 2 and 3). However, in women aged 70+, HTN incidence in 2009 was 17% (11%–22%) lower for the least deprived areas. SES differences were higher in women than men for both comparison points (eg, 1.52 (1.41–1.65) vs 1.12 (1.05–1.20) for women vs men, respectively, in 2009), suggesting a potential interaction between sex and SES. Considering per cent change, incidence levels have similarly decreased for both the most and least deprived areas across all age groups in women. In men, inequalities become evident: in the 40–54 age group, HTN incidence increased by 2.7% in the most deprived areas, while decreasing by 4.2% in the least deprived areas. On the other hand, in the 70+ age group, incidence in the most deprived areas decreased by 53.9% compared with 39.6% in the least deprived areas.

Type 2 diabetes mellitus incidence

T2DM incidence decreased in the 55–69 and 70+ age groups for both sexes (eg, in men aged 70+ from 17.8 in 2009 to 10.66 in 2018), while slightly increasing in the 40–54 age group (eg, in men from 8.3 in 2009 to 9.5 in 2018) (figure 1 and online supplemental table S2). In women aged 40–54, T2DM incidence was 9% (6%–13%) higher in 2016–2018 compared with 2009–2012 (figure 2). However, T2DM incidence decreased slightly more sharply for men than women in the 55–69 age group and more sharply in the 70+ age group in both sexes. In women aged 70+ years, incidence was 37% (34%–39%) lower in 2016–2018 and 14% (12%–17%) lower in 2013–2015, suggesting a sharper decrease in incidence in more recent years. This same differential in cumulative incidence between 2013 and 2015 and 2016 and 2018 was also observed in the 55–69 age group in both sexes.

Higher incidence levels were observed in the most deprived areas, especially in the 40–54 and 55–69 age groups (figure 3). For example, in women aged 40–54 years, the 2009 T2DM incidence was 2.85 (2.46–3.29) times higher in the most deprived areas compared with the least deprived areas, and 2.19 (1.90–2.52) times higher in the most deprived areas compared with the least in 2018 (table 2). A potential interaction between sex and SES was once again observed, as differences were higher in women than men for both comparison points (eg, 2.85 (2.46–3.29) vs 1.74 (1.57–1.93) for women vs men, respectively, in 2009).

Discussion

In this cohort study of over 3.2 million adults, we found that overall CVD incidence increased, while HTN and T2DM incidences decreased in both women and men from 2009 to 2018, despite differences by age group and specific time period. For example, CVD incidence decreased in the 70+ age group, while T2DM incidence increased in the 40–54 age group. HTN incidence increased from 2009 to 2013, before decreasing until 2018. When stratifying by SES, higher incidence levels were observed in the most deprived areas, especially in the youngest two age groups, despite CVD, HTN and T2DM trends mirroring the overall trends. Moreover, differences in incidence across SES appear to be larger in women, suggesting a potential interaction between sex and SES.

Past studies have identified decreasing CVD incidence in most high-income countries during the late 20th and early 21st centuries.5 A study on European CVD mortality found that CVD mortality rates have decreased in past decades in adults,18 while a study from Girona, Spain in 2005 found an increase in adult heart attack incidence.19 However, the results of these studies have not been stratified by age group and, thus, are difficult to compare directly with our results.

Few studies have considered differences in CVD trends across multiple age groups in the same study. A study performed in the UK found increasing rates of coronary heart disease among men aged 35–44 years, despite steady decreases among the oldest age group.20 Furthermore, a review of CVD epidemiology in young adults found that, in contrast to older adults, trends in CVD incidence in young adults have been increasing or plateauing in recent decades.21 As these results are congruent with our study, it may be important to consider potential age differences in disease trends. Differences in CVD incidence trends by age group may be explained by the fact that in many developed countries, rates of important CVD risk factors such as substance abuse,22 physical inactivity23 and obesity24 are increasing among adolescents and young adults. Moreover, these risk factors do not affect all population subgroups equally, with low SES being associated with higher rates of substance abuse,25 physical inactivity,26 and obesity.27

Our study found a decrease in HTN incidence after 2013 across all age and sex groups. Our results are consistent with a study on global trends in HTN prevalence from 1990 to 2019 which showed that HTN prevalence has declined as health systems have achieved high control rates during recent decades in Spain as well as other comparable European countries.28

We found a marked decrease in HTN incidence starting in the year 2013. We hypothesise that this may be due to the implementation of a plan for HTN control in Catalonia around a similar time. In 2011, a plan was implemented by the Catalan Institute of Health (ICS) to better control HTN.29 This plan included specific objectives for HTN prevention, with economic benefits for healthcare clinics and providers that met the objectives of the plan.29 This plan and its associated benefits may have incentivised HTN prevention and control around 2011, perhaps leading to a decrease in HTN incidence and reporting in subsequent years.

Our study found that T2DM incidence decreased in both the 55–69 and 70+ year age groups, while increasing in the 40–54 year age group. Past studies have found that T2DM incidence is decreasing in a majority of global adult populations during recent decades,30 especially among high-income countries.7 Although a direct link between obesity and T2DM cannot be established with our study alone, the trends in T2DM incidence found by our study coincide with the increase in obesity in Spain.31 We hypothesise that the observed differences in T2DM trends by age group may be due to a greater increase in obesity prevalence in younger adults compared with older adults within Spain,31 given that overweight and obesity are among the greatest risk factors for T2DM.32 Furthermore, low SES is associated with higher levels of physical inactivity26 and obesity27 and we would therefore expect disproportionate T2DM incidence across SES.

Based on our results alone, it is difficult to assess whether incidence trends in HTN and T2DM are truly decreasing or if our results highlight a potential deficiency in HTN and T2DM prevention and control. Uncontrolled HTN33 and T2DM34 lead to higher CVD incidence in the long term given that they are main risk factors for CVD. Therefore, if prevention and early detection of HTN and T2DM are not effective, patients may never receive a HTN or T2DM diagnosis but may be diagnosed directly with CVD in the long run as a result of not having their HTN and/or T2DM effectively prevented, diagnosed or controlled. In this scenario, we would expect to see decreases in HTN and T2DM incidences paired with an increase in CVD incidence, as signalled by our results.

Our study identified higher incidences rates of CVD, HTN and T2DM in the most deprived areas, despite similar trends in incidence as in the least deprived areas. CVD, HTN and T2DM incidence was especially higher in the 40–54 and 55–69 age groups in both sexes in the most deprived areas compared with the least. This suggests that individuals residing in more deprived areas may be more likely to be diagnosed with CVD, HTN or T2DM at a younger age. Thus, it is important to consider the structural factors that impact individual health and individuals’ abilities to be healthy and take care of themselves. For example, low SES and education levels influence food behaviours,35 physical activity patterns and abilities36 and access to preventative healthcare,37 all of which influence CVD, HTN and T2DM risk, but also CVD mortality.38 Therefore, though individuals of a low SES may understand the components of a healthy lifestyle, they may lack the economic conditions and resources to attain it.

Inequalities in CVD, HTN and T2DM incidence may persist due in part to the fact that there have not been equitable reductions in risk factors for these conditions across all social classes.39 Our results add to the growing body of literature which highlights the importance of tailoring interventions to the needs of specific populations, such as by social class, age group or geographical location. Interventions that do not take population subgroups into account may improve overall trends for the general population, but may not be effective in reducing health inequalities.40 Health inequities can be combated through the creation of health policies that are adapted to meet the needs of vulnerable groups and which enable changes in the health service structure to provide sufficient resources to all populations.

Our study has some limitations. First, SIDIAP does not include data from health information from primary healthcare centres that are not associated with the ICS. Moreover, our definitions of CVD, HTN and T2DM only took into account diagnostic codes recorded in SIDIAP and not prescriptions or laboratory reports, leading to a possible underdiagnosis of cases. By nature of being an electronic health record data set, SIDIAP is not explicitly designed for calculations of population disease incidence, as the population included in the data set is not a random population sample, but rather a sample of public healthcare users. However, bias that may be reflected in the capture and recording of outcomes is likely systematic. Our study used the MEDEA index of deprivation, which is an ecological measure based on data from the 2001 census, given that we did not have individual-level data on SES. Additionally, the MEDEA is only calculated for urban areas, and, therefore, we were unable to include individuals who reside in rural areas in our study of trends by SES. Finally, though the results of our study indicate a potential interaction between sex and SES, testing for this interaction was outside the scope of our descriptive study. Future studies should consider this possible interaction and present detailed, experimental results.