https://orcid.org/0000-0002-1005-1579
Abstract
Cancer and heart failure (HF) share risk factors, pathophysiological mechanisms, and possibly genetics. Improved HF survival may increase the risk of cancer due to a competing risk. Whether the incidence of cancer has increased over time in patients with HF as survival has improved is unclear. Therefore, temporal trends of new onset cancer in HF patients between 1997 and 2016 were investigated.
Using Danish nationwide registers, 103 711 individuals alive, free of cancer, and aged 30–80 years 1 year after HF diagnosis (index date) were included between 1 January 1997 and 31 December 2016. A five-year incidence rate of cancer for each year after index date was calculated. The median age and proportion of women at the index date decreased with advancing calendar time [1997–2001: 70.3 interquartile range (Q1–Q3 62.5–75.7), 60.9% men; 2012–16: 67.6 (59.2–73.8), 67.5% men]. The five-year incidence rate of cancer was 20.9 and 20.2 per 1,000 person-years in 1997 and 2016, respectively. In a multivariable Cox regression model, the hazard rates between index years 1997 (reference) and 2016 were not significantly different [hazard ratio 1.09 (0.97–1.23)]. The five-year absolute risk of cancer did not change with advancing calendar year, going from 9.0% (1997–2001) to 9.0% (2012–16). Five-year cumulative incidence of survival for HF patients increased with advancing calendar year, going from 55.9% (1997–2001) to 74.3% (2012–2016).
Although cancer rates during 1997–2016 have remained stable within 1–6 years after the HF diagnosis, long-term survival following a HF diagnosis has increased significantly.
Analysis of 19 years of follow-up data from the Danish registries reveals that cancer rates during 1997–2016 have remained stable within 1–6 years after the HF diagnosis, and long-term survival following a HF diagnosis has increased significantly. HF, heart failure.
See the editorial comment for this article ‘Cancer is a comorbidity of heart failure’, by P. Ameri et al., https://doi.org/10.1093/eurheartj/ehac710.
Introduction
Worldwide, heart failure (HF) and cancer are leading causes of death, although there have been major improvements in the treatment of both diseases.1 With improving survival, many HF patients may die from non-cardiac causes such as cancer. Conversely, the occurrence of HF among patients treated for cancer is an increasingly common problem that has contributed to the emergence of the sub-speciality of cardio-oncology.2,3
HF has been hypothesized to predispose to cancer as a result of the ‘common soil’ of chronic inflammation and neurohormonal activation and shared risk factors (i.e. age, type 2 diabetes mellitus, hypertension, obesity, alcohol, and tobacco).4 Additionally, recent experimental studies found that HF stimulates tumour growth in animal models.5–7 There is clinical evidence suggesting that the incidence of cancer in patients with HF might be higher than in the background population,8–10 although this could simply be the result of improved longevity among these patients due to a reduction in cardiovascular death,11 a growing burden of comorbidities in patients with HF,12 or some combination of these. However, the majority of studies examining cancer and HF address the impact of cancer and its treatment on HF.13–16
Therefore, we investigated temporal trends in the incidence of cancer among patients with HF in Denmark between 1997 and 2016. This was done by studying the development of cancer in all newly presenting cases of HF who survived at least 1 year after the diagnosis of HF and who had no history of cancer within the preceding 10 years. We hypothesized that the incidence of cancer in patients with HF had increased over time due to an improved prognosis of HF, leading to longer life and increased risk of cancer.
Methods
Data sources
All Danish residents receive a personal and permanent civil registration number at birth or immigration that enables a link between the Danish nationwide registers. The data for this study were obtained from the Danish Civil Registration System registry (including information about sex, date of birth, immigration, emigration, and vital status), the Danish National Patient Registry [including discharge diagnoses coded according to the International Classification of Diseases (ICD)-8 from 1977–1993, the ICD-10 system since 1994, and the NOMESCO Classification of Surgical Procedures (NCSP) codes], the Danish National Prescription Registry [including claimed prescriptions since 1995 according to Anatomical Therapeutic Classification (ATC)], and the Danish National Causes of Death Registry (including the date of death). All registries have been validated previously.17–20
Data availability statement
Data obtained through the nationwide registers in Denmark can be made available only through research on Danish servers hosted in highly protected research environments where researchers can be granted access and permission with encrypted person identification. Access to raw data can be gained only through collaboration with the authors or other Danish institutions that already have been granted access. Please contact the first author with any questions on data access.
Study population
The index date was defined as 1 year following a first diagnosis of HF defined as a diagnosis of HF from following an ambulatory episode of care or a hospital overnight stay, whichever came first (ICD codes DI42, DI50, DJ81). The index date was set to 1 year after HF diagnosis to prevent possible surveillance bias. The study population comprised all patients with a first HF diagnosis in Denmark aged 30–80 years between 1 January 1997 and 31 December 2016, alive, and not emigrated at the index date (Figure 1). Patients with a diagnosis of cancer (ICD codes from outpatient visits and hospital overnight stays: DC00-DC96 (excluding basal cell cancer with the diagnosis code DC44)) within 10 years before the index date were excluded. All patients had 5 years of potential follow-up.
Study design. The marked area from the timeline named ‘Surveillance’ marks the first year after diagnosis of heart failure. Index is 1 year after heart failure diagnosis.
Comorbidities and co-medication
Comorbidities [ischaemic heart disease (IHD), atrial fibrillation, stroke, and chronic obstructive pulmonary disease (COPD)], medication [anticoagulants, antihypertensives, antidiabetes drugs, aspirin, adenosine diphosphate (ADP) receptor inhibitors, and cholesterol-lowering agents], and device therapy (including implantable cardioverter-defibrillator and pacemaker) were identified based on ICD, NCSP, and ATC codes for individuals before the index date. Hypertension was defined as treatment with at least two types of antihypertensive drugs within 180 days before the index date.21 Medications were defined as dispensed prescriptions within 180 days before the index date. The ICD, ATC, and NCSP codes are reported in Supplementary material online, Table S2.
Outcomes
The primary outcome was incident cancer of any type (ICD codes DC00-DC96, excluding non-melanoma skin cancer with the diagnosis code DC44) reported as the first discharge diagnosis of cancer from ambulatory care or hospital overnight stays in the National Patient Registry. Secondary outcome includes all-cause death.
Statistics
All patients were followed from the index date until a cancer diagnosis, death, emigration, or 5 years following the index date, whichever came first. Population characteristics were presented at the index date as medians with 25th and 75th percentiles for continuous variables and as counts with percentages for categorical variables. We calculated crude 5-year incidence rates (IRs: events/1000 person-years) of cancer according to each year at index (from 1997 to 2016 in 1-year bands). A multivariable Cox proportional-hazards model adjusted for sex, year, IHD, COPD, loop diuretics, and antidiabetes drugs was used to assess the annual change in the rate of cancer compared to the rate of cancer in patients with an index date in 1997. The hazards were smoothed using kernel-based methods.22 We calculated crude 5-year IRs (events/1000 person-years) of cancer according to each year at index (from 1997 to 2016 in 1-year bands) age (<60/≥60) and IHD (yes/no). We estimated the number of different cancer types over time that patients were diagnosed with during follow-up (from 1997 to 2016). We estimated the 5-year cumulative incidence of all-cause death for HF patients according to the year at index (1997–2001, 2002–06, 2007–11, 2012–16) using the Kaplan–Meier estimator. We estimated the 5-year cumulative incidence of cancer for HF patients according to the year at index (1997–2001, 2002–06, 2007–11, 2012–16) using the Aalen–Johansen estimator. We calculated the 5-year risk of death from cardiovascular and non-cardiovascular diseases during 1997–2016.
In supplementary analysis, we calculated crude 5-year IRs (events/1000 person-years) of cancer according to each year at index (from 1997 to 2016 in 1-year bands) age (<60/≥60) and IHD (yes/no). Second, we calculated the cumulative incidence of cancer and death over a 10-year period in groups defined by age (<60/≥60) at index using the Aalen–Johansen estimator. Third, for a matched population consisting of Danish patients without HF, we calculated crude 5-year IRs (events/1000 person-years) of cancer according to each year at index (from 1997 to 2016 in 1-year bands). Using exposure density matching, the matched population comprised all Danish patients aged 30–80 years between 1 January 1997 and 31 December 2016, alive, without HF, cancer, and not emigrated at the index date (Supplementary material online, Figure S3A–C). Fourth, we calculated crude 5-year IRs (events/1000 person-years) of cancer according to time from a diagnosis of HF in groups defined by age at HF diagnosis (<60/≥60) and IHD (yes/no). Fifth, we calculated crude and standardized 5-year risks of cancer according to each year at index (from 1997 to 2016) based on a cause-specific Cox model incorporating the competing risk of death (supplementary analysis). Adjusted for sex, year, IHD, COPD, loop diuretics, and antidiabetes drugs, the standardized risks were according to the baseline hazard rate and the hazard ratios fitted in 1997. Sixth, we calculated the 5-year absolute risk of cancer for HF patients since HF date (1996–2000, 2001–05, 2006–10, 2011–15). Seventh, we calculated the 5-year risk of death from cancer and HF during 1997–2016. Lastly, we calculated the rate of different cancer types according to the year at index in groups defined by sex in HF patients during 1997 to 2016. We chose to set the index date to 1 year after HF diagnosis to avoid possible surveillance bias based on earlier data.4 The level of statistical significance was set at 5%. All statistical analyses were completed using RStudio version 1.3.1093.23
Ethics
In Denmark, de-identified register-based studies that are conducted for the sole purpose of statistics and scientific research do not require ethical approval or informed consent by law. However, the study is approved by the data responsible institute (Capital Region of Denmark—approval number P-2019-191) in accordance with the General Data Protection Regulation.
Results
Baseline characteristics
Between 1997 and 2016, we included 103 711 patients aged 30–80 years who were cancer-free within 10 years before the index date (Figure 2). At index, there were a higher proportion of men in all calendar year groups, and the proportion of men increased with advancing calendar year. Similarly, the proportion of patients with IHD and atrial fibrillation, as well as medical treatment with statins, anticoagulants, beta-blockers, antidiabetes drugs, renin-angiotensin system inhibitors, ADP-receptor inhibitors, and device therapy, increased with advancing calendar year. In contrast, the median age decreased with advancing calendar year (Table 1).
Flowchart: inclusion and exclusion criteria for the study population.
Incidence rates and hazard ratios of cancer during 1997–2016
The 5-year incidence rate of cancer was 20.9 and 20.2 per 1,000 person-years in 1997 and 2016, respectively (Figure 3A). Similarly, when adjusting for sex, year, IHD, COPD, loop diuretics, and antidiabetes drugs, there was a small, non-significant increase in the 5-year hazard ratio of cancer when compared to the 5-year hazard ratio from the index year 1997. The hazard rates between index years 1997 (reference) and 2016 were not significantly different (hazard ratio 1.09 [0.97–1.23]) (Figure 3B). The same was observed for men (Figure 3B-E) and for women (Figure 3C-F). This was also seen when stratifying according to age and IHD (Supplementary material online, Figure S1A and B). We observed that the highest IR of cancer was during the first year after an HF diagnosis regardless of age group and IHD (Figure 4). We observed that HF patients >60 years with or without IHD had higher IR of cancer compared to HF patients aged <60 years (Supplementary material online, Figure S1). The observed high IR may be explained by surveillance, and it is probably not explained by exposure to HF for which reason we chose +1 year from the HF diagnosis as the index date for our statistical temporal trend analyses. Gastrointestinal, pulmonary and breast cancer were the most common subtypes of cancer during follow-up (Figure 5).
(A) Shows the temporal trends in cancer incidence after new-onset heart failure for Danish patients during 1997–2016. (B) Shows the temporal trends in cancer incidence after new-onset heart failure for Danish male patients during 1997–2016. (C) Shows the temporal trends in cancer incidence after new-onset heart failure for Danish female patients during 1997–2016. (D) Shows the temporal trends in hazard rates of cancer after new-onset heart failure for Danish patients during 1997–2016. (E) Shows the temporal trends in cancer incidence after new-onset heart failure for Danish male patients during 1997–2016. (F) Shows the temporal trends in hazard rates of cancer after new-onset heart failure for Danish female patients during 1997–2016. The red dashed line represents the reference index year (1997).
Incidence of cancer according to time since heart failure diagnosis in groups defined by age and ischaemic heart disease.
Number of different cancer types developed within 5 years after the index date during 1997–2016.
Risk of cancer and death
The 5-year absolute risk of cancer for HF patients was constant over time [1997–2001 9.0% (8.9–9.0); 2002–06 9.0% (8.9–9.0); 2007–11 9.0% (8.9–9.0); 2012–16 9.0% (8.9–9.0); Figure 6]. Five-year cumulative incidence of survival for HF patients increased with advancing calendar year [1997–2001 55.9% (55.3–56.5); 2002–06 64.9% (64.3–65.6); 2007–11 70.4% (69.8–71.1); 2012–16 74.3% (73.8–74.7); Figure 7]. The 5-year risk of death from cardiovascular diseases declined substantially, whereas 5-year risk of death from non-cardiovascular diseases slightly decreased during 1997 to 2016 (Figure 8). The difference in absolute risk of death was higher for HF patients >60 years compared to HF patients aged <60 years (Supplementary material online, Figure S3A and B). HF patients >60 years had a higher absolute 10-year risk of cancer [53.5% (53.1–53.8)) and death (15.6% (15.3–15.9)] compared to HF patients <60 years [cancer 24.9% (24.3–25.5); death 9.01% (8.63–9.38); Supplementary material online, Figure S2]. There was not any marked change in cancer incidence or incidence of death at 5–10 years for younger and older patients. The 5-year incidence of cancer in Danish patients without HF matched for age and sex declined between 1997 and 2016 (Supplementary material online, Figure S3). The crude 5-year risk of cancer in Danish patients with HF during follow-up was relatively constant across all calendar years, whereas we observed a slight increase in the standardized 5-year risk of cancer over calendar time (Supplementary material online, Figure S4). The 5-year absolute risk of cancer for HF patients since the HF date did not change according to calendar time groups (Supplementary material online, Figure S5). Proportionally, the 5-year risk of non-cardiovascular compared to cardiovascular death increased with calendar time. The 5-year risk of death from cancer slightly declined over time, whereas the risk of death caused by HF declined substantially during 1997 to 2016 (Supplementary material online, Figure S6). We observed no clear change in the rate of various cancer forms in women, while in men the rate of pulmonary cancers may have decreased slightly, and the rate of male gonad cancers may have increased with time (Supplementary material online, Figure S7).
Absolute risk of developing cancer for Danish heart failure patients and comparing following time periods 1997–2001, 2002–06, 2007–11, and 2012–16.
Temporal trends in survival for Danish heart failure patients during 1997–2016 and comparing following time periods 1997–01, 2002–06, 2007–11, and 2012–16.
Five-year risk of cardiovascular death and non-cardiovascular death for Danish HF patients during 1997–2016.
Discussion
Main findings
In this nationwide study, we investigated the temporal trends of cancer among cancer-free Danish HF patients <80 years of age in the period 1997 to 2016. We confirmed that long-term survival after HF has improved substantially over the years, but reassuringly, we observed no change in the cancer rates over the same period, translating into an increased survival for cancer-free HF patients (Structured Graphical Abstract).
Temporal trends in incidence of cancer
We observed no change in the incidence of cancer among HF patients over time, and we did not observe a change in the incidence of cancer among patients without HF. We do not consider our findings contrasting with previous studies that observed an increased risk of cancer for HF patients compared to the background population,8,10,24 which is probably explained by an increased burden of comorbidities and shared risk factors.25 However, in contrast to our findings, a recent study by Bertero et al.26 found an increased risk of cancer in HF patients compared to matched controls. The study focused on comparing cancer incidence in patients with and without HF and was not investigating temporal trends but had inclusion criteria similar to the present study. Patients were included if they had been cancer free for 3 years and all patients had at least 5 years of follow-up. Similarly, a Danish study found an increased rate of cancer incidence in patients with HF compared to patients without HF.8 Though, the study by Banke et al.8 investigated a highly selected population of HF patients, which consisted of a cohort of outpatient HFrEF [predominantly left ventricular ejection fraction (LVEF) <45%] patients referred for optimization in guideline-directed therapies with unclear duration of their HF disease. In contrast, our study investigated a more unselected cohort of incident HF patients. Further, in our study, surveillance bias was managed, and start of follow-up was chosen to be 1 year after the HF diagnosis (Figure 4). However, other studies have reported an increased incidence of cancer among patients with cardiovascular disease,6 and Koelwyn et al.,27 that investigated former breast cancer patients, observed an increased risk of recurrence of breast cancer after a cardiovascular event.
We observed that the risk of cancer was highest during the first year after being diagnosed with HF regardless of age group and IHD. The strong association between HF and cancer within the first year after the diagnosis of HF may be due to greater medical surveillance. It may illustrate that a diagnosis of HF leads to more contact with the healthcare system and thereby an increased risk of finding occult cancers. In our study, we have, therefore, set the index date to 1 year after HF diagnosis to avoid surveillance bias and identification of cancers unrelated to exposure of HF. In accordance with this, a study from the Netherlands showed that early detection of cancer may be caused by a visit to an outpatient clinic or an admission to the emergency department where a cancer could be detected by the physical, biochemical, or radiological examinations.4
Temporal trends in survival
The temporal trends in survival of HF patients were improved over time. The improved survival was due to reduced mortality risk (Figure 7), probably explained by implementation of HF therapy.28 The 5-year risk of cancer (≈ 9%) remained unchanged over time (Figure 6). In randomized clinical trials, cancer death may account for 6% to 14% of all deaths without a temporal trend.29 Our analyses based on real-life administrative data in patients without a history of cancer within 10 years do therefore not differ from results obtained in well described patients from randomized clinical trials. In the study by Bertero et al.,26 they observed that HF patients had higher cancer mortality compared with matched controls. Differences in study design and populations may explain these differences.
It is important to differentiate studies primarily focusing on the development of cardiovascular disease in cancer survivors30,31 and others which take the opposite approach to examine cancer incidence in patients with baseline cardiovascular disease or other comorbidity.32 A prior analysis focused on temporal trends in the incidence of death and hospitalization in patients with HF in the UK and assessed these in relation to several baseline comorbidities. It was observed that the prevalence of cancer at baseline was increasing, and mortality rates due to cancer were constant or slightly decreasing from 2002 to 2013. The decline in overall mortality risk was only modest in the whole UK cohort, but it decreased for HF patients <80 years of age in accordance with our results.32 Noteworthy, Stoltzfus et al.30 and Sturgeon et al.31 reported in another study that patients with different cancer types developed significant cardiovascular disease over time in accordance with the baseline data from the study by Conrad et al.32 Cardiovascular disease in cancer survivors seems, therefore, to be a larger clinical challenge than the development of cancer in patients with HF taking our analyses into consideration.
Sturgeon et al.31 carried out a study looking at the opposite of ours. They found that in a cancer population the highest risk of cardiovascular death was during the first year following a cancer diagnosis, which also could be caused by surveillance. Secondly, the incidence of death was constant for a long period of time and thereafter patients who survived their cancer then developed cardiovascular diseases and died from it.31 In that context, it should be noted that we did not observe an increase in the risk of cancer from year 5–10 years in our study (Supplementary material online, Figure S3A and B). Oppositely, the high risk of cardiovascular death during the first year in the study by Sturgeon et al.31 could also be caused by cardiotoxicity from cancer treatments. Furthermore, other studies have shown that a multitude of cardiac biomarkers were already raised at cancer diagnosis even before treatment, which can be interpreted as cardiovascular disease triggered by the cancer inflammation.33 Similarly, to the study by Sturgeon et al.,31 a recent American temporal trend study found temporal increases in the proportion of HF deaths across all cancer types. Raisi-Estabragh et al.34 found that HF deaths were most common among patients with breast or haematological cancers, but the greatest increase in HF death was seen among patients with lung cancer.
Interestingly, we observed a decrease in absolute mortality risk, but the number of patients who were diagnosed with cancer during follow-up before they died was constant. Whether the patients with both a HF and a cancer diagnosis died from cancer or cardiovascular reasons cannot be deduced from our administrative data, but the lack of increase in cancer diagnoses in patients who died support that cancer-related death has not increased over time in HF patients in Denmark. Previous Danish studies with other designs have also observed a decline in overall mortality risk over time,28,35 in particular in younger HF patients. Exclusion of HF patients with a history of cancer to investigate the incidence of cancer did not result in another trend.
Clinical perspectives
The present study has several clinical implications. First, since the incidence of cancer is stable and HF patients have increased survival, cardiologists will more frequently be met by HF patients with symptoms of cancer and HF patients who get cancer diagnosed.16 Therefore, in absolute numbers, diagnosed cancer represent an emerging challenge, and the proportion of patients who die from cardiovascular causes has declined accordingly (Figure 7). Second, physicians should be aware of cancer symptoms, particularly in elderly patients who have a high risk for developing this disease. Survival has improved which is an important message to HF patients, and their families frequently experience an increased concern receiving a diagnosis of a chronic disease. Finally, based on our data, it may be speculated that HF patients <80 years of age who initially survive without a history of cancer within the preceding 10 years are a phenotype where mode of death is likely cardiovascular reasons or infections32 e.g. pneumonia36 or influenza37. To improve patient selection for future randomized clinical trials in HF, this should be kept in mind.
Methodological considerations
The major strengths of this study are that there is limited loss to follow-up and little risk of selection bias due to the Danish registries and the large sample size of 103 711 persons. Furthermore, the large cohort with 5 years of follow-up for all 19 calendar year groups gave us a great possibility to investigate temporal trends in the risk of developing cancer after getting diagnosed with HF. To compare our results from the study population, we made a matched population. As seen for Danish HF patients, we did not observe any changes in the incidence of cancer for Danish patients without HF (Supplementary material online, Figure S3). Given the study design, there are a few limitations and caveats that need to be mentioned. To our knowledge, there have not been any studies with the same outcome as our study is investigating. We chose to set the index date to 1 year after diagnosis of HF and only include HF patients without cancer within the preceding 10 years. This was done to avoid inclusion of previously diagnosed cancers and to avoid possible surveillance bias. Setting the index date to 1 year after diagnosis of HF may be a limitation to the study since the possibility that onset of HF and simultaneously development of cancer caused by chronic inflammation and immune mechanisms cannot be ruled out. Furthermore, undetected tumour growth may precipitate to the development of HF and then leading to the detection and diagnosis of cancer. We also chose to investigate patients <80 years of age and follow them for 5 years. HF patients above 80 years of age were, therefore, not included since investigation of cancer in these patients may be performed less systematically due to advanced biological age and frailty, and clinical practice may also have changed over time. Inclusion of patients >80 years might, therefore, have biased the association between HF and cancer. With the Danish registries, we do not have access to information about smoking status, blood samples, blood pressure, and LVEF. Risk of unmeasured and residual confounding can never be excluded in studies based on data from administrative registries, because clinical variables like use of tobacco, alcohol consumption, body mass index, natriuretic peptides, LVEF, and functional class are not accessible. The Danish population consists mainly of White individuals; therefore, our results may not be generalizable to other groups. It may be argued that we followed the patients for too short a period. For that reason, we did a sensitivity analysis in the whole cohort and extended follow-up to 10 years in age-stratified analyses. Here, we observed that the risk of cancer did not increase abruptly in the period from 5–10 years. In the current study, the incidence of different cancer forms over time was examined, however, not the survival of specific cancer forms. Furthermore, the diagnoses on HF and cancer used in this study rely on discharge diagnoses made in hospitals. In theory, cases of HF and cancer may have been overlooked, or wrong diagnoses may have been given (misclassification). Further, the aetiology of HF may have changed over time as well as HF may have been defined and diagnosed differently in 1997 and 2016. Due to improvements in examinations of HF, it might be that e.g. a HF diagnosis given in 1997 was not as exact as in 2016. Accordingly, patients with e.g. dyspnoea may have been classified as HF in 1997 but may have had other aetiologies. The misclassifications are non-systematic and do not influence the validity of the data from the Danish registries. Furthermore, the HF and cancer diagnoses both have a high diagnostic accuracy in Danish administrative registries.38–41
Conclusions
We did not observe an increase in the incidence of cancer for Danish HF patients <80 years of age during 1997–2016, who had survived at least 1 year after the HF diagnosis and without a previous cancer within the preceding 10 years, despite an increase in survival over time. The risk of cancer was highest during the first year after being diagnosed with HF regardless of the age group. The temporal trends in survival of HF patients were improved over time, whereas the 5-year risk of cancer remained unchanged over time.
Population characteristics according to year at index
| Variable | 1997–2001 (n = 25 826) | 2002–06 (n = 27 812) | 2007–11 (n = 24 312) | 2012–16 (n = 25 761) | Total (n = 103 711) |
|---|---|---|---|---|---|
| Age (IQR) | 70.3 (62.5–75.7) | 69.1 (60.6–75.0) | 67.6 (59.6–74.1) | 67.6 (59.2–73.8) | 68.7 (60.4–74.8) |
| Men (%) | 15 720 (60.9) | 17 335 (62.3) | 16 056 (66.0) | 17 393 (67.5) | 66 504 (64.1) |
| RASi | 13 824 (53.5) | 18 058 (64.9) | 17 816 (73.3) | 19 163 (74.4) | 68 861 (66.4) |
| MRA | 4305 (16.7) | 7648 (27.5) | 5514 (22.7) | 7267 (28.2) | 24 734 (23.8) |
| Statin | 3988 (15.4) | 11 316 (40.7) | 14 166 (58.3) | 14 968 (58.1) | 44 438 (42.8) |
| Calcium channel blockers | 5518 (21.4) | 5229 (18.8) | 4717 (19.4) | 4584 (17.8) | 20 048 (19.3) |
| Thiazide diuretics | 3644 (14.1) | 4056 (14.6) | 2991 (12.3) | 2025 (7.9) | 12 716 (12.3) |
| Anticoagulants | 4746 (18.4) | 6345 (22.8) | 6204 (25.5) | 8365 (32.5) | 25 660 (24.7) |
| Aspirin | 11 148 (43.2) | 14 977 (53.9) | 14 197 (58.4) | 11 550 (44.8) | 51 872 (50.0) |
| ADP-receptor inhibitors | 122 (0.5) | 2398 (8.6) | 4013 (16.5) | 5601 (21.7) | 12 134 (11.7) |
| Beta blockers | 7404 (28.7) | 15 292 (55.0) | 16 794 (69.1) | 19 227 (74.6) | 58 717 (56.6) |
| Loop diuretics | 17 872 (69.2) | 16 677 (60.0) | 12 764 (52.5) | 11 278 (43.8) | 58 591 (56.5) |
| Anti-diabetes drugs | 3751 (14.5) | 4589 (16.5) | 4707 (19.4) | 5386 (20.9) | 18 433 (17.8) |
| Ischaemic heart disease | 13 113 (50.8) | 15 139 (54.4) | 13 609 (56.0) | 14 088 (54.7) | 55 949 (53.9) |
| Atrial fibrillation | 7232 (28.0) | 8352 (30.0) | 7806 (32.1) | 8947 (34.7) | 32 337 (31.2) |
| Stroke | 3458 (13.4) | 3764 (13.5) | 3170 (13.0) | 3481 (13.5) | 13 873 (13.4) |
| COPD | 5726 (22.2) | 6056 (21.8) | 4645 (19.1) | 4603 (17.9) | 21 030 (20.3) |
| Device therapy | 105 (0.4) | 875 (3.1) | 2089 (8.6) | 2882 (11.2) | 5951 (5.7) |
| Variable | 1997–2001 (n = 25 826) | 2002–06 (n = 27 812) | 2007–11 (n = 24 312) | 2012–16 (n = 25 761) | Total (n = 103 711) |
|---|---|---|---|---|---|
| Age (IQR) | 70.3 (62.5–75.7) | 69.1 (60.6–75.0) | 67.6 (59.6–74.1) | 67.6 (59.2–73.8) | 68.7 (60.4–74.8) |
| Men (%) | 15 720 (60.9) | 17 335 (62.3) | 16 056 (66.0) | 17 393 (67.5) | 66 504 (64.1) |
| RASi | 13 824 (53.5) | 18 058 (64.9) | 17 816 (73.3) | 19 163 (74.4) | 68 861 (66.4) |
| MRA | 4305 (16.7) | 7648 (27.5) | 5514 (22.7) | 7267 (28.2) | 24 734 (23.8) |
| Statin | 3988 (15.4) | 11 316 (40.7) | 14 166 (58.3) | 14 968 (58.1) | 44 438 (42.8) |
| Calcium channel blockers | 5518 (21.4) | 5229 (18.8) | 4717 (19.4) | 4584 (17.8) | 20 048 (19.3) |
| Thiazide diuretics | 3644 (14.1) | 4056 (14.6) | 2991 (12.3) | 2025 (7.9) | 12 716 (12.3) |
| Anticoagulants | 4746 (18.4) | 6345 (22.8) | 6204 (25.5) | 8365 (32.5) | 25 660 (24.7) |
| Aspirin | 11 148 (43.2) | 14 977 (53.9) | 14 197 (58.4) | 11 550 (44.8) | 51 872 (50.0) |
| ADP-receptor inhibitors | 122 (0.5) | 2398 (8.6) | 4013 (16.5) | 5601 (21.7) | 12 134 (11.7) |
| Beta blockers | 7404 (28.7) | 15 292 (55.0) | 16 794 (69.1) | 19 227 (74.6) | 58 717 (56.6) |
| Loop diuretics | 17 872 (69.2) | 16 677 (60.0) | 12 764 (52.5) | 11 278 (43.8) | 58 591 (56.5) |
| Anti-diabetes drugs | 3751 (14.5) | 4589 (16.5) | 4707 (19.4) | 5386 (20.9) | 18 433 (17.8) |
| Ischaemic heart disease | 13 113 (50.8) | 15 139 (54.4) | 13 609 (56.0) | 14 088 (54.7) | 55 949 (53.9) |
| Atrial fibrillation | 7232 (28.0) | 8352 (30.0) | 7806 (32.1) | 8947 (34.7) | 32 337 (31.2) |
| Stroke | 3458 (13.4) | 3764 (13.5) | 3170 (13.0) | 3481 (13.5) | 13 873 (13.4) |
| COPD | 5726 (22.2) | 6056 (21.8) | 4645 (19.1) | 4603 (17.9) | 21 030 (20.3) |
| Device therapy | 105 (0.4) | 875 (3.1) | 2089 (8.6) | 2882 (11.2) | 5951 (5.7) |
ADP, adenosine diphosphate receptor inhibitors; COPD, chronic obstructive pulmonary disease; IQR, interquartile range; MRA, mineralocorticoid receptor antagonists; RASi, renin-angiotensin system inhibitors.
Population characteristics according to year at index
| Variable | 1997–2001 (n = 25 826) | 2002–06 (n = 27 812) | 2007–11 (n = 24 312) | 2012–16 (n = 25 761) | Total (n = 103 711) |
|---|---|---|---|---|---|
| Age (IQR) | 70.3 (62.5–75.7) | 69.1 (60.6–75.0) | 67.6 (59.6–74.1) | 67.6 (59.2–73.8) | 68.7 (60.4–74.8) |
| Men (%) | 15 720 (60.9) | 17 335 (62.3) | 16 056 (66.0) | 17 393 (67.5) | 66 504 (64.1) |
| RASi | 13 824 (53.5) | 18 058 (64.9) | 17 816 (73.3) | 19 163 (74.4) | 68 861 (66.4) |
| MRA | 4305 (16.7) | 7648 (27.5) | 5514 (22.7) | 7267 (28.2) | 24 734 (23.8) |
| Statin | 3988 (15.4) | 11 316 (40.7) | 14 166 (58.3) | 14 968 (58.1) | 44 438 (42.8) |
| Calcium channel blockers | 5518 (21.4) | 5229 (18.8) | 4717 (19.4) | 4584 (17.8) | 20 048 (19.3) |
| Thiazide diuretics | 3644 (14.1) | 4056 (14.6) | 2991 (12.3) | 2025 (7.9) | 12 716 (12.3) |
| Anticoagulants | 4746 (18.4) | 6345 (22.8) | 6204 (25.5) | 8365 (32.5) | 25 660 (24.7) |
| Aspirin | 11 148 (43.2) | 14 977 (53.9) | 14 197 (58.4) | 11 550 (44.8) | 51 872 (50.0) |
| ADP-receptor inhibitors | 122 (0.5) | 2398 (8.6) | 4013 (16.5) | 5601 (21.7) | 12 134 (11.7) |
| Beta blockers | 7404 (28.7) | 15 292 (55.0) | 16 794 (69.1) | 19 227 (74.6) | 58 717 (56.6) |
| Loop diuretics | 17 872 (69.2) | 16 677 (60.0) | 12 764 (52.5) | 11 278 (43.8) | 58 591 (56.5) |
| Anti-diabetes drugs | 3751 (14.5) | 4589 (16.5) | 4707 (19.4) | 5386 (20.9) | 18 433 (17.8) |
| Ischaemic heart disease | 13 113 (50.8) | 15 139 (54.4) | 13 609 (56.0) | 14 088 (54.7) | 55 949 (53.9) |
| Atrial fibrillation | 7232 (28.0) | 8352 (30.0) | 7806 (32.1) | 8947 (34.7) | 32 337 (31.2) |
| Stroke | 3458 (13.4) | 3764 (13.5) | 3170 (13.0) | 3481 (13.5) | 13 873 (13.4) |
| COPD | 5726 (22.2) | 6056 (21.8) | 4645 (19.1) | 4603 (17.9) | 21 030 (20.3) |
| Device therapy | 105 (0.4) | 875 (3.1) | 2089 (8.6) | 2882 (11.2) | 5951 (5.7) |
| Variable | 1997–2001 (n = 25 826) | 2002–06 (n = 27 812) | 2007–11 (n = 24 312) | 2012–16 (n = 25 761) | Total (n = 103 711) |
|---|---|---|---|---|---|
| Age (IQR) | 70.3 (62.5–75.7) | 69.1 (60.6–75.0) | 67.6 (59.6–74.1) | 67.6 (59.2–73.8) | 68.7 (60.4–74.8) |
| Men (%) | 15 720 (60.9) | 17 335 (62.3) | 16 056 (66.0) | 17 393 (67.5) | 66 504 (64.1) |
| RASi | 13 824 (53.5) | 18 058 (64.9) | 17 816 (73.3) | 19 163 (74.4) | 68 861 (66.4) |
| MRA | 4305 (16.7) | 7648 (27.5) | 5514 (22.7) | 7267 (28.2) | 24 734 (23.8) |
| Statin | 3988 (15.4) | 11 316 (40.7) | 14 166 (58.3) | 14 968 (58.1) | 44 438 (42.8) |
| Calcium channel blockers | 5518 (21.4) | 5229 (18.8) | 4717 (19.4) | 4584 (17.8) | 20 048 (19.3) |
| Thiazide diuretics | 3644 (14.1) | 4056 (14.6) | 2991 (12.3) | 2025 (7.9) | 12 716 (12.3) |
| Anticoagulants | 4746 (18.4) | 6345 (22.8) | 6204 (25.5) | 8365 (32.5) | 25 660 (24.7) |
| Aspirin | 11 148 (43.2) | 14 977 (53.9) | 14 197 (58.4) | 11 550 (44.8) | 51 872 (50.0) |
| ADP-receptor inhibitors | 122 (0.5) | 2398 (8.6) | 4013 (16.5) | 5601 (21.7) | 12 134 (11.7) |
| Beta blockers | 7404 (28.7) | 15 292 (55.0) | 16 794 (69.1) | 19 227 (74.6) | 58 717 (56.6) |
| Loop diuretics | 17 872 (69.2) | 16 677 (60.0) | 12 764 (52.5) | 11 278 (43.8) | 58 591 (56.5) |
| Anti-diabetes drugs | 3751 (14.5) | 4589 (16.5) | 4707 (19.4) | 5386 (20.9) | 18 433 (17.8) |
| Ischaemic heart disease | 13 113 (50.8) | 15 139 (54.4) | 13 609 (56.0) | 14 088 (54.7) | 55 949 (53.9) |
| Atrial fibrillation | 7232 (28.0) | 8352 (30.0) | 7806 (32.1) | 8947 (34.7) | 32 337 (31.2) |
| Stroke | 3458 (13.4) | 3764 (13.5) | 3170 (13.0) | 3481 (13.5) | 13 873 (13.4) |
| COPD | 5726 (22.2) | 6056 (21.8) | 4645 (19.1) | 4603 (17.9) | 21 030 (20.3) |
| Device therapy | 105 (0.4) | 875 (3.1) | 2089 (8.6) | 2882 (11.2) | 5951 (5.7) |
ADP, adenosine diphosphate receptor inhibitors; COPD, chronic obstructive pulmonary disease; IQR, interquartile range; MRA, mineralocorticoid receptor antagonists; RASi, renin-angiotensin system inhibitors.
Supplementary data
Supplementary data is available at European Heart Journal online.
Funding
Department of Cardiology, Herlev-Gentofte Hospital, University of Copenhagen
References
Author notes
Conflict of interest: C.H.G. reports personal fees from lectures honorarium from Astra Zeneca unrelated to the work in question; C.T.-P. reports grants from Bayer and Novo Nordisk unrelated to the work in question; J.V.M. reports grants from Abbott, Alkem Metabolics, Eris Lifesciences, Hikma, Lupin, Sun Pharmaceuticals, Medscape/Heart.Org, ProAdWise Communications, Radcliffe Cardiology, Servier, the Corpus unrelated to the work in question; N.N.L. reports grants from Roche Diagnostics, Boehringer Ingelheim, AstraZeneca, NIHR and British Heart Foundation. Further, reports lecturer fees from Roche, Pfizer, and Novartis and consulting fees from AstraZeneca. All unrelated to the work in question; L.K. reports personal fees from speakers honorarium from Novo Nordisk, Novartis, AstraZeneca and Boehringer Ingelheim outside the submitted work; M.S. reports personal fees from lectures honorarium from Novo Nordisk, Novartis, AstraZeneca, and Boehringer Ingelheim outside the submitted work.
