Circulation: Cardiovascular Quality and Outcomes

ORIGINAL ARTICLE

Type 2 Diabetes Mellitus and Impact of Heart

Failure on Prognosis Compared to Other
Cardiovascular Diseases
A Nationwide Study

BACKGROUND: Heart failure (HF) in patients with type 2 diabetes Bochra Zareini , MD
mellitus (T2D) has received growing attention. We examined the effect Paul Blanche, PhD
of HF development on prognosis compared with other cardiovascular or Maria D’Souza, PhD
renal diagnoses in patients with T2D. Mariam Elmegaard Malik,
MD
METHODS AND RESULTS: Patients with new T2D diagnosis patients Caroline Holm Nørgaard,
were identified between 1998 and 2015 through Danish nationwide MD
registers. At yearly landmark timepoints after T2D diagnosis, we estimated Christian Selmer, MD, PhD
the 5-year risks of death, 5-year risk ratios, and decrease in lifespan Gunnar Gislason, MD, PhD
within 5 years associated with the development of HF, ischemic heart Søren Lund Kristensen,
disease, stroke, peripheral artery disease, and chronic kidney disease. MD, PhD
A total of 153 403 patients with newly diagnosed T2D were followed Lars Køber, MD, DMSc
for a median of 9.7 years (interquartile range, 5.8–13.9) during which Christian Torp-Pedersen,
Downloaded from http://ahajournals.org by on July 11, 2020

MD, DMSc
48 087 patients died. The 5-year risk ratio of death associated with HF
Morten Schou, MD, PhD
development 5 years after T2D diagnosis was 3 times higher (CI, 2.9–3.1)
Morten Lamberts, MD,
than patients free of diagnoses (CI, 2.9–3.1). Five-year risk ratios were PhD
lower for ischemic heart disease (1.3 [1.3–1.4]), stroke (2.2 [2.1–2.2]),
chronic kidney disease (1.7 [1.7–1.8]), and peripheral artery disease (2.3
[2.3–2.4]). The corresponding decrease in lifespan within 5 years when
compared with patients free of diagnoses (in months) was HF 11.7 (11.6–
11.8), ischemic heart disease 1.6 (1.5–1.7), stroke 6.4 (6.3–6.5), chronic
kidney disease 4.4 (4.3–4.6), and peripheral artery disease 6.9 (6.8–7.0).
HF in combination with any other diagnosis imposed the greatest risk
of death and decrease in life span compared with other combinations.
Supplemental analysis led to similar results when stratified according to
age, sex, and comorbidity status, and inclusion period.
CONCLUSIONS: HF development, at any year since T2D diagnosis, was
associated with the highest 5-year absolute and relative risk of death, and
decrease in lifespan within 5 years, when compared with development of
other cardiovascular or renal diagnoses.

Key Words:  chronic kidney disease

◼ epidemiology ◼ heart failure
◼ mortality ◼ myocardial infarction

© 2020 American Heart Association, Inc.

https://www.ahajournals.org/journal/
circoutcomes

Circ Cardiovasc Qual Outcomes. 2020;13:e006260. DOI: 10.1161/CIRCOUTCOMES.119.006260 July 2020 1

 Zareini et al; Heart Failure and Diabetes Mellitus

Data Sources
WHAT IS KNOWN In Denmark, every citizen is provided with a unique per-
sonal identification number given at birth or immigration
• Patients with type 2 diabetes mellitus have a
and used throughout the Civil Registration System, which
higher risk of developing cardiovascular disease
allows for cross-linkage of health and administrative data-
and chronic renal disease.
bases at the individual level and enables near-complete fol-
• Patients with type 2 diabetes mellitus enrolled in
low-up. In Denmark, equal access to the health care system
trials have higher rates of death following cardio-
is granted to every citizen, including primary and hospital
vascular or renal events.
care. Data were collected from 4 nationwide registers made
available through Statistics Denmark, a government-based
WHAT THE STUDY ADDS institution responsible for maintenance of multiple registers
• We describe the absolute risk of survival among after anonymization and encryption of the personal identifi-
real-life patients with type 2 diabetes mellitus who cation number. The Danish National Patient Registry entails
develop cardiovascular and renal disease. information on all hospital admissions from 1977 onwards.
• While heart failure is not the most frequent Each hospital contact is coded with a primary diagnosis
comorbid disease, it is the most fatal condition in and one or more secondary diagnoses according to The
a real-world cohort of patients with type 2 diabe- International Classification of Disease, Eighth Revision (ICD-
tes mellitus. 8) until 1993, and ICD-10 from 1994 onwards. Surgical
• This association is present regardless of when the procedures are coded using the Nordic Medico-Statistical
patients develop the cardiovascular or renal dis- Committee Classification of Surgical Procedures. The
ease after diagnosis of type 2 diabetes mellitus. Danish National Prescription Registry holds information
(dosage, dates, and Anatomic Therapeutic Chemical codes)
on all prescriptions dispensed from a pharmacy since 1995.

H
eart failure (HF) has in recent years been rec- The Danish Cause of Death Registry entails information on
ognized as an important clinical end point in date, cause, and place of death from 1970 onwards. The
Danish Civil registry holds information on age, gender, and
randomized clinical trials of patients with type
date of birth. The type of registry and data extracted from
2 diabetes mellitus (T2D), in particular, after the results each registry is described in Table I in the Data Supplement.
from randomized controlled trials of sodium-glucose-
2-transporter (SGLT2) inhibitors showed benefit on
Study Population
cardiovascular death and HF hospitalisation.1–3 Sev-
Downloaded from http://ahajournals.org by on July 11, 2020

We defined patients with incident T2D in patients above

eral observational studies have highlighted increased 18 years according to the following criterion: a first-time
mortality following myocardial infarction,4–7 stroke,8 redeemed prescription of a noninsulin antidiabetic drug.
chronic kidney disease (CKD),9,10 and HF11–14 in patients Patients were eligible for inclusion in the period between
with T2D. Interestingly, a post hoc analysis from the January 1, 1998, and December 31, 2015. We excluded
ALTITUDE (Aliskiren Trial in Type 2 Diabetes Using Car- patients with a prior diagnosis of HF, IHD, stroke, CKD, PAD,
dio-Renal End Points) suggested considerable mortality and gestational diabetes mellitus (Figure 1 and Table II in the
Data Supplement). Comorbidities were identified from prior
following a cardiovascular (including HF) or renal event
hospitalizations and ambulatory contacts up to 10 years
in patients with T2D.7 But, trial patients are often en- before the date of inclusion and continuously throughout
rolled with a long duration of diabetes mellitus and the follow-up for the landmark analyses. Information on
considered high-risk subjects, the risk profile following concomitant medical therapy was obtained from dispensed
a cardiovascular or renal diagnosis in real-life patients prescriptions as listed in the Danish National Prescription
with newly diagnosed T2D remains unknown.1,15 In ad- Registry and defined by at least one redeemed prescription
dition, little is known about the prognostic importance 6 months before the inclusion date. The following drugs
were recorded at inclusion: ACE (angiotensin-converting
in terms of mortality risk of different cardiovascular
enzyme) inhibitors, ARB (angiotensin II receptor blockers),
and renal diagnoses in patients with T2D. Our objec- calcium channel blockers, loop diuretics, thiazides, acetyl-
tive was to estimate, at every year since T2D diagnosis, salicylic acid, mineralocorticoid receptor antagonists, statins,
the absolute risk of death, the risk ratio (RR) of death, and β-blockers (Table II in the Data Supplement).
and the decrease in lifespan within 5 years following
the development of HF, ischemic heart disease (IHD), Definitions of Cardiovascular and Renal
stroke, CKD, and peripheral artery disease (PAD). Diagnoses
Cardiovascular and renal diagnoses in patients with T2D were
METHODS identified from hospital discharge records and ambulatory
contacts in the Danish National Patient Registry. A cardiovas-
Data cular or renal disease was defined as a diagnosis of HF, IHD,
The data upon which this analysis is done will not be made stroke, PAD, or CKD. The diagnoses used are listed in Table II
publicly available due to the sensitive nature of the data. in the Data Supplement.

Circ Cardiovasc Qual Outcomes. 2020;13:e006260. DOI: 10.1161/CIRCOUTCOMES.119.006260 July 2020 2

 Zareini et al; Heart Failure and Diabetes Mellitus

Figure 1. Flowchart of the study cohort

showing exclusion and inclusion of pa-
tients.
T2D indicates type 2 diabetes mellitus.

Outcome medication, and present comorbidities at inclusion and calcu-

The main outcome was all-cause death. Patients were fol- lated the 1-year risk of death following 1 or 2 cardiovascular
lowed until emigration, death, or end of the study period or renal diagnoses according to each landmark year.
(December 31, 2015). Analyses were performed using SAS (version 9.4 for
Windows, SAS Institute, NC) and R, especially the R packages
survRM2 and prodlim.18–20
Statistical Analysis
Baseline characteristics were described by use of proportions
for categorical variables and means and SD or medians and
Ethics
interquartile ranges for continuous variables. To avoid immor- The study was approved by the Danish Data Protection
tality bias and to supply absolute risk estimates, we used a Agency Danish Data Protection Agency (j-nr. 2007-58-0015 /
landmark approach to assess the risk of death associated local.j.nr. GEH-2014-015 I-Suite nr: 02733). In Denmark, ethi-
with the development of cardiovascular or renal disease.16 For cal approval is not required for register-based studies.
every t years of follow-up since T2D diagnosis (t years mean-
ing any year during follow-up, 0 to 10 years), we grouped
patients still alive at landmark year t according to the cardio- RESULTS
vascular or renal diagnosis(s) they had received so far. For fur- A total of 153 405 patients with newly diagnosed T2D
ther information the landmark method, please see Figure I in
were included between 1998 and 2015 with no prior
Downloaded from http://ahajournals.org by on July 11, 2020

the Data Supplement. RR and restricted mean time loss were

cardiovascular or renal diagnoses (Figure  1). Patient
computed as the ratio of Kaplan-Meier risk estimates and
as the restricted area under the Kaplan-Meier curve, respec- characteristics at time of inclusion are presented in the
tively.17 The landmark approach provides a method for assess- Table. Included patients were more likely to be diag-
ing development of cardiovascular or renal disease according nosed with hypertension, while atrial fibrillation, can-
to time since T2D diagnosis. We choose to apply the non- cer, and chronic obstructive pulmonary disease were
parametric due to the assumption free nature of the analysis, diagnosed in <5% of the cohort at time of inclusion
and the inherent ability to provide absolute risk estimates, (Table). During a median follow-up of 9.7 years (inter-
which are easily interpreted in a real-world clinical setting. quartile range, 5.8–13.9) 69  201 patients (45.1%)
In a first analysis, we focused on groups of patients who had received a cardiovascular or renal diagnosis. Analyzing
received only one type of cardiovascular or renal diagnosis these 43 155 (62.3%) received one diagnosis, 17 309
before each landmark year t. In a second analysis, we focused
(25.0%) received 2 diagnoses, and 8737 (12.6%)
on groups of patients who had received 2 diagnoses before
patients received 3 or more. The average number of
each landmark year t. Due to a small number of patients at
each landmark year for some of the diagnoses, we chose to diagnoses per patient was 0.7.
focus on results from the most common patient groups. RR
and restricted mean survival time differences were computed
Prevalence and Duration of
using the group of patients with T2D not diagnosed with any
cardiovascular or renal disease as reference. Prespecified sub- Cardiovascular or Renal Disease
group analyses were also performed to assess whether the Within 10 years of T2D diagnosis, IHD (0.6%–8.0%
risk of death differed by age, sex, and comorbidity status. We of all patients) and HF (0.4%–4.5% of all patients)
also performed an additional analysis in which we included were the most common combination. Stroke, CKD,
only patients who received their diagnoses within 1 year
and PAD were present among 0.1% to 3.1% of all
before the landmark year t. The purpose was to minimize the
patients within 10 years of T2D diagnosis (Table III in
effect of longer duration of T2D complications on the results
and to ensure comparability of groups across landmark years. the Data Supplement). At all landmark years, IHD was
To investigate differences in prognosis and use of pharma- the most frequent disease, while HF was the least
cotherapy over time, we stratified the inclusion period in 2 frequent (Table III in the Data Supplement). The age
separate time periods: from 1998 to 2008 and 2009 to 2015. and comorbidity distribution among these groups were
For each time period, we recorded demographic details, similar except for stroke patients, who were more likely

Circ Cardiovasc Qual Outcomes. 2020;13:e006260. DOI: 10.1161/CIRCOUTCOMES.119.006260 July 2020 3

 Zareini et al; Heart Failure and Diabetes Mellitus

Table.  Patient Characteristics of Patients With T2D Free of Diagnoses years following onset of T2D was 3 times higher (CI,
at Inclusion
2.9–3.1) than that of patients without cardiovascular
T2D Population and renal disease. Corresponding RR estimates were
N 153 405 lower for patients who developed IHD (RR, 1.3 [95%
Age, median (IQR) 64.0 (55–72) CI, 1.3–1.4]), stroke (RR, 2.2 [95% CI, 2.1–2.2]), CKD
Sex (male), n (%) 82 204 (53.6)
(RR, 1.7 [95% CI, 1.7–1.8]), and PAD (RR, 2.3 [95%
CI, 2.3–2.4]; Figure 2 and Table VII in the Data Supple-
Comorbidities at inclusion, n (%)
ment). Patients who developed HF within the 5 years
 Atrial fibrillation 4972 (3.2)
after diagnosis of T2D lived on average 11.7 months
 Cancer 5909 (3.9)
less than patients free of cardiovascular and renal dis-
 COPD 3748 (2.4) ease (95% CI, 11.6–11.8). By comparison, we esti-
 Hypertension 16 662 (10.9) mated smaller decreases in lifespan for other diagno-
Medication at inclusion, n (%) ses: IHD (1.6 months [95% CI, 1.5–1.7]), stroke (6.4
 Statins 52 787 (34.4) months [95% CI, 6.3–6.5]), CKD (4.4 months [95%
 ACE inhibitor/ARBs 64 031 (41.7)
CI, 4.3–4.6]), and PAD (6.9 months [95% CI, 6.8–7.0];
Figure 2 and Table VIII in the Data Supplement).
 β-blockers 32 921 (21.5)

 ASA 36 381 (23.7)

 Loop diuretics 20 548 (13.4) Prognosis After Developing a
 MRA 5621 (3.7) Combination of 2 Cardiovascular or Renal
 Thiazide 31 033 (20.2) Diseases
 Calcium channel blockers 34 234 (22.3) The highest risks of death 5 years after T2D diagnosis
ACE indicates angiotensin-converting enzyme; ARB, angiotensin receptor were found in the patients who had HF in combina-
blockers; ASA, acetylsalicylic acid; COPD, chronic obstructive pulmonary tion with stroke (54.1% [95% CI, 44.7–63.5]) or CKD
disorder; IQR, interquartile range; MRA, mineralocorticoid receptor (63.7% [95% CI, 53.7–73.7]) while the risk was some-
antagonists; and T2D, type 2 diabetes mellitus.
what lower in patients with HF in combination with
to have concomitant atrial fibrillation (Figures II and III in PAD (48.4% [95% CI, 36.4–60.5]) and IHD (45.5%
the Data Supplement). The median duration of cardio- [95 CI, 42.3–48.7]; Figure  3 and Table VI in the Data
Downloaded from http://ahajournals.org by on July 11, 2020

vascular or renal disease in the landmark cohorts was Supplement). Patients who developed a combination of
similar among groups at early landmarks and increased IHD and stroke, CKD, or PAD 5 years after T2D diagno-
at later landmarks, especially among patients with IHD sis had a 5-year risk of death below 40%. The 5-year
(Table V in the Data Supplement). In patients having 2 RR of death, with patients with T2D free of cardiovas-
diagnoses, the most frequent combination at all land- cular or renal disease as reference, was highest among
mark years was IHD and HF. Other combinations did patients with HF in combination with stroke (3.4 [95%
not reach above 2 percent of the landmark population CI, 3.3–3.5]), CKD (4.0 [95% CI, 3.9–4.1]), or PAD (3.1
each year (Table IV in the Data Supplement). Patients [95% CI, 3.0–3.1]) while other combinations did not
with HF in combination with stroke, CKD, or PAD had exceed a RR estimate above 3.0 (Figure  3 and Table
similar age distribution and were more likely to suffer VII in the Data Supplement). The number of months
from cancer, atrial fibrillation, and chronic obstructive lost in a 5-year span for patients with T2D developing
pulmonary disease compared with patients with IHD HF in combination with stroke, CKD, PAD, or IHD was
in combination with stroke, CKD, or PAD (Figures IV 16.2 (95% CI, 16.1–16.4), 18.2 (95% CI, 18.1–18.3),
and V in the Data Supplement). The median duration 14.3 (95% CI, 14.2–14.4), and 11 (95% CI, 10.9–11.2)
of the diagnoses before the landmark year was similar months, respectively. Corresponding estimates among
between groups (Table V in the Data Supplement). patients developing IHD in combination with stroke,
CKD, or PAD was 8 to 9 months (Figure 3 and Table VIII
in the Data Supplement).
Prognosis After Developing 1
Cardiovascular or Renal Disease
Prognosis According to Early Versus Late
The highest 5-year risk of death among patients alive
5 years after T2D diagnosis was found among those Inclusion Period
who had developed HF (47.6% [95% CI, 44.8–50.3]). A total of 92 837 patients were included between 1998
By comparison, the risk was <35% for patients who and 2008 and 60 568 patients were included between
developed IHD, stroke, CKD, and PAD (Figure  2 and 2009 and 2015. Most patients were males in both
Table VI in the Data Supplement). The 5-year RR of cohorts (54.1% compared with 52.8%). Patients includ-
death of patients who developed HF within the 5 ed in 2009 to 2015 were more likely to be in active treat-

Circ Cardiovasc Qual Outcomes. 2020;13:e006260. DOI: 10.1161/CIRCOUTCOMES.119.006260 July 2020 4

 Zareini et al; Heart Failure and Diabetes Mellitus

Figure 2. The 5-y risk of death, the 5-y risk ratio of death when compared with patients with type 2 diabetes mellitus (T2D) free of cardiovascular or
renal disease, and the expected decrease in lifespan in relation to development of 1 cardiovascular or renal diagnosis.
Vertical bars represent 95% CIs. CKD indicates chronic kidney disease; HF, heart failure; IHD, ischemic heart disease; and PAD, peripheral artery disease.

ment with statins (50.3% versus 24.1%), ACE inhibitors/ analysis. HF alone or in combination with other cardio-
ARBs (52.8% versus 34.5%), β-blockers (25.4% versus vascular or renal disease conferred the highest 1-year
18.9%), acetylsalicylic acid (25.5% versus 22.6%), and risk of death (Tables X and XI in the Data Supplement).
Ca channel blockers (27.7% versus 18.8%; Table IX in
Subgroup Analysis
Downloaded from http://ahajournals.org by on July 11, 2020

the Data Supplement). When examining the 1-year risk

of death following a cardiovascular or renal diagnosis in To ensure comparability in the duration of cardiovascu-
the 2 time periods, the findings were similar to the main lar or renal disease up to each landmark year, we per-

Figure 3. The 5-y risk of death, the 5-y risk ratio of death when compared with patients with type 2 diabetes mellitus (T2D) free of cardiovascular or
renal disease, and the expected decrease in lifespan in relation to development of 2 cardiovascular or renal diagnoses.
Vertical bars represent 95% CI. CKD indicates chronic kidney disease; HF, heart failure; IHD, ischemic heart disease; and PAD, peripheral artery disease.

Circ Cardiovasc Qual Outcomes. 2020;13:e006260. DOI: 10.1161/CIRCOUTCOMES.119.006260 July 2020 5

 Zareini et al; Heart Failure and Diabetes Mellitus

formed an additional analysis, where we only included of death compared with patients with T2D free of car-
patients, who received a diagnosis 1 year before the diovascular disease. While IHD was the most prevalent
landmark year. The risk of death associated with the disease in the landmark population, risk of death associ-
receiving of 1 or 2 diagnoses did not differ from our ated with IHD development alone or in combination with
main results (Figures VI and VII in the Data Supplement). stroke, CKD, or PAD was lower than any disease combi-
We stratified patients at each landmark year accord- nation that included HF or HF alone. This association was
ing to gender, age (above and below 65 years), and present, when we stratified the cohort according to age,
presence of comorbidities. HF alone or in combination gender, comorbidities at inclusion, and year of inclusion.
with CKD, stroke, or PAD was associated with the high- Excess risk of death in individuals with T2D has
est absolute 5-year risk of death in all subgroups (Fig- been well documented and has been attributed to an
ure 4A and 4B). increased risk of cardiovascular disease.21 Indeed, life
expectancy in patients with T2D has been estimated
as 5 to 12 years lower than the general population.22
DISCUSSION Additionally, several studies report that patients with
In a nationwide real-life cohort of >150 000 patients with T2D have a poorer prognosis following cardiovascu-
newly diagnosed T2D, we found that HF development lar and renal complications than the general popula-
alone or in combination with stroke, CKD, or PAD con- tion. Most studies focus on a single complication only,
ferred the highest 5-year risk of death. This translated hence, information on the impact of more than one
into an average reduction of 12 to 25 months lived within diabetic complication or the relative magnitude of each
the next 5 years and a 3- to 5-fold increased 5-year risk complication is lacking.8,9,23–25
Downloaded from http://ahajournals.org by on July 11, 2020

Figure 4. The 5-y risk associated with development of a cardiovascular or renal diagnosis 5 y after type 2 diabetes mellitus (T2D) diagnosis accord-
ing to sex and comorbidities at inclusion in different subgroups of T2D and cardiovascular or renal diagnoses.
(A) in patients developing one diagnosis, (B) in patients developing IHD in combination with other diagnoses and (C) in patients developing HF in combinations
with other diagnoses. Vertical bars represent 95% CI. AF indicates atrial fibrillation; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; HF,
heart failure; HTN, hypertension; IHD, ischemic heart disease; and PAD, peripheral vascular disease.

Circ Cardiovasc Qual Outcomes. 2020;13:e006260. DOI: 10.1161/CIRCOUTCOMES.119.006260 July 2020 6

 Zareini et al; Heart Failure and Diabetes Mellitus

We included HF and assessed the risk of all-cause tor agonists have shown promising results in patients
death according to time since T2D diagnosis, a novel with T2D.1,15 Treatment effects of different antidiabetic
way of addressing prognosis in an incident T2D popu- drugs are not covered in the present study as focus was
lation and the registries ensures lifelong follow-up of on comparison of different diagnoses. Future studies
real-world patients. We defined T2D as the first-time could potentially separate different antidiabetic drugs
redeemed prescription of a noninsulin antidiabetic drug. effect on certain patient’s risk profile for a more per-
We wanted to select a homogenous low-risk population sonalized approach in diabetes mellitus management.
primarily diagnosed in the primary sector. Our findings
were consistent with Swedish data, showing the haz-
ard ratio estimates of death in patients with T2D, when Limitations
compared with patients with myocardial infarction, was Several limitations of the present study should be
highest among patients developing HF, followed by acknowledged. First, lack of information on clinical
CKD, stroke, and amputation.26 In another study, the factors reflecting HF and T2D severity including clinical
risk of death within 1 month after a T2D complication assessment of functional class and smoking status, vital
was highest among patients with myocardial infarc- parameters, for example, heart rate and blood pres-
tion, followed by stroke, CKD, HF, and amputation.27 sure, biochemical parameters such as N-terminal pro-
We included unstable and stable angina pectoris in the b-type natriuretic peptide, glucose levels, and hemo-
definition of IHD, arguably ending with a broader low- globin A1c. Second, we did not have any information
risk population not comparable with acute myocardial on the presence of other microvascular complications,
infarction. A similar argument extends to patients with which serve as to highlight the progression and severity
CKD in contrast to using patients with end-stage renal of diabetes mellitus illness in patients.
disease or renal replacement therapy. Additionally, we
did not find major differences in 5-year risks of death
among patients with HF and IHD (45.5%, at landmark Clinical Implications
year 5) and HF alone (47.6%, at landmark year 5). This Although not the most frequent complication, HF was
is in contrast to existing literature, where ischemic HF is clearly associated with the most unfavorable prognosis
associated with a higher risk of death compared with in patients with T2D. When providing care for patients
nonischemic HF among patients with T2D.28–30 Howev- with T2D, we hope that our findings contribute to
er, the findings of the present study must be interpreted
Downloaded from http://ahajournals.org by on July 11, 2020

assessing risk profiles and prognosis, especially con-

with caution since we did not have access to echocar- cerning the importance of evaluating patients with T2D
diographic measurements and cannot be fully certain of regularly for HF.
the HF etiology. To ensure comparability across groups,
we performed an additional analysis only including
patients diagnosed in the year before the landmark. Conclusions
With similar duration of cardiovascular or renal disease, In this nationwide cohort of newly diagnosed patients
the results did not differ from our primary analysis. with T2D, we found that HF development alone or in
Across the landmark years, we observed continu- combination with stroke, CKD, or PAD imposed the
ously smaller differences in 5-year risks of death for all highest absolute and relative risk of death along with
complications, especially for HF (Figure  3). There are the greatest decrease in lifespan compared with other
several potential explanations: First, increasing age of combinations of cardiovascular and renal disease.
the study population during follow-up together with
only inclusion of patients alive at each landmark will
result in a healthy survivor effect in the later landmark ARTICLE INFORMATION
years. Second, patients diagnosed at early landmark Received October 9, 2019; accepted April 23, 2020.
The Data Supplement is available at https://www.ahajournals.org/doi/
years cannot reflect true exposure of T2D and other suppl/10.1161/CIRCOUTCOMES.119.006260.
cardiovascular risk factors due to undiagnosed diabetes
mellitus for a longer period. Third, differences in dura- Correspondence
tion of complications could result in different prognoses Bochra Zareini, MD, Department of Cardiology, University Hospital Herlev and
for each complication. However, we did not find major Gentofte, Kildegårdsvej 28, 2900 Hellerup, Denmark. Email bochra.zareini.03@
variation in median duration of complications at each regionh.dk
landmark (Table V in the Data Supplement).
Inconsistencies of improved or worsened cardio- Affiliations
vascular risk profile with different antidiabetic drugs Department of Cardiology, Herlev and Gentofte University Hospital, Copenha-
from clinical trials and observational studies have been gen, Denmark (B.Z., M.D., M.E., G.G., M.S., M.L.). Danish Heart Foundation,
Copenhagen, Denmark (G.G.). Section of Biostatistics, Department of Public
debated.31 Contemporary trials with an intervention Health, University of Copenhagen, Denmark (P.B.). Unit of Epidemiology and
with SGLT-2 inhibitors or glucagon-like peptide-1 recep- Biostatistics (C.H.N.) and Department of Cardiology (C.T.-P.), Aalborg Univer-

Circ Cardiovasc Qual Outcomes. 2020;13:e006260. DOI: 10.1161/CIRCOUTCOMES.119.006260 July 2020 7

 Zareini et al; Heart Failure and Diabetes Mellitus

sity Hospital, Denmark. Department of Endocrinology, Amager and Hvidovre 12. Kristensen SL, Preiss D, Jhund PS, Squire I, Cardoso JS, Merkely B,
University Hospital, Copenhagen, Denmark (C.S.). Department of Cardiology, Martinez F, Starling RC, Desai AS, Lefkowitz MP, Rizkala AR, Rouleau JL,
Rigshospitalet University Hospital, Copenhagen, Denmark (S.L.K., L.K.). Depart- Shi VC, Solomon SD, Swedberg K, Zile MR, McMurray JJV, Packer M.
ments of Clinical Investigation and Cardiology, Nordsjaellands Hospital, Hill- Risk related to pre–diabetes mellitus and diabetes mellitus in heart fail-
erød, Denmark (C.T.-P.). ure with reduced ejection fraction: insights from prospective compari-
son of ARNI with ACEI to determine impact on global mortality and
morbidity in Heart Failure Trial. Circ Heart Fail 2016;9:e002560. doi:
Sources of Funding 10.1161/CIRCHEARTFAILURE.115.002560
The study was supported by the Danish Heart Association and an unrestricted 13. McMurray JJ, Gerstein HC, Holman RR, Pfeffer MA. Heart failure: a cardio-
hospital grant from Boehringer Ingelheim. The funding sources had no role in vascular outcome in diabetes that can no longer be ignored. Lancet Diabe-
study design, data collection, data analysis, interpretation of data, writing the tes Endocrinol. 2014;2:843–851. doi: 10.1016/S2213-8587(14)70031-2
article or the decision to submit for publication. The corresponding author had 14. Benjamin EJ, Virani SS, Callaway CW, Chamberlain AM, Chang AR,
full access to all data and final responsibility for the submission. Cheng S, Chiuve SE, Cushman M, Delling FN, Deo R, de Ferranti SD,
Ferguson JF, Fornage M, Gillespie C, Isasi CR, Jiménez MC, Jordan LC,
Judd SE, Lackland D, Lichtman JH, Lisabeth L, Liu S, Longenecker CT,
Disclosures Lutsey PL, Mackey JS, Matchar DB, Matsushita K, Mussolino ME, Nasir K,
Unrelated to the present work, Dr Køber reports honorarium as speaker from O’Flaherty M, Palaniappan LP, Pandey A, Pandey DK, Reeves MJ, Ritchey MD,
Novartis, AstraZeneca, and Boehringer. Dr D’Souza reports grants from the Rodriguez CJ, Roth GA, Rosamond WD, Sampson UKA, Satou GM,
Danish Heart Foundation and the Velux Foundation. The other authors report Shah SH, Spartano NL, Tirschwell DL, Tsao CW, Voeks JH, Willey JZ,
no conflicts. Wilkins JT, Wu JH, Alger HM, Wong SS, Muntner P. Heart disease and stroke
statistics—2018 update: a report from the American Heart Association.
Circulation. 2018;137:e67–e492. doi: 10.1161/CIR.0000000000000558
15. Nauck MA, Tornøe K, Rasmussen S, Treppendahl MB, Marso SP; LEADER
REFERENCES Publication Committee on behalf of the LEADER Trial Investigators. Car-
1. Zinman B, Wanner C, Lachin JM, Fitchett D, Bluhmki E, Hantel S, diovascular outcomes in patients who experienced a myocardial infarction
Mattheus M, Devins T, Johansen OE, Woerle HJ, Broedl UC, Inzucchi SE; while treated with liraglutide versus placebo in the LEADER trial. Diab Vasc
EMPA-REG OUTCOME Investigators. Empagliflozin, cardiovascular out- Dis Res. 2018;15:465–468. doi: 10.1177/1479164118783935
comes, and mortality in type 2 diabetes. N Engl J Med. 2015;373:2117– 16. Dafni U. Landmark analysis at the 25-year landmark point. Circ
2128. doi: 10.1056/NEJMoa1504720 Cardiovasc Qual Outcomes. 2011;4:363–371. doi: 10.1161/
2. Wiviott SD, Raz I, Bonaca MP, Mosenzon O, Kato ET, Cahn A, CIRCOUTCOMES.110.957951
Silverman MG, Zelniker TA, Kuder JF, Murphy SA, Bhatt DL, Leiter LA, 17. Uno H, Claggett B, Tian L, Inoue E, Gallo P, Miyata T, Schrag D, Takeuchi M,
McGuire DK, Wilding JPH, Ruff CT, Gause-Nilsson IAM, Fredriksson M, Uyama Y, Zhao L, Skali H, Solomon S, Jacobus S, Hughes M, Packer M,
Johansson PA, Langkilde AM, Sabatine MS; DECLARE–TIMI 58 Investiga- Wei LJ. Moving beyond the hazard ratio in quantifying the between-group
tors. Dapagliflozin and cardiovascular outcomes in type 2 diabetes. N Engl difference in survival analysis. J Clin Oncol. 2014;32:2380–2385. doi:
J Med. 2019;380:347–357. doi: 10.1056/NEJMoa1812389 10.1200/JCO.2014.55.2208
3. Neal B, Perkovic V, Mahaffey KW, de Zeeuw D, Fulcher G, Erondu N, 18. R Core Team. R: A Language and Environment for Statistical Computing
Shaw W, Law G, Desai M, Matthews DR; CANVAS Program Collaborative [Internet]. Vienna, Austria: R Foundation for Statistical Computing. Avail-
Group. Canagliflozin and cardiovascular and renal events in type 2 diabe- able at: https://www.R-project.org/. Accessed June 10, 2020.
Downloaded from http://ahajournals.org by on July 11, 2020

tes. N Engl J Med. 2017;377:644–657. doi: 10.1056/NEJMoa1611925 19. A. Gerds T. prodlim: Product-Limit Estimation for Censored Event History
4. Chun BY, Dobson AJ, Heller RF. The impact of diabetes on survival among Analysis. 2018. Available at: https://CRAN.R-project.org/package=prodlim.
patients with first myocardial infarction. Diabetes Care. 1997;20:704– Accessed June 10, 2020.
708. doi: 10.2337/diacare.20.5.704 20. Uno H, Tian L, Cronin A, Battioui C, Horiguchi M. survRM2: Comparing
5. Norgaard ML, Andersen SS, Schramm TK, Folke F, Jørgensen CH, Restricted Mean Survival Time. 2017. Available at: https://CRAN.R-project.
Hansen ML, Andersson C, Bretler DM, Vaag A, Køber L, Torp-Pedersen C, org/package=survRM2. Accessed June 10, 2020.
Gislason GH. Changes in short- and long-term cardiovascular risk of in- 21. Roper NA, Bilous RW, Kelly WF, Unwin NC, Connolly VM. Excess mortal-
cident diabetes and incident myocardial infarction–a nationwide study. ity in a population with diabetes and the impact of material deprivation:
Diabetologia. 2010;53:1612–1619. doi: 10.1007/s00125-010-1783-z longitudinal, population based study. BMJ. 2001;322:1389–1393. doi:
6. Koek HL, Soedamah-Muthu SS, Kardaun JW, Gevers E, de Bruin A, 10.1136/bmj.322.7299.1389
Reitsma JB, Bots ML, Grobbee DE. Short- and long-term mortality after 22. Franco OH, Steyerberg EW, Hu FB, Mackenbach J, Nusselder W. Associa-
acute myocardial infarction: comparison of patients with and without dia- tions of diabetes mellitus with total life expectancy and life expectancy with
betes mellitus. Eur J Epidemiol. 2007;22:883–888. doi: 10.1007/s10654- and without cardiovascular disease. Arch Intern Med. 2007;167:1145–
007-9191-5 1151. doi: 10.1001/archinte.167.11.1145
7. Jhund PS, McMurray JJ, Chaturvedi N, Brunel P, Desai AS, Finn PV, 23. Sprafka JM, Burke GL, Folsom AR, McGovern PG, Hahn LP. Trends in
Haffner SM, Solomon SD, Weinrauch LA, Claggett BL, Pfeffer MA. Mor- prevalence of diabetes mellitus in patients with myocardial infarction and
tality following a cardiovascular or renal event in patients with type 2 effect of diabetes on survival. The Minnesota Heart Survey. Diabetes Care.
diabetes in the ALTITUDE trial. Eur Heart J. 2015;36:2463–2469. doi: 1991;14:537–543. doi: 10.2337/diacare.14.7.537
10.1093/eurheartj/ehv295 24. Miettinen H, Lehto S, Salomaa V, Mähönen M, Niemelä M, Haffner SM,
8. Icks A, Claessen H, Morbach S, Glaeske G, Hoffmann F. Time-dependent Pyörälä K, Tuomilehto J. Impact of diabetes on mortality after the first myo-
impact of diabetes on mortality in patients with stroke: survival up to 5 cardial infarction. The FINMONICA Myocardial Infarction Register Study
years in a health insurance population cohort in Germany. Diabetes Care. Group. Diabetes Care. 1998;21:69–75. doi: 10.2337/diacare.21.1.69
2012;35:1868–1875. doi: 10.2337/dc11-2159 25. Malmberg K, Yusuf S, Gerstein HC, Brown J, Zhao F, Hunt D, Piegas L,
9. Adler AI, Stevens RJ, Manley SE, Bilous RW, Cull CA, Holman RR; UKPDS Calvin J, Keltai M, Budaj A. Impact of diabetes on long-term prognosis
GROUP. Development and progression of nephropathy in type 2 diabetes: in patients with unstable angina and non-Q-wave myocardial infarction:
the United Kingdom Prospective Diabetes Study (UKPDS 64). Kidney Int. results of the OASIS (Organization to Assess Strategies for Ischemic Syn-
2003;63:225–232. doi: 10.1046/j.1523-1755.2003.00712.x dromes) Registry. Circulation. 2000;102:1014–1019. doi: 10.1161/01.
10. Andrésdóttir G, Jensen ML, Carstensen B, Parving HH, Rossing K, cir.102.9.1014
Hansen TW, Rossing P. Improved survival and renal prognosis of patients 26. Hayes AJ, Leal J, Kelman CW, Clarke PM. Risk equations to predict life ex-
with type 2 diabetes and nephropathy with improved control of risk fac- pectancy of people with type 2 diabetes mellitus following major compli-
tors. Diabetes Care. 2014;37:1660–1667. doi: 10.2337/dc13-2036 cations: a study from Western Australia. Diabet Med. 2011;28:428–435.
11. MacDonald MR, Jhund PS, Petrie MC, Lewsey JD, Hawkins NM, Bhagra S, doi: 10.1111/j.1464-5491.2010.03189.x
Munoz N, Varyani F, Redpath A, Chalmers J, MacIntyre K, McMurray JJ. 27. Kelly PJ, Clarke PM, Hayes AJ, Gerdtham UG, Cederholm J, Nilsson P,
Discordant short- and long-term outcomes associated with diabetes in Eliasson B, Gudbjornsdottir S. Predicting mortality in people with type
patients with heart failure: importance of age and sex: a population study 2 diabetes mellitus after major complications: a study using Swedish
of 5.1 million people in Scotland. Circ Heart Fail. 2008;1:234–241. doi: National Diabetes Register data. Diabet Med. 2014;31:954–962. doi:
10.1161/CIRCHEARTFAILURE.108.794008 10.1111/dme.12468

Circ Cardiovasc Qual Outcomes. 2020;13:e006260. DOI: 10.1161/CIRCOUTCOMES.119.006260 July 2020 8

 Zareini et al; Heart Failure and Diabetes Mellitus

28. Johansson I, Dahlström U, Edner M, Näsman P, Rydén L, Norhammar A. 30. Cubbon RM, Adams B, Rajwani A, Mercer BN, Patel PA, Gherardi G,
Prognostic implications of type 2 diabetes mellitus in Ischemic and Non- Gale CP, Batin PD, Ajjan R, Kearney L, Wheatcroft SB, Sapsford RJ, Witte KK,
ischemic Heart Failure. J Am Coll Cardiol. 2016;68:1404–1416. doi: Kearney MT. Diabetes mellitus is associated with adverse prognosis in
10.1016/j.jacc.2016.06.061 chronic heart failure of ischaemic and non-ischaemic aetiology. Diab Vasc
29. Dries DL, Sweitzer NK, Drazner MH, Stevenson LW, Gersh BJ. Prognos- Dis Res. 2013;10:330–336. doi: 10.1177/1479164112471064
tic impact of diabetes mellitus in patients with heart failure according 31. Hippisley-Cox J, Coupland C. Diabetes treatments and risk of heart failure,
to the etiology of left ventricular systolic dysfunction. J Am Coll Cardiol. cardiovascular disease, and all cause mortality: cohort study in primary
2001;38:421–428. doi: 10.1016/s0735-1097(01)01408-5 care. BMJ. 2016;354:i3477. doi: 10.1136/bmj.i3477
Downloaded from http://ahajournals.org by on July 11, 2020

Circ Cardiovasc Qual Outcomes. 2020;13:e006260. DOI: 10.1161/CIRCOUTCOMES.119.006260 July 2020 9