risk factors, mortality, and cardiovascular … engl j med 379;7 nejm.orgaugust 16, 2018 635 risk...

T h e n e w e ngl a nd j o u r na l o f m e dic i n e

n engl j med 379;7 nejm.org August 16, 2018 633

From the Department of Molecular and Clinical Medicine, Institute of Medicine (Aidin Rawshani, Araz Rawshani, B.E., A. Rosengren, S.G.), and the Health Metrics Unit, Sahlgrenska Academy (S.F.), Uni-versity of Gothenburg, and the Swedish National Diabetes Register, Center of Registers in Region (Aidin Rawshani, Araz Rawshani, S.F., B.E., A.-M.S., M.M., S.G.), Gothenburg, and the Department of Pub-lic Health and Caring Sciences–Geriat-rics, Uppsala University, and the Swedish Medical Products Agency, Uppsala (B.Z.) — all in Sweden; the Institute of Cardio-vascular and Medical Sciences, Universi-ty of Glasgow, Glasgow, United Kingdom (N.S.); and the Division of Cardiology, Department of Internal Medicine, Univer-sity of Texas Southwestern Medical Cen-ter, Dallas (D.K.M.). Address reprint re-quests to Dr. Aidin Rawshani at the Swedish National Diabetes Register Re-gion of Vastra Götaland, Medicinaregatan 18G, Gothenburg 413 45, Sweden, or at aidin . rawshani@ gu . se.

N Engl J Med 2018;379:633-44.DOI: 10.1056/NEJMoa1800256Copyright © 2018 Massachusetts Medical Society.

BACKGROUNDPatients with diabetes are at higher risk for death and cardiovascular outcomes than the general population. We investigated whether the excess risk of death and cardio-vascular events among patients with type 2 diabetes could be reduced or eliminated.

METHODSIn a cohort study, we included 271,174 patients with type 2 diabetes who were reg-istered in the Swedish National Diabetes Register and matched them with 1,355,870 controls on the basis of age, sex, and county. We assessed patients with diabetes according to age categories and according to the presence of five risk factors (ele-vated glycated hemoglobin level, elevated low-density lipoprotein cholesterol level, albuminuria, smoking, and elevated blood pressure). Cox regression was used to study the excess risk of outcomes (death, acute myocardial infarction, stroke, and hospitalization for heart failure) associated with smoking and the number of vari-ables outside target ranges. We also examined the relationship between various risk factors and cardiovascular outcomes.

RESULTSThe median follow-up among all the study participants was 5.7 years, during which 175,345 deaths occurred. Among patients with type 2 diabetes, the excess risk of outcomes decreased stepwise for each risk-factor variable within the target range. Among patients with diabetes who had all five variables within target ranges, the hazard ratio for death from any cause, as compared with controls, was 1.06 (95% confidence interval [CI], 1.00 to 1.12), the hazard ratio for acute myocardial in-farction was 0.84 (95% CI, 0.75 to 0.93), and the hazard ratio for stroke was 0.95 (95% CI, 0.84 to 1.07). The risk of hospitalization for heart failure was consistently higher among patients with diabetes than among controls (hazard ratio, 1.45; 95% CI, 1.34 to 1.57). In patients with type 2 diabetes, a glycated hemoglobin level outside the target range was the strongest predictor of stroke and acute myocardial infarction; smoking was the strongest predictor of death.

CONCLUSIONSPatients with type 2 diabetes who had five risk-factor variables within the target ranges appeared to have little or no excess risk of death, myocardial infarction, or stroke, as compared with the general population. (Funded by the Swedish Associa-tion of Local Authorities and Regions and others.)

A BS TR AC T

Risk Factors, Mortality, and Cardiovascular Outcomes in Patients with Type 2 Diabetes

Aidin Rawshani, M.D., Araz Rawshani, M.D., Ph.D., Stefan Franzén, Ph.D., Naveed Sattar, M.D., Ph.D., Björn Eliasson, M.D., Ph.D., Ann-Marie Svensson, Ph.D.,

Björn Zethelius, M.D., Ph.D., Mervete Miftaraj, M.Sc., Darren K. McGuire, M.D., M.H.Sc., Annika Rosengren, M.D., Ph.D.,

and Soffia Gudbjörnsdottir, M.D., Ph.D.

Original Article

The New England Journal of Medicine Downloaded from nejm.org by JULES LEVIN on August 16, 2018. For personal use only. No other uses without permission.

Copyright © 2018 Massachusetts Medical Society. All rights reserved.

n engl j med 379;7 nejm.org August 16, 2018634


Type 2 diabetes is a complex disease that leads to continuous medical care with comprehensive, multifactorial strategies for

reducing cardiovascular risk. Patients with type 2 diabetes have risks of death and cardiovascular events that are 2 to 4 times as great as the risks in the general population.1 Results from random-ized trials support a range of interventions that target isolated risk factors such as elevated levels of glycated hemoglobin, blood pressure, and cho-lesterol to prevent or postpone complications of type 2 diabetes. The Steno-2 Study investigated the effects of multifactorial risk-factor control by means of behavior modification and pharmaco-logic therapy and showed long-lasting reductions in the risks of death and cardiovascular events among patients in whom these risks were reduced, as compared with patients who had been ran-domly assigned to usual care.2,3

The extent to which the excess risk associated with type 2 diabetes may be mitigated, or poten-tially eliminated, by contemporary evidence-based treatment and multifactorial risk-factor modifica-tion is unclear. In a nationwide cohort, we evalu-ated the association between the excess risks of death and cardiovascular outcomes among pa-tients with type 2 diabetes, according to the num-ber of risk-factor variables within therapeutic guideline levels, as compared with controls who were matched for age, sex, and county in Sweden. Risk-factor data were not available for controls.

In ancillary analyses, we estimated the strength of the associations between various risk factors and the incremental risks of death and cardiovascular outcomes associated with diabe-tes. Moreover, we examined the association be-tween selected risk-factor variables such as levels of glycated hemoglobin, systolic blood pres-sure, and low-density lipoprotein (LDL) choles-terol within evidence-based target ranges and these outcomes.

Me thods

Study Design and Support

The Regional Ethics Review Board of Gothen-burg, Sweden, approved the study. All the patients provided written informed consent before inclu-sion in the Swedish National Diabetes Register. The Swedish Association of Local Authorities and Regions and other nonprofit agencies supported the study; no industry support was provided.

Data Sources and Study Cohort

The Swedish National Diabetes Register has been described previously.1,4 Type 2 diabetes was de-fined according to epidemiologic criteria — treat-ment with diet, with or without the use of oral antihyperglycemic agents, or treatment with in-sulin, with or without the use of oral antihyper-glycemic agents. The latter category (insulin use) applied only to patients who were 40 years of age or older at the time of diagnosis. Patients with at least one entry in the register between January 1, 1998, and December 31, 2012, were included in the study. At baseline (defined as the first entry in the register), each patient with type 2 diabetes was matched for age, sex, and county with five controls without diabetes who were randomly selected from the Swedish population register by Statistics Sweden.

We constructed two cohorts of patients with type 2 diabetes. One cohort excluded patients with previous stroke, acute myocardial infarction, or amputation; those who had undergone dialy-sis or renal transplantation; and those with a body-mass index (the weight in kilograms divided by the square of the height in meters) of less than 18.5. The second cohort of patients with type 2 diabetes had exclusion criteria that were similar to those of the first cohort but also excluded pa-tients with previous coronary heart disease, atrial fibrillation, or heart failure. Controls who met any of these criteria were excluded without the exclusion of their matched patient.

Outcomes

We assessed death from any cause, fatal or non-fatal acute myocardial infarction (henceforth re-ferred to as acute myocardial infarction), fatal or nonfatal stroke (henceforth referred to as stroke), and hospitalization for heart failure. Outcomes were identified in hospital discharge records with use of codes in the International Classification of Diseases, 9th Revision and 10th Revision. The specific codes are listed in Table S1 in the Supplementary Appendix, available with the full text of this ar-ticle at NEJM.org. Patients were followed until an event occurred or until December 31, 2013, for all the outcomes except for death from any cause, for which follow-up ended on December 31, 2014.

Statistical Analysis

Crude incidence rates were calculated according to the number of risk-factor variables within tar-




Risk Factors and Cardiovascular Outcomes in Type 2 Diabetes

get ranges and are presented as events per 10,000 patient-years of observation with exact Poisson confidence intervals of 95%. We constructed Cox regression models and included baseline values in the models. All the models were adjusted for so-cioeconomic variables (income, marital status, immigrant status, and educational level), with stratification according to sex to allow for dif-ferent underlying baseline hazards among men and women, and age was used as the time scale. Some models were adjusted for preexisting con-ditions. For death from any cause, we adjusted for coronary heart disease and heart failure; for myocardial infarction, we adjusted for atrial fi-brillation and heart failure; for stroke, we adjust-ed for atrial fibrillation, heart failure, and coro-nary heart disease; and for hospitalization for heart failure, we adjusted for atrial fibrillation and coronary heart disease.

For the main analyses, we estimated the risk of each outcome among patients with type 2 diabetes, according to the number of risk-factor variables within target ranges, as compared with the controls matched for age, sex, and county. We defined whether the risk-factor variables were within target ranges on the basis of guideline-recommended target levels.5,6 The regression mod-els included the covariable “category,” which de-noted the number of risk-factor variables that were not within the target range (scale, none to five variables). The following five risk factors were considered: the glycated hemoglobin level (cutoff value, ≥7.0% or ≥53 mmol per mole), systolic and diastolic blood pressure (cutoff value, ≥140 mm Hg for systolic blood pressure or ≥80 for diastolic blood pressure), albuminuria (the presence of microalbuminuria or macroalbuminuria), smok-ing (being a current smoker at study entry), and the LDL cholesterol level (cutoff value, ≥2.5 mmol per liter [97 mg per deciliter]).

We constructed a separate Cox model for each age category, according to age at baseline: young-er than 55 years of age, 55 years to younger than 65 years of age, 65 years to younger than 80 years of age, and 80 years of age or older. We adjusted for the duration of diabetes by assigning matched controls a duration of 0 years, and patients with type 2 diabetes had their duration of diabetes centralized around the grand mean (the mean duration among all the study participants).

Among patients with type 2 diabetes, we ana-lyzed the relative importance of the risk factors.

Relative importance provides an estimate of how important each risk factor is in terms of predict-ing the outcome. Several methods are available for this task; we chose to calculate the relative importance as measured by the R2 values of the models,7 and we tested the consistency of the re-sults by calculating the explainable log-likelihood that was attributable to each risk factor.

We assessed the level associated with the low-est risk, given the glycated hemoglobin level, systolic blood pressure, and LDL cholesterol level, by modeling the association between each risk factor and the outcomes using restricted cubic splines. The evidence-based target level was set as the reference for each risk factor; not smok-ing was set as the reference for smoking status.

Missing data were imputed with the multi-variate imputation by chained equations (MICE) algorithm. We imputed five complete data sets; a list of the variables that were used in the im-putation model is provided in Table S3 in the Supplementary Appendix. The estimates from each imputed data set were combined into one overall estimate with the use of Rubin’s rule. Figures S4 and S5 in the Supplementary Appendix show the frequency of missing-data elements and the dis-tribution of each variable before and after the imputation. For the main analysis, we pooled the results across all five imputed data sets; the re-sults of the analyses with the imputed data sets were virtually identical to those in the cohort with complete cases and were thus consistent with a complete-case analysis. For the ancillary analyses, we used the first imputed data set. All the analy-ses were performed with the use of RStudio soft-ware, version 3.2.3.

R esult s

Study Population

A total of 433,619 patients with type 2 diabetes and 2,168,095 controls were identified, and 271,174 patients with type 2 diabetes and 1,355,870 matched controls were included in the study (Fig. S1 in the Supplementary Appendix). The median follow-up among all the study partici-pants was 5.7 years, during which 175,345 deaths occurred. The baseline characteristics of the pa-tients with complete data on all five risk factors (96,673 patients with diabetes [35.6%]) and their matched controls are presented in Table 1. The number of patients in each risk-factor group in





the imputed data sets is also shown in Table 1. The complete data regarding the characteristics of the participants at baseline are presented in Table S4 in the Supplementary Appendix.

The mean age of the patients was 60.60 years, and 47,777 of 96,673 patients (49.4%) were wom-en. Table S5 in the Supplementary Appendix shows the baseline characteristics of the patients for whom data were missing for at least one risk fac-tor. Figure S6 in the Supplementary Appendix shows the trends in risk factors over the period from 1998 through 2012, and Figure S7 in the Supplementary Appendix shows how causes of death varied among the groups according to the number of risk-factor variables in the target ranges.

Risk of Cardiovascular Events

A total of 37,825 patients with diabetes (13.9%) and 137,520 controls (10.1%) died during the study period. The numbers of events, incidence rates, and hazard ratios for all the outcomes among patients with diabetes, as compared with controls, at increasing numbers of risk-factor vari-ables, as well as the risks of death among men and women, are presented in Table S2 in the Supple-mentary Appendix.

Figure 1 shows the adjusted hazard ratios for the outcomes, according to age category and the number of risk-factor variables within target rang-es, among patients with diabetes as compared with matched controls. The results show a step-wise increase in the hazard ratios for each ad-ditional variable that was not within the target range among patients with diabetes, and the incre-mental risks of cardiovascular events and death that were associated with diabetes decreased in a stepwise fashion from younger to older age groups. Patients with diabetes who were 80 years of age or older at baseline had the lowest incre-mental risk of cardiovascular events and death, as compared with controls. In the overall cohort, patients with type 2 diabetes who had no risk-factor variables outside the target ranges had a marginally higher risk of death than the controls (hazard ratio, 1.06; 95% confidence interval [CI], 1.00 to 1.12).

Patients with diabetes who were 80 years of age or older at baseline and had no risk factors outside target ranges had the lowest hazard ratio, as compared with controls, for acute myocardial infarction across all the groups (hazard ratio,

0.72; 95% CI, 0.49 to 1.07) (Fig. 1B). Overall, patients with type 2 diabetes who had no risk-factor variables outside the target ranges had a lower risk of acute myocardial infarction than the matched controls (hazard ratio, 0.84; 95% CI, 0.75 to 0.93). Corresponding estimates for the excess risk of stroke are shown in Figure 1C. The overall hazard ratio for stroke among patients with no risk-factor variables outside the target ranges, as compared with controls, was 0.95 (95% CI, 0.84 to 1.07) (Table S2 in the Supple-mentary Appendix). Similar to the findings with acute myocardial infarction, there was a higher incremental risk of stroke in the younger age categories and for each variable that was not within the target range. Estimates regarding hos-pitalization for heart failure are shown in Fig-ure 1D. Patients with type 2 diabetes who were younger than 55 years of age and had all five risk-factor variables outside the target ranges had the highest excess risk of hospitalization for heart failure of all the outcomes assessed (haz-ard ratio vs. control, 11.35; 95% CI, 7.16 to 18.01). The overall hazard ratio for hospitalization for heart failure among patients with no risk-factor variables outside the target ranges, as compared with controls, was 1.45 (95% CI, 1.34 to 1.57) (Table S2 in the Supplementary Appendix).

Risk-Factor Strength and Levels in Patients with Type 2 Diabetes

Figure 2A shows the predictors with the apparent greatest importance with regard to death from any cause. The five strongest predictors regard-ing the risk of death among patients with type 2 diabetes were smoking, physical activity, marital status, glycated hemoglobin level, and use of statins (lipid-lowering medication). The data shown in Figure 3A suggest that lower glycated hemo-globin levels than are currently recommended in guidelines were associated with a lower risk of death.

The strongest predictors regarding the risk of acute myocardial infarction were the glycated hemoglobin level, systolic blood pressure, LDL cholesterol level, physical activity, and smoking (Fig. 2B). These risk factors showed a linear as-sociation with the risk of acute myocardial in-farction (Fig. 3B).

The strongest predictors regarding the risk of stroke were the glycated hemoglobin level, systolic blood pressure, duration of diabetes, physical ac-





Tabl

e 1.

Bas

elin

e C

hara

cter

istic

s of

the

Patie

nts

with

Typ

e 2

Dia

bete

s an

d M

atch

ed C

ontr

ols.

*

Var

iabl

eM

atch

ed C

ontr

ols

Patie

nts

with

Dia

bete

s w

ith C

ompl

ete

Dat

a on

All

Ris

k Fa

ctor

s

Ove

rall

No.

of R

isk-

Fact

or V

aria

bles

out

side

Tar

get R

ange

01

23

45

No.

of p

artic

ipan

ts

Com

plet

e-ca

se d

ata

set

483,

365

96,6

734,

852

22,5

8439

,673

24,3

4149

2729

6

Impu

ted

data

set

11,

355,

870

271,

174

11,6

1257

,000

107,

840

76,3

2517

,227

1,17

0

Impu

ted

data

set

21,

355,

870

271,

174

11,5

6957

,033

107,

709

76,4

5517

,213

1,19

5

Impu

ted

data

set

31,

355,

870

271,

174

11,6

8557

,164

107,

521

76,5

1717

,102

1,18

5

Impu

ted

data

set

41,

355,

870

271,

174

11,6

6957

,217

107,

551

76,4

3117

,137

1,16

9

Impu

ted

data

set

51,

355,

870

271,

174

11,5

6257

,246

107,

551

76,3

9217

,219

1,20

4

Fem

ale

sex

— n

o. (

%)

238,

885

(49.

4)47

,777

(49

.4)

2,52

5 (5

2.0)

11,5

28 (

51.0

)19

,799

(49

.9)

11,7

13 (

48.1

)2,

085

(42.

3)12

7 (4

2.9)

Age

— y

r60

.58±

10.8

960

.58±

10.8

960

.96±

12.0

661

.04±

11.2

360

.93±

10.7

960

.00±

10.5

458

.32±

10.0

657

.27±

10.2

1

Dur

atio

n of

dia

bete

s —

yr

—4.

53±5

.75

4.09

±5.2

84.

08±5

.35

4.29

±5.6

15.

19±6

.18

5.51

±6.4

16.

09±6

.60

Age

at d

iagn

osis

of d

iabe

tes

— y

r—

56.0

9±11

.11

56.8

4±12

.20

56.9

7±11

.44

56.6

9±11

.02

54.8

7±10

.64

52.7

6±10

.20

51.3

4±11

.11

Gly

cate

d he

mog

lobi

n†

Mill

imol

es p

er m

ole

—53

.22±

13.9

644

.31±

5.14

46.7

5±8.

6751

.10±

12.3

261

.58±

15.2

566

.28±

15.0

170

.51±

15.4

5

Perc

ent

—7.

02±1

.28

6.21

±0.4

76.

43±0

.79

6.83

±1.1

37.

79±1

.39

8.22

±1.3

78.

60±1

.41

LDL

chol

este

rol

Mill

imol

es p

er li

ter‡

—3.

00±0

.95

1.98

±0.3

92.

55±0

.87

3.09

±0.9

23.

38±0

.85

3.50

±0.8

43.

65±0

.81

Mill

igra

ms

per

deci

liter

—11

6.0±

36.7

76.5

±15.

198

.6±3

3.6

119.

5±35

.513

0.7±

32.8

135.

3±32

.414

1.1±

31.3

Tota

l cho

lest

erol

— m

mol

/lite

r‡—

5.11

±1.0

54.

04±0

.58

4.64

±0.9

55.

20±1

.01

5.51

±0.9

75.

67±0

.98

5.88

±0.9

9

Cur

rent

sm

oker

— n

o. (

%)

—16

,486

(17

.1)

088

6 (3

.9)

4,26

8 (1

0.8)

7,12

5 (2

9.3)

3,91

1 (7

9.4)

296

(100

)

Bod

y-m

ass

inde

x§—

30.3

6±5.

5329

.53±

5.48

29.8

5±5.

3730

.40±

5.48

30.8

5±5.

6830

.74±

5.67

30.8

5±5.

86

Blo

od p

ress

ure

— m

m H

g

Syst

olic

—13

7.94

±17.

0112

3.02

±9.3

813

1.58

±15.

3813

9.37

±16.

6314

2.96

±16.

6314

4.84

±17.

1714

7.79

±18.

71

Dia

stol

ic—

79.1

6±9.

5769

.54±

5.88

75.0

8±8.

8980

.08±

9.17

82.3

5±8.

8883

.89±

8.92

84.5

3±9.

88

Mac

roal

bum

inur

ia —

no.

(%

)—

4,69

5 (4

.9)

017

7 (0

.8)

908

(2.3

)1,

802

(7.4

)1,

512

(30.

7)29

6 (1

00)

Estim

ated

GFR

— m

l/m

in/1

.73

m2 ¶

—84

.19±

21.5

282

.29±

20.7

183

.01±

20.5

983

.43±

20.9

485

.90±

22.3

988

.76±

24.7

291

.93±

29.1

6

Trea

tmen

t — n

o. (

%)‖

Stat

in—

57,9

45 (

61.5

)2,

907

(61.

5)13

,312

(60

.4)

24,2

06 (

62.6

)14

,449

(61

.0)

2,86

2 (5

9.8)

209

(73.

1)

Ant

ihyp

erte

nsiv

e ag

ent

—40

,553

(42

.4)

2,93

4 (6

1.0)

11,0

35 (

49.3

)15

,839

(40

.4)

8,80

9 (3

6.7)

1,82

3 (3

7.6)

113

(38.

4)

* Pl

us–m

inus

val

ues

are

mea

ns ±

SD. C

ontr

ols

wer

e m

atch

ed fo

r ag

e, s

ex, a

nd c

ount

y, o

n th

e ba

sis

of d

ata

from

Sta

tistic

s Sw

eden

. All

info

rmat

ion

rega

rdin

g pa

tient

s w

ith ty

pe 2

dia

bete

s w

as

obta

ined

from

the

Nat

iona

l Dia

bete

s R

egis

ter.

Mis

sing

dat

a w

ere

impu

ted

with

the

mul

tivar

iate

impu

tatio

n by

cha

ined

equ

atio

ns (

MIC

E) a

lgor

ithm

. We

impu

ted

five

com

plet

e da

ta s

ets

(for

de

tails

, see

Tab

le S

3 an

d Fi

gs. S

4 an

d S5

in th

e Su

pple

men

tary

App

endi

x). P

erce

ntag

es in

this

tabl

e ar

e ba

sed

on th

e co

mpl

ete-

case

dat

a se

t. LD

L de

note

s lo

w-d

ensi

ty li

popr

otei

n.†

Con

cent

ratio

ns o

f gly

cate

d he

mog

lobi

n w

ere

base

d on

val

ues

from

the

Int

erna

tiona

l Fed

erat

ion

of C

linic

al C

hem

istr

y an

d La

bora

tory

Med

icin

e. P

erce

ntag

es fo

r th

e gl

ycat

ed h

emog

lobi

n le

vel w

ere

base

d on

val

ues

from

the

Nat

iona

l Gly

cohe

mog

lobi

n St

anda

rdiz

atio

n Pr

ogra

m.

‡ T

o co

nver

t va

lues

for

chol

este

rol t

o m

illig

ram

s pe

r de

cilit

er, d

ivid

e by

0.0

2586

.§

The

body

-mas

s in

dex

is t

he w

eigh

t in

kilo

gram

s di

vide

d by

the

squ

are

of t

he h

eigh

t in

met

ers.

¶ T

he g

lom

erul

ar fi

ltrat

ion

rate

(G

FR)

was

est

imat

ed w

ith t

he u

se o

f the

Mod

ifica

tion

of D

iet

in R

enal

Dis

ease

equ

atio

n.‖

Trea

tmen

t w

as a

sses

sed

only

in p

atie

nts

with

dia

bete

s. O

vera

ll, d

ata

on s

tatin

use

wer

e m

issi

ng fo

r 11

06 p

atie

nts,

and

dat

a on

the

use

of a

n an

tihyp

erte

nsiv

e ag

ent

wer

e m

issi

ng fo

r 24

45. P

erce

ntag

es w

ere

calc

ulat

ed o

n th

e ba

sis

of p

atie

nts

with

ava

ilabl

e da

ta.





2 3 4 6 8

Control≥80 yr≥65 to <80 yr≥55 to <65 yr<55 yr

No risk factors≥80 yr≥65 to <80 yr≥55 to <65 yr<55 yr

1 Risk factor≥80 yr≥65 to <80 yr≥55 to <65 yr<55 yr

2 Risk factors≥80 yr≥65 to <80 yr≥55 to <65 yr<55 yr




1

Hazard Ratio (95% CI)

A Excess Mortality in Relation to Range of Risk-Factor Control

0.99 (0.84–1.17)1.01 (0.92–1.12)1.15 (1.00–1.34)1.29 (0.94–1.77)

0.94 (0.88–1.00)1.05 (1.02–1.09)1.23 (1.16–1.31)1.56 (1.34–1.81)

0.99 (0.94–1.04)1.17 (1.13–1.20)1.32 (1.27–1.38)1.68 (1.56–1.80)

1.13 (1.06–1.21)1.46 (1.42–1.50)1.63 (1.55–1.71)2.21 (2.05–2.37)

1.47 (1.28–1.70)2.10 (1.96–2.26)2.53 (2.37–2.70)2.80 (2.51–3.13)

1.39 (0.51–3.80)3.10 (2.53–3.80)3.88 (3.07–4.92)4.99 (3.43–7.27)

ReferenceReferenceReferenceReference

2 3 4 6 8 10








1


B Excess Acute Myocardial Infarction in Relation to Range ofRisk-Factor Control

0.72 (0.49–1.07)0.80 (0.69–0.93)0.93 (0.73–1.18)0.91 (0.62–1.35)

1.05 (0.93–1.19)1.05 (0.97–1.14)1.14 (1.04–1.25)1.46 (1.26–1.69)

1.38 (1.27–1.49)1.44 (1.39–1.50)1.54 (1.44–1.65)2.08 (1.90–2.27)

1.78 (1.60–1.98)2.11 (2.02–2.20)2.16 (2.02–2.31)3.02 (2.80–3.27)

2.32 (1.78–3.01)2.87 (2.62–3.14)3.32 (3.02–3.66)4.56 (4.01–5.18)

3.19 (1.23–8.28)4.60 (3.37–6.29)4.84 (3.78–6.21)7.69 (5.02–11.77)

2 3 4 6 8 10








1


C Excess Stroke in Relation to Range of Risk-Factor Control

0.95 (0.74–1.22)0.90 (0.76–1.06)0.94 (0.72–1.23)1.22 (0.70–2.13)

1.06 (0.95–1.18)1.11 (1.04–1.18)1.27 (1.14–1.41)1.55 (1.23–1.95)

1.13 (1.04–1.24)1.32 (1.26–1.38)1.59 (1.50–1.69)2.04 (1.76–2.36)

1.35 (1.21–1.51)1.73 (1.65–1.82)2.13 (2.01–2.27)2.78 (2.46–3.16)

1.54 (1.12–2.11)2.31 (2.09–2.55)2.66 (2.30–3.08)3.34 (2.72–4.10)

2.65 (0.96–7.30)3.54 (2.36–5.31)2.79 (1.88–4.14)6.23 (3.22–12.05)

2 3 4 5 7 9








1


D Excess Heart Failure in Relation to Range of Risk-Factor Control

1.12 (0.89–1.41)1.42 (1.28–1.58)1.61 (1.31–1.97)2.40 (1.63–3.54)

1.17 (1.08–1.27)1.46 (1.39–1.53)1.80 (1.63–1.98)2.37 (1.99–2.82)

1.23 (1.15–1.32)1.62 (1.56–1.68)2.11 (1.98–2.26)2.71 (2.40–3.05)

1.42 (1.31–1.54)2.01 (1.92–2.10)2.82 (2.63–3.02)3.93 (3.50–4.42)

1.81 (1.42–2.30)2.88 (2.64–3.14)3.85 (3.47–4.26)5.70 (4.84–6.71)

2.76 (0.82–9.25)3.93 (2.75–5.60)6.54 (4.85–8.81)

11.35 (7.16–18.01)








tivity, and atrial fibrillation (Fig. 2C). Levels be-low the guideline target levels for glycated hemo-globin and systolic blood pressure were associated with lower risks of stroke (Fig. 3C).

Hospitalization for heart failure was predict-ed primarily by atrial fibrillation and a body-mass index outside the target range; a low estimated glomerular filtration rate and high glycated he-moglobin level were also strong predictors of this outcome (Fig. 2D). The risk of hospitaliza-tion for heart failure was marginally lower at glycated hemoglobin levels of less than 53 mmol per mole (Fig. 3D).

The glycated hemoglobin level was the stron-gest or the second strongest predictor regarding the risk of the outcomes in five of the eight mod-els (Fig. 2). Smoking was the strongest predictor of death.

Relative Risk-Factor Strength and Explained Log Likelihood

Figures S2 and S3 in the Supplementary Appen-dix show the relative strength of the associations for predictors of cardiovascular outcomes in pa-tients with type 2 diabetes, with or without pre-existing conditions, with the use of explained log-likelihood. The results were broadly consis-tent with the results obtained with the use of explained relative risk (R2) models.

Discussion

Our analysis of Swedish nationwide registry data from 1998 through 2012 showed that patients with type 2 diabetes and five selected risk-factor variables within target range had, at most, mar-ginally higher risks of death, stroke, and myocar-dial infarction than the general population. The study indicates that having all five risk-factor vari-ables within the target ranges could theoretically eliminate the excess risk of acute myocardial in-farction. However, there was a substantial excess risk of hospitalization for heart failure among patients who had all the variables within target ranges. We identified a monotonic relationship among younger age, increasing number of vari-ables not within target ranges, and a higher rela-tive risk of adverse cardiovascular outcomes. The results suggest that there may be greater poten-tial gains from more aggressive treatment in younger patients with diabetes.

The following risk factors were considered to be the strongest predictors for cardiovascular out-comes and death: low physical activity, smoking, and glycated hemoglobin, systolic blood-pressure, and LDL cholesterol levels outside the target rang-es. Using real-world data, we found that levels of glycated hemoglobin, systolic blood pressure, and LDL cholesterol that were lower than target levels were associated with lower risks of acute myocar-dial infarction and stroke.

Randomized trials investigating the effect of multifactorial cardiovascular risk-factor interven-tion in patients with type 2 diabetes are scarce, and contemporary studies were designed to mea-sure the cumulative incidence of cardiovascular events among patients with various risk factors (e.g., hyperglycemia, hypertension, dyslipidemia, and microalbuminuria) who received intensive therapy, as compared with those who received conventional therapy.2,3,8,9 Observational studies and randomized trials have shown inconsistent evidence of effects of glycated hemoglobin levels below contemporary guideline levels (<7.0%) with regard to cardiovascular events and death.10-15

In the present analyses, a glycated hemoglo-bin level outside the target range was a strong predictor for all outcomes, especially for athero-thrombotic events, which shows the importance of dysglycemia with regard to these complications.

Figure 1 (facing page). Adjusted Hazard Ratios for Outcomes, According to Age Category and Number of Risk-Factor Variables outside Target Ranges, among Patients with Type 2 Diabetes, as Compared with Matched Controls.

Hazard ratios show the excess risk of each outcome among patients with type 2 diabetes, as compared with matched controls from the general population, according to age categories and to the number of risk-factor variables (scale, none to five) that were outside target ranges currently recommended in guidelines. The analysis included patients with type 2 diabetes and controls matched for age, sex, and county in Swe-den. We constructed a Cox hazards model for each age category, and these models were adjusted for the covariable “category”; this covariable denotes the number of risk-factor variables that were within target ranges. These Cox model analyses were performed on five imputed data sets for each age category, and haz-ard ratios were pooled from all the data sets with the use of Rubin’s rule.





Figure 2. Relative Importance of Risk Factors for Predicting Death from Any Cause, Acute Myocardial Infarction, Stroke, and Hospitalization for Heart Failure among Patients with Type 2 Diabetes, with or without Preexisting Conditions.

The estimated explained relative risk (i.e., relative importance) shows the strength of the association for various risk-factor variables (with values outside the target ranges) for predicting death (Panel A), acute myocardial infarc-tion (Panel B), stroke (Panel C, facing page), and hospitalization for heart failure (Panel D, facing page) among pa-tients with type 2 diabetes. Results were obtained from the first imputed data set; there were no significant differ-ences between the sets. The analysis was restricted to patients with type 2 diabetes. We constructed a Cox hazard model for each outcome, which included every predictor. We then constructed a separate Cox model for each pre-dictor and permutated covariables from each of these Cox models to estimate the explained relative risk (R2). R2 was generated by developed applications for the Cox model and is bounded between 0 and 1. Risk factors show-ing a clear and substantial R2 measure, as compared with other adjacent predictors, are considered to be relevant. Full definitions of the risk factors and the values that were considered to be outside the target ranges are provided in the Supplementary Appendix. The body-mass index is the weight in kilograms divided by the square of the height in meters. LDL denotes low-density lipoprotein, and GFR glomerular filtration rate.

0.000 0.005 0.010 0.015 0.020

Increasing Importance Increasing Importance


A Death from Any Cause

SmokingPhysical activityMarital statusGlycated hemoglobinLipid-lowering medicationEstimated GFRBody-mass indexIncomeHeart failureCoronary heart diseaseSystolic blood pressureAlbuminuriaEducationDuration of diabetesImmigrantLDL cholesterolDiastolic blood pressureBlood-pressure medication

All Patients

0.000 0.005 0.010 0.015 0.020 0.025

Patients without CoexistingConditions at Baseline

SmokingPhysical activityMarital statusGlycated hemoglobinEstimated GFRLipid-lowering medicationBody-mass indexIncomeAlbuminuriaEducationSystolic blood pressureImmigrantDuration of diabetesLDL cholesterolDiastolic blood pressureBlood-pressure medication

0.000 0.005 0.010 0.015 0.020 0.000 0.005 0.010 0.015 0.020

B Acute Myocardial Infarction

Glycated hemoglobinSystolic blood pressureLDL cholesterolPhysical activitySmokingDuration of diabetesEstimated GFRIncomeDiastolic blood pressureHeart failureBlood-pressure medicationMarital statusEducationAlbuminuriaLipid-lowering medicationImmigrantAtrial fibrillationBody-mass index

All PatientsPatients without Coexisting

Conditions at BaselineGlycated hemoglobinLDL cholesterolSystolic blood pressureSmokingPhysical activityEstimated GFRDuration of diabetesIncomeDiastolic blood pressureMarital statusEducationBlood-pressure medicationAlbuminuriaImmigrantLipid-lowering medicationBody-mass index

R2 R2

R2 R2





Low physical activity was also a strong predictor of cardiovascular outcomes and death, but ran-domized trials have not shown long-lasting ben-eficial effects from increased physical activity in patients with diabetes.16-18

With regard to hospitalization for heart fail-ure, the present analyses showed that the pres-ence of atrial fibrillation, a high body-mass index, and a glycated hemoglobin level and renal func-tion outside the target ranges were the strongest predictors. These findings indicate that cardio-

renal mechanisms may contribute to the develop-ment of heart failure in patients with type 2 dia-betes. A high body-mass index was a stronger risk factor for heart failure than for other out-comes, which may explain why the risks associated with this outcome may continue to be higher among patients with type 2 diabetes than among controls, since patients with diabetes are, on average, heavier than compared controls.

Our study shows, in accordance with previous studies, that lower systolic blood pressure is asso-

D Heart Failure

0.000 0.010 0.020 0.030


Atrial fibrillationBody-mass indexPhysical activityEstimated GFRGlycated hemoglobinCoronary heart diseaseIncomeDuration of diabetesBlood-pressure medicationSystolic blood pressureDiastolic blood pressureMarital statusSmokingEducationLipid-lowering medicationAlbuminuriaLDL cholesterolImmigrant

All Patients

0.00 0.01 0.02 0.03

Body-mass indexGlycated hemoglobinPhysical activitySmokingDuration of diabetesIncomeMarital statusEstimated GFRSystolic blood pressureBlood-pressure medicationDiastolic blood pressureAlbuminuriaEducationLipid-lowering medicationLDL cholesterolImmigrant

0.04


R2 R2

C Stroke

0.000 0.005 0.010 0.015


Glycated hemoglobinSystolic blood pressureDuration of diabetesPhysical activityAtrial fibrillationIncomeMarital statusSmokingEstimated GFRLipid-lowering medicationBlood-pressure medicationLDL cholesterolDiastolic blood pressureBody-mass indexHeart failureAlbuminuriaEducationImmigrant

All Patients

0.000 0.005 0.010

Glycated hemoglobinSystolic blood pressurePhysical activityDuration of diabetesIncomeSmokingMarital statusLipid-lowering medicationEstimated GFRBlood-pressure medicationDiastolic blood pressureLDL cholesterolEducationAlbuminuriaBody-mass indexImmigrant

0.015


R2 R2





Hazard Ratio

2.0

1.6

1.0

1.2

0.8

5075

100

Gly

cate

d H

emog

lobi

n (m

mol

/mol

)

100

150

200

Syst

olic

Blo

od P

ress

ure

(mm

Hg)

2.5

5.0

0.0

7.5

LDL

Cho

lest

erol

(mm

ol/l

iter)

AD

eath

from

Any

Cau

se

Hazard Ratio

2.5

3.0

2.0

1.0

1.5

0.8

1.0

5015

010

0G

lyca

ted

Hem

oglo

bin

(mm

ol/m

ol)

100

150

200

Syst

olic

Blo

od P

ress

ure

(mm

Hg)

2.5

0.0

5.0

7.5

LDL

Cho

lest

erol

(mm

ol/l

iter)

BA

cute

Myo

card

ial I

nfar

ctio

nHazard Ratio

2.0

1.6

1.0

1.2

0.8

5015

010

0G

lyca

ted

Hem

oglo

bin

(mm

ol/m

ol)

100

150

200

Syst

olic

Blo

od P

ress

ure

(mm

Hg)

2.5

5.0

7.5

LDL

Cho

lest

erol

(mm

ol/l

iter)

CSt

roke

Hazard Ratio

4.0

3.0

2.0

5015

010

0G

lyca

ted

Hem

oglo

bin

(mm

ol/m

ol)

100

150

200

Syst

olic

Blo

od P

ress

ure

(mm

Hg)

2.5

0.0

0.0

5.0

7.5

LDL

Cho

lest

erol

(mm

ol/l

iter)

DH

eart

Fai

lure





ciated with lower risks of cardiovascular outcomes and death.19 The Systolic Blood Pressure Interven-tion Trial (SPRINT) showed that systolic blood-pressure targets below guideline levels in patients without diabetes were associated with a lower risk of cardiovascular outcomes and death.20 However, the Action to Control Cardiovascular Risk in Dia-betes (ACCORD) trial examined the same systolic blood-pressure targets in patients with type 2 diabetes (<120 mm Hg vs. <140 mm Hg) and did not show a significant effect on cardiovascular mortality.15,21 Our analysis implies that systolic blood pressure is a central factor for virtually all outcomes in patients with diabetes, and lower lev-els of systolic blood pressure are associated with significantly lower risks of acute myocardial in-farction and stroke among patients with diabetes. The assessment of systolic blood pressure and its relation to death and heart failure is more diffi-cult, owing to potential reverse causality. More-

specific trials of blood-pressure reduction to dif-ferential targets in patients with type 2 diabetes may be warranted.20,22

Our observational study has several strengths but also some notable limitations. Almost all the patients with type 2 diabetes in Sweden were included. The epidemiologic definitions of type 2 diabetes and the outcomes are well validated. We did not consider changes in the risk-factor variables during follow-up, and although this would have some advantages, the approach we used minimizes the risk of reverse causation in the interpretation of the results. In addition, we did not distinguish between patients with all or some variables within target range without any specific intervention and patients who had been medically treated to attain the observed risk-fac-tor levels. We also acknowledge that residual confounding and reverse causation are impossible to overcome fully. Finally, given the observational nature of this work, this cannot be a complete comparison of the effects of treating risk factors; rather, because some patients may have had risk-factor variables in the target ranges without treat-ment, the findings represent the prognostic im-portance of such risk factors for persons with diabetes.

In conclusion, patients with type 2 diabetes who had five risk-factor variables within target ranges appeared to have little or no excess risks of death, myocardial infarction, and stroke as com-pared with the general population.

The views expressed in this article are those of the authors and are not necessarily the views of the Swedish Medical Prod-ucts Agency.

Supported by grants from the Swedish Association of Local Authorities and Regions, the Swedish State under the agreement concerning research and education of doctors, Region Vastra Götaland, the Swedish Diabetes Foundation, the Swedish Heart and Lung Foundation, the Swedish Research Council (2013-5187 SIMSAM), the Wilhelm and Martina Lundgren Science Founda-tion, and the Swedish Council for Health, Working Life, and Welfare.

Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.

References1. Rawshani A, Rawshani A, Franzén S, et al. Mortality and cardiovascular disease in type 1 and type 2 diabetes. N Engl J Med 2017; 376: 1407-18.2. Gaede P, Lund-Andersen H, Parving H-H, Pedersen O. Effect of a multifacto-rial intervention on mortality in type 2 diabetes. N Engl J Med 2008; 358: 580-91.3. Gaede P, Vedel P, Larsen N, Jensen GVH, Parving H-H, Pedersen O. Multifac-

torial intervention and cardiovascular disease in patients with type 2 diabetes. N Engl J Med 2003; 348: 383-93.4. Eliasson B, Gudbjörnsdottir S. Diabe-tes care — improvement through mea-surement. Diabet Res Clin Pract 2014; 106: Suppl 2:: S291-S294.5. Perk J, De Backer G, Gohlke H, et al. European Guidelines on cardiovascular disease prevention in clinical practice

(version 2012): the Fifth Joint Task Force of the European Society of Cardiology and Other Societies on Cardiovascular Dis-ease Prevention in Clinical Practice (con-stituted by representatives of nine societ-ies and by invited experts). Eur Heart J 2012; 33: 1635-701.6. Inzucchi SE, Bergenstal RM, Buse JB, et al. Management of hyperglycemia in type 2 diabetes, 2015: a patient-centered

Figure 3 (facing page). Association between Levels of Glycated Hemoglobin, Systolic Blood Pressure, and LDL Cholesterol and Death from Any Cause, Acute Myocardial Infarction, Stroke, and Heart Failure in Patients with Type 2 Diabetes.

We constructed a Cox model for each outcome and applied a prediction function to assess the relation-ship between glycated hemoglobin, systolic blood-pressure, and LDL cholesterol levels and the risks of death from any cause (Panel A), acute myocardial in-farction (Panel B), stroke (Panel C), and hospitaliza-tion for heart failure (Panel D). Reference values were the contemporary guideline levels: 53 mmol per mole for the glycated hemoglobin level, 140 mm Hg for the systolic blood pressure, and 2.5 mmol per liter (97 mg per deciliter) for the LDL cholesterol level. The dark lines indicate the hazard function, and the shaded ar-eas 95% confidence intervals. Continuous variables were modeled with restricted cubic splines, whereas all the categorical variables were stratified. This analy-sis was restricted to patients with type 2 diabetes. To convert values for cholesterol to milligrams per decili-ter, divide by 0.02586.





approach: update to a position statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care 2015; 38: 140-9.7. Heller G. A measure of explained risk in the proportional hazards model. Bio-statistics 2012; 13: 315-25.8. Wan EYF, Fung CSC, Yu EYT, et al. Ef-fect of multifactorial treatment targets and relative importance of hemoglobin A1c, blood pressure, and low-density lipo-protein-cholesterol on cardiovascular dis-eases in Chinese primary care patients with type 2 diabetes mellitus: a popula-tion-based retrospective cohort study. J Am Heart Assoc 2017; 6(8): e006400-e006414.9. Gæde P, Oellgaard J, Carstensen B, et al. Years of life gained by multifactorial intervention in patients with type 2 diabetes mellitus and microalbuminuria: 21 years follow-up on the Steno-2 randomised trial. Diabetologia 2016; 59: 2298-307.10. Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HAW. 10-Year follow-up of intensive glucose control in type 2 diabetes. N Engl J Med 2008; 359: 1577-89.11. Stratton IM, Adler AI, Neil HA, et al. Association of glycaemia with macrovas-cular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ 2000; 321: 405-12.12. Ray KK, Seshasai SR, Wijesuriya S, et al.

Effect of intensive control of glucose on cardiovascular outcomes and death in pa-tients with diabetes mellitus: a meta-analysis of randomised controlled trials. Lancet 2009; 373: 1765-72.13. Stratton IM, Cull CA, Adler AI, Mat-thews DR, Neil HAW, Holman RR. Addi-tive effects of glycaemia and blood pres-sure exposure on risk of complications in type 2 diabetes: a prospective observa-tional study (UKPDS 75). Diabetologia 2006; 49: 1761-9.14. Patel A, MacMahon S, Chalmers J, et al. Effects of a fixed combination of per-indopril and indapamide on macrovascu-lar and microvascular outcomes in pa-tients with type 2 diabetes mellitus (the ADVANCE trial): a randomised controlled trial. Lancet 2007; 370: 829-40.15. Ismail-Beigi F, Craven T, Banerji MA, et al. Effect of intensive treatment of hy-perglycaemia on microvascular outcomes in type 2 diabetes: an analysis of the ACCORD randomised trial. Lancet 2010; 376: 419-30.16. The Look AHEAD Research Group. Association of the magnitude of weight loss and changes in physical fitness with long-term cardiovascular disease out-comes in overweight or obese people with type 2 diabetes: a post-hoc analysis of the Look AHEAD randomised clinical trial. Lancet Diabetes Endocrinol 2016; 4: 913-21.

17. Zethelius B, Gudbjörnsdottir S, Elias-son B, Eeg-Olofsson K, Cederholm J. Level of physical activity associated with risk of cardiovascular diseases and mor-tality in patients with type-2 diabetes: re-port from the Swedish National Diabetes Register. Eur J Prev Cardiol 2014; 21: 244-51.18. Wing RR, Reboussin D, Lewis CE. In-tensive lifestyle intervention in type 2 dia-betes. N Engl J Med 2013; 369: 2358-9.19. Stevens RJ, Kothari V, Adler AI, Strat-ton IM. The UKPDS risk engine: a model for the risk of coronary heart disease in Type II diabetes (UKPDS 56). Clin Sci Lond 2001; 101: 671-9.20. The SPRINT Research Group. A ran-domized trial of intensive versus standard blood-pressure control. N Engl J Med 2015; 373: 2103-16.21. Margolis KL, O’Connor PJ, Morgan TM, et al. Outcomes of combined cardio-vascular risk factor management strate-gies in type 2 diabetes: the ACCORD ran-domized trial. Diabetes Care 2014; 37: 1721-8.22. Buckley LF, Dixon DL, Wohlford GF IV, Wijessinghe DS, Baker WL, Van Tas-sell BW. Intensive versus standard blood pressure control in SPRINT-eligible par-ticipants of the ACCORD-BP Trial. Diabe-tes Care 2017; 40: 1733-8.Copyright © 2018 Massachusetts Medical Society.

clinical trial registration

The Journal requires investigators to register their clinical trials in a public trials registry. The members of the International Committee of Medical Journal Editors (ICMJE) will consider most reports of clinical

trials for publication only if the trials have been registered. Current information on requirements and appropriate registries

is available at www.icmje.org/about-icmje/faqs/.



risk factors, mortality, and cardiovascular … engl j med 379;7 nejm.orgaugust 16, 2018 635 risk...

Documents