Changes in levels of haemoglobin A1c during the first 6 years after

diagnosis of clinical type 2 diabetes

Clinical implications

Niels de Fine OlivariusVolkert SiersmaLars J. Hansen

Thomas DrivsholmMogens Hørder

Glucose-lowering trials in type 2 diabetes

UGDP1970

: 2.9 times higher death rate from CVD in tolbutamide group (p=0.005 vs. placebo)

UKPDS1998

: 96% increase in DM-related death for those treated with both metformin and sulfonylurea (p=0.04 vs. conventional / sulphonylurea)

Metaanalysis 2007

: Rosiglitazone: 64% increase in deaths from CVD (p=0.06 vs. placebo or other agents)

ACCORD2008

: 35% increase in deaths from CVD (p=0.02 vs. less intensive glucose-lowering therapy)

Lesson to be learned

Treatment regimens aiming at normalizing blood glucose for all patients may be detrimental,

at least in middle-aged and older adults with diabetes of long duration.

Against generalized thought

Average figures used to describe changes in HbA1c may conceal a clinically important diversity in the way the blood glucose level deteriorates or even improves, especially immediately after diabetes diagnosis.

Objective

To assess the variability in levels of glycosylated haemoglobin (HbA1c) during the first six years after diagnosis of clinical type 2 diabetes in relation to possible predictors.

Flow chart

Newly diagnosed diabetic patients from the intervention

group, n=718

Patients who did not have at least three measurements of HbA1c, n=97

Missing HbA1c-measurement within 90 days after diabetes diagnosis, n= 40

Patients for analysis of predictors of course of HbA1c, n=621

Patients for figures showing the course of HbA1c, n=581

All measurements of HbA1c

Upper limit of reference range

HbA1c for 50 randomly selected patients

Modelling the course of HbA1c

Years since diabetes diagnosis

Hae

mog

lobi

n A1

c (%

)

Level of HbA1c

one year afterdiabetes diagnosis

Slope of theHbA1c-curve

Predictors of level and slope of HbA1c

Gender

Age

BMI

HbA1c

Systolic BP

Total cholesterols

Urinary albumin

Multivariate linear regression models

Predictors of level and slope of HbA1c

Gender

Age

BMI

HbA1c

Systolic BP

Total cholesterols

Urinary albumin

Diabetic retinopathy

Coronary heart disease

Peripheral vascular disease

Peripheral neuropathy

Familial disposition to diabetes

Smoking

Physical activity

Marital status

Residence

GP's acquaintance with the patient

Multivariate linear regression models

Predictors of level and slope of HbA1c

Gender

Age

BMI

HbA1c

Systolic BP

Total cholesterols

Urinary albumin

Diabetic retinopathy

Coronary heart disease

Peripheral vascular disease

Peripheral neuropathy

Familial disposition to diabetes

Smoking

Physical activity

Marital status

Residence

GP's acquaintance with the patient

Multivariate linear regression models

SLOPE

LEVEL

Changes in HbA1c according to HbA1c at diagnosis

Years since diabetes diagnosis

HbA1c > 10%

Hae

mog

lobi

n A 1c

(%)

HbA1c ≤ 10%

Changes in HbA1c according to age

Years since diabetes diagnosis

Hae

mog

lobi

n A 1c

(%)

Age ≤ 65 y.

Age > 65 y.

Years since diabetes diagnosis

HbA1c > 10 at diagnosis

Hae

mog

lobi

n A 1c

(%)

Patients are grouped according to quartiles of the slope of HbA1c after 1-year follow-up

Changes in HbA1c (I)From generalisation to individualisation

Changes in HbA1c (II)From generalisation to individualisation

Years since diabetes diagnosis

HbA1c ≤ 10 at diagnosis

Hae

mog

lobi

n A 1c

(%)

Patients are grouped according to quartiles of the slope of HbA1c after 1-year follow-up

Conclusion 1

ACCORD – ADVANCE – VADT – UKPDS – UGDP

Lowering glucose levels for all patients with T2DM

has limited benefit and can lead to significant harm

Conclusion 2

Diabetes Care in General PracticeIndividualization:

A more appropriate strategy to reduce the level and variability of HbA1c could be to

monitor changes in HbA1c more closely and intensify treatment of those patients who

actually experience the beginning of an apparently inexorable deterioration of their

glycaemic control

Years since diabetes diagnosis

HbA1c > 10 at diagnosis

Hae

mog

lobi

n A 1c

(%)

Patients are grouped according to quartiles of the slope of HbA1c after 1-year follow-up

Changes in HbA1c (I)From generalisation to individualisation

DCGP vs. UKPDS

Years since diabetes diagnosis

Fasti

ng p

lasm

a gl

ucos

e (m

mol

/l)

conventional treatment

intensive treatment

structured care

comparisongroup (=conventionaltreatment)

UKPDSDCGP

DCGP vs. UKPDS

% treated with diet alone

Fasti

ng p

lasm

a gl

ucos

e (m

mol

/l)

UKPDSDCGP

90%70%

45%

17%7%

13%

55%40%

29% 32%

Years since diabetes diagnosis

HbA1ccross-sectional, median values

06

7

8

9

0 3 6 9 12 15

HbA

1c

(%)

Years from randomisation

Conventional

Intensive

UKPDS