Diabetes. Professor Philip Clarke Centre for Health Policy - - PowerPoint PPT Presentation

Conducting economic analyses alongside clinical studies in Type 1 Diabetes.

Professor Philip Clarke Centre for Health Policy Melbourne School of Population of Global Health

Inter-generational Report 2015 Does this make you “fall of your chair”?

• Longer life expectancy:

– Males 91.5 years at birth; – females 93.6 years

• Australian Government real health expenditure per

person is projected to more than double over the next 40 years ($2,800 to around $6,500).

• Key drivers of health expenditure is not aging, but

– demand from rising income, – technological change. How long will a child born today with Type 1 diabetes live?; How much will it cost to treat/ cure them?

Lung 2014 (PLOS One)

• The relative mortality of people

with Type 1 Diabetes has improved

• Relative risk (pre-1971): 6

times general population

• Relative Risk (post-1990) : 3

times general population

• There is still a life expectancy

gap of 11-13 years

Lung PLOS One 2014

2 0 4 0 6 0 8 0 2 0 4 0 6 0 8 0 1 0 0

A ll C a u s e M o rta lity

A ge P ro p o rtio n a liv e (% )

2 0 4 0 6 0 8 0 2 0 4 0 6 0 8 0 1 0 0

A ll C a u s e M o rta lity

A ge P ro p o rtio n a liv e (% ) N D R 2 0 0 2 -0 6 N D R 2 0 0 7 -1 1 G e n e ra l p o p u la tio n 2 0 0 2 -0 6 G e n e ra l p o p u la tio n 2 0 0 7 -1 1

2 0 4 0 6 0 8 0 2 0 4 0 6 0 8 0 1 0 0

C V D M o rta lity

A ge P ro p o rtio n e v e n t fre e (% )

2 0 4 0 6 0 8 0 2 0 4 0 6 0 8 0 1 0 0

C V D M o rta lity

A ge P ro p o rtio n e v e n t fre e (% )

M en W o m e n

Current trends in mortality (unpublished)

• Evidence of

LE gap closing in men due to reductions in CVD mortality

• No evidence in

women for improvements in

• Rel. Survival.

Case study: Insulin analogues Insulin analogues “afford more flexible treatment regimens with a lower risk of the development of hypoglycemia” (NEJM 2005) Benefits:

• Rapid acting analogues

reduce postprandial hyperglycemia (high blood sugar after meals)

• Long acting analogues

reduce the risk of hypoglycemia

Now a look at the cost…

• Australian prices:

INSULIN ISOPHANE HUMAN: $224 per script INSULIN GLARGINE: $433 per script

• The increase in expenditure reflects:

Expanded use for people with Type 2 Diabetes The higher price of insulin analogues

What are the benefits of reducing “hypos”?

1)Quality of life

a) Patients value not having hypoglycemic episodes – they have reduced quality of life during an episode and long-term due to increased complications. b) Perhaps all people with Diabetes are affected by a “fear of hypos” 2) Clinical impact There is emerging evidence of increased mortality (particularly after cardiovascular events)

Recent evidence of survival post CVD event

Lung et. al., Diabetes Care, 2014

Are Analogues “value for money”?

• In Australia these decisions are made by the

Pharmaceutical Benefits Advisory Council (PBAC) on the basis of cost-effectiveness.

• Example of Insulin Glargine:

– Considered 5 times by the economic sub-committee of the PBAC – First considered by the PBAC in 2003 – Finally listed in 2006 – Company projected to cost $145 million over first four years (actual cost $263million)

What the PBAC thought…

Comments from Oct 2005 Meeting: “A number of problems with this analysis were identified during the evaluation, and the PBAC considered that the trial-based incremental costs per extra hypoglycaemic event avoided could be higher than estimated in the submission” “The PBAC did not accept other assumptions in the economic model” Recommendation: Reject

Food for thought: health system performance

Source: McKeon Review

UKPDS – Example of economic evaluation alongside an RCT

UKPDS overview

• First of the large scale RCTs in Type 2

diabetes – involving 10 years of follow-up

• n 5102 patients
• Economics integrated into the study

prospectively

• Included collection of cost and quality of

life information

• Data formed the basis of development of

the UKPDS Outcomes model

UKPDS – intermediate outcomes

• First major RCT

in Type 2 diabetes

• Intensive vs

conventional Blood Glucose control

• Very long term

follow-up (median 10.3 years)

UKPDS – final outcomes

At the end of follow-up At the end of follow-up

Example CEA

Blood Glucose & Blood pressure Study: Aims To determine the cost per Quality Adjusted Life Year (QALY) gained of three UKPDS policies:

• Intensive blood glucose control: is intensive blood

glucose control with sulphonylurea, or insulin, cost- effective in preventing clinical complications in people with Type 2 Diabetes?

• Metformin: is intensive blood glucose control with

Metformin in overweight patient cost-effective in preventing complications?

• Tight blood pressure control: is tight blood pressure

control cost-effective in preventing complications?

Main economic evaluation

Comparison: intensive versus conventional blood glucose and blood pressure control policies in Type 2 diabetic patients Data: patient level data on costs and outcomes from UKPDS Outcome measure: Quality adjusted life years Time period: within trial effects extrapolated over lifetime (median follow-up 10 years for Blood Glucose and 8.4 years for Blood Pressure Trial) Perspective: UK health care system Result: Incremental cost per QALY gained in 2004 £s (£1 = 1.5 Euros)

Costs and resource use

• Therapy:
• drug doses (antidiabetic, antihypertensive, other)
• tests (blood glucose, HbA1C)
• standard practice (cost of implementing UKPDS policies

in a general health care setting)

• Complications:
• hospital admissions: length of stay, diagnosis
• non-inpatient services: home, clinic & telephone

contacts with GPs, nurses, dieticians etc. Cost of complications is modelled over each patient’s lifetime

Unit costs

Costs: Blood pressure study

Total discounted costs

Mean costs per patient over trial period, £s, 3.5% discount Item Mean (SD) conventional Mean (SD) intensive Mean difference (95% C.I.)

Intensive BG 14,984 (17,888) 15,868 (14,465) 884(-483, 2250) Metformin 16,941 (23,193) 15,920 (13,678)

• 1021(-4291, 2249)

Blood pressure 15,786 (16,378) 15,895 (16,025) 108(-2347, 2563)

Estimating Outcomes

• Estimate QALYs for people with Type 2 diabetes, based on

profile of complications of each patient over their remaining lifetime

• Use UKPDS Outcomes Model which is based on an

integrated set of parametric proportional hazard models to predict absolute risk of first occurrence of seven major diabetes-related complications, using: – patients’ characteristics (e.g. age and sex) – time varying risk factors (e.g. HbA1c)

Model equations: Complications

Simulating lifetime outcomes

Average QALY for UKPDS patients by blood glucose policy

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Years (from diagnosis of diabetes) QALY (Utility weight)

Intensive Therapy Conventional therapy Median duration of the trial Minimum follow-up Maximum follow-up

Total outcomes

Mean QALYs per patient, undiscounted Item Mean (SD) conventional Mean (SD) intensive Mean difference (95% C.I.)

Intensive BG 16.35 (8.36) 16.62 (8.35) 0.27 (-0.48, 1.03) Metformin 16.44 (8.49) 17.32 (7.94) 0.88 (-0.54, 2.29) Blood pressure 13.71 (8.00) 14.16 (7.81) 0.45 (-0.70, 1.60)

Cost-utility results

• £4,000
• £2,000

£0 £2,000 £4,000

• 2.0
• 1.0

0.0 1.0 2.0 3.0

Intervention more effective, more costly Intervention less effective, more costly Intervention less effective, less costly Intervention more effective, less costly

Tight blood pressure control £370 per QALY INCREMENTAL QALYs INCREMENTAL COST

Costs & QALYs discounted at 3.5% p.a.

Intensive blood glucose control with sulphonylurea/insulin £6000 per QALY Intensive blood glucose control with metformin

Development of a Type 1 diabetes simulation model

Rationale for a specific Type 1 Diabetes simulation model

• Much longer durations of diabetes
• Absolute risk is different
• Different effect of risk factors on
• utcomes

Figure 1: Risk of CVD event for two age Swedish NDR cohorts people with Type 1 diabetes:

Person aged 35-45 years

50-60 years

(Unpublished)

NHMRC Project Grant (2012-14): Collaboration to build a type 1 diabetes model

• University of Melbourne
• University of Tasmania (Prof Andrew Palmer)
• DCCT/EDIC collaboration
• Swedish National Diabetes Register
• Glasgow University
• Data sources:

– Swedish National Diabetes Register (NDR) – DCCT/EDIC

• US trial of intensive versus conventional control

– Scottish SCI-Diabetes

Swedish National Diabetes Registry

Clinic Data (1996-2010)

• Demographics
• Clinical risk factors
• self-reported events

Hospitalisation Data (1986-2010)

• Length of stay
• Primary and secondary ICD admission
• Operation codes

Mortality Data (until 13/12/2012)

• ICD codes for cause of death

(available until 31/12/2011)

Prescription Data (2005-2011)

• Drug
• Date filled
• Volume

Inclusion criteria

• Diagnosed with diabetes
• Aged 29 or under at diabetes diagnosis (recorded as insulin only)
• 1 clinic visit between 2002-2011

Exclusion criteria

• Prescribed metformin
• Alive but no insulin prescriptions recorded
• Myocardial Infarction
• Stroke
• Heart Failure
• Amputation
• Hypoglycaemia
• Hyperglycaemia
• End-stage renal disease
• Ischaemic heart disease & unstable

angina

• Percutaneous coronary intervention
• Coronary artery bypass graft
• Age at Diabetes
• HbA1c
• BMI
• Total Cholesterol ratio
• High density lipoprotein
• Low density lipoprotein
• Triglycerides
• Systolic blood pressure
• Estimate glomerular filtration rate
• Smoking
• Microalbuminuria
• Macroalbuminuria

N=27,964 Age (first visit) 37.37 Males 54.7% Deaths 2387(8%) Average follow- up 8 years

Event Risk Equations (clinical risk factors)

Time at risk based on age

Event equations

Mortality Macrovascular (Weibull) Microvascular (Wei) Significant clinical risk factors Gompertz MI/PCI/CABG IHD Stroke CHF Amp Renal failure

Male

+ +

Age of Diagnosis

• ln(HbA1c)
• +

ln(wHbA1c)

+ + + + + + +

BMI

• Blood Pressure (systolic)

+ + +

Triglycerides

+ +

• HDL cholesterol

+

• LDL cholesterol

+

Micro albuminuria

+ + + + + +

Macro albuminuria

+ + + +

ln(eGFR)

• Year
• Smoker

+ + +

Former Smoker

+ + +

Sweden (NDR), DCCT/EDIC (Conventional & Intensive)

0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 30 35 40 45 50 55 60 65 70

CVD free NDR DCCT Conventional DCCT Intensive

Using Swedish NDR Risk Equations to predict DCCT groups

0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 30 35 40 45 50 55 60 65 70

CVD free

Age

NDR DCCT Conventional Conventional NDR predict DCCT Intensive

NDR equation over predicts events in both DCCT groups but results in a similar separation between groups

Main Points

• Increasing need to demonstrate that new

technologies represent value for money

• Type 1 diabetes simulation models are

tools to summarize epidemiology and trial evidence to predict long-term outcomes

• Economic evaluation is best built in

prospectively

• Clinicians & economists need to work on

ways to close the life-expectancy gap for people with Type 1 diabetes