Genetics and Underwriting in Health Insurance Angus Macdonald - - PowerPoint PPT Presentation

Genetics and Underwriting in Health Insurance

Angus Macdonald

Heriot-Watt University, Edinburgh

Outline

N Genetic epidemiology N Example 1 – Alzheimer’s disease N Example 2 – Coronary heart disease

Single-Gene Disorders

N Extra morbidity and mortality - high N Age at onset or death - often much younger

than average, with high probability

N Comparatively rare - about 1% of births N Many sufferers will not survive to insuring

ages

N Probably a strong family history

Multifactorial Disorders

N Associated with common causes of death N May occur in healthy lives N May be a family history N May interact with environmental factors N Extra mortality - variable, often quite low N Age at death/illness - not known with much

greater probability

What Do Genetic Epidemiologists Study?

N Penetrance: p(x) = the probability that a

given genotype will cause disease by age x

N Mutation frequency in the population N Survival after onset of disease N Progression or stages of disease N Gene-environment interactions

Ascertainment Bias

N A significant feature is ascertainment bias

– Search the world for unusual families with many affected members in several generations – Estimate rates of onset based on affected members of these families

N Result: Penetrance estimates may be greatly

• verstated

Multifactorial Epidemiology

N Molecular epidemiology – find out which

genes are “switched on” during specific biomedical processes

N Population epidemiology – find out which

combinations of genetic variants and environment influence disease

UK Biobank

N Recruit 500,000 healthy people age 45-69 N Obtain DNA N Obtain data on environmental exposures N Follow up via doctors, hospitals and disease

registers

N Use as source for large scale case-control

studies from 5 years onwards

UK Biobank – Value for Money?

N Cost of data collection alone ~ £40 million

– Medical Research Council, Department of Health and Wellcome Trust

UK Biobank – Value for Money?

N Cost of data collection alone ~ £40 million

– Medical Research Council, Department of Health and Wellcome Trust

N Power to detect associations?

– 50% of population with “interesting” genotype and “interesting” exposure – 4 controls per case – 3,600 cases needed to detect 30% extra risk

UK Biobank – Value for Money?

N Cost of data collection alone ~ £40 million

– Medical Research Council, Department of Health and Wellcome Trust

N Power to detect associations?

– 10% of population with “interesting” genotype and “interesting” exposure – 4 controls per case – 12,600 cases needed to detect 60% extra risk

Pharmacogenomics

N (1) Use molecular medicine to develop new

drugs

N (2) Use genetic profiling to target specific

drugs to persons who will respond well

N Unknown impact on:

– Drugs costs, in public and private medicine – Rationing of expensive medicine – Affordability of pooled medical costs

Alzheimer’s Disease

N Alzheimer’s Disease

– progressive dementia – onset mostly at ages > 65 – accounts for a significant part of LTC costs – no known cure or effective treatment

Alzheimer’s and the ApoE Gene

N The ApoE gene has 3 alleles: e2, e3, e4 N e4 allele predisposes to earlier onset of AD N e4/e4

– about 2% of the population – highest risk - up to 10-12 x, males and females

N e3/e4

– about 20% of the population – females at up to 4-5 x normal risk

Learning About ApoE

N Early 1980s – three variants of the ApoE

protein identified

N 1980s – e4 allele linked to heart disease N 1991 – e4 allele linked to Alzheimer’s

disease

N 1997 – Age-dependent risk estimates

published

A Model of AD and Care Costs

Healthy Onset of AD In Institution Dead

A Model of AD and Care Costs

e2/e2, e2/e3 e3/e3 e2/e4 e3/e4 e4/e4

Features of the Model

N The “normal” level of insurance N The extent of genetic testing N The probability of a positive result N The behaviour of “adverse selectors” N The behaviour of insurers N The amount and incidence of medical costs N The amount and incidence of care costs

LTC Insurance Policies

N LTC policies pay out while suffering:

– typically, loss of 3 or more Activities of Daily Living (ADLs) ; or – cognitive impairment, such as AD

N Payments are usually linked to an index N Fewer than 30,000 policies sold in the UK N AD accounts for 1/2 - 1/3 of costs (USA)

A Model of AD and Care Costs

Healthy Onset of AD In Institution Dead

Extra Premiums? Female age 60

Proportion of Relative Risk e4/e4 e3/e4 e2/e4 % % % 1.00 37 23 21

Population Risk: Parameter m

N Rates of onset based on case-control studies N Expect strong selection bias, lower genetic

risk in a population sample

N Modelled by reducing excess e4 onset rate

– by 50% (m=0.5) – by 75% (m=0.25)

N Delphic estimates may be even lower

Extra Premiums? Female age 60

Proportion of Relative Risk e4/e4 e3/e4 e2/e4 % % % 1.00 37 23 21 0.50 21 13 12 0.25 11 7 6

Long-Term Care Costs of AD

N Holmes et al. (1998) model:

– £41,794 p.a. PLUS – £436.6 p.a, for each year since onset MINUS – £336.0 p.a., for each year of age PLUS – £17,840 p.a., if in an institution

N Includes costs of unpaid care

AD Costs at Age 60 (£,000)

e2/e3 e3/e3 e2/e4 e3/e4 e4/e4 Ave M 17 44 28 37 95 39 F 37 56 93 95 116 64

Population Risk: m=0.25

e2/e3 e3/e3 e2/e4 e3/e4 e4/e4 Ave M 33 38 35 37 50 37 F 55 58 71 70 78 62

Combined Pension and LTC?

N AD has opposite effects on pensions and

LTC:

– Increases LTC costs – Decreases pension costs

N Pension/LTC (AD) benefit in ratio 1:3

– Average pension = £3,200 p.a. – Average LTC benefit = £9,600 p.a.

Extra Premiums for Combined Pension + LTC? Female age 60

Prop’n

• f RR

e4/e4 e3/e4 e2/e4 % % % LTC only 1.00 37 23 21 0.25 11 7 6 LTC+Pen 1.00 3 3 2 0.25 1 1

Conclusions

N e4/e4 as a separate risk group? N Timescale of ~10 years between discovery

and reliable assessment

N Great uncertainty about results: N Recommend continuing research:

– Development of actuarial models – More collaboration/interdisciplinary work

Coronary Heart Disease (CHD)

N Major common cause of death N Caused by fatty deposits in coronary

arteries

N Outcome myocardial infarction (MI,

meaning heart attack)

N Genetic component is multifactorial

CHD

N “Lifestyle” risk factors:

– Sex, smoking, body mass index (BMI)

N “Clinical” risk factors:

– Hypertension – Hypercholesterolaemia – Diabetes

N These are also risk factors for stroke

Examples of Extra Premiums

N Model from Macdonald, Waters &

Wekwete (2004) based on Framingham study

N Compare a mutation that increases the

intensity of:

– Progression through a clinical risk factor – MI or stroke directly

Examples of Extra Premiums

30% Stroke (directly) 192% CHD (directly) 5% Diabetes type 2 2% Diabetes type 1 6% Hypercholesterolaimia 14% Hypertension Extra Premium 5 x risk of

Male, 35, N-S, Normal BMI, 10-year term

Examples of Extra Premiums

203% Stroke (directly) 597% CHD (directly) 105% Diabetes type 2 99% Diabetes type 1 113% Hypercholesterolaemia 97% (basic) Extra Premium 5 x risk of

Male, 35, N-S, Normal BMI, Severe Hypertension, 10-year term

Conclusion

N Genotypes that confer high additional risks

• f risk factors in complex disorders do not

imply large additional risks of the disease endpoints

N The genetic contributions to complex

disorders will mostly act in this way

N Important message for public debate

What is the Purpose of Research?

N NOT to allow more discrimination N Main purpose is to obtain quantitative

information to:

– inform policymakers; – identify potential problems; – allay unnecessary fears; and – help to achieve fairness in provision

Genetics and Underwriting in Health Insurance

Angus Macdonald

Heriot-Watt University, Edinburgh