05Mar14 - Missing 90 year olds

Missing 90 year olds

An ILC-UK debate supported by Partnership as part of the Population

Patterns Series

Wednesday 5th March 2014

This event is kindly supported by Partnership

#missing90s #populationpatterns

Baroness Sally Greengross

Chief ExecutiveILC-UK



Richard Willets

Director of LongevityPartnership



‘Missing’ 90 year oldsAn Introduction

Richard Willets

International Longevity Centre – 5 March 2014

Introduction

April 8, 2023

5

• Source: Wall Street Journal

• Population estimates for England & Wales based on the 2011 Census were published in July 2012

• Prior to this our population estimates were based on the 2001 Census with adjustments made to allow for subsequent migration and deaths

Impact of the 2011 Census on population estimates

April 8, 2023

6

• Source: own calculations using ONS data

Revision to England & Wales population estimates for mid-year 2011 following 2011 Census, by ageRevision to England & Wales population estimates for mid-year 2011 following 2011 Census, by age

40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89 90-94 95-99 100+

-20%

-15%

-10%

-5%

0%

5% Females

Impact of the 2011 Census on population estimates

April 8, 2023

7


Revision to England & Wales population estimates for mid-year 2011 following 2011 Census, by ageRevision to England & Wales population estimates for mid-year 2011 following 2011 Census, by age

40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89 90-94 95-99 100+

-20%

-15%

-10%

-5%

0%

5% FemalesMales

Approx 30,000 fewer individuals

More significant revisions in the US

April 8, 2023

8

• Source: Wall Street Journal

• In 2004 the US Census Bureau projected there would be:-• 114,000 Americans aged 100 plus in 2010• 1.1 million centenarians in 2050

• Following the 2010 Census, figures revised to:-• 53,364 Americans aged 100 plus in 2010• 0.59 million centenarians in 2050

Impact on apparent mortality improvement rates

April 8, 2023

9


Average annual rate of mortality improvement, males in England & Wales, 2001-2011, by age group, before and after publication of the 2011 Census results


80-84 85-89 90-94 95-990.0%

0.5%

1.0%

1.5%

2.0%

2.5%

3.0% 2010-based


April 8, 2023

10




80-84 85-89 90-94 95-990.0%

0.5%

1.0%

1.5%

2.0%

2.5%

3.0% 2010-based

revised


April 8, 2023

11




80-84 85-89 90-94 95-990.0%

0.5%

1.0%

1.5%

2.0%

2.5%

3.0% 2010-based

revised

2001 to 2013

Impact on projected mortality rates

April 8, 2023

12


Average mortality rate for males in the 90-99 age range, England & Wales, actual and projected figures


1990

1992

1994

1996

1998

2000

2002

2004

2006

2008

2010

2012

2014

2016

2018

2020

2022

2024

2026

2028

2030

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40Actual

Simple extrapo-lation

Impact on projected mortality rates

April 8, 2023

13




1990

1992

1994

1996

1998

2000

2002

2004

2006

2008

2010

2012

2014

2016

2018

2020

2022

2024

2026

2028

2030

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40ActualSimple extrapo-lation2010-based pro-jection

+50%

Consequences

April 8, 2023

14

• *Source: ONS • **Source: own calculation

• Greater appreciation of the challenges in producing accurate data for the highest age groups

• Small but material reductions in the projected lifespans of pensioners/annuitants

• Significant revisions in the projected number of elderly individuals

The ONS publication “What are the chances of reaching age 100?” published in Spring 2012 projected that 9.5% (37,000) of men aged 65 in the UK in 2012 would reach age

100*

Arguably, a more realistic survival probability could be 5-6%**

The equivalent publication in 2013 had a principal projection of 8% (31,000) of men aged 65 in 2013 reaching age 100*

• Raises more fundamental questions (e.g. why is rate of mortality improvement so much lower at high ages?)

Thank you

April 8, 2023

15

Partnership is a trading style of the Partnership group of Companies, which includes; Partnership Life Assurance Company Limited (registered in England and Wales No. 05465261), and Partnership Home Loans Limited (registered in England and Wales No. 05108846).

Partnership Life Assurance Company Limited is authorised by the Prudential Regulation Authority and regulated by the Financial Conduct Authority and the Prudential Regulation Authority. Partnership Home Loans Limited is authorised and regulated by the Financial Conduct Authority. The registered office for both companies is Sackville House, 143-149 Fenchurch Street, London EC3M 6BN.

Dave Grimshaw FIA

PartnerBarnett Waddingham LLP



UK Actuarial Advisory Firm of the Year

Understanding mortality at 90+ years

ILC-UK Population Patterns Series event: ‘Missing 90 year olds’

Dave Grimshaw, Partner, Barnett Waddingham LLP

[email protected]

05 March 2014

mailto:[email protected]

BACKGROUND

18

Why are actuaries interested in mortality?

19

Where a payment is contingent on survival• Valuing liabilities for pension schemes• Pricing and valuing insurance products:

• Where a benefit is payable on death, and• Where a benefit is payable throughout life

We are interested in current mortality and how it might change in future

Mortality rates and mortality improvements

Mortality rate Deaths ÷ Exposure• Example:

• Age 83 in 2005 9,227 ÷ 92,381 = 9.99%

• Age 83 in 2006 8,779 ÷ 89,958 = 9.76%

Mortality improvement• Percentage reduction in mortality rate compared to

the same age one year earlier• Example:

• Mortality reduction: 0.23%• Mortality improvement: 0.23% ÷ 9.99% = 2.3%

20

What data do actuaries use?

21

• Specific portfolio• Data for the (pension scheme or insurance company)

• Pooled studies• e.g. the CMI produces analyses of pension schemes and

various insurance products

• Population data

Why is population data useful?

22

Preference for similar data i.e. the specific group of pensioners or insured lives but…• Pension schemes can be small and heterogeneous• CMI pensioners dataset is approximately 24% of the

population for Males and only 8% for Females• Population data adds credibility, especially where

data is sparse (e.g. oldest or youngest ages)• Insurance/pensioner data may be unreliable• Mortality tends to converge at older ages• Mortality improvements need especially large

datasets

Unreliability of insured and pensioner data at older ages?

23

Comparison of observed experience with graduations

Source: CMI Working Paper 35 (for S1PML) and CMI Report 23 (for PCML00).

65 70 75 80 85 90 95100

105 65 70 75 80 85 90 95100

1051.00

10.00

S1PML PCML00

Low GateCrude RateHigh GateS1PMLPCML00

Age

Mo

rta

lity

Ra

te

(log

m)

Convergence of mortality rates at older ages?

24

Relative crude mortality rates by age band and socio-economic class for males in England and Wales, 1982-2005.

Source: BW calculations using data from ONS Longitudinal Study 1982-2005 classified by RGSC

40−44 45−49 50−54 55−59 60−64 65−69 70−74 75−79 80−84 85−89 90+0%

20%

40%

60%

80%

100%

120%

140%

160%

180%

200%

I - Professional

II - Managerial

IIIN - Skilled non-manual

IIIM - Skilled manual

IV - Partly skilled

V - Unskilled

All classes combined

Age band

% o

f m

ort

alit

y ra

te f

or

all c

lass

es c

om

bin

ed

UNDERSTANDING POPULATION MORTALITY AT AGES 90+

25

2011 Census and E&W population estimates

26

2011 Census count compared with roll-forward estimateSource: ONS supplemented by BW calculations for ages 90+; estimates as at mid-2011

0-4 5-9 10-14

15-19

20-24

25-29

30-34

35-39

40-44

45-49

50-54

55-59

60-64

65-69

70-74

75-79

80-84

85-89

90-94

95-99

100+

-20%

-15%

-10%

-5%

0%

5%

10%Males

Females

Age Band

Ch

an

ge

in p

op

ula

tion

est

ima

te (

%)

Population estimates for older ages between Censuses

27

• ONS use Census roll-forward method to estimate population at individual ages to age 89, plus a total for ages 90+

• Use Kannisto-Thatcher survivor ratio method to allocate 90+ total to individual ages

• Extinct cohort method• Reconstruct population working back along cohort’s path, starting

from last survivor and adding back deaths, assuming no migration

• Non-extinct cohorts• Estimate current population using survivor ratio

• Survivor ratio = Survivors ÷ deaths in cohort in last 5 years• Assume stability or predictability of survivor ratio across cohorts

• Then work back along cohort’s path as above

Population estimates for older ages between Censuses

28

• As ONS apply a constraint on the K-T method, the total 90+ estimates are determined by the roll-forward method

• These roll-forward estimates are highly sensitive at older ages to accuracy of initial estimate

• 1% error in estimates for ages 80-89 results in:• 4% error for females at ages 90-99 at next

census; and• 7% error for males at ages 90-99 at next census.

Population estimates for older ages – an alternative approach

29

• ONS use Kannisto-Thatcher method to allocate 90+ total to individual ages (but 90+ total based on roll-forward estimates)

• But can also apply K-T methodology without constraint to 90+ total

• Key assumptions:• Death registration data is accurate• No net migration• Survivor ratio is stable / predictable.

Alternative estimates compared with roll-forward estimate

30

Comparison of estimates of mid-2011 population for E&WSource: BW calculations using population estimates and death registrations data to 2011 published by ONS.

80-84 85-89 90-94 95-99 100+-25%

-20%

-15%

-10%

-5%

0%

2011 Census (M)

BW (M)

2011 Census (F)

BW (F)

Age Band

% r

ed

uct

ion

to th

e r

oll-

forw

ard

es-

tima

te fr

om

the

20

01

Ce

nsu

s

K-T estimate using 2012 deaths compared with roll-forward estimate

31

Comparison of estimates of female population for E&WSource: BW calculations using population estimates and death registrations data to 2012 published by ONS.

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

0

50,000

100,000

150,000

200,000

250,000

300,000

350,000Females

ONS

Barnett Waddingham (K-T)

K-T estimate using 2012 deaths compared with roll-forward estimate

32

Comparison of estimates of male population for E&WSource: BW calculations using population estimates and death registrations data to 2012 published by ONS.

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

0

30,000

60,000

90,000

120,000

150,000Males

ONS

Barnett Waddingham (K-T)

Summary

33

• Census estimates at older ages appear reasonably accurate…• But some doubt regarding 2001 for males

• Inter-Censal estimates may become less reliable, as any inaccuracy in the previous Census is amplified in the current approach

Regulatory Information• The information in this presentation is based on our understanding of

current taxation law, proposed legislation and HM Revenue & Customs practice, which may be subject to future variation.

• This presentation is not intended to provide and must not be construed as regulated investment advice. Returns are not guaranteed and the value of investments may go down as well as up.

• Barnett Waddingham LLP is a limited liability partnership registered in England and Wales.

• Registered Number OC307678. • Registered Office: Cheapside House, 138 Cheapside, London, EC2V

6BW

• Barnett Waddingham LLP is authorised and regulated by the Financial Conduct Authority and is licensed by the Institute and Faculty of Actuaries for a range of investment business activities.

34

Angele Storey

Demographic Analysis UnitONS



How accurate are official high age population estimates?

Angele Storey, Demographic Analysis Unit, Population Statistics Division, Office for National Statistics

Official high age estimates

• Annual mid-year population estimates by single year of age and sex up to age 89 and 90 and over population- component cohort method

• Life tables and population projections require population estimates at single year of age for the 90 and over population- Kannisto-Thatcher method (form of survivor ratio

methodology)

.

Kannisto -Thatcher method

• Age at death data is used to build profiles of the distribution of older people in back years

• Average of the last 5 years ‘age at death’ data for each cohort is used to produce an estimate of the number of survivors for the current year.

• KT estimates constrained to the official 90+ total in final stage of the method

• Back years are recalculated annually as more recent deaths data becomes available

19811983

19851987

19891991

19931995

19971999

20012003

20052007

20092011

75

80

85

90

95

100

105

Males Females

Sum of deaths

KT 90+ totals as percentage of official MYE 90+ totals

19811983

19851987

19891991

19931995

19971999

20012003

20052007

20092011

80

82

84

86

88

90

92

94

96

98

100

2012 based KT estimates 2001 based KT estimates 1991 based KT estimates

1991, 2001 and 2012 based KT 90+ totals as a percentage of official 90+ totals - Males

19811983

19851987

19891991

19931995

19971999

20012003

20052007

20092011

80

85

90

95

100

105

2012 based KT estimates 2001 based KT estimates 1991 based KT estimates

1991, 2001 and 2012 based KT 90+ totals as a percentage of official 90+ totals - Females

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 201270,000

80,000

90,000

100,000

110,000

120,000

130,000

90+ on Patient Register DWP 90+ claimants

Official 90+ MYE 90+ KT estimate (unconstrained)

Comparison of 90+ totals across data sources - Males

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012240,000

260,000

280,000

300,000

320,000

340,000

360,000

90+ on Patient Register DWP 90+ ClaimantsOfficial 90+ MYE 90+ KT estimate (unconstrained)

Comparison of 90+ totals across data sources - Females

Possible sources of error in data sources

• Census estimates:- D.o.b. exaggeration and mis-reporting- Proxy responses - Some incorrect recording of year of birth / processing issues

• Inter-censal estimates: (Annual Mid-Year Estimates)- Any errors in 90+ age group at Census are rolled forward - Any over-count in those in their 80s comprises a larger

proportion as these non-existent people are aged on

• Patient Register and DWP claimants - People not removed / time lag in removing?

- People not on registers

-

1) Deaths data is accurate and complete- No validation of d.o.b on death certificate- Occurrences recorded in the correct year but those not captured

at time are not added back- Registrations capture all deaths but not necessarily in correct year

2) No migration at oldest ages- Internal to E & W (around 2.1% of people aged 90+ in 2005)

- international migration – in and out flows too small to measure

KT method assumptions

2012 BASED MALES

KT estimates constrained to official 90+ total

KT estimates constrained to 92% of official 90+ total

KT estimates unconstrained

Life expectancy at age 0

79.2 79.2 79.1


18.4 18.4 18.3


4.0 3.8 3.5

What would be the impact of lower estimates of the very old on life expectancy?

2012 BASED FEMALES

KT estimates constrained to official 90+ total

KT estimates constrained to 95% of official 90+ total

KT estimates unconstrained


82.9 82.8 82.7


20.9 20.8 20.7


4.7 4.5 4.1

What would be the impact of lower estimates of the very old on life expectancy?

Review of 90+ totals and age distributions

• Investigate effects of ‘tweaking the method’ - optimum survivor ratios; death registrations v. death occurrences; calendar year to mid-year

• Investigate effect of no allowance being made for improvements to the mortality rate in KT method

• Produce UK set of 90+ estimates on basis of combined constituent countries deaths data (to eliminate any internal migration factor)

• Assess optimum age to ‘join’ KT estimates to official 90+ estimates

Review of 90+ totals and age distributions cont….

• Investigate extent of inaccuracies in reporting of d.o.b. in the Census , - validate a sample of 90+ deaths by matching to birth

certificates

- use ONS LS to trace those aged 90 to 115 back thorough the 5 available Censuses

- quantify imputation rates at oldest ages

• Detailed investigation of comparative data sources- age distributions

• Discussion – Your ideas – what else could we do?

For further information…

For information or queries on…..

- population ageing, contact: [email protected]

- national life expectancy, contact: [email protected];

- sub-national life expectancy ,contact: [email protected]

- population projections ,contact: [email protected]

- 2011 Census, contact: [email protected]






David Sinclair

Assistant Director, Policy and CommunicationsILC-UK



The International Longevity Centre-UK is an independent, non-partisan think-tank

dedicated to addressing issues of longevity, ageing and population change.

The oldest old

David Sinclair, International Longevity

Centre – UK @ilcuk @sinclairda



Life is pretty tough. (frailty/isolation/health/income)

Dependency is not inevitable

Irrespective on the exact numbers. There will be more.

We don’t actually know a lot about the oldest old.

Today’s oldest old are not necessarily representative of tomorrows (genetics/environment/poor health = withdrawal from longitudinal studies)









Life is not easy for the oldest old

Three quarters of the oldest old

suffer from limiting longstanding

illnesses, and one out of three

perceive themselves as being in

poor health. (Tomassini C, 2005)

“almost 50% of men and women

aged 80-84 report severe

limitations in activities” (IFS,

2010)

http://www.flickr.com/photos/pondspider/4170990903/sizes/m/in/

photostream/



Falls

60% of interviewees aged over 90 had had a fall and that of these, 4 in five were unable to get up after at least one fall and almost a third had lain on the ground for an hour or more.

Call alarms were widely available but not used.(Fleming and Brayne, 2008; Cambridge City over 75-Cohor. BMJ)



And many find it difficult to do day to day tasks

Sixty per-cent of over 90s

report difficulties shopping for

groceries, almost a quarter

report difficulties making

telephone calls and 35% report

difficulties managing money.

(Sinclair, 2010/ELSA)

http://www.flickr.com/photos/pinkchocolate/3039589789/sizes/m/in/photostream/



A relatively high proportion live alone

Of those living in private households, four in ten very old men and seven out of ten very old women live alone. One out of five very old people live in communal establishments. (Tomassini C, 2005). http://www.flickr.com/photos/sbeebe/

5154169795/sizes/m/in/photostream/



Living together is good for us

Those who moved from living alone to living as part of a couple (with no children) exhibited a 68% fall in the odds of becoming multiply excluded between 2002 and 2008 compared to those who stayed living alone;

Those who moved from being resident in a couple household to living alone were over three times more likely to become multiply excluded. For this age group (50+), becoming a widow is one of the most common reasons for starting to live alone.

http://www.flickr.com/

photos/anabadili/

2963913137/sizes/m/in/

photostream/



The oldest old remain the most excluded

Almost 38% of those aged 85 or older faced some kind of social exclusion, an encouraging decline of 10% from the 2002 levels

As people age, they are more likely to become more socially excluded than less

Almost two-fifths (38%) of those aged 85 and older were excluded from two or more domains of exclusion in 2008

http://www.flickr.com/photos/pinkchocolate/3039589789/sizes/m/in/photostream/



Quality of Life falls with age



Depression

“23% of those aged 85 and over had levels of depressive symptoms indicative of clinical relevance”

“Almost 13% of men and women aged 80 and over had high levels of depressive symptoms in 2008-09 but not in 2002-03” (IFS, 2010) ELSA http://www.flickr.com/photos/

junglearctic/3002442666/sizes/m/in/photostream/



Most centenarians consult their GP

98% of centenarians and near centenarians consulted a GP and received prescription medicine during follow up. (Roughead, Kalisch et al, 2010)

http://www.flickr.com/photos/rwjensen/2288339230/sizes/m/in/photostream/



Centenarians do use drugs heavily

A study of 602 centenarians in Italy found that a very high proportion of this age group were users of drugs.

5% no drugs. 13% one drug a day 16% took 2 drugs per day 65% took three drugs a day 5.5% more than 3 drugs a

day.



Some evidence of longer hospital stays

Centenarians who had suffered from a hip fracture between 2000 and 2007 compared to a randomly selected control group of 50 hip fracture patients aged between 75 and 85. “the mean stay in acute orthopaedic wards for centenarians was 20.7 days and for the control group was 14.9 days”.

The longer acute hospital stay in the centenarian cohort would amount to a mean extra cost of £ 2511 per patient. (Verma et al)



Dementia among centenarians

The prevalence of

dementia-free

survival past 100

years of age varied

between 0 and 50

percent.

http://www.flickr.com/photos/thousandshipz/4679235/sizes/m/in/photostream/



Poverty is a very real challenge

There is evidence that the oldest old (aged 85 and over) are, as a group, at greater risk of poverty than younger older people (aged 65-85)

Up to 10% of the oldest old have total net wealth of £3,000 or less.



Britons ageing quicker than their parents



Dependency is not inevitable

Dependency is not inevitable and a ”considerable proportion of the centenarians maintain a good level of auto sufficiency for the basic performance of the everyday life”. (Antonini et al, 2008)

http://www.flickr.com/photos/driever/5525684658/sizes/m/in/photostream/



Some of the oldest old become more active



Prevention of ill health

Physical Activity

Smoking and alcohol consumption

Nutrition

Immunisation

A move to prevention is vital



We need to get housing right

1. Extra care housing is a home for life

2. Extra care translates into fewer falls

3. Extra care is associated with a lower uptake of inpatient hospital beds



Many thanks

David Sinclair

Head of Policy and Research

International Longevity Centre

[email protected]

02073400440

Twitter: @ilcuk and @sinclairda


Andrew Latto

Deputy Director, State Pensions DirectorateDepartment for Work and Pensions



Missing 90 year olds

An ILC-UK debate supported by Partnership as part of the Population

Patterns Series

Wednesday 5th March 2014



05Mar14 - Missing 90 year olds

News & Politics

financial conduct authority

prudential regulation authority

allocate 90 total

official 90 totals

international longevity centre

uk debate supported

death registrations data

national life expectancy