Disability free life expectancy (DFLE) in the European Union from 1995 to 2003

using the European Community Household Panel (ECHP)

EUROSTAT DEVELOPMENTS

Structural indicators : background & inclusion of health issues

23-23/03/2000 : Lisbon European Council invited “the Commission to draw up an annual synthesis report on progress on the basis of structural indicators (SI) relating to employment, innovation, economic reform & social cohesion”

SI are proposed by policy DGs, the methodology developed using recognised methods & when necessary with support of academic experts, their values calculated, stored (SI database) & disseminated by Eurostat (validation by MS)

2003 : shortlist of headline indicators + public database (large number of indicators to fully monitor structural reforms)

08/12/2003 : according to the EU Sustainable Development Strategy & the 6th Environmental Action Plan, Council conclusions request : indicators on biodiversity & health need to be included in the SI database

Structural indicators :inclusion of HLY-DFLE as 1rst health

indicator July 2003 : support of DG SANCO Network of

Competent Authorities 08/10/2003 : Commission Communication COM(2003)

585 final on SI included in part “indicators under development” the “healthy life years” (HLY-DFLE)

26-27/11/2003 : information of the Eurostat WG on Public Health statistics meeting

21/01/2004 : Commission COM (2004) 0029 - report from the Commission to the Spring European Council delivering Lisbon reforms for the enlarged Union : priority to examine possibilities for integrating Public Health into the Lisbon strategy by 2005, as a contribution to growth & sustainable development

HLY-DFLE : definition & methodology

Health expectancy = number of years that a person of a specific age is still expected to live in a healthy condition; breakdown by gender

The Sullivan method combines information on mortality & morbidity, i.e. it is a LE calculation weighted by morbidity prevalence : information on mortality from life tables : age-specific

probability of dying (Eurostat demographic statistics)

information on morbidity is based on prevalence measures of the age specific proportion of the population in healthy and unhealthy conditions

For HLY, healthy condition defined by the absence of limitations in functioning / disability; consequently it is also called disability-free life expectancy (DFLE)

HLY-DFLE as SI is calculated at birth

HLY-DFLE : Eurostat activities for the development of this structural indicator

Eurostat data sources on limitations in functioning / disability : Minimum European Health Module (MEHM) in the

Statistics on Income and Living Conditions survey SILC carried out from (2003)/2004/2005 depending on MS

For 1995-2001, use of the similar variable in the European Community Households Panel (ECHP) for EU15 (excl. LU)

1rst half 2004 : on the basis of the 2003 requirements & supports as well as sources above, Eurostat D/6 : took over the existing methodology (Sullivan method) extracted input data from ECHP & assessed its quality, prepared documentation & computed initial values

1995-2001

HLY-DFLE : Eurostat activities for the development of this structural indicator

20/04/2004 & 01/07/2004 meetings of Commission interservice group on SI : Eurostat development (support by SANCO C/2 leader, + EMPL + ENV) acknowledged; however, the group stated that the inclusion of HLY-DFLE in the SI was subject to 2 conditions : preparation of a “quality profile” on assessment &

validation of the quality & methodology of the indicator data should be provided for 2002 & as much as

possible 2003 (SI timeliness minimum requirements) 19/07/2004 letter sent to all national representatives

in the Eurostat WG on Public Health statistics : draft “quality profile” draft values of the HLY calculated by Eurostat from the

ECHP data over the period 1995-2001 for the EU15 Member States (excluding LU)

HLY-DFLE : Eurostat activities for the development of this structural indicator

Asking, by 20/08/2004, for : agreement on the document “quality profile” validation on the national data 1995-2001 (EU15 MS) suggestions concerning national data for new MS suggestions concerning 2002 & if possible 2003

Answers : high support; problems only for some national values; no solution for missing data

31/08/2004 : Meeting EHEMU-Eurostat-SANCO calculation by EHEMU of estimates on 2002 &

2003 and for the few problematic values on the basis of trends 1995-2001 in the limitation prevalences or mortality data (high inertia in evolutions)

Calculation of HLY by Eurostat for new MS/EFTA national sources when sufficiently comparable; also EU15 aggregate (i.e. EU14 excl. LU)

Adoption & state of play on provision of HLY-DFLE data by

Eurostat

Final discussions within the SI interservice group & with Council in 09/2004 HLY ADOPTED by Communication to the Commission on 14/10/2004 in the SI data base for report to 2005 Spring EU Council

Quality report officially approved by MS in 11/2004 final check & improvements of calculations with EHEMU,

final preparation of metadata documents (“SDDS”) in 12/2004

Data already available in Eurostat web-site (ex-Newcronos) in the sections of SI & public health

EU15 MS (excl. LU) 1995- 2001 ECHP data from variable PH003A :“are you hampered in your daily activities by any physical or mental health problem, illness or disability ?”

severely / to some extend / not limited”available for people in sample 16 years old & over

5 new MS + NO, different sources : 18 HIS items (2002/2003): CZ HU CY MT PL(1996)

others : PL 2002 census, NO 2003 pilot SILCdata from EE & LV not used (not sufficiently comparable ?)

Aggregate EU15 (EU14) using the ECHP double weights : sample + « EU » weight (sample population level)

Eurostat calculations

Eurostat calculations

HLY as structural indicator is calculated at birth but the value have also been calculated (according to ECHP age breakdown) at 15, 20, 25, … 70, 75, 80, 85, 90+ years; are disseminated in the Eurostat web site : Values in years at birth (SI part of reference

database) values in years & in % of LE, both at birth & at 65

years (« health » part of reference database) Corrections needed for few countries : DE & UK due to use

of another source than ECHP to provide “ECHP” health data from 1997 onwards, also some difficulties for probabilities of dying for few countries & years, etc. EHEMU estimates;

EU15 (14) aggregate calculated only when having received corrected values from EHEMU for the relevant MS / years

Eurostat calculations

Assumptions made : average limitations/ disability prevalence 0-14

years old supposed to be half of the ECHP prevalence of the next age group (16-19 = 15-19)

institutional households are not included in the ECHP, i.e. it is assumed that the population living in private households is representative for the total population : not significant at birth (but effect at old ages)

Some sample size issues or missing values (prevalence or mortality) for very old ages corrections or assumptions made (e.g., Log. regression, life expectancy at the last age taken from another year, etc.)

HLY-DFLE on ECHP 1995-2003 :comparability restrictions

Cultural differences from country to country : same questions on limitation / disability might be interpreted differently from a country to another one

No common survey instruments in the new MS : comparability among them as well as between them and the EU15 Member States is restricted

However restrictions on the comparability across countries apply on the “level” of the HLY-DFLE but not on its evolution

Due to transition between end-ECHP and start-EU-SILC, there will be breaks in series between 2003 & 2005, the extent to be examined once the data have become available

HLY-DFLE on ECHP 1995-2003 : main results

In EU15 in 2001 : women could expect to live to 65.0 years … women 65 years old could still expect to live

19.9 years of which 10.4 (52%) … men to 63.6 years … men 65 years old could still expect to live 16.3

years of which 9.9 (61%) …… without any limitation / disability

In EU15 over the period 1999-2001 HLY-DFLE at birth : increased from 63,9 to 65,0 years for women

(about 79% of LE) from 63,2 to 63,6 for men (84% of LE)

HLY-DFLE on ECHP 1995-2003 : main results

HLY-DFLE in 2003 ranged : less than 60 years in Hungary, the Netherlands

(women only), Portugal (men only) and Finland to more than 70 years in Italy

In % of LE, over the period 1995-2001, HLY-DFLE at birth : increased of 2,7% in BE & IT for women and more than 3% in BE & DE for men but decreased - 4,9% in NL for women and - 2,8% in PT for men

But cultural effects, better comparability with SILC data in future

EHEMU DEVELOPMENTS

Data and Methods

• Estimation of Life Expectancy (LE) and 95% CI

• Estimation of DFLE and 95% CI, using Sullivan method

age specific probability of death age specific disability prevalence

Data and Methods

1) Data base• Probable data errors• Replacement with other sources• Missing

1) Data base• Linear imputation of age specific

probabilities according to trends

2) Interruption of data collection• No data for 2002 and 2003

Problems in both mortality and the panel data Solutions

Example…

• Imputation of age spercific probabilities according to observed trends

6

7

8

9

10

11

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Year

Exp

ecte

d y

ear

6

7

8

9

10

11

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Year

Exp

ecte

d y

ear

Ex: Female LE at age 65 in UK

Data and Methods

Problems in both mortality and the panel data

1) Data base• Probable data errors• Replacement with other sources• Missing

Solutions

1) Data base• Linear imputation of age specific

probabilities (death and disability)

• Shift of the prevalence trend to the ECHP level

• Imputation of data according to observed trends

2) Interruption of data collection• No data for 2002 and 2003

Example…

77

78

79

80

81

82

83

1994 1995 1996 1997 1998 1999 2000 2001 2002

Year

Exp

ecte

d y

ear

• Linear imputation of missing probabilities of death

77

78

79

80

81

82

83

1994 1995 1996 1997 1998 1999 2000 2001 2002

Year

Exp

ecte

d y

ear

Ex: Female LE at birth in Denmark

Data and Methods

Problems in both mortality and the panel data

1) Data base• Probable data errors• Replacement with other sources• Missing

2) Interruption of data collection• No data for 2002 and 2003

Solutions

1) Data base• Linear imputation of age pecific

probabilities (death and disability)

• Shift of the prevalence trend to the ECHP level

• Imputation of data according to observed trends

2) Interruption of data collection • Linear extrapolation of the

disability prevalence

Example…

• Linear extrapolation of the disability prevalence up to 2003

0,0%

0,1%

0,2%

0,3%

0,4%

0,5%

0,6%

0,7%

0,8%

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Year

Dis

abil

ityp

reva

len

ce

Ex: Male disability prevalence in Greece (65 years and older):

6

8

10

12

14

16

18

20

22

1995 1996 1997 1998 1999 2000 2001 2002 2003

Year

Ex

pe

cte

d y

ea

r LE

DFLE

Example: Trends in LE and DFLE at age 65, 1995-2003, Male, Austria

Main results

Trends in LE and DFLE at birth in European countries, 1995-2003

Male

52

54

56

58

60

62

64

66

68

70

72

74

76

78

80

82

84

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Year

Exp

ecte

d ye

ar

Austria Belgium Denmark Finland France Germany Greece

Ireland Italy Netherland Portugal Spain Sweden United Kingdom

LE

DFLE

Trends in LE and DFLE at birth in European countries, 1995-2003

Female

52

54

56

58

60

62

64

66

68

70

72

74

76

78

80

82

84

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Year

Exp

ecte

d ye

ar

Austria Belgium Denmark Finland France Germany Greece

Ireland Italy Netherland Portugal Spain Sweden United Kingdom

LE

DFLE

Distribution of life and disability free life expectancy EU (14), 2003

DFLE

Males Females

sex

55.00

60.00

65.00

70.00

75.00

80.00

85.00LE

SWE

54%

60%58% 59%

57%

63%

59%61% 61%

0

2

4

6

8

10

12

14

16

18

1995 1996 1997 1998 1999 2000 2001 2002 2003

Year

Ex

pe

cte

d y

ea

r

LE with disability

DFLE

Trends in the percentage of expected life free of disability at bith

Austria, 1995-2003

Female population

Trends in expected life free of disability at age 65

DEU, GRC, IRL, NDL, PRT

AUT, BEL, DNK, ITA, ESP, SWE

FIN, FRA, GBR

Women

4

6

8

10

12

14

16

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Year

Expe

cted

yea

r

Austria Belgium Denmark Finland France Germany Greece

Ireland Italy Netherland Portugal Spain Sweden United Kingdom

Men

4

5

6

7

8

9

10

11

12

13

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Year

Expe

cted

yea

r

Austria Belgium Denmark Finland France Germany Greece

Ireland Italy Netherland Portugal Spain Sweden United Kingdom

Trends in expected life free of disability at age 65

AUT, BEL, FIN, FRA, DEU, IRL, ITA, ESP

GRC, NDL, PRT, GBR

DNK, SWE

1) Life expectancy:• Small variation in life Expectancy between these 14 MS• increase over 1995-2003

Trends in DFLE using the ECHPSome conclusions

2) Disability Free Life Expectancy and % of years free of disability

• Large variation between these 14 MS• Diverging trends over 1995-2003: reduction / stagnation /

increase in the expected life with reported disability while LE increases

3) Gender differences in DFLE trends in some countries

4) Gender differences in DFLE are smaller than gender differences in LE

The population aging has a different impact in the 14 european populations:

- different levels of reported disability (larger dispersion than LE) - variation in the magnitude of the gender difference- different trends over time

Trends in DFLE using the ECHP Issues and inference

Need to further document cross-national differences in self-reported disability:

- improved harmonisation of the instruments- using different levels of severity (SILC)- documenting differences in reporting- documenting differences in selection in the panel…